Wikipedia:Village pump (technical)/Archive 110

Village pump
- Policy
- Technical
- Proposals (persistent)
- Idea lab
- WMF
- Miscellaneous

Village pump (technical) archive

This page contains discussions that have been archived from Village pump (technical). Please do not edit the contents of this page. If you wish to revive any of these discussions, either start a new thread or use the talk page associated with that topic.

< Older discussions · Archives: A, B, C, D, E, F, G, H, I, J, K, L, M, N, O, P, Q, R, S, T, U, V, W, X, Y, Z, AA, AB, AC, AD, AE, AF, AG, AH, AI, AJ, AK, AL, AM, AN, AO, AP, AQ, AR, AS, AT, AU, AV, AW, AX · 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216

Which edit counter

Ok back in 2009 I created User:Nthep/Editcounter obviously in connection with an edit counter but for the life of me I can't remember which one. Does anyone know which and is it still functioning? I know it's not tparis's as that relies on User:Nthep/EditCounterOptIn.js. Thanks. NtheP (talk) 22:40, 30 March 2013 (UTC)

See User talk:X!/Archives/6/2010#OptIn. The links there to Interiot's tool don't work - "Account expired". -- John of Reading (talk) 07:21, 31 March 2013 (UTC)

References

In an article i use note references but i also have a real reference i wish to use to back up the note, Is there anyway to add a reference to the note and if not whats the best way to ref a note?Blethering Scot 00:13, 31 March 2013 (UTC)

If I understand correctly, Help:Footnotes#Embedding references within footnotes is what you are looking for. If that's not it, please provide the article you're working on and I can give you better answer. Cheers. 64.40.54.208 (talk) 06:55, 31 March 2013 (UTC)

Incorrect interwiki

In articles which I've written in Polish and English Wikipedia happened something wrong with the interwikis. It has something to do with recent activity of Addbot, I suppose. I mean the series of articles about Vetulani family - interwiki redirects not to the article about particular person in Wikipedia in different language, but to the template with whole family tree (for example interwikis in article Adam Vetulani - Adam Vetulani). I have no idea how to fix it. Can someone help? Francesco 13 (talk) 09:30, 31 March 2013 (UTC)

Templates must have interwiki links inside <noinclude>...</noinclude> to prevent the interwiki link from also applying to pages transcluding the template. Fixed in [1] and [2]. PrimeHunter (talk) 12:24, 31 March 2013 (UTC)

Thank you very much! Francesco 13 (talk) 15:28, 31 March 2013 (UTC)

Want to block myself from going on Wikipedia forever due to addiction to it, tried Wikibreak Enforcer, it isn't working

I set a ridiculously long Wikibreak of 100 years in the Wikibreak Enforcer to make sure I can quit my current addictive behaviour to Wikipedia completely. But it is not working, I thought I entered the right material, this is what I entered for the Wikibreak Enforcer here: [3]. Could someone please correct it so that the Wikibreak is in place?--R-41 (talk) 18:46, 31 March 2013 (UTC)

I recommend that you use the new Wikibreak enforcer:

/*** BEGIN WIKIBREAK ENFORCER ***/
addOnloadHook(function() {
 
        /*** Start editing here ***/
 
        // When you want to end your break?
        // no leading zeroes. (example: 7 - correct, 07 - incorrect)
 
        var date = { year: 2100, month: 1, day: 1};
        var time = { hours: 0, minutes: 0, seconds: 0 };
 
        /*** Stop editing here ***/
 
        var currentDate = new Date();
        var enforcedBreakEnd = new Date(
                date.year,date.month-1,date.day,time.hours,time.minutes,time.seconds);
        if (currentDate <= enforcedBreakEnd) {
                alert("Enforced wikibreak until "+enforcedBreakEnd.toLocaleString()
                        + "\n(now is "+currentDate.toLocaleString()+")\n\nBye!");
                location = "//"+location.host+"/w/index.php?title="
                        + "Special:Userlogout&returnto=Main_Page";
        }
});
/*** END WIKIBREAK ENFORCER ***/

Sorry to see you go. —Theopolisme (talk) 19:03, 31 March 2013 (UTC)

"the crap on this page"

For some reason, Indigenous peoples is displaying with the sentence "please don't fall for the crap on this page" in the lead paragraph--but the wiki text doesn't show this comment when I went in to edit it out. Has a template or something been compromised? Thanks, -- Khazar2 (talk) 21:05, 31 March 2013 (UTC)

There was vandalism of the page, which was automatically reverted by a bot. apparently, bot edits do not trigger purging of the page, so you saw the cached, vandalized version. "Purge" or simply null-edit (i.e., pressing "edit" and then saving without actually changing anything on the page) cures it. peace - קיפודנחש (aka kipod) (talk) 21:20, 31 March 2013 (UTC)

I got it corrected.by removing the mis-formated [ethnic minorities]].that had been left there by an IP edit. It was deliberate vandalism, and perhaps some hidden code in what I removed. Cluebot tried to delete it and didn't. — Maile (talk) 21:23, 31 March 2013 (UTC)

there was no "mis-formatting". you removed a perfectly good internal link - cluebot reverted it just fine. peace - קיפודנחש (aka kipod) (talk) 21:34, 31 March 2013 (UTC)

I say it was "mis-formatted", because it only had one bracket on the left instead of the double bracket. It WAS...— Maile (talk) 21:54, 31 March 2013 (UTC)

Whatever the case, the page is again displaying correctly. Thanks for the fix. -- Khazar2 (talk) 21:56, 31 March 2013 (UTC)

Resolved

I just took care of it. — User:Technical 13 ( C • M • Click to learn how to view this signature as intended ) 21:59, 31 March 2013 (UTC)

View count bot

The page Aho–Corasick string matching algorithm has for the past week been experiencing a view count bot. I hope someone can fix this. Pass a Method talk 21:52, 31 March 2013 (UTC)

Can someone disambiguate the names affected by tl:sortname?

Some disambig errors are reported at Talk:List_of_Shadowrun_books#Author_links. I tried to fix them, but the article now uses {{sortname}} and I don't know how to tell it to use a red link. --_{Piotr Konieczny aka Prokonsul Piotrus| reply here} 11:19, 1 April 2013 (UTC)

{{sortname}} shows documentation. {{sortname|Tom|Dowd}} gives Tom Dowd with the bad link to the disambiguation page. {{sortname|Tom|Dowd|Tom Dowd (author)}} gives Tom Dowd. It's the same syntax whether the link is blue or red. {{sortname|Tom|Dowd|nolink=1}} gives Tom Dowd. PrimeHunter (talk) 11:59, 1 April 2013 (UTC)

Bots and the sudden loss of communication

Hello. I ran a bot add_text.py. But his work was terminated error 503. I stopped at the article on the letter E. Tell me, is it possible to continue the action of the script from the page where the failure occurred? If not, then how should I complete the planned my action? Thank you.--Ворота рая Импресариата (talk) 15:38, 1 April 2013 (UTC)

I think you can use the argument -start:E. —Theopolisme (talk) 00:36, 2 April 2013 (UTC)

Yes, I certainly tried, but in one case the bot is still starting from the beginning of the list, in the other - to make transactions in the space of articles, even though I asked to perform tasks in templates, and the third - did nothing.

So, I completed the task with AWB - this program is always less hassle - but what if I need to do something that is not in the standard functional AWB and there is the same error - I do not know. Thanks.--Ворота рая Импресариата (talk) 05:55, 2 April 2013 (UTC)

Convert HTML into an infobox

Can someone convert the HTML at Joel Goffin into an infobox and put his image in the infobox rather than the center of the article? Ryan Vesey 21:46, 1 April 2013 (UTC)

Done -- WOSlinker (talk) 22:03, 1 April 2013 (UTC)

Thank you so much. Ryan Vesey 23:28, 1 April 2013 (UTC)

Question for those familiar with MediaWiki template parameters / variables

Resolved

– Bug filed - bugzilla:46768 7 04:15, 2 April 2013 (UTC)

Can I ask anyone who is an expert in this topic to take a look at the question here. Thanks 7 01:59, 2 April 2013 (UTC)

Is there a tool to update an infobox automatically?

I was looking at the infobox at Open Knowledge Foundation. It seems to be using old parts of the syntax such as | Non-profit_name = | Non-profit_logo = | vector_logo = (you can see them in the 2007 version but not in the modern {{Infobox non-profit}}). Is there any tool that could update this in the article replacing old syntax with new or do we have to do this by hand? --_{Piotr Konieczny aka Prokonsul Piotrus| reply here} 02:34, 2 April 2013 (UTC)

You could either use WP:AWB for that, or request a bot operator to do this at WP:BOTREQ (multiple bots do these kind of replacements and removals on a regular basis). mabdul 04:52, 2 April 2013 (UTC)

Wikidata Bug ?

Hello,

I have a questions on this article Cape Esperance. Using Chrome (and Firefox), I can not edit the interwiki links / have access to Wikidata at the bottom of the interwiki links. Is it a known bug ? What can I do ? Thanks in advance. Poppy (talk) 18:25, 28 March 2013 (UTC)

"Edit links" appeared when I purged the page. PrimeHunter (talk) 18:57, 28 March 2013 (UTC)

It depends: if there isn't any wikidata item, the edit links button doesn't appear. But I sometimes also have the problem, that the link is not displayed in Opera 12.14... I don't know when it happens, but it drives me somehow sometimes crazy, resulting in empty wikidata items... mabdul 02:06, 30 March 2013 (UTC)

I'd recommend asking on http://www.wikidata.org/wiki/Wikidata:Contact_the_development_team whether it's a bug and whether to create a request in the bugtracker. --AKlapper (WMF) (talk) 11:46, 2 April 2013 (UTC)

Link to copy page content to clipboard?

I don't even know if this is the right place to ask about this, but it seems more appropriate than any other place I can find. What I'm trying to do is see if there's a way to copy a page's contents to the clipboard - so you can paste in a page you're editing - by clicking a link on another page. Is this something that can even be done with the current wikimedia software? ~~Should I put in a Lua request?~~ Is this just not technically feasible at all for security reasons? I would like to implement it as a template for use on template/module documentation pages to help editors transclue them with the right parameters, eg infoboxes. Any pointers or thoughts would be appreciated. Van Isaac_{WS Vex}^contribs 14:04, 31 March 2013 (UTC)

Copying content to the clipboard can only be done via JavaScript; lua (as a server-side script) has no access to local clipboards. — Edokter (talk) — 14:55, 31 March 2013 (UTC)

Well, that's good to know; struck that option. Is JavaScript even allowed on Wikipedia pages? Van Isaac_{WS Vex}^contribs 15:18, 31 March 2013 (UTC)

JavaScript cannot be used directly in wiki pages. What you can do is write scripts that are loaded on every page you view that modify the loaded page if it contains certain content (e.g. it could add a copy link where it finds  in the page). You can place JavaScript that only affects yourself in Special:MyPage/common.js. To write a script for other users to use, create a subpage in your userspace with a name that ends .js. Other users can load the script by adding something like importScript( 'User:Yourname/yourscript.js' ); to their own common.js file. See Wikipedia:WikiProject User scripts/Guide for more information about user scripts.

Administrators can write scripts that other users can use without needing them to muck around in their common.js file. Admins can create gadgets, which registered users can turn on and off in Preferences. Admins can also edit MediaWiki:common.js to make changes that affect everyone without any option to turn them off (without turning off JavaScript altogether, that is). – PartTimeGnome (talk | contribs) 16:05, 31 March 2013 (UTC)

So I could actually do this completely within my own user space without any special permissions, just make my own JS script, and import it to my common.js, and only once it works would it get included in global files so it works for everyone? Van Isaac_{WS Vex}^contribs 16:39, 31 March 2013 (UTC)

Yes. I think a lot of gadgets start life as a user script. Once other people have had a chance to test it and find that they like it, you can propose that it be added as a gadget at Wikipedia:Gadget/proposals. – PartTimeGnome (talk | contribs) 16:54, 31 March 2013 (UTC)

Well, that pretty much answers that. Thanks for all your help. Van Isaac_{WS Vex}^contribs 17:16, 31 March 2013 (UTC)

The way I do it is click to edit the page and then click anywhere in the edit box. Then press CTRL+A then CTRL+C. This copies the content of the page to the clipboard. Then I go to the new page, put the cursor where I want to add the content and press CTRL+V. — User:Technical 13 ( C • M • Click to learn how to view this signature as intended ) 21:21, 31 March 2013 (UTC)

Yeah, I'm looking for a bit more functionality than that, so even though it started out as an idea about copying code I store on test pages in my user space, I'm trying to make an interface for editors to copy template parameters from /doc pages. Something along the lines of "Click here to copy the basic infobox code to your clipboard, here for extra code X, here for extra code Y, or here for extra code Z." Van Isaac_{WS Vex}^contribs 01:48, 1 April 2013 (UTC)

We do have preloads for page creation and input box. They are limited in functionality, but often enough. —TheDJ (talk • contribs) 08:31, 2 April 2013 (UTC)

Convert HTML into an infobox Part 2

I'm very active at lowp at the moment and cleaning the complete wiki (only ~850 article at the moment). Would somebody help me and cleaning up the infoboxes and converting them to "wikimarkup infoboxes? Here is a link for them for some [4]. More will likely follow in a few days if I find some in articlespace which aren't covered by this link. If there are still interwiki links, feel free to help there, too. ;-)

Regards,

mabdul 22:30, 1 April 2013 (UTC)

Some more infoboxes: [5]. mabdul 11:42, 2 April 2013 (UTC)

badly positioned magnifying glass in search bar using Firefox

Tracked in Phabricator
Task T32525

I have seen a problem with the positioning of the magnifying glass in the search bar using Firefox for a while now (at least about a half a year). I have uploaded a screenshot that shows the problem here. Notice how the magnifying glass is shifted down to the point of being partly outside the search box. Do other Firefox users see this? Not sure if problem here or with Firefox's rendering yet. This is the Linux version of Firefox. That shouldn't but may be important. Jason Quinn (talk) 19:48, 2 April 2013 (UTC)

Actually, it does seem to be important that it's the Linux version of Firefox. I don't see it in the Windows version. Most likely a bug in Firefox now. Jason Quinn (talk) 20:13, 2 April 2013 (UTC)

There's already a Wikimedia bug report (T32525) regarding the positioning of the icon, although it seems that your case is particularly bad. May I ask which version of Firefox are you running? On my Linux installation, Firefox 19 seems to match the "Epiphany" screenshot. PleaseStand (talk) 20:54, 2 April 2013 (UTC)

It's 19.0.2. Jason Quinn (talk) 21:13, 2 April 2013 (UTC)

For what it's worth, my Firefox 19.0.2 on Ubuntu 12.10 also has the magnifying glass shifted down slightly, but not so severely as to leave the box. —Remember the dot ^(talk) 01:40, 3 April 2013 (UTC)

Vector.js not working

For some reason my js page User:Reywas92/vector.js is no longer working. Popups are the default style rather than my variant settings, and the scripts I have imported do nothing. Any suggestions how to fix it? I have refreshed the page several times and javascript is working in my browser, so I don't know what could be causing it. Thanks, Reywas92^Talk 22:41, 2 April 2013 (UTC)

Sure. Get rid of all the lines starting with "<!--". Those are not valid JavaScript, those are XML comments. Lupo 22:56, 2 April 2013 (UTC)

Replace the  lines with /**/ (note the /* and */ on the ends. I also suggest moving your importScript() lines to the top to make them load first assuming there has been no issues with them in the past. Move them up one at a time and see if they work... If one of them got broken, everything after it linearly in that .js will be broken. — User:Technical 13 ( C • M • View signature as intended) 23:34, 2 April 2013 (UTC)

Thanks, that worked. I did them all at once, but none of the imported pages have been edited and they all just suddenly stopped a few days ago for me so who knows. Reywas92^Talk 01:19, 3 April 2013 (UTC)

"Blacklisted" page isn't blacklisted

I try to move L 20 α class battleship to L 20 α-class battleship (with hyphen) it says the latter is on the title blacklist. Is it just being racist against Greek letters? Marcus Qwertyus (talk) 04:16, 3 April 2013 (UTC)

It is probably triggering on the mixed Latin, Greek characters. Werieth (talk) 04:17, 3 April 2013 (UTC)

Super admin powers activate, form of ~~megazord~~page moved. - The Bushranger _{One ping only} 05:49, 3 April 2013 (UTC)

Lua Citation deployment

In the not too distant future, we are likely to deploy the Lua version of {{cite encyclopedia}}, {{cite encyclopedia/lua}}, as a moderate scale test of Module:Citation/CS1. Lua-based citations are expected to reduce citation formatting time by about 80% without sacrificing any formatting options. The current set of test cases for this template can be seen at Module talk:Citation/CS1/test/encyclopedia. As shown, there are a number of cases where errors in the current {{cite encyclopedia}} have been corrected, as well as small formatting changes (often motivated by a desire make the different families of citation templates more consistent). Dragons flight (talk) 01:20, 17 March 2013 (UTC)

I've gone ahead and deployed the Lua version for {{cite encyclopedia}}. Dragons flight (talk) 03:46, 17 March 2013 (UTC)

{{cite news}} has now also been replaced with a Lua version. Test cases at Module talk:Citation/CS1/test/news. Dragons flight (talk) 14:20, 19 March 2013 (UTC)

{{cite journal}} has now also been deployed using Lua. Test cases as Module talk:Citation/CS1/test/journal. Dragons flight (talk) 01:00, 23 March 2013 (UTC)

{{cite book}} has now been deployed using Lua. Test cases at Module talk:Citation/CS1/test/book. Dragons flight (talk) 23:33, 24 March 2013 (UTC)

{{cite web}} has now been deployed using Lua. Test cases at Module talk:Citation/CS1/test/web. Dragons flight (talk) 00:51, 29 March 2013 (UTC)

Thanks very much for all the updates, DF. I must admit I was a bit worried about the conversion to Lua, but everything has turned out great from all the checking I've done. I certainly appreciate all the work you've done. 64.40.54.208 (talk) 05:53, 31 March 2013 (UTC)

{{citation}} has now been deployed using Lua. Test cases at Module talk:Citation/CS1/test/citation. Dragons flight (talk) 23:51, 3 April 2013 (UTC)

Error mesage

Just for documentation purposes, I got sent to a page that said the Wiki was having a problem. It was not the normal "Wikimedia error" page that usually appears.— Vchimpanzee · talk · contributions · 20:49, 1 April 2013 (UTC)

We had three outages in the last days, the latest on Sunday 2013-03-31 around 21:20 UTC, and the operations team is aware of it and has fixed them. Are there specific steps to reproduce? Did this happen with any page on en.wikipedia.org, is this still reproducible for you, and if yes what is the exact error message? (Screenshot or copying the text welcome). Thanks & sorry for the problems. --AKlapper (WMF) (talk) 11:52, 2 April 2013 (UTC)

It just happened once and I didn't think to do a screen shot.— Vchimpanzee · talk · contributions · 19:45, 3 April 2013 (UTC)

Bugday today (16:30 - 20:30UTC) on MediaWiki skin and page rendering issues

Hi everybody, today is a Wikimedia Bugday from 16:30-20:30 UTC in #wikimedia-office on Freenode IRC. Everyone is welcome to join, and no technical knowledge needed! Just say Hi and that you're here for the bugday when joining at any time. It's an easy way to get involved in the community or to give something back. This even is part of the QA weekly goals. This information and more can be found on the wiki. There is also more information available on Bug triaging in general. Looking forward to seeing you there! --AKlapper (WMF) (talk) 14:32, 2 April 2013 (UTC)

Thanks to everybody who joined. We will continue this week to triage Skin and page rendering issues, and everybody is welcome to join. I'm andre__ on IRC. --AKlapper (WMF) (talk) 08:48, 3 April 2013 (UTC)

Change in logged-in status

In the past, I'd get a warning that I was not logged in every day or two, while attempting to save an edit, even though I was definitely logged in when I started, as evidenced by the top menus, nav popups, WikEd use, edit settings, etc.

I've noticed that, in the last couple of weeks, this has changed, and I haven't needed to log in again during that time. I generally leave a WP page open in a browser (FFox 15), and generally hibernate the machine between sessions.

I also noticed another change today. I log in and use the https (secure) web interface to WP. When I Google search (i.e. from Google) in another tab or window and click on a link to a WP article, it is usually to the non-secure http page. In the past, even though I am logged in and viewing an https page in another tab/window, the http page I clicked on would not be logged in. If I proceeded to log into that page, I believe (but am not sure) it would log me out of the https page. Now, it seems that has changed. When I click on an http link, it still shows me as logged in (based on the top menu, nav popups, etc.).

If this is a fix, thanks! I thought I'd mention it for the benefit of those who were routinely logging in every day as a (now un-necessary) preventative measure. —[AlanM1(talk)]— 18:36, 2 April 2013 (UTC)

If you log-in on HTTPS, you are only logged-in for using HTTPS and will appear logged-out on plain HTTP. If you log-in on HTTP, you are logged-in for using both HTTP and HTTPS. If you log-in on HTTP while already logged-in on HTTPS, the HTTP log-in replaces the HTTPS one, so you appear logged-in on both. (Replacing an existing login like this might cause a "loss of session data" if you were editing a page in another tab. You remain logged-in, and can complete the edit by clicking "Save page" again.)

If you appear logged-in on HTTP, you must have logged-in using HTTP. If you are concerned someone could capture your unencrypted login cookie on HTTP and use it to impersonate you, I suggest you log out then log back in again using HTTPS.

As for remaining logged in for longer, you might have checked the "Remember my login on this browser (for a maximum of 30 days)" check box when logging in. (I'm sure this used to be 180 days, so something's changed here...) Alternatively, your browser might have a function where it regularly deletes HTTPS cookies for security reasons, while keeping the non-secure HTTP ones. Hence, remaining logged-in for longer could be linked to being logged-in on HTTP.

Personally (using Opera 12), I only have to log-in once in a blue moon. I always login using HTTPS and check "Remember my login". I logged out and logged back in just now, and can confirm that I still appear logged out using HTTP (which is how I like it). – PartTimeGnome (talk | contribs) 22:45, 3 April 2013 (UTC)

Changes and queries

Can I show my ignorance and enquire where the discussions are held that lead to changes in editing interfaces, such as the addition of 'user rights management' to the sidebar, and the apparent attempted 'unification' of the American and Brit English admin action screens? (At present, the Brit screen is without the lengthy palaver that the American screen has.) I have noticed as well that in the Brit English screen for general pages I get a button 'edit', while in the American screen it is 'edit this page'. I use a compression thing that changes 'discussion' to 'talk' (don't ask me what it's called), but 'edit this page' remains long (and somewhat unnecessarily so, IMO - is anyone really going to think that while they're on the Ninja Ocelot page, clicking 'edit' will enable them to edit Gertie the Aardvark?). In a discussion continuing my thread above 'Just out of curiosity' (unanswered here except for Kww) over at AN, Kww has raised this point of the location of the discussions that lead to these changes, and I would like to know too. Peridon (talk) 11:07, 3 April 2013 (UTC)

Things are discussed in different places. Interface messages made by pages in the MediaWiki namespace are often discussed on the corresponding talk page, for example MediaWiki talk:Blockiptext. Only admins can change user rights so only admins have the "User rights management" link. I don't know whether the addition of the link was discussed, but it's a practical link and I already had it via personal js. Many interface messages at the English Wikipedia are customized but usually only for users with the default language setting "en - English" at Special:Preferences. Other languages, including British and Canadian English, get the default MediaWiki messages when there is no customization for that language, for example the default MediaWiki:Blockiptext/en-gb instead of the customized MediaWiki:Blockiptext. For this reason, I once added this to Help:Preferences: "It is not recommended to select the British English or Canadian English language options". I would say this especially holds for admins, not only for their own information but also so they see the same messages as most users they interact with. PrimeHunter (talk) 12:10, 3 April 2013 (UTC)

Your only personal js page is User:Peridon/monobook.js so I assume you use MonoBook. You import User:Edokter/CompactTabs.js. As it says, it's intended for Vector, but it's activated in all skins. In Vector the page edit tab only says "Edit" by default, so the script has no code to compact it. In MonoBook it says "edit this page" by default. Perhaps "this page" was included to explain the difference from the section edit links. You can change it to "edit" with this in User:Peridon/monobook.js:

$( document ).ready( function() {
    $( 'a', '#ca-edit' ).text( 'edit' );
});

PrimeHunter (talk) 12:22, 3 April 2013 (UTC)

Thanks. I do use Monobook. I'm also opted out of the new colours for diffs. In both cases, I find the older version far easier to work with. (And I keep Windows in Classic view too...) Could you have a look at the comments at the Repost from VP (Tech) thread at AN? Thanks. Peridon (talk) 14:48, 3 April 2013 (UTC)

Pink BG Citation needed

As in 2013 Canning riots,, is it a new change? --Tito Dutta (contact) 17:30, 3 April 2013 (UTC)

It's {{citation needed span}}. the wub "?!" 17:36, 3 April 2013 (UTC)

There's also the similar {{Clarify span}}. These "span" templates are quite nice when used properly. Jason Quinn (talk) 17:59, 3 April 2013 (UTC)

Easyblock

Resolved

– Another script has an error and stopped subsequent script execution. Amalthea 10:02, 4 April 2013 (UTC)

I use the very useful easyblock script for admins. Initially the "Block" button would show on a contributor's User, Talk, Contribution and Diffs, but now it only appears on Contributions and Diffs. I tried to fix this by adding some extra lines, but no joy. I don't really know what I'm doing, please help! The current script in my commons.js is importScript("User:Animum/easyblock.js"); //User:Animum/easyblock.js ebPrefs = {showOnPages : ["user_usertalk", "contribs", "diffs", "ipblocklist"]};

thanks Jimfbleak - talk to me? 10:01, 26 March 2013 (UTC)

Wouldn't ebPrefs = {showOnPages : ["user", "talk", "contribs", "diffs", "ipblocklist"]}; make more sense? — User:Technical 13 ( C • M • Click to learn how to view this signature as intended ) 23:36, 28 March 2013 (UTC)

That seemed the obvious thing to try, didn't make any difference though Jimfbleak - talk to me? 09:48, 31 March 2013 (UTC)

Bulleted lists and the signpost subscription template

I noticed that the multi-level bulleted list on my talk page we're displaying correctly, level 2 was inline with level 1 (see below, the second "L2. Bulleted list item 2 asterisks" line is not indented, but directly under the L1 line). It seems to have something to do with the signpost subscription template. See below, after the signpost, the two asterisks indentation isn't done, but it is OK before the template.

L1. Bulleted list item 1 asterisk
- L2. Bulleted list item 2 asterisks
  - L3. Bulleted list item 3 asterisks

L2. Bulleted list item 1 colon, 1 asterisk

L3. Bulleted list item 2 colons, 1 asterisk

- L3. Bulleted list item 1 colon, 2 asterisks

Num list
1. Num list level 2


18 November 2024
News and notes: Open letter to WMF about court case breaks one thousand signatures, big arb case declined, U4C begins accepting cases In the media: Summons issued for Wikipedia editors by Indian court, "Gaza genocide" RfC close in news, old admin Gwern now big AI guy, and a "spectrum of reluctance" over Australian place names Recent research: SPINACH: AI help for asking Wikidata "challenging real-world questions" News from the WMF: Wikimedia Foundation and Wikimedia Endowment audit reports: FY 2023–2024 Traffic report: Well, let us share with you our knowledge, about the electoral college
Single page Front page About Subscribe Suggestions Archives

L1. Bulleted list item 1 asterisk
- L2. Bulleted list item 2 asterisks
  - L3. Bulleted list item 3 asterisks

L2. Bulleted list item 1 colon, 1 asterisk

L3. Bulleted list item 2 colons, 1 asterisk

- L3. Bulleted list item 1 colon, 2 asterisks

Num list
1. Num list level 2

The :* method seems to work OK. No problems with # either. Is there something "wrong" with the template? As a side question, is ** or :* the preferred/recommended way of having an indented bulleted list? The-Pope (talk) 17:38, 31 March 2013 (UTC)

It displays fine for me in regular viewing and breaks only on edit-preview. I don't know what could be wrong. jcgoble3 (talk) 18:40, 31 March 2013 (UTC)

I've noticed that any template that uses the MediaWiki's method of short handing Unordered Lists using the MWML shorthand confuses the parser. It has something to do with the wikicore I think... I've been meaning to put in a ticket if one doesn't already exist. I guess I'll make sure to do that tomorrow (I have the same problem with {{Tbullet}}) As far as the sign-post goes, you could probably change the template to use the HTML tags for the bulleted list and prevent the problem spawning from that template (<ul><li>Item content</li><li>Item content</li><li>....</li></ul>) To answer your other question, if you want the bullet indented all by itself without being nested in another list, than you use :*. ::Examples:

List
- Nested list

- Not nested **

Not nested but properly indented :*

— User:Technical 13 ( C • M • Click to learn how to view this signature as intended ) 21:17, 31 March 2013 (UTC)

This is the issue explained in detail at Template talk:Shortcut#Changing wiki-markup to pure HTML in lists. The proper fix is to correct the wikitext in the template to not result in tag soup. Not that it shouldn't be done, but be aware that filing a bug isn't likely to help. Changing the behavior of wikitext lists would require some fairly drastic parser changes that few are able to do and likely none are willing.

Regarding the "double bullet" effect: Any subitem should begin with the same sequence of characters as its parent item. This is more clear when using numbered lists rather than bulleted:

	Wikitext	Rendering	Wikitext	Rendering
Correct:	* Number 1 Subitem with "" * Number 2	Number 1 Subitem with "**" Number 2	# Number 1 #* Subitem with "#*" # Number 2	Number 1 Subitem with "#*" Number 2
Incorrect:	* Number 1 :* Subitem with ":" Number 1 again?!	Number 1 Subitem with ":*" Number 1 again?!	# Number 1 ::* Subitem with "::" # Number 1 again?! (Note the above uses two colons because the indent for "#" is twice that of "" or ":")	Number 1 Subitem with "::*" Number 1 again?!

Using differing "prefixes" causes separate lists to be generated, as can be seen by the resetting of the numbering. Those multiple separate lists are also what causes the "double bullet" effect. Also, I am told, the multiple separate lists can be annoying to screenreader users even when the visible page appears fine. I'm not sure what tangent User:Technical 13 is going off on above. Anomie ⚔ 02:50, 1 April 2013 (UTC)

Re: Screen readers, see this section of the accessibility guideline. Graham87 07:37, 1 April 2013 (UTC)

What tangent do you mean Anomie? Technical 13 (talk) 15:54, 4 April 2013 (UTC)
"if you want the bullet indented all by itself without being nested in another list". Bringing up additional things no one asked about in the middle of a discussion like this tends to confuse the issue. Anomie ⚔ 21:30, 4 April 2013 (UTC)

User:The-Pope asked above, " As a side question, is ** or :* the preferred/recommended way of having an indented bulleted list?" I was responding to a question, not going off on a tangent. :) Technical 13 (talk) 00:17, 5 April 2013 (UTC)

I have commented at User talk:Technical 13/2013#Encyclopedia that the above is unhelpful. Johnuniq (talk) 00:56, 5 April 2013 (UTC)

Lua implementation of Template:Portal

I've written a Lua version of Template:Portal at Module:Portal, and I hope it can be rolled out fairly soon. {{Portal}} has more than four million transclusions, though, so I want to make sure it is done right. There are test cases at Template:Portal/testcases and I've outlined the details of the implementation at Template talk:Portal#Lua implementation. If people could take a look over my code and comment over there, I'd be most grateful. — Mr. Stradivarius ^{♪ talk ♪} 19:06, 1 April 2013 (UTC)

I have just made the change. If anyone notices any issues, please report them to Template talk:Portal. Thanks — Mr. Stradivarius ^{♪ talk ♪} 13:45, 4 April 2013 (UTC)

Weekend Testing Americas focusing on the new account creation and login this Saturday

You are invited to join the Weekend Testing Americas session on the new Account creation user experience: Saturday, April 6, 2013 5pm UTC - 1pm EDT - 10am PDT. We will play with the new mw:Account creation user experience for Wikimedia sites and also with the new Login user experience. Check the full test plan. This event is also part of Wikimedia's QA weekly goals. Needless to say, you are also invited to our upcoming activities.--Qgil (talk) 15:21, 4 April 2013 (UTC)

Hey everyone. For context: this event is where a group of experienced software QA testers are volunteering their time to help us test the changes our team are building as permanent design changes based on previous testing. (Most recently mentioned here). This testing will be on Labs, not English Wikipedia. The next step after that, other than smashing any bugs they identify, is to enable the login and account creation changes on an opt-in basis across the wikis, so editors can try it and help test the localization in the relevant language, etc. If you're fairly technical or willing to give the slightly wonky environment of Labs a try, you're more than welcome to join us for testing this Saturday. :) Steven Walling (WMF) • talk 18:43, 4 April 2013 (UTC)

Mediawiki amendment date

How do I find (through bugzilla, gerrit or something else) when a particular change to the MediaWiki software occurred? Specifically, this is no longer a problem, since Media:A-104.jpg now works - a year or so ago, it would be a redlink; so, when was it fixed? --Redrose64 (talk) 17:10, 4 April 2013 (UTC)

A year ago is hard, since some of the practices/infrastructure for tracking this stuff is different now. Searching Bugzilla for the relevant keywords to find if there was a bug is a good start, since that will have dates associated with it. Gerrit is easier to search if you know who made the change and/or to which part of the software (specifically, whether it was to core, a configuration change, or to an extension you know the name of). Steven Walling (WMF) • talk 21:46, 4 April 2013 (UTC)

Bugzilla would be easier, but I did it the hard way via a git bisect on a local MediaWiki install. The commit that fixed this -- at least, turned the redlink blue -- was made on 4 November 2011 by Aaron Schwarz. Crossreferencing with SVN, that would be rev:102073. - Jarry1250 ^{[Vacation needed]} 22:19, 4 April 2013 (UTC)

November 2011 is odd. It definitely wasn't working in January 2012, which is why I put that bullet into WP:REDIR. --Redrose64 (talk) 22:33, 4 April 2013 (UTC)

We weren't on fortnightly deploys at that point, the code would only have been deployed in February/March 2012, bundled into MediaWiki 1.19. Admittedly it took me a few minutes to work that out that I was searching in the wrong timeframe too :) - Jarry1250 ^{[Vacation needed]} 22:44, 4 April 2013 (UTC)

Back around 2010/2011, the de facto deployment schedule on the Wikimedia cluster for non-emergency issues was around every 3 to 6 months, so it could easily take several months between when a fix was patched in the version control system and when it got into production. Thanks to the effort that WMF has put into improving this area, we are now on a roughly once every two week deployment schedule, so updates are much more rapid. Dragons flight (talk) 22:44, 4 April 2013 (UTC)

{{PAGENAME}} in the address bar

Hello. I created a template. If I use articles with one word, it is functioning properly [6]. But if in the title a few words with spaces, template is no longer functioning properly [7]. What should I do to have turned out to use the template and articles whose title contains spaces? Thanks.--Ворота рая Импресариата (talk) 08:09, 1 April 2013 (UTC)

You need to use {{urlencode:{{PAGENAME}}}} in order to replace spaces with underscores for bare urls, like your links to en.wiki. Your code currently interprets the first space in the page name as the end of the URL. Van Isaac_{WS Vex}^contribs 08:25, 1 April 2013 (UTC)

Or {{PAGENAMEE}} may be better (depending on situation) -- WOSlinker (talk) 08:38, 1 April 2013 (UTC)

Well, he's trying to link to an en.wiki article in another wiki with [en.wikipedia.org/wiki/{{PAGENAME}} {{PAGENAME}}], which doesn't work when the article name has a space. So the first {{PAGENAME}} needs to get urlencoded with underscores. Van Isaac_{WS Vex}^contribs 09:12, 1 April 2013 (UTC)

Thank you all!! I'm sorry, but where can I read more about these and similar templates? Up to this point I did not know about their existence. Thanks.--Ворота рая Импресариата (talk) 09:51, 1 April 2013 (UTC)

WP:PF Van Isaac_{WS Vex}^contribs 10:58, 1 April 2013 (UTC)

I humbly thank you. :-)--Ворота рая Импресариата (talk) 11:27, 1 April 2013 (UTC)

I agree that [http://en.wikipedia.org/wiki/{{PAGENAMEE}} {{PAGENAME}}] would be more efficient than [http://en.wikipedia.org/wiki/{{urlencode:{{PAGENAME}}}} {{PAGENAME}}]. Just my — User:Technical 13 ( C • M • View signature as intended) 12:39, 1 April 2013 (UTC)

"efficient " - yes; useful - no! Try it with a page containing a whitespace... mabdul 09:00, 2 April 2013 (UTC)

That is why they are using {{PAGENAMEE}} or {{urlencode: it will change the whitespaces to usable characters for the URL string. On a side note, I went back and looked at your template again and it is possible to have characters that won't work the way you have it now (? & = to name a few)...

{{PAGENAME}} → Village pump (technical)/Archive 110
{{PAGENAMEE}} → Village_pump_(technical)/Archive_110
{{URLENCODE:{{PAGENAME}}}} → Village+pump+%28technical%29%2FArchive+110
{{URLENCODE:{{PAGENAMEE}}}} → Village_pump_%28technical%29%2FArchive_110

I'd use {{URLENCODE:{{PAGENAMEE}}}} instead to make sure that "every" page name was properly converted. — User:Technical 13 ( C • M • View signature as intended) 11:02, 2 April 2013 (UTC)

If you use that, you'll have incorrect output. {{PAGENAMEE}} for "10 & 6 = 2?" returns "10_%26_6_%3D_2%3F"; your example returns the double-encoded "10_%2526_6_%253D_2%253F". Anomie ⚔ 00:18, 3 April 2013 (UTC)

Mabdul: Note there is a difference between "{{PAGENAME}}" (with one E at the end) and "{{PAGENAMEE}}" (with two E's at the end). Anomie ⚔ 00:18, 3 April 2013 (UTC)

I've also just recently come across {{PAGENAMEU}}, but haven't found documentation on it yet, maybe Anomie can tell us more about it? Village+pump+%28technical%29%2FArchive+110. — User:Technical 13 ( C • M • View signature as intended) 00:28, 3 April 2013 (UTC)
{{PAGENAMEU}} is a template, equivalent to {{urlencode:{{PAGENAME}}}}. Anomie ⚔ 00:38, 3 April 2013 (UTC)

Note, "(" and ")" are safe to use in http(s) urls, and in particular when linking to wiki page names. {{PAGENAMEE}} is specificly made for using in links. The extra urlencode is unnecessary. Bawolff (talk) 17:32, 5 April 2013 (UTC)

Watchlist group by section?

I'm playing with the WP api and wondering if a watchlist variant that grouped by section name (when available) as well as by article would be useful. Has this been done already in some user script or extension or something? Just wondering if it would be useful. Mainly for busy pages like VP or teahouse or Jimbo talk. Silas Ropac (talk) 20:27, 2 April 2013 (UTC)

There's probably no such grouping variants in the watchlist as such. You'll likely need to make your own grouped lists of linked articles and use the related changes workaround. Watchlisting in general does have numerous limitations which need some technical attention, e.g. lack of multiple/grouped watchlists, fixing of bugs especially those related to large sizes (not easy to put a watchlist on a diet if both the normal and raw watchlist edits fail, Catch-22). Dl2000 (talk) 23:56, 2 April 2013 (UTC)

So far I'm just doing it with single article histories, not actually using watchlists or even psuedo-watchlists. I figure if it doesn't work nicely with a single article, there is not much point in showing multiple articles. So yesterday Village pump (policy) had 61 changes, but only 6 sections were involved. So rather than look at a flat list of 61 changes mine would present just 6 sections, if you wanted to expand the section with 42 changes you can, but if you don't care about it you can ignore it. So that is the goal anyway. One thing you lose is the overall time ordering. I could still order by time within a section, but if you want to peek at this throughout the day, it would be harder to see what's new. But for someone checking once a day it seems like it would tame the complexity somewhat. I don't know just an experiment. Which is why I was wondering if anyone has done it already, save me some time! Silas Ropac (talk) 15:17, 4 April 2013 (UTC)

Haven't found a solution, although a "section" based watchlisting can be done if a page consists of transcluded pages as sections. For example, Wikipedia:Good article reassessment lists various sections, but each article is transcluded, such as Wikipedia:Good article reassessment/Levitsky versus Marshall/1. You could watch the Levitsky versus Marshall sub article to see the action in just that "section". Note that watching the GAR page will not indicate what sections are being added or purged - apparently you would have to watch the transcluded User:VeblenBot/C/Wikipedia good article reassessment for those. As mentioned, watchlisting in general is in need of much fixing and functional upgrading, and section-based watching would be useful. However, drawbacks may involve usability issues and complications with the extended functionality e.g. accidentally setting a watch only a section when the entire article is intended, or vice-versa. Dl2000 (talk) 22:28, 4 April 2013 (UTC)

That's interesting about transclusion. But I was hoping to make this work for any page. I was really thinking only about displaying the changes by section, not about subscribing certain sections. But maybe one will lead to the other. But I see an issue. If you grab enough history for a page that is actively archived like a Village Pump page, you are going to get changes for a lot of sections which have already been archived. But from the history of the page there is no obvious way to distinguish between sections which have been archived or not nor to figure out the correct archived links. I could grab the full text of the head revision, and search through it for section names to determine which sections are still on the page. And then just drop any section which has fallen off. This might be good enough for my purposes. So this would be like "for the sections that are currently on the page, show me the changes in those sections" which I think is useful, even if it is limited to only recent stuff.

So does archiving just blatantly break links and nothing is automatically repaired or maintained? Because I have linked to often-archived pages before and find the link just breaks ones archiving happens. If it's then fixed by hand presumably it's stable from then on. But really even a one-time breaking is not cool, seems like you want some kind of permalink which survives the archiving. I have no idea how that would be implemented. Silas Ropac (talk) 02:51, 5 April 2013 (UTC)

It depends upon how the archiving is done. Most archiving is done by bots, and there are several of these, which use different techniques. For example, the MiszaBot family are popular, but these do a simple cut&paste, which often breaks section links - and the bot leaves these in a broken state.

By contrast, when ClueBot III (talk · contribs) archives a thread, it also uses cut&paste, but then goes around fixing links to that section which would otherwise have become broken - see, for example, these six edits. I don't know if it's 100% successful though. If you know PHP, you could check the bot source: search for $forktasklist

Permalinks might help, but the disadvantage of those is that they link to a thread as it was at a particular instant in time, and there may have been subsequent posts prior to the archiving. --Redrose64 (talk) 13:07, 5 April 2013 (UTC)

That's neat that ClueBot III (talk · contribs) tries to fix links, I didn't know that. For "permalink" I was thinking not a link to a old revision of a page, that wouldn't show new content, but some kind of "permanent" link to the section on whichever page it ended up. I'm probably asking for magic here, because I have no idea how to implement it given that sections can be renamed, moved, duplicated, copied, etc. Seems like section would have to be given some kind of ID which was independent of the section name. And that ID would have to have to be in the wikitext but not get edited. And then the server would have to be able to take a request for page&sectionid=123 and return page#foo or page#bar or page/archive_109#foo or page/archive_109#bar depending if the section had been renamed or moved. But it would apply to normal articles, not just discussions, because today those incoming section links break if sections are renamed or articles are split or merged. So the dream is being able to make a link to a chunk of a content within a page, and have that link permanently take me to the future chunk which best matches or was derived from the old chunk, even if the chunk moved to a new page. I'm thinking that might be 5-10 years away, nothing I'm waiting on certainly. Silas Ropac (talk) 14:29, 5 April 2013 (UTC)

Potentially useful side effect of MathJax

In August 2012 there were some discussions about changing the preferred font in class="texhtml". Sysop Edokter declined to make changes in Commons.css on a pretext that

“

The installed base of the MathJax fonts amongst our readers is virtually nil, so it makes no sense to put this in Common.css

”

Later I noticed that, when MathJax uploads its fonts, the appearance of "texhtml" (a.k.a. {{math}}) switches to these font, even if they are not installed, if the CSS refers to names of these fonts. This can serve as a cheap improvement of appearance of simple formulae. Namely:

If a reader uses the low-quality PNG renderer, then s/he will see serif, but if s/he uses this renderer, then the typesetting quality will not be high in any case.
If a reader enables MathJax (one of available MathJaxes), but the page does not contain any <math>, then the font of class="texhtml" will depend on whether MathJax fonts were previously downloaded (or installed locally). But it is not important to mimic the MathJax appearance with {{math}}, because there are no MathJax formulae on the page.
If a reader enables MathJax and the page loads it, then the proposed setup will make class="texhtml" formulae uniform with MathJax formulae.

This can encourage editors to use more template formatting ({{math}}, {{sfrac}}, {{sqrt}}, {{radic}}, {{closed-open}}, {{mvar}}, {{vec}}…) which will allow articles to render faster and will diminish other MathJax’s adverse effects (such as page jerks).

Opinions? Incnis Mrsi (talk) 09:34, 4 April 2013 (UTC)

There are some issues with this approach:

The font used for texhtml will depend on wether MathJax is actually used on a page. This results in inconsistent behaviour of these templates, which I think is a bad thing.
Consistency aside, the mathjax fonts are still no match in quality when compared to native system fonts. The webfonts are not TrueType (they're based on OTF), and lack the hinting that is needed for optimal display (at least on Windows). Using MathJax would be detrimental IMO. — Edokter (talk) — 16:33, 5 April 2013 (UTC)

Help with some boxes

I made some changes on the WP U2 page but then it chaged its original format. the the Wikiproject box, the Quick links box, the discussion pages box and the Statistics box went to the bottom instead of staying the right where it always were. Please, help me to undo this but keeping the new adds

See, this is where I add and I need to keep it there:

To-do list for Wikipedia:Village pump (technical)/Archive 110:

To get Bono or The Edge to GA status.
Addition of pictures and/or photographs to illustrate the relevant topics/sections.
Improve and/or create articles related to U2 (Ex. Paul McGuinnes, Brian Eno, Daniel Lanois, Arthur Fogel, Larry Mullen Jr., Adam Clayton, Walk In My Shoes Project, From the Ground Up (new U2 material), Eve Hewson, etc)
To launch The Collaboration of the Month Department
To gather more Wikipedian to colaborate in the project List of Active Members.
Update and improve the Outreach Department
Update the Newsletter's pages.
Improve the Wikiproject U2 Barnstars.

Miss Bono (zootalk) 19:27, 4 April 2013 (UTC)

This has already been resolved (I guess). Under the project section at Wikipedia:WikiProject U2/leftpanel, the to do list is included. --Ushau97 ^talk 09:56, 5 April 2013 (UTC)

Collapsed information does not remain in section

In my sandbox (User:MarshalN20/Sandbox4), in the "Players" section, I am trying to collapse the "Recent Call-ups" to the bottom of the section. Yet, the template skips the section and mixes with the template in the "Managers" section. Please help.--MarshalN20 | ^Talk 03:02, 6 April 2013 (UTC)

Fixed--Make cat 05:36, 6 April 2013 (UTC)

Problematic CAPTCHA

Here someone has reported that they faced trouble with the verification code while creating an account! (you may need an ACC account to view this) --Tito Dutta (contact) 10:42, 4 April 2013 (UTC)

Most of us don't have an ACC account. Can you summarize or repost it here? Steven Walling (WMF) • talk 18:32, 4 April 2013 (UTC)

Here is another one! Both the editors attempted to answer the CAPTCHA (one editor mentioned "more than 10 times") but could not answer correctly no matter what I did I could not get the code word right to create my account . --Tito Dutta (contact) 19:05, 4 April 2013 (UTC)

So it was just that they found it too difficult to read? Steven Walling (WMF) • talk 21:49, 4 April 2013 (UTC)

Requesting an account right now. I'll let you know what I find. Technical 13 (talk) 19:04, 4 April 2013 (UTC)

Third one. They also told that they could not create an account for CAPTCHA error! --Tito Dutta (contact) 19:18, 4 April 2013 (UTC)

Still haven't gotten a response to my request for an account... They must have a good size back-log there... Technical 13 (talk) 00:18, 5 April 2013 (UTC)

<aside>"CAPTCHA Error" would be a great name for a band.</aside>--ukexpat (talk) 02:49, 5 April 2013 (UTC)

Using the self-checkout at the local library a while back, the screen showed Fatal Error. Took me a moment to figure out it was the title of the book. -- Gadget850 (Ed)^talk 17:54, 6 April 2013 (UTC)

"I'm sorry, but, your account has not been approved by a site administrator yet. Please stand by." Technical 13 (talk) 20:19, 6 April 2013 (UTC)

Encoding section names into anchors

From an external site how can I encode the string of a section name using javscript into a url which jumps to the anchor for that section. I read on stack overflow the recipe is "UTF-8 percent-encoding, but with . instead of %, and spaces replaced with underscores; additionally, multiple consecutive whitespaces are collapsed, and : is preserved (not encoded into .3A)" Is this correct, is it officially documented somewhere, anyone have a javascript snippet that does this? Silas Ropac (talk) 02:25, 5 April 2013 (UTC)

The relevant code is at CoreParserFunctions::anchorencode -> Parser::guessSectionNameFromWikiText -> Sanitizer::escapeId. But be aware:

"The problem there is that the output of guessSectionNameFromWikiText depends on the configuration of the particular wiki. $wgHtml5 and $wgExperimentalHtmlIds are used, and maybe others." [8]

So the only safe method is to use the API. In order to use the API off-site client side, you usually need to do a callback. The best way I can see to encode an anchor via the API with a callback is a quick action=parse: action=parse&callback=foo making sure to encodeURIComponent() the anchor parameter. You then create a <script> object with that URL as the SRC, which bypasses any sort of XSS problems. Hacky, but effective (except in most versions of IE where there are limited numbers of <script> tags allowed). --Splarka (rant) 07:49, 5 April 2013 (UTC)

Thanks for thorough answer, it is appreciated. For now I just care about en.wikipedia.org and so I'm hoping I can do something off-site, encode it one fixed way. I haven't yet used mediawiki API callbacks and don't really understand XSS issues. If I pull down a history with 30 sections and want to link them all, would that be 30 API calls or just 1 somehow? I guess I could make the links not point to WP at all, but point to my site in such a way that they are resolved into section links only if they are clicked on. Like google's result page which contains links not to external sites but to google, which then forward to the external sites. Silas Ropac (talk) 12:44, 5 April 2013 (UTC)

You could just parse them all in one go via text=, but if you just plan to emulate en.wp, probably you should port the above linked PHP code from Sanitizer::escapeId into javascript and use it as you originally intended. The configuration for en.wp is mostly visible in the mess at http://noc.wikimedia.org/conf/ (unless these got obsoleted while I wasn't looking).--Splarka (rant) 07:50, 6 April 2013 (UTC)

One case puzzles me, if two or more sections on a page have the same name, someone gives their ID's numeric suffixes to distinguish them. Seems like there is no way to duplicate that from just a section name, you have to see the whole page. Probably a very rare case, but still. However I realize now I'm going to have to call action=parse, prop=sections anyway, and it conveniently returns both the text and the anchor for each section. So I have to use the server after all, so I will be right on any wiki. Although I think if I did do it on my end, I could get it 99% right anyway, duplicates seem rare.Silas Ropac (talk) 18:56, 6 April 2013 (UTC)

Article assessment template problem

I'm not too good with template coding as it is and I'm rather sleepy at this point, so I know better than to try to solve this myself. Would somebody please take a look at Template:WPBreakfast assessment level category-main and quite possibly Template:WPBreakfast importance level category-main as well. The first one was creating a problem at Category:WikiProject Breakfast articles before it was removed. The second could easily be problematic too, I really don't know. AutomaticStrikeout (T • C • Sign AAPT) 03:45, 6 April 2013 (UTC)

Both templates require a parameter to be passed to it (|class= or |importance=) in order to format correctly. Thus, they are designed to be used only on categories for specific classes/importance ratings, not on the main category. jcgoble3 (talk) 05:18, 6 April 2013 (UTC)

Okay. Thanks. AutomaticStrikeout (T • C • Sign AAPT) 13:50, 6 April 2013 (UTC)

Saving default watchlist settings

Is there a way for one to set default watchlist settings (i.e., "hide my edits" by default)? I feel like I'm missing something very simple here. —Theopolisme (talk) 16:29, 6 April 2013 (UTC)

Yes, on the "Watchlist" tab of your Preferences. -- John of Reading (talk) 16:47, 6 April 2013 (UTC)

Problem with toolbar

The table icon for "insert table" and the new line icon have fallen down into the edit window and are displaying there instead of on the toolbar. It's distracting because it overlays the first line or so of text in the edit window. Don't know how long this has been going on - I was gone for a week and it was displaying like this when I returned. I haven't changed preferences recently. Browser is Safari. Truthkeeper (talk) 18:13, 6 April 2013 (UTC)

See Wikipedia:Village pump (technical)/Archive 109#Strange error in edit mode, particularly kipod's comment. – PartTimeGnome (talk | contribs) 20:08, 6 April 2013 (UTC)

That worked. Now I can see the top of the edit window; thanks. Truthkeeper (talk) 21:50, 6 April 2013 (UTC)

Classic skin and CSS

The WMF has announced that in two weeks time, it will stop supporting the Classic skin. As per this XKCD strip, this will severely impede my workflow.

I've been told that the majority of Classic's features can be replicated via .css tinkering. However, this would require a skillset I don't have. Can anyone help out?

(I want serif fonts - preferably Times New Roman. I know how to set my browser so that all pages everywhere will be in Times New Roman, but I'd rather it was just all WMF pages. Because those are the pages that are affected by my user skin. This is doable, yes?

(And I prefer the 'toolbox' links to be off to the left side, not the top, and I don't want collapsibles.) DS (talk) 19:06, 2 April 2013 (UTC)

toolbox on the left: unless you mean something else in "toolbox" than what i think you mean, this is already the default for vector.
non-collapsible: this is settable from Preferences => Appearance => Enable collapsing of items in the sidebar in Vector skin
fonts: if you want times all over, add the following to your CSS page
```
body { font-family: "Times new roman";}
```
if you want it for content only, use
```
#bodyContent { font-family: "Times new roman";}
```

peace - קיפודנחש (aka kipod) (talk) 19:22, 2 April 2013 (UTC)

Although it will only affect a small minority of users, I believe that by making this change, without any form of discussion and only minimal notice, the Foundation is seriously overstepping its mark. This should be fully discussed, on all Wikis, and there should only be change if there is good reason and consensus to do so. Objections to this heavy-handed approach can be made [9]. An optimist on the run! 19:34, 2 April 2013 (UTC)

Ironic that you would call for making objections on that page. The purpose of that Meta page is to get translations for advance notifications to each local community about the reasons for the change, who it will impact and so on. Oliver Keyes and others are working on making sure there are announcements etc., so please assume good faith instead of rushing to get your pitchfork. There will be an opportunity to comment about the change without needing to go to Meta. Steven Walling (WMF) • talk 20:23, 2 April 2013 (UTC)

Then please tell us where we can make objections, and let us know if such objections will be listened to. An optimist on the run! 20:46, 2 April 2013 (UTC)

I let Oliver know over IRC so he can explain the plan. Steven Walling (WMF) • talk 20:49, 2 April 2013 (UTC)

Both users have now commented on the meta page provided, which explains the plan - that seems to resolve that :). Okeyes (WMF) (talk) 21:30, 2 April 2013 (UTC)

I've seen this brought up somewhere else as well. I'm curious, what "other" skin is "classic" the least different from? I would like to be able to offer an alternative that isn't that different to some of the questions I've seen like "what other skin uses the full page width" that I think was on WP:HD or WP:TH. I'd also like to have available the differences in those two skins so I can quickly offer css they can add to their common.css or skin.css to make it feel like they still are using classic. Any ideas here? — User:Technical 13 ( C • M • View signature as intended) 23:39, 2 April 2013 (UTC)
- I think that Cologne Blue would be the best starting point. --Carnildo (talk) 00:21, 3 April 2013 (UTC)
  - If you like full width use Vector and a custom css like mine. - Thomas d stewart (talk) 00:36, 3 April 2013 (UTC)

The skins that will remain are

Those in the process of removal are

Click any of the nine links to see what WP:VPT looks like in each skin. --Redrose64 (talk) 15:24, 3 April 2013 (UTC)

Aw, nostalgia! Not that I use it, but I'm .... nostalgic :) -- phoebe / (talk to me) 19:12, 7 April 2013 (UTC)

And how do I get the yellow (hex code FF FF EC) background on all non-article pages? DS (talk) 22:29, 3 April 2013 (UTC)

I think something like this in your common.css will do what you want (untested):

div#content, div#content div.thumb {
    background-color: #FFFFEC;
}
.ns-0 div#content, .ns-0 div#content div.thumb {
    background-color: white;
}

– PartTimeGnome (talk | contribs) 22:53, 3 April 2013 (UTC)

I'm working out the details with the help of user:Coren; however, he tells me that it will also require modifying the layout, which is a .js issue. Is anyone here sufficiently competent with .js to help me do this? DS (talk) 00:08, 4 April 2013 (UTC)

Unwatch bug?

I've been using this unwatch script for five years - it suddenly stopped working. Any ideas how to fix this? Simple instructions, please - I am not a coder. A note on my talk would be great. Thanks Tvoz/talk 08:22, 7 April 2013 (UTC)

The culprit in that script seems to be the !document.forms[0].namespace; apparently namespace is no longer being defined, so the script aborts at that point. You could try replacing with User:Dl2000/Unwatcher.js with omits the namespace test and seems to show the unwatch links. The only issue is that in recent times the Wikimedia software added an extra step to the unwatch process, showing a sort of are-you-sure page which completes the unwatch when the button is clicked. Not sure if there is an option available in the unwatch call to skip that formality. Dl2000 (talk) 18:52, 7 April 2013 (UTC)

Thank you- it worked! Tvoz/talk 02:55, 8 April 2013 (UTC)

API and uccontinue

I've been playing around with the Special:ApiSandbox feature lately, trying to learn some of its myriad uses, but I can't for the life of me figure out what the uccontinue option is for under list=usercontribs. It seems like it should be a way to get the next set, but what should I enter there? ~ Amory (u • t • c) 00:32, 8 April 2013 (UTC)

If it is passed back to you in the "query-continue" node (or the "continue" node, for the new-style continuation), you enter the value given when you make the next query. Otherwise you don't use it.

For example, this query (using the new-style continuation) includes a "continue" node of <continue continue="-||" uccontinue="AnomieBOT|2013-04-07T23:39:45Z" />; you would merge that into your original query like this. On the other hand, this query has a continue node of <continue continue="-||" ucstart="2013-04-07T23:39:45Z" />, so you wouldn't use uccontinue; instead you'd use ucstart as given, merging it into that original query like this. Note, of course, that whether "uccontinue" or "ucstart" happens to be returned is liable to change; your code should just use whatever is given without concern for what the keys or values are. Anomie ⚔ 01:22, 8 April 2013 (UTC)

Okay, that's beginning to make a fair amount of sense. Thanks a bunch. ~ Amory (u • t • c) 05:39, 8 April 2013 (UTC)

Problems with images

I am having problems getting images to display corectly when a new version of an image is uploaded or when reverting to an older revision. An example is at File:Kurz Weber.JPG, where I made a tighter crop of the subject. The image is displaying as the older, wider crop even though the tighter crop is the one selected. I just can't seem to get the correct image to be displayed. Others are having trouble as well, for example see User talk:Nowyouseeme#Question. -- Dianna (talk) 15:01, 6 April 2013 (UTC)

Sounds like you need to WP:REFRESH the page. Happy editing! Technical 13 (talk) 15:11, 6 April 2013 (UTC)

It appears that you are not the only having this issue. See WP:HD#Purge image thumbnail --Ushau97 (talk) 16:18, 6 April 2013 (UTC)

How and where exactly can I *see* the problem? All images except for the initial one seem to have the same size (363 x 272)? --AKlapper (WMF) (talk) 23:05, 6 April 2013 (UTC)

This is a browsercache issue. Browsers don't fetch the new image with the 'source' server because the URL of the old and the new image are identical. I believe this was a rather difficult issue to solve, without increasing either cache hitcount (by forcing cache revalidation), or regenerating pages (changing urls of thumbs). Easy to replicate. Upload an image, view it, upload a new version on top of that image. Go back to your 'view page' and you will see that you will get a thumbnail of the old image, instead of your new upload. —TheDJ (talk • contribs) 10:09, 8 April 2013 (UTC)

No, it is not. It is trivial to override the browsercache on any modern browser with either Ctrl-F5 or just opening the image separately and clicking refresh. You can also check in other browsers. The problem is with certain Wikimedia image servers. It's such a common problem that I've even developed a template for the situation, so that image curators have a link to the current image rather than the one displayed on the page. — trlkly 13:39, 24 April 2013 (UTC)

Talk page tab

Tracked in Phabricator
Task T3495

A few months ago, there was a large discussion regarding whether the tab should be renamed 'Talk' to avoid user confusion. However, this change has only taken place in the en version of Wikipedia. Would it be possiable to roll it out to other English versions (for example when a user sets their language to en-GB). Oddbodz (talk) 20:18, 6 April 2013 (UTC)

We normally discourage people from selecting both "en-GB - British English" and "en-CA - Canadian English" because so many of the messages are out of step with the vanilla "en - English". Keeping three variants of the same language in step is a nightmare. --Redrose64 (talk) 21:00, 6 April 2013 (UTC)

Do users who select those options know this? If not, they should be removed. Or kept "in step". --Dweller (talk) 21:46, 6 April 2013 (UTC)

I don't know how to inform these users. Nor am I willing to synchronise the lists. What I do know is that this list is shorter than this list, which itself is much shorter than this list. --Redrose64 (talk) 22:24, 6 April 2013 (UTC)

I once added this to Help:Preferences: It is not recommended to select the British English or Canadian English language options ("en-GB - British English" or "en-CA - Canadian English") because many interface messages have been customized for U.S. English Wikipedia ("en - English"). The help page is linked from the top of Special:Preferences via MediaWiki:Preferences-summary – but only if you have the default "en - English" as language! We could create MediaWiki:Preferences-summary/en-gb and MediaWiki:Preferences-summary/en-ca to add the link for those languages, or maybe make another message which says directly that the language is not recommended. For the general issue, MediaWiki namespace messages permit transclusion like {{MediaWiki:Histlegend}} at MediaWiki:Histlegend/en-gb which ensures that en-gb users get the current version of MediaWiki:Histlegend. Creating such transclusions for both en-gb and en-ca for every customized message would be possible, but maybe a bit excessive. We could also make a bugzilla request for wikis to get an option to say for selected language pairs (X, Y): If X has a customized message and Y doesn't then Y should automatically display the X message. PrimeHunter (talk) 23:50, 6 April 2013 (UTC)

The developers recently had a go at doing this (i.e. so en-GB and en-CA users saw customised en messages). The change was reverted as it caused non-English users to see a confusing mix of English and their native language. This was discussed at Wikipedia:Village pump (technical)/Archive 109#Language preferences getting mishandled. There is ongoing discussion on how best to handle message translation in bug 1495. – PartTimeGnome (talk | contribs) 14:54, 7 April 2013 (UTC)

Is there any reason not to just disable the GB and CA variants as options entirely here? While having english variants available might have uses on some projects, here it just seems to complicate things. -— Isarra ༆ 07:52, 7 April 2013 (UTC)

I would support this (and en-us, if it's enabled). Wikidata is experiencing quite dramatic similar issues. Andrew Gray (talk) 16:19, 8 April 2013 (UTC)

Special:Random

Is it possible to change the behavior of Special:Random without getting the developers involved? I'd like to ask for a modification (to prevent it taking you to disambiguation pages), but there's no point in requesting something that can't be done. Nyttend (talk) 04:21, 8 April 2013 (UTC)

Not possible as far as I know —TheDJ (talk • contribs) 08:07, 8 April 2013 (UTC)

It would take developers, I'm afraid. But then again, MediaWiki does already track disambiguation pages (Special:Disambiguations), and I'm pretty sure Special:Random already skips redirects, so it may be pretty trivial to make it treat disambiguations the same way. It's a minor thing, but if someone's bored, why not? File a bug, perhaps. -— Isarra ༆ 05:20, 9 April 2013 (UTC)

Wikidata phase 2 deployment delayed

Heya folks :)

Just a quick note that we decided to delay the deployment of Wikidata phase 2 to fix some technical issue we are experiencing and to give you more time to decide on some initial groundrules. I'll let you know as soon as I have more info on the new deployment date. Please please do make use of the time until then. I am here to answer questions and all but I can't lead this. --Lydia Pintscher (WMDE) (talk) 17:57, 8 April 2013 (UTC)

Raw HTML in error message

Tracked in Phabricator
Task T40894

I am sure, I have seen it before too and this is not the first experience. From ACC I went to create an account and got exactly the following error message

The user name "The username" <a href="/wiki/MediaWiki_talk:Titleblacklist" title="MediaWiki talk:Titleblacklist">has been blacklisted</a> from creation. Wikipedia <a href="/wiki/Wikipedia:U" title="Wikipedia:U" class="mw-redirect">username policy</a> does not allow names that are misleading, promotional, offensive or disruptive. Please select another username that complies with <a href="/wiki/Wikipedia:U" title="Wikipedia:U" class="mw-redirect">policy</a>, or if you want to seek approval for a username, you can do so by filing a request at <a href="/wiki/Wikipedia:Request_an_account" title="Wikipedia:Request an account">Wikipedia:Request an account</a>.

--Tito Dutta (contact) 22:30, 8 April 2013 (UTC)

Perhaps related to T40894? Technical 13 (talk) 22:51, 8 April 2013 (UTC)

New vector image won't update

I've tried the purging steps mentioned in Wikipedia:Purge and haven't had any luck in getting File:Duke Energy logo.svg's 200px version to purge itself. Subsequently the wrong image is being displayed in the article Duke Energy. Any help would be great. Shep^Talk 03:39, 6 April 2013 (UTC)

I applied a very temporary fix, but someone should look into the underlying issue. jcgoble3 (talk) 05:05, 6 April 2013 (UTC)

The 200px thumbnail version is the updated one for me (I'm accessing it from Europe). Is this still an issue for you, and from which continent? (On a general note, it's welcome to enter reproducible software or server issues in the bug tracker by following the instructions How to report a bug. This is to make developers of the software aware of the issue.) For your interest, a list of known software problems with thumbnails can be found here. --AKlapper (WMF) (talk) 10:09, 6 April 2013 (UTC)

I checked from another device just to be sure and it's still an issue for me, I'm in the USA. Shep^Talk 02:58, 7 April 2013 (UTC)

This problem should be fixed now, thanks to Leslie of the Operations team (might need a purge though). If it is not, please leave a comment in bugzilla:46976. We're sorry for the inconvenience caused. --AKlapper (WMF) (talk) 02:36, 10 April 2013 (UTC)

Deleted user pages don't give a 404 error

If an account exists then a url in its userspace never gives a HTTP 404 Not Found in my tests at http://404checker.com/404-checker. Deleted pages and pages that never existed both give 200 OK. Is this a bug or intentional? I definitely think bug. The current behaviour can cause deleted user pages to remain in search engines. See for example the Google search "This page has been deleted. The deletion and move log for the page are provided below for reference" site:en.wikipedia.org ("About 139,000 results"). Examples of red links versus 404 checks: User:Shastri09 was deleted (404 check). User:Shastri09/test never existed (404 check). If an account doesn't exist then it gives 404, for example User:Shastri09test (404 check). Deleted and never-existed pages in other tested namespaces both give 404 as they should. Examples: Lunatic labs was deleted (404 check). Lunatic test never existed (404 check). I think deleted user pages once gave a 404 but I'm not sure. PrimeHunter (talk) 12:24, 8 April 2013 (UTC)

The list from Google... How long ago were those pages deleted? I believe Google has an archive of pages cached that expires after (iirc from the report I did on Google in one of my classes last summer) three months. So, any page deleted within that time frame would still show up there. The other stuffs I'm in no position to comment on. Technical 13 (talk) 13:02, 8 April 2013 (UTC)

If Google visits a page and gets a 404 error from the server then the page should be removed right away. The search terms show that Google has visited the pages after they were deleted, but they didn't give a 404 error. Deleted pages in many other namespaces like Lunatic labs and Category:Use mdy dates from May 2010 display "A page with this title has previously been deleted". The Google search "A page with this title has previously been deleted" site:en.wikipedia.org gives no deleted pages as far as I can tell (only existing pages quoting the message). That supports my experiments with the 404 checker: It is only in userspace that deleted pages don't give a 404. PrimeHunter (talk) 13:40, 8 April 2013 (UTC)

I did some research on Google that I hope can be helpful: FAQ→Remove a page or site from Google's search results and Meta tags→Using the robots meta tag. Technical 13 (talk) 14:00, 8 April 2013 (UTC)

http://lists.wikimedia.org/pipermail/wikitech-l/2013-February/066748.html Bawolff (talk) 14:33, 8 April 2013 (UTC)

Thanks for the link. It sounds like a poor decision to me, and why does it include user subpages? Google says in [10]: "We recommend that you always return a 404 (Not found) or a 410 (Gone) response code in response to a request for a non-existing page." If we want to claim that non-existing pages exist then we should at least noindex them. We don't do that now for many of the deleted userspace pages (we do for some, I don't know the system, maybe it depends on when they were deleted). A deleted userpage with a company name may remain the first Google hit on the company. That sounds like an unreasonable "punishment" for creating a promotional userpage. PrimeHunter (talk) 15:01, 8 April 2013 (UTC)

But the page does exist (links to the user logs are valid for instance) It's an empty page, but existing. You name the exact problem. You want to 'punish' the user for doing something we don't approve of in the Google statistics. I guess adding NOINDEX to deleted user pages is a way to do that. That doesn't fix promotional username without a userpage though. —TheDJ (talk • contribs) 15:24, 8 April 2013 (UTC)

ttbomk, it really does not matter if the page exists or not - if the url structure is correct (<WP>/wiki/something in this case), WP will never return a 404. instead it returns a page that politely informs you that such an article/template/user page/whatever does not exist on this wiki, and will you be interested in creating one?. adding "noindex" magic word will not induce 404 either, but it will cause the returned page to come back with the "noindex" meta tag, which may induce google to drop this page from its DB. peace - קיפודנחש (aka kipod) (talk) 17:19, 8 April 2013 (UTC)

Kipod, Wikipedia does return 404 errors for non-user pages. For example, use a tool like Wireshark to monitor your network traffic, then go to a deleted page like this one. You should see a 404 status code returned to your browser. (If you don't get a 404, then some misbehaving proxy server is getting in the way.) Most web browsers only show the page content, not the status code. The status code can still be 404 even if the page content doesn't mention "404". – PartTimeGnome (talk | contribs) 22:12, 8 April 2013 (UTC)

you are correct - i wrote my previous response and then i did edit it to add "ttbomk" because i was not 100% sure. i did try to find the message status code before i wrote it, i just did not look in the right place... BTW: you do not need to download any external tool like wireshark if you use chrome, or FF with firebug - the status is available under F12-Network => Status. peace - קיפודנחש (aka kipod) (talk) 22:50, 8 April 2013 (UTC)

There are web services like http://web-sniffer.net/ that show web page headers. That service shows status "HTTP/1.0 404 Not Found" for a Wikipedia page that does not exist (and "HTTP/1.0 302 Moved Temporarily" for the red link above, which does show a page). Johnuniq (talk) 00:05, 9 April 2013 (UTC)

The issue I'm posting about is that the behaviour is different for non-existing user pages like http://en.wikipedia.org/wiki/User:Shastri09 (deleted) and http://en.wikipedia.org/wiki/User:Japloessl (never created). They both say "Status: HTTP/1.0 200 OK". There is no noindex tag on the pages so there is nothing stopping search engines from indexing them. Google is apparently indexing more than 100,000 of them. It seems the only way to remove the pages from Google with the current software is to create them and add __NOINDEX__ to them (or add a template which adds __NOINDEX__). Users cannot be expected to know that, and it seems illogical that you have to create a page to avoid it being indexed. And if you do create it then mirrors may copy it without including noindex, so it ends up being indexed elsewhere. PrimeHunter (talk) 01:10, 9 April 2013 (UTC)

But why is being indexed a problem ? These users exist. They should be indexed in my opinion... (And that was also the point of why the change was made). —TheDJ (talk • contribs) 08:06, 9 April 2013 (UTC)

Or do you want to add a user preference for it ? "Include my user pages in search engines" ? —TheDJ (talk • contribs) 08:09, 9 April 2013 (UTC)

As an aside, the original justification (for openid) I don't think applies anymore (given the concerns about user renames when using user pages as open id identifiers). Bawolff (talk) 11:41, 9 April 2013 (UTC)

There is an arguable case for the main user page of a registered user to return 200 OK (and be indexed) even if the page doesn't exist, since the page title also refers to the user who does exist. However, that shouldn't apply for all possible sub-pages of the user page (of which there are practically infinite possibilities). The title of a sub-page only represents a page, which doesn't exist, so should return 404 to my mind.

You might be interested to know that user pages of IP addresses also return 200 OK (and hence can be indexed by search engines), even if the address has never edited, and even if it is technically impossible for the IP address to ever edit. I really don't think User:0.0.0.0/sandbox should give a 200 OK response, but it does at present. – PartTimeGnome (talk | contribs) 20:34, 9 April 2013 (UTC)

Two very unrelated questions

On File talk:Breakfast!.jpg, why does the tag for WikiProject Breakfast list the file as NA-class instead of File-class? AutomaticStrikeout (T • C • Sign AAPT) 00:49, 9 April 2013 (UTC)
- It actually seems to have it listed as both. Technical 13 (talk) 00:55, 9 April 2013 (UTC)
Well, it lists it as a file for WikiProject Food and Drink, but not for WikiProject Breakfast. AutomaticStrikeout (T • C • Sign AAPT) 01:06, 9 April 2013 (UTC)
If it was I that was curious, I would ask them on WikiProject breakfast. :) Technical 13 (talk) 01:14, 9 April 2013 (UTC)

It's because Template:WikiProject Breakfast has an incorrect value at "QUALITY_SCALE" and is accidentally using the default quality scale. See Template:WPBannerMeta/doc#Assessment. -- John of Reading (talk) 06:55, 9 April 2013 (UTC)
John of Reading is correct. {{WikiProject Breakfast}} has |QUALITY_SCALE=Wikipedia:WikiProject Food and drink/Assessment which is invalid (see Template:WPBannerMeta#Assessment). For File-class to be recognised, it should be set to |QUALITY_SCALE=extended --Redrose64 (talk) 11:44, 9 April 2013 (UTC)
Does this solve the problem? AutomaticStrikeout (T • C • Sign AAPT) 15:31, 9 April 2013 (UTC)
Judging by this, yes. Please note that it may take some time before they all show up in Category:File-Class Breakfast articles, because of the job queue. --Redrose64 (talk) 17:45, 9 April 2013 (UTC)
As you can see at my signature history, the word "Automatic" is red (like a redlink) in two of the more recent regular versions. However, I am quite certain that it was actually blue, so why is it showing as red? It is also appearing this way on other pages. AutomaticStrikeout (T • C • Sign AAPT) 00:49, 9 April 2013 (UTC)
- I was actually somewhat befuddled by that behavior myself when I was working on helping you rework your signature. I think it has something to do with the fact that one of the styles was invalid or syntactically incorrect so the whole style definition was being dropped, but I'm not entirely sure. Technical 13 (talk) 00:55, 9 April 2013 (UTC)
Is there any way to fix that? AutomaticStrikeout (T • C • Sign AAPT) 01:06, 9 April 2013 (UTC)
It shows as the color blue that I thought you wanted now, are you asking if there is anyway to go back and fix all of your previous signatures? If so, yes, but it would require bot access to go through all of the pages you've ever signed and replace the old signature with the new one without disrupting anything else. Technical 13 (talk) 01:12, 9 April 2013 (UTC)
I'm talking about my old signatures, some of which contain red where blue should be. I'm trying to figure out why that is. It would appear that something got deleted. AutomaticStrikeout (T • C • Sign AAPT) 01:15, 9 April 2013 (UTC)
You can write darkblue (one word) instead of dark blue. PrimeHunter (talk) 01:29, 9 April 2013 (UTC)
Or just use hex like you do now since "#00008B" is shorter than "darkblue" anyways. :) Technical 13 (talk) 01:36, 9 April 2013 (UTC)
I'd really prefer not to go back and change my signature a bunch of times. My opinion is that something must have been deleted in order to cause the blue in my signature to become red. Is there some way to correct this, perhaps by undeleting something? AutomaticStrikeout (T • C • Sign AAPT) 04:09, 9 April 2013 (UTC)
I've been back through a bunch of your old posts and can find plenty of sigs that are blue and orange, also some that are orange and blue (mainly from Oct/Nov 2012), but cannot find signatures which are partly red. Please give links to examples of these. --Redrose64 (talk) 12:35, 9 April 2013 (UTC)

He's talking about the ones he linked here From March 11, 2013 to April 3, 2013. Technical 13 (talk) 12:56, 9 April 2013 (UTC)

I've put my verdict at User talk:AutomaticStrikeout/signature history#Markup problems. --Redrose64 (talk) 14:13, 9 April 2013 (UTC)
It's because  isn't valid code in any way, shape, or form. I dunno why it suddenly doesn't show up correctly; I remember always seeing your signature as red, and it never occurred to me that it was displaying correctly at all. Maybe your browser updated, and so it fixed whatever bug that caused it to show up as you intended (when it shouldn't have)? EVula // talk // ☯ // 05:55, 9 April 2013 (UTC)

⇑ is exactly what I meant when I said "I think it has something to do with the fact that one of the styles was invalid or syntactically incorrect so the whole style definition was being dropped..." Technical 13 (talk) 12:03, 9 April 2013 (UTC)
Yes, my browser was updated. I guess that explains it. Oh well, I had red in my signature the whole time without knowing it! Thanks everyone for your help. AutomaticStrikeout (T • C • Sign AAPT) 14:49, 9 April 2013 (UTC)

man

can you make a pattern kits with commons?

[11]

[12] --2.6.124.215 (d) 8 avril 2013 à 18:43 (CEST) — Preceding unsigned comment added by 2.6.124.215 (talk) 12:16, 9 April 2013 (UTC)

when is the soocer/football pub?--2.6.124.215 (talk) 12:29, 9 April 2013 (UTC)

You can use {{football kit}} to make those. If you need help, try asking on the talk page of whichever team that is. Cheers, — Bility (talk) 16:05, 9 April 2013 (UTC)

Wikidata phase 2 is coming soon

Hi :)

2 days ago the first 11 Wikipedias got the ability to include data from Wikidata in their articles. These are the Italian, Hebrew, Hungarian, Russian, Turkish, Ukrainian, Uzbek, Croatian, Bosnian, Serbian and Serbo-Croatian Wikipedias. This means they are now able to make use of the structured data that is available on Wikidata. It includes things like conservation status for a species, ISBN for a book or the top level domain of a country.

The next step is to roll out phase 2 of Wikidata on English Wikipedia. We are currently planning this for Wednesday next week (April 3). We are currently carefully monitoring performance on the first 11 Wikipedias to make sure everything is ok. If problems arise we might have to postpone the deployment a bit longer but I wanted to give you a heads-up that this is the current plan.

We have prepared an FAQ for this deployment and are looking forward to your testing and feedback. You can already test it on test2.

And some more details about how this will work: There are two ways to access the data:

Use a parser function like {{#property:p169}} in the wiki text of the article on Yahoo!. This will return “Marissa Mayer” as she is the chief executive officer of the company.
For more complicated things you can use Lua. The documentation for this is here.

We are working on expanding the parser function so you can for example use {{#property:chief executive officer}} instead of {{#property:p169}}. The complete plan for this is here.

Thank you so much for helping us bring Wikidata to all Wikipedias!

Cheers Lydia Pintscher (WMDE) (talk) 14:45, 29 March 2013 (UTC)

Hey Lydia. I was going to email you and Denny about this, but I figure here is as good a place as any to ask, since some community members might be wondering the same thing. To be clear: this is a pretty broad question, not one that I think should hold up enabling Phase II.

When it comes to implementing use of Wikidata properties in infoboxes and the current state of the syntax, what is the actual, step-by-step workflow for updating content? I see description of the syntax linked from the FAQ, but not a clear description of the steps involved in editing something should we implement properties in infoboxes or directly in articles willy-nilly. I am of course looking at this through the eyes of my own editing as an admin on enwiki, and with my editor engagement hat on. This is my understanding: Previously, if an editor wanted to update an infobox, all they'd need to do is click Edit and change the value. For new editors, the hurdle was scary infobox wikitext, but in theory you could figure out the relationship between the markup, values, and what the article looked like. Now, when you click edit, you can't actually change the content in a wiki page by clicking edit here, you have to know to go to Wikidata and update the property.

That workflow is extremely problematic from a user experience design perspective, because it adds a number of potentially annoying and complicated steps to do what once a two-step "edit-save". There seems to be zero indicator of how to do so that is obvious too, which is bound to confuse even experienced editors until everybody across the wikis learns the new system.

How are we going to solve this problem? I don't think asking all editors to head over to a completely different wiki to update infobox values is okay at all, because it forces end users to conform to the needs of the Wikidata software architecture, rather than using software to help us keep the content up to date. The main solution I am thinking is that we allow local updating of Wikidata values which the software then pulls back in to the main revision history on wikidata.org. This would be best accomplished with VisualEditor of course, so it's probably not going to happen immediately. However, I'd really like us to consider how we can reduce user confusion and introducing multiple new steps in to the workflow of updating infobox content. Steven Walling (WMF) • talk 18:29, 29 March 2013 (UTC)

My take: Ideally we should work towards an inline invocation of Wikidata editing for properties. Til then, "edit" links in infoboxes that point straight to wikidata.org may be an acceptable solution. Property invocations in the article namespace should probably be avoided completely and perhaps made impossible. On the plus side, having parameter-less templates will make wikitext less cluttered.

Cf. an example of a parameter-less template invocation.--Eloquence * 18:50, 29 March 2013 (UTC)

Inline edit links to Wikidata (ala the interwiki links implementation) is definitely a good interim solution, since it is an obvious pointer to where the content is coming from. That is a good point about templates without huge parameters, and it's actually quite likely to reduce headaches for first time editors around seeing massive amounts of infobox markup at the head of articles. Steven Walling (WMF) • talk 18:54, 29 March 2013 (UTC)

Eloquence, in your example, for property with multiple value ("It's named after ... Linus Torvalds, Unix, Linux kernel"), they are not linkable (Linus Torvalds, Unix, Linux kernel). Has anyone raised this issue before? Bennylin (talk) 14:45, 1 April 2013 (UTC)

This will be done. I've filed it as bugzilla:46788 so we don't forget. --Lydia Pintscher (WMDE) (talk) 14:29, 2 April 2013 (UTC)

Thanks :) . Bennylin (talk) 17:13, 5 April 2013 (UTC)

(ec) One quick-and-dirty solution would be a discreet [edit] link in the top of the infoboxes similar to the (V D E) links which currently exist for a lot of navigational templates; rather that take you to the template page, however, it would send you to Wikidata.

That aside, remember that this depends on how the infobox is set up; there won't be any changes to enwiki infoboxes until someone actually codes the explicit use of wikidata into them. (It's not like the interwiki links, which effectively superseded existing content). As such, we can't really write guidance until we have some functioning examples to explain how it'll work - wikidata can't dictate that behaviour centrally. Andrew Gray (talk) 19:00, 29 March 2013 (UTC)

One thing for sure: integrating Wikidata to infoboxes will change the dynamic of revision diff. For example let say Yahoo's infobox is automated as of today, and a couple years from now (2020) it's CEO has changed, but looking back at today's revision, you won't find Marissa Mayer as it's past (2013) CEO, instead you'll find 2020's CEO in 2013 text.. Automated infoboxes, while it surely has its advantages, loses it's ability to be checked at exactly what diff the change occur (Wikiblame came to mind). And, as a consequence, automated infoboxes would rely on other website (Wikidata) to be vandal-free, or at least have the critical mass to revert problematic edits which could affect hundreds of projects.. Bennylin (talk) 20:19, 29 March 2013 (UTC)

We already have this problem to some degree with templates, of course! Andrew Gray (talk) 21:35, 29 March 2013 (UTC)

With template,s the diff could still be checked on the corresponding transcluded template's history, which is located on the same site (and listed on the bottom of the edit page). Having said that, it would be great to have list of property used in an article on the bottom of edit page. Bennylin (talk) 03:33, 30 March 2013 (UTC)

Isn't there the same issue with Commons as well? Alterations to the file or even deletions could easily happen without a corresponding change to the revision history locally. They just don't happen often because files are less prone to mutability than text. Steven Walling (WMF) • talk 22:50, 31 March 2013 (UTC)

Yes, and that is recognized as enough of a problem that the {{keep local}} template exists. Regards, Orange Suede Sofa (talk) 22:57, 31 March 2013 (UTC)

No. No matter you view the page locally or in Commons, they all have their file history on the image information screen. It's working great already. See if Wikidata can do the same thing, for example, for every "data"/"property" provided by Wikidata using parserfunction, one can click a different colored link (like, black?) which ideally go to a page complete with the revision history on the page itself, akin to image namespace. Bennylin (talk) 14:42, 1 April 2013 (UTC)

Update: We encountered some small issues that we'll have to sort out. Deployment will probably not happen on Wednesday as initially planned but hopefully soon after. I'll keep you posted. --Lydia Pintscher (WMDE) (talk) 20:16, 29 March 2013 (UTC)

I would hope that the wiki page history would reflect changes to both the wiki page and any data which it fetches. Perhaps at the same time, the similar issue with template changes not being reflected in wiki page histories could be handled. There is also the related issue of source citation, the elements of which probably need to be properties of the data (covering multiple data items), which could then be pulled into the wiki page cites. —[AlanM1(talk)]— 23:13, 29 March 2013 (UTC)

I have a request for anyone involved in updating infobox templates: do not make the templates completely parameterless, but instead allow optional parameters to override data from Wikidata. It's going to take time for the data we have in infoboxes to be added to Wikidata, so we still need to use the data in wiki pages. As with the interwiki links, there will probably be some situations where we need to override data from Wikidata. Also, since Wikidata is not so easily edited as Wikipedia (both in terms of usability and technical requirements), editing the page on Wikipedia is the only way some editors can change infobox content. (Ideally, a bot could pick up such edits and transfer them to Wikidata. The infobox should still display local overrides while waiting for the bot to transfer them to Wikidata.) – PartTimeGnome (talk | contribs) 22:32, 30 March 2013 (UTC)

As just an editor: having stuff like {{#property:p169}} is as un-wiki as it can be - (wiki as in "everyone can edit", without much trouble) It should definitely be only a short temporary use, probably betrer not to be used at all (if if it short and temporary let'«s skip such a incredibly nasty looking thing). Something like {{#property:chief executive officer}} is still hardly easy to understand to the non-programmer mind (I presume, I have a programmer mind, I think it is easy :-), but the gains may outweight that. But please don't go to "The president of {{#property:p123}} is {{#property:p456}}, since {{#property:p789}}", not even "The president of {{#property:company-short-name}} is {{#property:president}}, since {{#property:term-begin}}". - Nabla (talk) 23:18, 31 March 2013 (UTC)

Indeed. In a lot of cases it is probably not a good idea to use it in the text of the article as opposed to the infobox. --Lydia Pintscher (WMDE) (talk) 12:00, 5 April 2013 (UTC)

Update: Heya folks :) It seems things are fine again now. We have scheduled a new deployment for Monday (April 8) evening to night UTC. I'll let you know if anything changes about this plan. --Lydia Pintscher (WMDE) (talk) 11:58, 5 April 2013 (UTC)

Actually, I am quite strongly opposed to this implementation until such time as the English Wikipedia community decides that it wants this, and frankly I'm doubtful that we do. I cannot understand why we will be sourcing our articles to another site not under our control. I cannot understand why updating an article will require users to go to another website. And I cannot understand why a change made by a user who has never edited English Wikipedia will be automatically viewable on English Wikipedia. This project is already heavily conflicted over the use of infoboxes in the first place. This does not help. How about deploying on projects that actually want this software before it's activated on one that hasn't even had a discussion about it? Risker (talk) 01:30, 6 April 2013 (UTC)
- The feature will be available for us to use. Whether we use it or not is up for us to decide. --Rs chen 7754 01:34, 6 April 2013 (UTC)

Anyone who's spent any amount of time on English Wikipedia knows that statement is untrue. As soon as it is enabled, it will be used, even over the objections of other editors, because there's no rule against it. We already have edit wars about infoboxes: both their existence and the content within it (and in some memorable cases, which infobox(es) should be present on an article and in what order). We know from experience that a handful of editors can, in only a few days, take a new extension and completely negate existing policies and practices (q.v. revision-deletion, which is constantly being used to delete content against policy), and in this case we don't even have existing policy. Today, if a new editor sees an infobox that says "Spouse:Marry Smith" instead of "Spouse:Mary Smith", they can click on the edit button and find the misspelling and probably fix it - if the wikitext doesn't scare them off. With Wikidata, they won't be able to find it, they'll just see "Spouse: #property:P169" instead. What's the point of making editing easier with Visual Editor (a very worthy endeavour that recognizes the challenges involved in editing), when we then make infoboxes impenetrable and unfixable unless someone goes to another site with an even more incomprehensible editing interface? How does this attract new users, or keep the old ones?

I'll repeat what I've said in another forum: I'm supportive of offering the extension to any project that wants it. I can see it being particularly useful for smaller wikis with a very tiny community, to help them build up content. But deploying this change means that the WMF is actively affecting content on every project where it is linked, and I don't believe that the good people behind Wikidata and the WMF haven't realised that: it's the entire purpose of the extension. The moment that this extension is turned on, it will be an irreversible fait accompli. Major change should never be initiated without advance preparation: community discussion, policies and processes developed, education available, resources established; that's Change Management 101, and it's being done well with Visual Editor and is practically non-existent on this project with Wikidata. I suppose this is what is to be expected from an organization that has decided to put "tech" before "information" or "community". Risker (talk) 05:24, 6 April 2013 (UTC)

Risker once again is speaking eminently sensibly—although in a community not dominated by the mindset of "invent a technological feature, implement it widely, then decide if it has any benefit", her perspective is also entirely obvious. Why she is, so far, the only person to come close to broaching the change management and big-picture issues here is a mystery to me (or is it?).

I also oppose the implementation of Wikidata's functionality [1] on the English Wikipedia until such time as the community holds a discussion to determine what benefit, if any, Wikidata functionality provides, and what good use cases there are, if any. (And can that discussion even be held with available information? See second paragraph.) (I can tell you now that there are many bad use cases. I don't see anyone making that distinction, except Risker.) Wikidata is an interesting concept, but from the perspective of any single, highly active Wikipedia project, the cost–benefit analysis of retrieving data from it upon first impression is so obviously weighted to disadvantages that the default position should be "do not use until we figure out where it provides benefit". I am retired from this project, but this topic interests me very much—it has the potential to further reduce the editability of Wikipedia—and I may continue to comment. I hope that someone more active in the community will begin discussions about goals, use cases, and caveats, before adding a single Wikidata invocation to this project.

I'd also like to note that a) Wikidata still does not support basic property values like "date" and "number". It is, frankly, ridiculous to rush to implement "Phase 2" on the client side (the client being a particular Wikipedia project) when the basic infrastructure to store data is not fully in place. Everyone should slow way down here. (How much of a rush can you be in to replace "birthplace: London" on a biography with "birthplace: [magic-word invocation to retrieve Wikidata's birthplace property for that biography, equal to London]?) b) Wikidata is still being populated (duh--and will be, for a long time), and provides no contract, not even a loose one, that any data set for any particular range of topics, or particular range of properties, exists on Wikidata. Ironically, since the majority of data on Wikidata have been copied from Wikipedias, you could draw some pretty Dilbert-esque modelling diagrams of the information flow at work here, that, to any independent observer, would seem circular and nonsensical. [Note 1: turning on the Wikibase extension in software is not the same as "using it", of course, but it would seem more sensible to have the community ask to turn it on, to prevent inconsistent use, until the community makes its decisions. Otherwise, this is another round of the theme we increasingly see--provide an option by flicking a software switch; have everyone use it differently or not at all; force ad hoc responses from the community; the community then decide months later what to keep and what not to keep; then clean up.] Riggr Mortis (talk) 23:27, 6 April 2013 (UTC)

Risker's view is wise. I too find Risker's view to be sensible and having much merit. The community should decide if this is wanted or not. It should not be deployed by fiat, or through mostly off-English Wikipedia decision. I haven't actually seen a single example yet of a page with an infobox filled out by Wikidata. Why aren't examples being linked? I couldn't even find any linked from Wikidata. I'm sure there there somewhere but why are they hard to find in the first place? (If somebody could find a dozen or so examples from the 11 languages using it, that'd be great.) I also find Steven's questions about usability very important. How is this anything but 100x harder for new editors to figure out? I'm an experienced editor but since I haven't seen any examples yet I'm still confused about it. Using Wikidata for the language links made HUGE sense. It's one of the greatest things to happen around here in a long time. Intertwining Wikidata with the articles however raises many questions. They should be discussed thoroughly. What I really hate to read are myopic comments "The feature will be available for us to use. Whether we use it or not is up for us to decide.", as made above. Risker himself pointed out the problems with simple take. We have an amazingly valuable resource here. We need to treat it with care and caution, and proceed slowly and only when the risks/rewards are well understood. Jason Quinn (talk) 03:45, 7 April 2013 (UTC)

But being realistic here, the infoboxes would still have to be recoded. We don't have a functional demo of an infobox ready yet that I know of, anywhere on WMF sites. So that takes time. Secondly, a lot of our infoboxes are full-protected, or at least they're supposed to be. Full-protected pages generally require consensus before being edited - admins who don't get consensus before throwing in the Wikidata fields are taking a bit of a risk. I really don't see the need to panic. I don't think a sitewide RFC is necessarily the answer; I think this sort of decision should be made by subject area. However, if folks want a RFC about this whole thing, rather than complaining about WMDE doing this, it's probably better to spend that time drafting a RFC. --Rs chen 7754 04:20, 7 April 2013 (UTC)

Yes by all means if anyone wants to start an RFC about how to make use of it please do. I am even happy to link to it in the announcement with a text like "The English Wikipedia community asks you to not make use of this in the main articles until RFC foo is closed and it is decided how to make use of this initial deployment of Wikidata phase 2." or something similar. --Lydia Pintscher (WMDE) (talk) 07:24, 7 April 2013 (UTC)

Maybe as some starting point for someone who wants to tackle this: The Hebrew Wikipedia already has phase 2 and they decided on some groundrules. They're not using the parser function in the article itself but only in templates. Template changes to move to Wikidata need to be discussed on a specific page. Controversial data isn't taken from Wikidata for now. This seems to be working quite well for all I can tell. --Lydia Pintscher (WMDE) (talk) 07:49, 7 April 2013 (UTC)

Hebrew Wikipedia, with its small community, had that discussion well before Phase 2 was deployed, and it was a conscious decision on the part of their community to do so. I have no objection to deploying it for projects that make such a decision. However, this is being done without any discussion with the English Wikipedia community (believe me, this page is not representative of the community at all), without any action plans, and based on what has been pointed out below on the workflow, at a stage of development where it is not useful and potentially very harmful to this project at a time when we are trying to retain editors and attract new ones. There are projects that are happy to help you to develop this software, and I encourage you to work with them until you have something that is ready to consider on this project. Risker (talk) 03:14, 8 April 2013 (UTC)

It could be good for the {{Authority control}} data. -- WOSlinker (talk) 14:14, 6 April 2013 (UTC)

I was thinking about this today, as it happens; it's a pretty good case of a template where we can "silently" call in a variety of cross-language data without needing to embed them in the wikitext, where they're not likely to be changed on a regular basis (unlike infoboxes, which are reasonably frequently edited), and where there's a clear benefit to a central db. In many ways, the focus on infoboxes is somewhat unhelpful for considering the use of Wikidata on Wikipedia; it's not where the major benefits to us are likely to come from. (Consider, for example, running a sortable "list of cities in X" and being able to dynamically populate the properties of each line). Andrew Gray (talk) 20:57, 8 April 2013 (UTC)

Could this be the reason why there has been a flood of new pages in the last few days, each listing a completely unnotable village in Turkey/Serbia/Poland etc.? Is Wikipedia to become a gazetteer? eg Nowe Staroźreby, eg Sığırlıhacı, Çubuk, eg Polu John of Cromer in Philippines (talk) mytime= Sun 08:09, wikitime= 00:09, 7 April 2013 (UTC)

I don't think so. --Lydia Pintscher (WMDE) (talk) 07:24, 7 April 2013 (UTC)

None of those three pages was created "in the last few days". Nowe Staroźreby was created 23 October 2008‎; Sığırlıhacı, Çubuk was created 13 July 2012‎; Polu was created 1 October 2012‎. I suggest that you contact each of the three different individuals as to their motives (and of the three, Dr. Blofeld (talk · contribs) is one of the most prolific creators of new articles that we have); but you can be sure that the presence (or not) of an entry on Wikidata was not the reason for creation. --Redrose64 (talk) 14:24, 7 April 2013 (UTC)

It was a manifestation of bugzilla:46978 I think, when whole swathes of similar articles suddenly popped up on the error list, even though unchanged. Sorry

John of Cromer in Philippines (talk) mytime= Tue 14:48, wikitime= 06:48, 9 April 2013 (UTC)

Risker's concerns are spurious, technically illiterate and nothing but spreading a cloud of FUD. At this thread, you can see multiple technical people trying to painstakingly explain this to her - that the deployment does not alter anything about the syntax, that no editor need use it unless they want to, and that nothing need change unless the editor base vote with their feet. This is all electioneering to build a base for December's arbcom elections. It is spreading FUD in bad faith, and needs to be called out for what it is - David Gerard (talk) 20:45, 8 April 2013 (UTC)

That's complete nonsense, but ABF asides are not worth pursuing here so I'll just reiterate that there are two important points from above (from Steven Walling and from Risker):

"what is the actual, step-by-step workflow for updating content?" (and see section just below).
"As soon as it is enabled, it will be used, even over the objections of other editors, because there's no rule against it."

Risker's concern expresses the situation precisely: as soon as some technical gizmo is available, it will be forced on everyone by the small but enthusiastic crowd of editors who aim to make Wikipedia readable by machines. Johnuniq (talk) 23:55, 8 April 2013 (UTC)

I obviously don't fully agree with Risker, but to accuse her of bad faith in this is quite inappropriate. --Rs chen 7754 07:14, 9 April 2013 (UTC)

When it comes to deciding what to implement, I don think it makes sense to go separately by subject area. Rather we should go step by step with types or items of data, starting with the most obvious. I don't know what he wikidata team is ready for first, but one obvious place is dates in infoboxes, and then geodata in infoboxes. I would be reluctant to use personal name early on, because the conventions here are different for each wiki. What I really hope for, but I understand is going to be technically difficult, is references to sources.

I don't think it true or relevant that the enWP has little to gain. We have the greatest diversity of editors, who will be greatly helped by standardized data if it is done in the proper editable fashion. We have an immense number of articles on other language areas--I would suppose surely more in the aggregate than any other single WP. And, since our articles are often copied into other WPs, particularly the smaller ones, we have a responsibility for the data they use--we have to gain by being a up-to-date and trustworthy source for this. DGG ( talk ) 17:16, 10 April 2013 (UTC)

I love what the idea of Wikidata and the results of its work so far. We should all be happy about the reductions in duplicated labour it has brought do far and promisses to bring in the future. For this reason I want further implementation of WD to be as seemless, positive and as user friendly as possible. In this regard I think Risker has a point in that including people in a consultation and changing WD for the better in any way which is practical is a good idea, one which I'm surprised the WD team didn't think of themselves. This should be done. I really don't see a down side to it, unless there's some sort of factor of urgency present which I'm unaware of. Having said that, I feel I have to object to the idea that it would be perfectly acceptable for the english wikipedia community to say: "This doesn't benefit us, so the rest of you can go to hell." A big part of why I contribute to the English version of Wikipedia raughter than to the Slovene one is because it's always been clear to me that work done on EnWP would, over time, difuse into the other versions. This is a big step in the right direction and it shouldn't get messed up for the sake of speed! --U5K0'sTalk_{Make WikiLove not WikiWar} 19:28, 10 April 2013 (UTC)

Workflow

I started a page at Wikipedia:Wikidata/Workflow. I personally consider some of the issues mentioned there to be blockers to deployment here. --MZMcBride (talk) 16:18, 6 April 2013 (UTC)

Thanks MZ. I think focusing on the practical problems with the direct experience (for editors) is the right way to look at this issue, instead of dickering about the way deployment might or might not have been communicated about to everyone's personal satisfaction. With a list like this, which could easily translate in to bugs to be addressed, the community can make a decision not based on generalities. Steven Walling (WMF) • talk 22:08, 8 April 2013 (UTC)

Navboxes and edit tools broken

Just how soon will all the navboxes on Wikipedia be fixed? While we're at it, when will be get the tools at the top of the pages back? -------User:DanTD (talk) 00:22, 31 March 2013 (UTC)

They're all working fine here... - The Bushranger _{One ping only} 00:31, 31 March 2013 (UTC)

Not really. The navboxes haven't been collapsible for the past several days, and there's no edit tools above my page. I used to be able to click on an icon for my signature, or bold or italic type, or images, etcetera, and that tool bar is gone. -------User:DanTD (talk) 15:18, 31 March 2013 (UTC)

I'll bet you inadvertently disabled javascript on your browser.קיפודנחש (aka kipod) (talk) 15:48, 31 March 2013 (UTC)

What kipod said. To check, click on this link; if it says "NO", then you've disabled JavaScript accidentally. jcgoble3 (talk) 18:26, 31 March 2013 (UTC)

Nope. It said "Yes." There were other complaints about this earlier this week, but I can't find them. -------User:DanTD (talk) 22:39, 31 March 2013 (UTC)

Sounds like you have something stuck in your browser cache. —TheDJ (talk • contribs) 08:43, 2 April 2013 (UTC)

Well, whatever it is, clearing it didn't work. -------User:DanTD (talk) 01:30, 5 April 2013 (UTC)

UPDATE - All it really did was get me signed off of other webpages that I'm on, including the commons. -------User:DanTD (talk) 13:54, 5 April 2013 (UTC)

Your description certainly sounds like JavaScript isn't working, but maybe it's only here. Does it work when you are logged out? Does http://commons.wikimedia.org work? How about https://en.wikipedia.org? Which browser do you have? Can you try another? PrimeHunter (talk) 14:02, 5 April 2013 (UTC)

(edit conflict)

At this point, I might be willing to bet that you have some kind of custom JavaScript that is causing an error OR you changed one of your Special:Preferences:

On your "Editing" tab
- Show edit toolbar (requires JavaScript)
- Enable enhanced editing toolbar
- Enable dialogs for inserting links, tables and more
On your "Gadgets" tab (In the "Editing" section)
- CharInsert, adds a toolbar under the edit window for quickly inserting wiki markup and special characters

⇑ Those are the next things I would check, as they may have accidentally got unchecked somewhere on the way. If those all look normal, I'll try to think of some more things! Technical 13 (talk) 14:35, 5 April 2013 (UTC)

My commons page works fine, and all things that should be checked, are checked. If any changes were made in my preferences, they were made by Wikipedia itself, not by me. The thing is, I don't have an edit toolbar, despite the fact that it's checked in my preferences. And what's worse, is that it's really not just the navboxes and edit tools. It's everything that's normally collapsible. BTW, I just checked my JavaScript, and it claimed it was okay. -------User:DanTD (talk) 03:24, 6 April 2013 (UTC)

If you want me to keep guessing then please answer all my questions. When I ask whether something works I mean things requiring JavaScript, for example collapsible table of contents. PrimeHunter (talk) 13:31, 6 April 2013 (UTC)

Some of your questions I just don't know the answer to. I have Internet Explorer, but I'm not sure if it's 8, 9, or 10, and the PC won't give me any way of finding out(unfortunately, this is one of the hassles with Windows 8). I don't log out on this PC, so I can't tell you if it works when I'm logged out or not. -------User:DanTD (talk) 07:02, 7 April 2013 (UTC)

If you're using Windows 8, then unless you've purposely downgraded, you're probably using IE 10, which is what comes with Windows 8. To check for sure: using the desktop version of IE, press the Alt key to bring up the menu bar, choose Help → About Internet Explorer, and it should tell you what version you're using. If you're using the Metro app, on the other hand, then that is IE 10. jcgoble3 (talk) 07:12, 7 April 2013 (UTC)

I had to struggle to get the info, but it's Windows 10. -------User:DanTD (talk) 13:25, 7 April 2013 (UTC)

I think you mean IE 10. Windows 10 doesn't exist (at least not yet, outside Redmond). LeadSongDog come howl! 20:33, 10 April 2013 (UTC)

It's hard to help without more answers. The first thing to check should often be whether a problem persists when you log out and get the default interface. If wikipedia.org has been marked an unsafe website in your browser then the browser may disable JavaScript here. If you have collapsible table of contents at other Wikimedia sites but not here when you are logged out then this is a possibility. PrimeHunter (talk) 10:47, 7 April 2013 (UTC)

I logged out briefly last night, and tried cleaning my cache again, then logged back in. Wikipedia.org has not been marked as an unsafe, and as far as I can tell my JavaScript is okay, but everything else still doesn't work. -------User:DanTD (talk) 13:25, 7 April 2013 (UTC)

Table right floating next to TOC and after image problem

Can someone please try to get the table of issue organizations ratings I just added to the end of the intro on Political positions of Joe Biden to come after the photo on the far right instead of in between the Table of Contents and the photo, squishing the TOC? I could not see how to do this in the Table Help page. Thanks in advance. EllenCT (talk) 01:59, 6 April 2013 (UTC)

Thanks very much to User:Wbm1058 for trying, but there was no obvious solution that worked very well, so I punted by moving the photo down to replace one from the 1980s. Everyone looking at that article probably knows what he looks like anyway, and he has an icon-sized photo in the navbox. EllenCT (talk) 03:18, 6 April 2013 (UTC)

I posted another solution for you. See Help:Table#Floating images in the center, which also discusses far right behavior. Using class="infobox wikitable" or class="wikitable infobox" seems to work, although that isn't documented at Help:Table. – Wbm1058 (talk) 15:51, 6 April 2013 (UTC)

That looks much, much nicer, too. EllenCT (talk) 19:39, 6 April 2013 (UTC)

Please don't use 'infobox' unless something actually IS an infobox. Classes like these have scope and meaning, they are not tools to apply functionality. In this case, you can use the class "floatright" which has similar behavior to infobox and a right side thumb, but without the special meaning and the special styling of those elements. —TheDJ (talk • contribs) 10:14, 8 April 2013 (UTC)

What is a class? Can you disambiguate it? Given that I don't understand what a "class" is, I have no idea what you mean by the "scope" and "meaning" of classes. I have a general idea of what a (wiki)table is supposed to look like. Lacking decent documentation, what else can editors do but trial & error to see what works? Wbm1058 (talk) 04:42, 9 April 2013 (UTC)

A CSS class is a reusable group of CSS styling instructions. We have a lot of common classes used for various purposes. To use them in wikicode, you use the HTML class attribute, like this: class="nameofclass". This applies the class "nameofclass" to the scope of that block of code. You can even have multiple classes apply to the same block class="nameofclass anotherclassname".

This is the simple part. The more complex part is that class="nameofclass anotherclassname" not ONLY applies the styling. In fact, it is more a 'naming scheme' to identify blocks in the page, that has automatic inheritance of styling information with identical names. So using class="infobox" does two things. It designates the block as being an infobox AND implicitly makes it receive the styling named 'infobox'. To designate something to be an infobox, just to get similar styling is therefore incorrect. Most of our classes have designated meanings like this, there are only a few exceptions; floatleft, floatright are some of those 'purely cosmetic' classes that would be an exception.

When there is no designated or cosmetic class that suits your need, you usually should create a template that does what you want, using style="" statements that contain individual CSS styling instructions to achieve the behavior you want. If the use case is common, usually the people who work on these kinds of things on the English Wikipedia will eventually turn a template with style statements into a new class. If you have trouble understanding such styling instructions, then just ask for help here, that is better than reusing a class for a different purpose than what is was intended for. —TheDJ (talk • contribs) 09:52, 9 April 2013 (UTC)

By CSS class, do you mean pseudo-class? Cascading Style Sheets doesn't define class, just pseudo-class. See this. Some have been uncomfortable with the title of Wikipedia:Catalogue of CSS classes (see Wikipedia talk:Catalogue of CSS classes#Suggested move). Given that Wikipedia:Catalogue of CSS classes#Classes has columns for both CSS and HTML, perhaps rename to Wikipedia:Catalogue of classes, which catalogues both CSS pseudo-classes and HTML class attributes? I've been working on updating related documentation, which still needs work.

Is Help:Table#Classes talking about HTML class attributes, CSS pseudo-classes, or both? I'm unclear on whether this edit was an improvement. Wbm1058 (talk) 23:04, 9 April 2013 (UTC)

Pseudo-classes are really only of interest when you want certain portions of a page (such as links) to change styling (including, but not confined to, colour) when you do interactive things like click on a link - they're only encountered when you're building a style sheet (there are some in MediaWiki:Common.css in the section marked "Highlight clicked reference in blue to help navigation" - it's the three instances of :target). In fact, don't worry about what pseudo-classes are, since they don't come up in wikicode - in this, there's a six-line table; ignore the last three lines and concentrate on the first three. Therefore, in the context of wikicode like {| class="wikitable infobox" the class attribute is class="wikitable infobox" and the classes (or class names) are wikitable and infobox. --Redrose64 (talk) 07:13, 10 April 2013 (UTC)

"Added photo to page" edit summary

Lately on RC, I've noticed a lot of new editors with the edit summary of "Added photo to page" where they add in an off-topic photo or a broken photo link. For example, [13], [14], [15], or [16] (topic NSFW, image off-topic). Obviously, this is a new "feature" from somewhere, but I'm not sure where. Where is it coming from and can we turn it off or at give some editor assistance on what is appropriate? -- Gogo Dodo (talk) 00:13, 9 April 2013 (UTC)

Wasn't mobile uploading just enabled? I dealt with something similar to this here[17] and on Wiktionary[18] (I deleted the image from Commons though, as it was just an illustration of Cyborg (comics)). EVula // talk // ☯ // 00:30, 9 April 2013 (UTC)

@EVula - yep, it's mobile uploading. —Theopolisme (talk) 01:17, 9 April 2013 (UTC)

Is there a good place to report problems? In a "providing feedback about a new feature" sort of way; all of the cited examples of mobile uploads have been crappy, but that doesn't mean all of them are bad. EVula // talk // ☯ // 05:37, 9 April 2013 (UTC)

In addition to submitting bug reports, you're welcome to come hang out/lurk/yell at us in our IRC channel, #wikimedia-mobile ^connect – that's the fastest way to get in touch. I read the Village Pump with an eye to mobile stuff, so dropping a note here works, too; I'm also happy to receive notes on my talk page :) Lastly, the mobile team has a mailing list, mobile-l

lists.wikimedia.org, but it's pretty quiet. Though maybe this is a good time to plug it, in order to solicit, erm, spirited conversation... Maryana (WMF) (talk) 20:54, 9 April 2013 (UTC)

So far, I haven't seen very many good ones. A lot of them have either been possible copyright violations with the wrong licensing [19] or people snapping photos with their cellphones: William of Normandy's TV or random things around the house. -- Gogo Dodo (talk) 07:30, 9 April 2013 (UTC)

There is a good place to report problems: In general (not here as Maryana has commented on Commons that this is already being worked on), problems that sound like a potential issue in the code of the MediaWiki software or the server configuration should be send to the 'Bugzilla' bug tracker by following the instructions on How to report a bug. This is to make developers of the software aware of the issue. --AKlapper (WMF) (talk) 08:39, 9 April 2013 (UTC)

See also lot's of discussion about this on Commons. —TheDJ (talk • contribs) 07:51, 9 April 2013 (UTC)

Thanks for the discussion link. I'm glad to see that it is going to be "fixed" RSN. I'm really disappointed in how poorly this seems to have been planned and monitored. It seems to me that somebody had one of those "this is a great idea to get new editors!" moments without considering that a lot of new editors are not as smart as you think they are. -- Gogo Dodo (talk) 08:13, 9 April 2013 (UTC)

AGF run amok. :) EVula // talk // ☯ // 17:07, 9 April 2013 (UTC)

Trying to modernize our archaic multi-domain, images-on-a-separate-wiki uploading infrastructure; crazy, I know! :) Thanks so much for helping out with tagging/triaging the uploads over on Commons. I'm watching the uploader numbers now as we deploy. They should drop back down in an hour or so, once we're done. Maryana (WMF) (talk) 20:42, 9 April 2013 (UTC)

Gogo Dodo, this feature wasn't intended solely for new editors, and, in fact, some experienced Wikipedians are taking advantage of it quite happily. Put simply, it's just a nicer, more modern uploading workflow. I'd urge you to try it out if you have a camera phone and let me know what you think.

As for your comment about this being not properly planned or monitored – true, a lot more new users flooded in to try out the feature than the mobile team anticipated. I thought most people wouldn't care/understand why they should add an image, which would keep the numbers of mobile uploads low enough for the community to handle. After just a few days of full release, however, we saw that this was not the case; the usage was high and the quality was low, so the mobile team is responding to this today (right now, in fact) by making sure the feature only appears to existing Wikimedia users. This will significantly reduce the number of newbie mobile uploaders/crappy uploads while we work on improving the educational UI to ensure less noise. You could argue that a week is too slow to deploy a feature, gather data and feedback, and throttle usage/iterate on UI based on that data/feedback – but I'd say those are some very high standards for software experimentation you've got there :) Maryana (WMF) (talk) 20:31, 9 April 2013 (UTC)

How about restricting mobile image uploads to autoconfirmed users? --Redrose64 (talk) 20:51, 9 April 2013 (UTC)

Good call... on Commons, that appears to be registered for four days, which isn't a terribly high bar but should be enough to cool the heels of obvious vandals.

Let's see how the situation looks after today's deployment. If the number of low quality uploads is still too high, that could potentially be an alternative method of bringing it down. Maryana (WMF) (talk) 21:03, 9 April 2013 (UTC)

It may not have been intended solely for new users, but when the team decided to display the upload button to logged out users, you were bound to get a majority of new users clicking the button. From my own experience, if I'm not logged in (different computer, login timed out, etc.), the very first thing I do as an editor with an account is to login. I'm not going to poke around and try to edit or click a button. I suspect a lot of existing editors do the same thing.

But going back to the topic at hand, it seems a bit better, but I still see an occasional vanity photo. One other thing that I think needs to be addressed as could be considered a bug instead of a misused feature is the number of "undefined" additions such as [20] [21] [22]. The form shouldn't allow an undefined file to be saved. -- Gogo Dodo (talk) 05:28, 10 April 2013 (UTC)

Hmm, that is very strange. I hadn't noticed this undefined bug; thanks for catching it. Let me know if you notice any more weirdness like that!

Tracked in Phabricator
Task T49103

FYI, we're working on a pre-upload EXIF data verification system that will allow us to filter out the majority of the obvious copyvio, which should cut down on the number of newbies messing around and uploading random things they found elsewhere on the Internet. Maryana (WMF) (talk) 22:10, 10 April 2013 (UTC)

SVG image thumbnail caching broken?

Tracked in Phabricator
Task T49087

one pixel wider makes the whole difference in the world.

If I go to Bird anatomy#Respiratory system, I see on the right hand side a grey schematic of the bird lungs, File:BirdRespiration.svg (shown to the right). If I click on that thumbnail, I get a different, updated schematic. The update happened already in June 2012, but the cached thumbnail still isn't updated. Do other people see the same problem; is it a known bug; is there a workaround? Thanks, AxelBoldt (talk) 16:52, 9 April 2013 (UTC)

I'm seeing the same problem. I have never looked at that file before, so I know it isn't a local caching issue. Technical 13 (talk) 17:15, 9 April 2013 (UTC)

it's an image caching issue on the imageservers. if it does not go away, there is a workaround: show the image in a slightly different size, which will entice the imageservers to create a new copy (even though the image is .svg. the imageservers create an actual .png copy in the correct size which is what's actually gets sent to the readers. i believe the reason is poor support for svg on some browsers, most notably older versions of ie).

to demostrate, i added a second image of same file, one pixel wider (221 instead of 220). hopefully, when i hit "save", you'll see the new image under the old one. peace - קיפודנחש (aka kipod) (talk) 19:39, 9 April 2013 (UTC)

Kipod, I'm using the most recent version of Firefox, yet I still see the difference. Perhaps there needs to be a tag or category created on commons to tell the image server to update the .png it creates when a new file is uploaded? Who would be the contact person for that, if you know? Technical 13 (talk) 19:43, 9 April 2013 (UTC)

You seem to have misunderstood the explanation. Since some browsers cannot handle SVG files directly, everybody is served PNG thumbnails instead of the original SVG files, so the version of your browser has no bearing on the issue.—Emil J. 20:23, 9 April 2013 (UTC)

This is a bug. The image thumbnail seems not to be purgeable. A possible cause is permission problems on the thumbnail server, or an image being stuck in the squid cache. Usually forcing thumbnail generation with thumb.php and then triggering another purge should fix the latter though, and I can't seem to get that working. This requires filing a bugzilla ticket and then attention of a system administrator I think. —TheDJ (talk • contribs) 20:06, 9 April 2013 (UTC)

bugzilla:46976 may be relevant.—Emil J. 20:23, 9 April 2013 (UTC)

Could you please explicitly state whether you have tried to purge already? See https://en.wikipedia.org/wiki/Wikipedia:Purge . As I constantly get the same (wrong) result for the 220px thumbnail version on https://upload.wikimedia.org/wikipedia/commons/thumb/f/f2/BirdRespiration.svg/220px-BirdRespiration.svg.png this seems to be a different problem than https://bugzilla.wikimedia.org/show_bug.cgi?id=46976 which is about "sometimes an old version of the thumbnail is delivered, sometimes the new version of the thumbnail is delivered". --AKlapper (WMF) (talk) 17:09, 10 April 2013 (UTC)

This is a different issue than bugzilla:46976. I've reported it as bugzilla:47087. --AKlapper (WMF) (talk) 17:43, 10 April 2013 (UTC)

Scripts etc not loading

Is there a problem with user scripts not loading at the moment, was OK a few minutes ago. Also in edit mode the edit toolbar is missing and the selectable items below the edit box are not selectable but all are all displayed. Keith D (talk) 22:17, 9 April 2013 (UTC)

Someone has broken the system! Jared Preston (talk) 22:23, 9 April 2013 (UTC)

What browser versions are you using? Some old browsers have recently been put on a JavaScript blacklist (T37906, gerrit:57693, gerrit:55446). I knew people were going to complain... PleaseStand (talk) 22:40, 9 April 2013 (UTC)

Extended content
function isCompatible( ua ) { if ( ua === undefined ) { ua = navigator.userAgent; } // MediaWiki JS or jQuery is known to have issues with: return !( // Internet Explorer < 6 ( ua.indexOf( 'MSIE' ) !== -1 && parseFloat( ua.split( 'MSIE' )[1] ) < 6 ) \|\| // Firefox < 4 ( ua.indexOf( 'Firefox/' ) !== -1 && parseFloat( ua.split( 'Firefox/' )[1] ) < 4 ) \|\| // BlackBerry < 6 ua.match( /BlackBerry[^\/]*\/[1-5]\./ ) \|\| // Open WebOS < 1.5 ua.match( /webOS\/1\.[0-4]/ ) \|\| // Anything PlayStation based. ua.match( /PlayStation/i ) \|\| // Any Symbian based browsers ua.match( /SymbianOS\|Series60/ ) \|\| // Any NetFront based browser ua.match( /NetFront/ ) \|\| // Opera Mini < 7 ua.match( /Opera Mini\/[0-6]\./ ) ); }

Extended content

function isCompatible( ua ) {
	if ( ua === undefined ) {
		ua = navigator.userAgent;
	}

	// MediaWiki JS or jQuery is known to have issues with:
	return !(
		// Internet Explorer < 6
		( ua.indexOf( 'MSIE' ) !== -1 && parseFloat( ua.split( 'MSIE' )[1] ) < 6 ) ||
		// Firefox < 4
		( ua.indexOf( 'Firefox/' ) !== -1 && parseFloat( ua.split( 'Firefox/' )[1] ) < 4 ) ||
		// BlackBerry < 6
		ua.match( /BlackBerry[^\/]*\/[1-5]\./ ) ||
		// Open WebOS < 1.5
		ua.match( /webOS\/1\.[0-4]/ ) ||
		// Anything PlayStation based.
		ua.match( /PlayStation/i ) ||
		// Any Symbian based browsers
		ua.match( /SymbianOS|Series60/ ) ||
		// Any NetFront based browser
		ua.match( /NetFront/ ) ||
		// Opera Mini < 7
		ua.match( /Opera Mini\/[0-6]\./ )
	);
}

Firefox 3.6.28, because I can't use a more-updated version on my computer. Jared Preston (talk) 22:45, 9 April 2013 (UTC)

Only FF < 2 would be blocked, and that block does not appear to be active. -- Gadget850 (Ed)^talk 22:48, 9 April 2013 (UTC)

In the "extended content" above, it shows any Firefox browser "< 4", which would include Version 3. Since that includes me and I'm experiencing problems like the author of this thread, it would seem that there's a problem... Jared Preston (talk) 22:52, 9 April 2013 (UTC)

gerrit:57693 indeed says in the commit message that Firefox < 4 was blacklisted, as does the actual code change. jcgoble3 (talk) 23:00, 9 April 2013 (UTC)

I am using Firefox version 3.6.28. Keith D (talk) 23:03, 9 April 2013 (UTC)

So can a fix be found for people who don't or can't use the newest, flashiest browsers? Jared Preston (talk) 23:06, 9 April 2013 (UTC)

If you are meant to be blocking only FF < 2 then the code is in error and should be changed. Keith D (talk) 23:10, 9 April 2013 (UTC)

There is a fix: Stop using an end-of-lifed browser that's no longer supported and get a modern one. And if your end-of-lifed operating system prevents you from using a newer browser, then buy an upgrade to a newer OS or even a whole new computer if that's what it takes. Developers cannot, repeat cannot, be expected to account for every browser and operating system that ever existed. Instead, they can, should, and do code for what's still supported. At some point the technology has to move on, and if you insist on using outdated browsers and/or OSes, you can and should expect things to break. jcgoble3 (talk) 00:44, 10 April 2013 (UTC)

It's harsh, but it's true; developers can only support outdated software for so long before it becomes counter-productive. Improvements and features become harder and harder to implement over time the longer legacy software is supported. Firefox 3.6 has apparently been out for more than three years, and had its support officially dropped almost a year ago. I realize "get a new browser/computer" isn't the answer that you'd prefer, but I don't think it's fair to expect the devs to continue supporting a browser that has been dropped by its own parent company. EVula // talk // ☯ // 02:36, 10 April 2013 (UTC)

Why is it apparently forcing us ffox3.6 folks to be completely turned off rather than just not supported? Everything was working fine yesterday, did every javascript bit suddenly get updated to do things my browser can't? "You're on your own, things might start to behave badly because we're adding new-browser-only stuff" is a lot more friendly than "screw off, don't even try, no bits for you." DMacks (talk) 03:53, 10 April 2013 (UTC)

It's not always that simple. I remember a while ago there was a problem with Wikipedia tending to lock up for readers due to Javascript not running properly on an old version of IE. If something impacts readers badly, then often the right response is to turn it off rather than telling visitors to just live with it. Anyway, the person who blocked FF less than 4 was Timo Tijhof ttijhof

wikimedia.org, you could ask him why. Dragons flight (talk) 04:38, 10 April 2013 (UTC)

The article "History of Firefox" quite rightly states that support for Firefox 3.6 expired not even 365 days ago. I don't know much about IE, but there are also a number of wp skins which are ancient and still supported, or at least still offered. At 22:00 UTC everything was still fine, so now to be "blacklisted" all of a sudden makes me feel like a criminal. I thought this was supposed to be "Wikipedia, the free encyclopedia that anyone can edit" – telling someone to update their OS or buy a new computer isn't very helpful. I'm really sorry, but despite having a job I simply cannot afford a new computer. And I can't for the life of me work out how one MediaWiki user can decide which browsers are to be supported, and which should not be. Well done for pissing someone off that has helped out here for almost eight years! Jared Preston (talk) 05:58, 10 April 2013 (UTC)

For what it is worth, I doubt very much that it was "one Mediawiki user". The change is traceable to a particular developer, but in all likelihood there was a significant discussion about it before hand as well as a specific reason for the change (i.e. some specific problem with the old browsers). However, those discussions likely happened within the WMF so we are unlikely to know about them unless we ask. The quickest way to get to the underlying reasons for the change would probably be to ask the developer who made it. Dragons flight (talk) 06:31, 10 April 2013 (UTC)

Firstly, let me address the "Wikipedia, the free encyclopedia that anyone can edit" argument. In all honesty, this is getting old and inappropriate. The encyclopedia is no more or less free or editable by the presence of the extra javascript features. Obviously things are prettier and easier to edit with the edit toolbar and gadgets than without, but it is by no means an access restriction for even a blocked user is still able to read the encyclopedia pages.

Now, as for Firefox 3.x and below, jQuery has not supported Firefox 3.6 for quite a few releases. They currently only actively support Firefox 18 and above (though consider that number with the knowledge that Firefox now has auto-updating built-in, like Google Chrome, which means that only users of before that version are stuck. There are practically 0 users of any Firefox version between 15 and 18). Also note that this was not Wikimedia's decision to make, it was jQuery's and since we depend on it (unless we start to fork and patch it) we can't control this. jQuery could break in Firefox 3 at any time as nobody is actively looking after it during jQuery development as far as I know (see jquery/testswarm for example).

Having said that, there is a difference between a browser not actively supported (e.g. tested for before deployment and reported bugs are prioritised) and browsers that are known to have fatal errors in script execution - in which case a blacklist in ResourceLoader is appropriate.

When updating our blacklist I jumped the gun in this and should've excluded Firefox below 3 not up until 3.x (< 4). Firefox 2 is known to be incompatible with our latest scripts, Firefox 3.0 - 3.5 is a bit of a wild card. Firefox 3.6 however is an LTS (Long Term Support) release by Mozilla that (last I checked) is compatible enough to not be a problem for most users most of the time. So I will lower our blacklist for now to tolerate Firefox 3.6.

However LTS is not ELS (Ever Lasting Support), Firefox has maintained 3.6 for many years, but that stopped in April 2012. I can't stress enough the importance of upgrading to a browser that the creator of that browser supports (in this case, a version of Firefox that Mozilla supports). When Mozilla finds out about security issues and releases updates, they do not create and release patches for users of old versions such as Firefox 3.6. As such you make yourself vulnerable to security leaks when browsing the web (not Wikipedia in particular).

Anyhow, as said, I've lowered ResourceLoader startup check from Firefox 4 to Firefox 3 (gerrit:58465). Krinkle (talk) 07:53, 10 April 2013 (UTC)

Well I knew I didn't have a problem using Wikipedia until late last night. And of course, it's not just Wikipedia, but without javascript support, you're pretty much excluded from editing many features of Wikidata too. I appreciate that using an outdated browser leaves yourself open to security issues, and fortunately I haven't had any since April 2012 (having an anti-virus program helps a bit, I guess) and, like, I said, I just can't update from here. It isn't great being stuck in a black hole, and didn't think it made such a big difference to Wikipedia which browser I use to edit in... Thank you, Timo, for lowering the bar for me, and other users experiencing the same problem. I don't know whether or not you got my e-mail from an hour or so ago. I don't know what "jQuery" is (maybe I should?) but having javascript open again makes a whole lot of editing here a lot, lot easier. So, as until a time that javascript breaks uncontrollably, thank you very, very much! Jared Preston (talk) 08:14, 10 April 2013 (UTC)

Thank you for explaining the situation! In the future, these sorts of user-impactful changes might be worthy of a sitenotice, or technical note in Signpost. DMacks (talk) 11:52, 10 April 2013 (UTC)

English Wikipedia servers going down around 02:30 UTC every day?

Tracked in Phabricator
Task T29320

For the last few days, I have noticed around 02:30 UTC, the English Wikipedia completely shuts down, and I receive a message that states "This wiki is having technical problems." However, the message disappears after reloading the site about a minute afterwards. I have been using this Wikipedia regularly for the last few months, and have never ran across this message regularly (or at all) until last week. Just wondering ... what is the cause of this, and is any action being done to resolve this intermittent issue? Steel1943 (talk) 02:40, 10 April 2013 (UTC)

It's not the normal site error notice either. Ryan Vesey 02:55, 10 April 2013 (UTC)

I think this is T29320. If you look at the Server Admin Log, you will see that LocalisationUpdate runs at about that time every day. Database server overload could certainly result in such error messages as database connection errors ("Sorry! This site is experiencing technical difficulties. Try waiting a few minutes and reloading." as opposed to the text shown at WP:WFEM). PleaseStand (talk) 03:42, 10 April 2013 (UTC)

I saw the same message last night from my mobile device on the desktop site. It fixed itself in less than 2 minutes though. Technical 13 (talk) 17:14, 10 April 2013 (UTC)

I've been seeing that error in the last week or so too. I didn't take notice if it only occurs at a specific time or not. What I do know if that I basically would never see this problem unless Wikipedia had a major outage so it's something relatively new. Jason Quinn (talk) 21:28, 10 April 2013 (UTC)

Show

There appears to be a glitch at this category where several entries such as James Wilson are missing. Can someone fix that please. Pass a Method talk 13:09, 10 April 2013 (UTC)

I can see Wilson's name there! --Tito Dutta (contact) 14:57, 10 April 2013 (UTC)

archivebot.py

Hello, somebody uses this script? At me it is impossible to archive pages with its help.

    C:\Python26\pywikipedia>archivebot.py Archive-bot/archive
    Fetching template transclusions...
    Getting references to [[Template:Archive-bot/archive]] via API...
    Processing [[:ru:Обсуждение участника:Парис "Анима" надаль]]
    4 Threads found on [[:ru:Обсуждение участника:Парис "Анима" надаль]]
    Looking for: {{Archive-bot/archive}} in [[:ru:Обсуждение участника:Парис "Анима"
    надаль]]
    Processing 4 threads
    There are only 0 Threads. Skipping
    Processing [[:ru:Участник:Парис "Анима" надаль/Черновик]]
    14 Threads found on [[:ru:Участник:Парис "Анима" надаль/Черновик]]
    Looking for: {{Archive-bot/archive}} in [[:ru:Участник:Парис "Анима" надаль/Черно
    вик]]
    Processing 14 threads
    There are only 0 Threads. Skipping
     
    C:\Python26\pywikipedia>

Repeatedly I repeated start of the bot, the same result turns out. Nobody knows, what it is necessary to make in order that the bot ceased to pass pages? Thanks.--Парис "Анима" надаль (talk) 10:31, 11 April 2013 (UTC)

I tweaked your talk page see if it works now. Werieth (talk) 12:19, 11 April 2013 (UTC)

Yes, I saw, thank you very much, but, unfortunately, the bot all the same doesn't carry out archiving. :-( --Парис "Анима" надаль (talk) 12:58, 11 April 2013 (UTC)

Probably, someone uses this script. May I look at the pages of discussions archived by this bot? Perhaps, I will find, in what my mistake. Thanks.--Парис "Анима" надаль (talk) 14:35, 11 April 2013 (UTC)

Looks like the bot wasnt programmed for ru.wiki's timestamp format. Werieth (talk) 15:53, 11 April 2013 (UTC)

I believe this is the script used by MiszaBots I, II, and III. jcgoble3 (talk) 23:11, 11 April 2013 (UTC)

I agree with Werieth. You should use a regex to search for timestamp in Russian (in English, it's in the form "23:11, 11 April 2013 (UTC)"), otherwise it can't find the date of each thread. Here is my code that used on the Chinese Wikipedia.--Make cat 02:25, 12 April 2013 (UTC)

Makecat, would you mind submitting a patch to our tracker so we can push it upstream for others to use?

Hi, most Pywikipediabot developers don't watch this page, you're better off filing a bug in our tracker. Legoktm (talk) 02:39, 12 April 2013 (UTC)

Sourceforge is quite slow here due to some well-known reasons. I found that I can't add thread to tracker and it seems that Sourceforge doesn't allow using proxies (log out just after logging in). I hope you will fix it as my code was published. --Make cat 03:21, 12 April 2013 (UTC)

Timeline question

I just created User:Werieth/sandbox, is there a way to highlight every other row (white/gray alternating) so that it is easier to follow the timeline? Werieth (talk) 16:36, 11 April 2013 (UTC)

I couldn't find anything of the sort listed on MW:Extension:EasyTimeline/syntax which is where I expect it would be.

Technical 13 (talk) 16:49, 11 April 2013 (UTC)

with timeline extension, almost everything is "possible", and almost nothing is easy or simple.

it is not clear what do you mean by "highlight". the simplest thing would be to simply assign a different color to the odd vs. even bars. if by "highlight" you mean using different background, i think that even though "background" per se is not supported as far as i could see, it should be possible to place different bars one on top of the other, in effect creating background "manually" (by making the "background bars" assume the full width of the row). for a very complex example, see mw:Template:Wikimedia Growth.

(side comment): at one point i was considering adding line-graph capabilities to Module:Chart, by studying "timeline" syntax and using it (the idea was to use the module to translate sane and more-or-less unified way to pass parameters, and translate it to the byzantine syntax used by timeline), but i am not sure i want to do it any more - timeline syntax may be too much for me. peace - קיפודנחש (aka kipod) (talk) 17:22, 11 April 2013 (UTC)

I was looking for something like [23] where I could color the rows and make reading it easier. Werieth (talk) 17:36, 11 April 2013 (UTC)

Remove page from categories

I'm sure there is a way to do this, does anybody know how to make it so that User:Jay8g/Duplicate templates is not included in the maintenance categories from the tags?--Jac16888 ^Talk 20:53, 11 April 2013 (UTC)

Well, if nothing else you could just substitute the templates with cats attached and manually remove the cats afterward.--Fyre2387 ^{(talk • contribs)} 23:08, 11 April 2013 (UTC)

Set |category= to "no" on the translation and Wikiquote templates, and to the empty string on the policy templates. jcgoble3 (talk) 23:21, 11 April 2013 (UTC)

It seems to have been removed from the now anyway, I suspect due to the edits User:WOSlinker has been making to various templates (because of this thread?). Thanks for the suggestions though--Jac16888 ^Talk 19:01, 12 April 2013 (UTC)

How do I categorize a template?

I made Category:Conflict of interest templates but I can't figure out how to put templates in it. When I go to edit {{uw-coi}} the only categories I see are ones that are for the pages the template would be put on, even though that template itself is already in Category:Standardised user warning templates and Category:User warning templates.--Atlantima (talk) 14:27, 12 April 2013 (UTC)

Most templates are categorised by putting the cat on the template's /doc page, enclosed in <includeonly>...</includeonly>. Where the template has no /doc page, the cat goes on the template page but enclosed in <noinclude>...</noinclude>. --Redrose64 (talk) 14:37, 12 April 2013 (UTC)

And the existing categories on that template are coming from the inclusion of {{Single notice}}. the wub "?!" 14:40, 12 April 2013 (UTC)

Ah, yes, the noinclude. Forgot about that. Thanks--Atlantima (talk) 14:45, 12 April 2013 (UTC)

Man down?

I cannot access the toolserver. Anyone else experiencing this? Do the hamsters need replacing? Thanks.--ukexpat (talk) 03:01, 6 April 2013 (UTC)

I first noticed it was inaccessible about 6 or 8 hours ago. I don't know why. Dragons flight (talk) 03:44, 6 April 2013 (UTC)

Me too. --Make cat 05:36, 6 April 2013 (UTC)

Seems to be back up now.--ukexpat (talk) 14:24, 6 April 2013 (UTC)

Man down?! Methinks technology is overvalued in that description. - Ac44ck (talk) 14:51, 6 April 2013 (UTC)

Wondering if it was taken down for a bit to catch up... Before it went down it had a 28 hour overhead for me, now it is 4 hours... Technical 13 (talk) 15:04, 6 April 2013 (UTC)

Down again! -(t) Josve05a (c) 11:30, 7 April 2013 (UTC)

Hopefully it might have cleared the replag when it comes back up.--Gilderien Chat|List of good deeds 12:22, 7 April 2013 (UTC)

Is toolserver broken?

I noticed that at Special:Contributions you enter a username, then at the bottom there appears a box containing a number of links. One of those links, Global contributions, doesn't seem to work. It seems the whole toolserver is unreachable, although most services listed at http://status.toolserver.org/ seem to work properly. -- Toshio Yamaguchi 12:49, 7 April 2013 (UTC)

It's been down since 04:00, 6 April 2013 (UTC). See Template:Toolserver. The information in that template is updated hourly by BryanBot. If there is no update after an hour, it is likely that the entire Toolserver is down and there have been no update since 04:00, yesterday. According to the template doc, it is possible that while the user bot (BryanBot) is up, the database (MySQL) is down. When MySQL is down, bots and web services that require MySQL will not function. Also see the post above #Man down? --Ushau97 (talk) 12:57, 7 April 2013 (UTC)

Well, I just found http://meta.wikimedia.org/wiki/Future_of_Toolserver. -- Toshio Yamaguchi 13:05, 7 April 2013 (UTC)

Well, that's sad news it's getting discontinued. Next time there is a discussion on how to use the funds, I will suggest some should be spent on maintaining the servers used by Toolserver. What do you say? --Ushau97 (talk) 13:11, 7 April 2013 (UTC)

It also means Template:Coords and related ones don't work. Since most articles about places use those templates, it's a matter of some urgency. Jim.henderson (talk) 13:25, 7 April 2013 (UTC)

Its a known issue, the toolserver roots are working on it. Werieth (talk) 13:36, 7 April 2013 (UTC)

Quick status update

The toolserver appears to be back up to full function, thanks to Nosy's hard work. For the record, Toshio is correct that the toolserver is slated to be decommissioned, but it's an orderly process – it's not going to just be turned off at some fixed date.

In the meantime, thew new home for tools, the Tool Labs, is almost ready to house every tool. Those maintainers who wish to move early can do so, provided their tools do not rely on the database replicas (which will be deployed in a few weeks); web tools and typical bots can already be moved (and, indeed, some already have). — MPelletier (WMF)^(talk) 13:41, 8 April 2013 (UTC)

I think that link was supposed to be mw:Wikimedia Labs/Tool Labs. --SoledadKabocha (talk) 20:51, 10 April 2013 (UTC)

Right now, coords are not working, though. So I have my doubts about "full function". - Jmabel | Talk 06:35, 13 April 2013 (UTC)
The toolserver has gone down again. See Toolserver is down again below. – PartTimeGnome (talk | contribs) 14:47, 13 April 2013 (UTC)

It comes and goes, I think. I was just doing a contribution search on an article. It was slow, but brought up 100 results. Then I clicked on "show next 100", it brought up the "404 not found" issue. — Maile (talk)

false negatives in category:Pages with missing references list

Tracked in Phabricator
Task T48978

There is a very annoying problem here, with many pages being added to the list even though they have no cite error. There was a discussion here with the final suggestion that the conversation moves here. NB user:John of Reading is not the same person as me user:John of Cromer in Philippines

From what I have seen so far, about 80% of the entries on the list are false negatives, i.e. there is nothing wrong with them, but the only way to get them off the list is by null edit. There is no way of knowing other than by inspection whether a page is correctly on the list. Inspecting and nullediting take a couple of minutes; meanwhile it is quite feasible (and often happens) that the list grows.

Having an error in a template is quite rare, usually when an editor includes <ref>s there and relies on there being a {{reflist}} on the invoking page. as soon as the template is fixed, all the using pages immediately drop out of the list.

Question is how/why do pages get on the list, and how to get the false negatives off?

As an aside, I think it would be a good move to ensure editors have reviewed, and resolved any errors, before enabling the save button.

John of Cromer in Philippines (talk) mytime= Sun 07:34, wikitime= 23:34, 6 April 2013 (UTC)

How they get into the category is beyond my level of knowledge, but getting them out of the category is simple: a mass forcelinkupdate-purge via the API (equivalent to null edits). I ran the API over them (using the API sandbox), and the category is now down to just 51 pages. Does that look better? jcgoble3 (talk) 00:22, 7 April 2013 (UTC)

It may have been, I was just out buying more internet time. Now it's back up to 144 pages.

I wish I knew what your answer meant.

John of Cromer in Philippines (talk) mytime= Sun 10:17, wikitime= 02:17, 7 April 2013 (UTC)

What I meant is that normally a standard purge of an article does not update the link tables (which is what category pages are built from), requiring a null edit. However, purging via the API has two advantages: first, you can purge many articles with one command, and secondly, the API offers a parameter, not available through this web interface, called forcelinkupdate, which causes both the article and the associated link table to be updated—the equivalent of a null edit. At any rate, I ran the category through the API again, and it's back down to 53 pages.

If you want to run it yourself, go to Special:ApiSandbox, select "purge" from the "Action" dropdown, check the box next to "forcelinkupdate" (first row of the table), select "categorymembers" from the dropdown next to "generator", then enter "Category:Pages with missing references list" next to "gcmtitle" (first row of the "Generator" table). Finally, find the "gcmlimit" row and enter a number larger than the number of pages in the category. Then click "Make request" and go fix yourself a sandwich; purging many articles at once can take a few minutes. If you get a red error message, give it another try. Once done, go back and purge the category page directly to see how many pages are left. You can't break anything with the purge action.

And in the time it took me to type that second paragraph, the category is back up to 125 pages, so I'm running it through the API again. There's gotta be a bug in here somewhere... jcgoble3 (talk) 02:46, 7 April 2013 (UTC)

Bug ticket filed: bugzilla:46978 jcgoble3 (talk) 03:00, 7 April 2013 (UTC)

Could anybody provide a hint when this problem started, approximately? Thanks! --AKlapper (WMF) (talk) 09:55, 8 April 2013 (UTC)

It was "always" a minor irritant, but then the Thursday (my time) before Easter it suddenly escalated - overnight the entries on the list jumped from less than 20 to 300 or more. (usually there are perhaps 40 true bad 'uns). This actually took more than a week to clear, as each entry had to be inspected and nulledited at a minimum. Then it suddenly, last Friday I think, jumped back from clear to in excess of 200, which is when I started this particular ball rolling. I was monitoring yesterday: it is not a steady trickle but a whole batch at once. There seemed to be three waves. Fortunately I am now able to purge the set through the sandbox. I ran this once in the(my) morning, which removed more than 50, then again later in the afternoon which removed a similar number. Finally in the evening the list jumped from around 10 to more than 50 and I ran it again. I still have the log for that, but it doesn't tell you much, doesn't have a timestamp or anything. Hope this helps. John of Cromer in Philippines (talk) mytime= Tue 06:57, wikitime= 22:57, 8 April 2013 (UTC)

Comment I noticed a single instance of a phantom entry at Category:Pages containing citation needed template with deprecated parameters for the article Environmental terrorism. (See this discussion.) ~~Near the beginning of April~~On March 20th, I started cleaning that category and have made about 1000 edits (roughly ~~100~~50 per day) as part of that. During this time, I ended up making quite a few minor changes to some inline templates. The biggest was that the tooltips now show the a date in the format "(April 2013)" whereas before it was "from April 2013" (see this discussion at {{fix}}, change made on April 4th). I have also made some minor changes to the wording and punctuation of the "title" parameter of a bunch of inline templates like {{citation needed}}. I don't see how any of this could could responsible for problems with missing references but the timing and number of phantom entries is coincidence enough to make me mention it. Jason Quinn (talk) 20:44, 13 April 2013 (UTC)

Forward slash

Hello. I have a problem with the symbol /. For example Talk:C/2006 P1. If i create a subpage it would be Talk:C/2006 P1/Archive. But if i used it in a template or someone else, its regonzise that this is a subpage of Talk:C. What can i do? Xaris333 (talk) 14:47, 10 April 2013 (UTC)

/Archive would be a sub-page of /2006 P1 which is in itself a sub-page of Talk:C. What template are you trying to use it in? Technical 13 (talk) 15:05, 10 April 2013 (UTC)

C/2006 P1 and C are different articles. Xaris333 (talk) 16:09, 10 April 2013 (UTC)

Yes, but whereas the software treats a slash in an article name as yet another character, it treats it as a subpage separator everywhere else. Thus Talk:C/2006 P1 is indeed a subpage of Talk:C, despite that humans think of it as the talk page of C/2006 P1. Notice the little “< Talk:C” below the heading of Talk:C/2006 P1, which reminds to this fact.—Emil J. 16:24, 10 April 2013 (UTC)

I think archive templates may have problems. {{Talk archive}} includes the code {{#titleparts:{{FULLPAGENAME}}|1}} if its used on Talk:C/2006 P1/Archive the result would be Talk:C rather than Talk:C/2006 P1. I probably suggest do archive links by hand, although it might be posible to fix the template with {{#titleparts:{{FULLPAGENAME}}|-1}}, but this might break some other pages. See parser function doc--Salix (talk): 16:43, 10 April 2013 (UTC)

{{#titleparts:{{FULLPAGENAME}}|-1}} would most definitely fix the problem... To avoid possible issues with other pages that have been using this for any duration of time, you might go {{#titleparts:{{FULLPAGENAME}}|{{#ifexpr:{{REVISIONDATE}}<{{subst:REVISIONDATE}}|1|-1}}}} and change the documentation to reflect the new way it is working. Alternatively, you could have someone with AWB access pull up a list of the pages that transclude the template and fix the usage for them. This would probably be preferred but less likely to happen. Technical 13 (talk) 17:08, 10 April 2013 (UTC)

The page C/2006 P1 is not a subpage of C. But the page Talk:C/2006 P1 is a subpage of Talk:C. This is wrong. How can i change that. Why C/2006 P1 is not a subpage and Talk:C/2006 P1 is; Where is the different? Xaris333 (talk) 16:51, 10 April 2013 (UTC)

The difference is that C/2006 P1 is in the main namespace, and Talk:C/2006 P1 is in the corresponding Talk namespace. You cannot change that, this is how the site is set up. See WP:SUBPAGE#Articles do not have sub-pages (main namespace) for more details.—Emil J. 17:03, 10 April 2013 (UTC)

in reality, in mw system "subpage" is pure fiction. it's a convention we use for convenience, and some features of the system pretend it's real, e.g. all the "titleparts" tools, and some protections/rights (e.g., a page in "User:" namespace whose name ends with ".js" can only be modified by users with "editinterface" rights, and the user that this page looks to be a subpage of her user page). however, as i said, this is pure fiction. you can create a page called No such page/No such subpage/This doesn't exist also/Faux subpage, without having to create any of the pages that nominally "contain" it, and all will be well (from the system's page storage POV, the forward slash is merely one more legal character in a page name. we actually have an article called [[//]], which, if you think of forward slash as "subpage separator" really should mean "a page with empty name which is a subpage of a page with empty name which is a subpage of "the" page with empty name". of course, this is not so - it's just a page with the name "//" (becaus it confuses the system, in order to link to it you'll need to write [[://]], which may or may not be a small legal program in Brainfuck). peace - קיפודנחש (aka kipod) (talk) 19:11, 10 April 2013 (UTC)

I do agree that subpages should be enabled on BOTH an article namespaces and its associated talk namespace, or neither of them. Given that the main namespace is not configured to support subpages, the main Talk namespace should not be configured to support it too (because for some articles in the main namespace it's impossible to link to their associated Talk page, as it will reach a Talk page associated to another article !)

In fact the situation would be safer if it was the reverse : article namespaces with support of subpages, but Talk pages without, allowing to group discussions related to a page and all their subpages (this would not prohibit the creation of archived Talk "subpages", but navigating between these false subpages and the main Talk page would be complicated a bit, and would require the cration and use of navigation templates).

The consequence of adding support for subpages is the possibility of creating relative links to a subpage using "/" as the first character in the wiki link, or the initial characters "../" to create a link based from the parent page.

Enabling subpages adds new restrictions on valid titles of pages, where they may contain slashes (such as "../" or "//" within any namespaces, which will be interpreted hierarchically instead of litterally): you cannot change this setting before testing the content of the complete set of pages in that namespace to see if they start with a "/" or with "../" or if they contain "//" or "/../", because these pages will no longer be reachable.

So these rare pages will have to be renamed, and the redirection link deleted, and all pages linking to them will need to be edited to not use the redirection link but the new name in the target parameter of links.

Note that some pages are creating links using URLs instead of wikilinks (sometimes via template helpers), and they are more difficult to detect because these sources pages using these URLs are not completely indexed ; and this won't be detected if source pages are from other Wikimedia wikis, or from other external sites). To solve this last issue, we would sometimes have to insert in the top of some pages a disambiguation notice, providing the new name of the page whose old name would now conflict with the title of another unrelated article (lots of pages may be affected if ever the database, before the change of setting, contained some article titles starting by "/" or "../" or containing "//" or "/../").

The "titleparts" parserfunction should also be able to test the namespace given to see if that namespace is configured to support subpages or not. If not, it should not split the given title into multiple parts. Unfortunately, this builtin function is also used in various templates to split random parameters into multiple parts easily, allowing to pass a list of items in a single template parameter: if namespaces are tested by the builtin "titleparts" parserfunction, these templates will stop working. The only solution would be to add an additional (optional) parameter to the "titleparts" parserfunction specifying that namespaces present in its 1st parameter should be tested.

For now, the best you can do is to write an utility template that will perform the namespace test itself before using "titleparts"... But this will still return incoherent results between the main articles namespace and the main Talk namespace, the way they are currently configured on English Wikipedia ! verdy_p (talk) 08:40, 11 April 2013 (UTC)

For the most part subpages are very useful for the Talk: namespace. A large number of articles have archives for talk pages (94464 transclude {{Talk archive}}[24]) and treating these as subpages makes a lot of sense. Yes there are a few problems when the corresponding article has a / in it but this only a tiny portion of articles and there are workarounds. --Salix (talk): 10:11, 11 April 2013 (UTC)

The problem i had is with a template in Greek Wikipedia. el:Πρότυπο:Κριτική λημμάτων. I had to change the parameter [[{{TALKSPACE}}:{{BASEPAGENAME}}/Κριτική λημμάτων|σχετική σελίδα]] than can recognize the talk page of an article has name like el:C/2006 P1. Can anyone help? Xaris333 (talk) 18:23, 13 April 2013 (UTC)

Problem with wgRedirectedFrom

I have some JavaScript code at User:Dudemanfellabra/NRHPstats/sandbox.js that does one thing if you've been redirected to a given page and another thing if you went straight there. The only way I know to check if the user has been redirected is

 if (wgRedirectedFrom != null) {
    // do stuff if you were redirected
 }
 else {
    // do stuff if you weren't redirected
 }

For some reason, though, the wgRedirectedFrom variable seems not to be functioning as expected. If I'm on a redirected page, it works (i.e. it isn't null), but if I'm not on a redirected page, I get an error from that check saying wgRedirectedFrom isn't defined. I thought it was always defined but null if you weren't redirected. If this is not the case, does anyone know how to check if you've been redirected to where you are now?--Dudemanfellabra (talk) 18:01, 11 April 2013 (UTC)

Sounds like wgRedirectedFrom doesn't exist if you weren't redirected... Try:

 if (wgRedirectedFrom) {
    // do stuff if you were redirected
 }
 else {
    // do stuff if you weren't redirected
 }

Technical 13 (talk) 18:08, 11 April 2013 (UTC)

Tried that earlier (and again just now, just in case); didn't work. Still said it wasn't defined, even though I was literally checking to see if it was defined. I have no idea what's going on.--Dudemanfellabra (talk) 18:11, 11 April 2013 (UTC)

This page says "wgRedirectedFrom | String | When redirected contains the title of the page we were redirected from. null when not redirected. Uses the same format as wgPageName." So it should be defined... Right?--Dudemanfellabra (talk) 18:18, 11 April 2013 (UTC)

When you try to use a variable name like that, Javascript tries to dereference it to get at its value, which will fail if the variable is undefined. Try:

if(typeof wgRedirectedFrom == "undefined")
{
   //handling of undefined redirect
}
else
{
   //handling of defined redirect
}

That should let you test to see if it's defined without actually looking at what's in it, thus avoiding the undefined error. Writ Keeper ⚇♔ 18:27, 11 April 2013 (UTC)

typeof worked. The script now does what it's supposed to do. That doesn't change the fact that the manual over at mediawiki is incorrect. That manual says it's null when it's actually undefined.--Dudemanfellabra (talk) 18:37, 11 April 2013 (UTC)

Yeah, that happens. WP:SOFIXIT. ;) Writ Keeper ⚇♔ 18:39, 11 April 2013 (UTC)

I didn't know I could edit mediawiki.. thought that required more credentials. I actually logged in for the first time there just now and was about to update the page, but I see you did. Since I've never edited there, though, I can't confirm the pending change. Oh well.--Dudemanfellabra (talk) 18:45, 11 April 2013 (UTC)

I guess it requires more credentials than either of us have to really edit it. :) Writ Keeper ⚇♔ 18:48, 11 April 2013 (UTC)

normally, you do not need to use "typeof". however, more and more scripts nowadays define "use strict", which may protest when encountering constructs like "if (wgRedirectedFrom)" i thing the standard way to access globals is "if (window.wgRedirectedFrom)", which should still work, even in strict mode, but in wikipedia code, we are supposed to use "if (mw.config.get('wgRedirectedFrom'))" instead. peace - קיפודנחש (aka kipod) (talk) 19:31, 11 April 2013 (UTC)

According to MW:ResourceLoader/Default_modules#mediaWiki.config the proper way to do it is if ( mw.config.exists( 'wgRedirectedFrom' ) ) { /* do stuff */ } Technical 13 (talk) 18:09, 13 April 2013 (UTC)

API help question

Tracked in Phabricator
Task T49173

Hello, I originally asked my question here, but realize it may be awhile to get a response there. So, I'm going to ask here too in hopes of getting a speedier answer. I have found that "api.php?action=query&list=backlinks&blnamespace=0&bltitle=" + wgPageName; will tell me if there are any articles that link to the page, but how can I get the full count (if greater than 500)? I don't care "what" articles link there, just how many. I'm trying to write myself a little JavaScript which in part will expand the "What links here" link in my p-tb to include an count of pages that link to a specific NS (or two or three) and a total count of pages that link to that article. I intend to use such a script to save me some time when I'm tagging articles for improvement on enwiki to quickly know if the article is an "orphan" or not. Thank you. Technical 13 (talk) 17:04, 12 April 2013 (UTC)

You cant, that information is not stored in the database, the only way to do that is to retrieve the lists in chunks of 500 until you dont have any more. Werieth (talk) 17:06, 12 April 2013 (UTC)

That doesn't make sense. It has to be in the database somewhere. If it isn't, how does Special:MostLinkedPages have the counts readily available. If I do have to do it like you say, some of the pages on that list would require 9,000+ API requests to populate the list. Technical 13 (talk) 17:24, 12 April 2013 (UTC)

Special:MostLinkedPages is updated infrequently and probably takes quite a bit of time to run. The results are then cached until the next update. Anomie ⚔ 18:10, 12 April 2013 (UTC)

Honestly, I'm not real good with ajax or json. Would someone take a look at my script and tell me why it's not even trying to run? User:Technical_13/Scripts/Enhanced "What links here".js‎‎ I'm not looking for anyone to re-write it for me, just point to things that are not right and where I might find the documentation to fix it... If it is easier to tell me "this" should be "that", I have no problem with that, just please link some documentation so I can learn why and get it right myself next time, if possible. Thanks Technical 13 (talk) 19:38, 12 April 2013 (UTC)
- On a quick glance, elseif isn't valid; JavaScript uses two words (else if) instead. :) jcgoble3 (talk) 19:51, 12 April 2013 (UTC)
  - I knew that one... It was a typo in the first section that got copy pasted to the second... Technical 13 (talk) 19:58, 12 April 2013 (UTC)
- Well, you're attaching a load event to an element—is the element already loaded? — Bility (talk) 21:11, 12 April 2013 (UTC)
  - I'll have to get back to you on this one... I'm thinking of changing the .load() to a .mouseover() or something. Technical 13 (talk) 00:48, 13 April 2013 (UTC)
- Aha! Plugged your script into my common.js, and my error console complained about a parsing error, namely an invalid return in the first if statement. return can only be used inside a function, to terminate the function, but your code isn't wrapped in a function. jcgoble3 (talk) 21:38, 12 April 2013 (UTC)
  - I've taken down the script and am "almost" entirely re-doing it. I'm working on it off-line. I'll keep that in mind though. Technical 13 (talk) 00:48, 13 April 2013 (UTC)
Okay, I've reworked my script, and put it back up. It "should" be building a 2-dimensional array of which each inner array should be comprised of [nsNum, nsName, backlinks] and each inner array is stored in a sequential outer array. Right now, it is set to just gather data and pop out an alert for each of the 24-26 namespaces. If someone familiar would be so kind as to take a look at it and tell me if I've got anything badly borked, I would be appreciative. Technical 13 (talk) 09:44, 13 April 2013 (UTC)
- Still throwing a SyntaxError: return must be in a function. You have two options here, as I see it: either wrap the script in a function and then call the function at the bottom of the script, or enclose the whole thing in an if-then-else statement. The latter would involve using the namespace/link presence check as the condition, only the alert in the first part of the if block, and the rest of the script in the else block. jcgoble3 (talk) 16:37, 13 April 2013 (UTC)
  - I forgot about that return up there. I'm importScript instead of using the resourceloader... Should I be using the resourceloader? I've commented out the return and replaced it with a break for now. Technical 13 (talk)
    - ResourceLoader is beyond my knowledge. As for break instead of return, that throws yet another SyntaxError: break can only occur inside a switch statement or loop. If you want that condition to cause the rest of the script to not run, you'll have to use one of the two methods I mentioned above. jcgoble3 (talk) 17:55, 13 April 2013 (UTC)

Display problem with sub/superscripts

Sorry for asking a question for the Chinese Wikipedia. I posted a request on {{su}} about display problems of related sub/superscript templates on zh.wiki. When viewed in-line, a big space is added before the sub/superscripts, like this: . It is displayed fine when put in a numbered list or table. An incomplete list of problematic templates are in the first post on this site. This problem occurs on Chrome 26.0 on Mountain Lion and Chrome app on iPad, but is not seen on other browsers. The same templates here on en.wiki don't have this issue, and a simple copy and paste onto the zh.wiki templates result in only the big space before the subscript being removed, but the superscript is still bad. Please help us solve the problem! Yinweichen (talk) 02:10, 13 April 2013 (UTC)

We can's solve the problem; it's a Webkit bug. You may want to post a bug report at Chromium. — Edokter (talk) — 09:48, 13 April 2013 (UTC)

On second thought, it seems you are using old code on zh. Try our code again and add margin-left: 0; to see if that fixes the problem. — Edokter (talk) — 09:56, 13 April 2013 (UTC)

No it doesn't help. The subscript returned to the correct position, but the superscript remains to be detached. Yinweichen (talk) 17:15, 13 April 2013 (UTC)

I think the key is to figure out why Chrome renders it differently than Safari, and why the same code works in en.wiki but not zh.wiki. Yinweichen (talk) 17:30, 13 April 2013 (UTC)

Here's the reason: it has nothing to do with the code. I turned off the option to indent the first line of every paragraph, and the problem is solved. Yinweichen (talk) 17:57, 13 April 2013 (UTC)

Good to know! — Edokter (talk) — 20:16, 13 April 2013 (UTC)

Articles randomly showing up in categories

If you look at this cat for unassessed articles, Talk:Robert Andrew Loughnan and Talk:Rangi Mawhete suddenly showed up a few days ago. However, both are assessed and neither talk page has been edited in months (and the unassessed cat is not on those talk pages). Ditto with this other unassessed cat has Talk:Jones (left fielder) which is also assessed. This has been happening a bit the last week or so. Often I can fix it by moving the talk page header below the project banners, but none of the aforementioned articles has the talk page header. Was there a software upgrade recently that caused this problem? Aboutmovies (talk) 07:20, 13 April 2013 (UTC)

Take a look up this page at false negatives in category:Pages with missing references list and bugzilla:46978. I expect the api procedure described would work for you too. John of Cromer in Philippines (talk) mytime= Sat 17:19, wikitime= 09:19, 13 April 2013 (UTC)

I saw possibly related problems here, with Harrison Brown being reported as re-assessed in every bot run, even though the article talk page wasn't touched. My first suspicion was that WP 1.0 bot keeps it's own database to track assessment changes, which could be affected by the recent toolserver problems. But I don't think that can explain your observations. — HHHIPPO 09:47, 13 April 2013 (UTC)

Update: the problem I mentioned first started with this bot-run on 27 October 2012, one month after the last talk page edit. (The similar behavior for Ernst Abbe was due to a duplicated WikiProject tag, but I don't see such problems on Talk:Harrison Brown.) — HHHIPPO 10:05, 13 April 2013 (UTC)

Another update: looking at this thread, my problem seems to be indeed a toolserver issue and unrelated to yours. — HHHIPPO 10:28, 13 April 2013 (UTC)

Main Page added to watch list

It has been recommended (via my talk page) that I post here: Is it just me but main page is being continually added to my watch list with no input from me. This occurs on iPad, smartphone and laptop. Is it a common fault? Asked this on talk page but that may not have been the best place for such a question. I think it is the smartphone I remove from my watchlist and see it returned after I have accessed wikipedia via my phone. Advice appriciated. Phone HTC one S android version 4.1.1 / browser chrome and one provided with phone - webkit/534.30. Ipad is usual, up to dat,e and laptop is windows 7 via firefox. Edmund Patrick – confer 11:10, 13 April 2013 (UTC)

Categories need a purge

Usage of Template:Convert was categorised. The cats were deleted. We now have redlinked cats in over 100,000 articles. See Northwest Lineman College for example (contains Category:Pages using Template Convert and Category:Pages using Template Convert/adj=on). An edit will clear out the redlinked cats. There is a way of clearing this problem by changing a flag or something. Another fine mess. -- Alan Liefting (talk - contribs) 12:03, 13 April 2013 (UTC)

Just give it some time, the Job Queue will take care of it. Werieth (talk) 12:04, 13 April 2013 (UTC)

How long will it take? -- Alan Liefting (talk - contribs) 12:10, 13 April 2013 (UTC)

likely a half a day. maybe a full day. That is exactly the reason why many templates (with high usage) are fully protectd mabdul 12:16, 13 April 2013 (UTC)

Is it is a function that can be done by a bot? That could speed it up? -- Alan Liefting (talk - contribs) 12:27, 13 April 2013 (UTC)

Bad idea, we have the job queue for a reason, it keeps server stress to a minimum. Werieth (talk) 12:37, 13 April 2013 (UTC)

Updated Wikipedia favicon

Hi. I didn't see this mentioned in many places around here, but the Wikipedia favicon is being updated this month (I think it's slowly rolling out to the various Wikipedias now). Further details are available here: m:Favicon#Wikipedia. --MZMcBride (talk) 19:10, 13 April 2013 (UTC)

Note that the mobile favicon, for when a wikipage is added to your homescreen on an iOS device, has been updated as well-previously it had no icon. —Theopolisme (talk) 19:39, 13 April 2013 (UTC)

Favicon

Since when did our favicon change? I just noticed it having rounded corners and being more squashed.--Gilderien Chat|List of good deeds 20:23, 13 April 2013 (UTC)

Look up. —Theopolisme (talk) 20:35, 13 April 2013 (UTC)

Oh thanks don't know why I didn't see that I did look for it :/ --Gilderien Chat|List of good deeds 20:37, 13 April 2013 (UTC)

Browser table rendering differences

Resolved

I am building a table with 18 equal width columns (plus one filler column). The text in the column should wrap if it is too wide. One browser does it nicely (Firefox on WinXP), but Safari on WinXP mixes up the sequence: some columns are minimised, then others are left too wide.

See {{Periodic table/sandbox}}. I used these css techniques: {{shy}} (soft hyphen) in the text, and set column width through ! style="min-width:5.35%; max-width:5.36%;" |. Suggestions anyone? -DePiep (talk) 15:26, 11 April 2013 (UTC)

I don't have any other browsers except for FF installed on this laptop, yet I suspect the edit I just made "should" have fixed the problem. I believe the problem was caused by undefined cells in the table. Technical 13 (talk) 15:57, 11 April 2013 (UTC)

This did not work (though the changes are to stay). Printscreen current Safari effect (same as previous) shown here. -DePiep (talk) 20:26, 11 April 2013 (UTC)

table column width is always problematic issue. you may want to try adding to the table style itself, "table-layout: fixed;" (picked this one from stack overflow - i do not know what other side-effects this may have, but i think it will make it do what you want). peace - קיפודנחש (aka kipod) (talk) 16:27, 11 April 2013 (UTC)

Will try this. You mean add the code to wikitable top row, like {| style="table-layout: fixed;" ? (or go ahead in the sandbox). -DePiep (talk) 20:26, 11 April 2013 (UTC)

exactly. i tried it in your sandbox (didn't hit "save"...), and it definitely makes a difference, though i'm not sure this is what you really wanted. peace - קיפודנחש (aka kipod) (talk) 22:36, 11 April 2013 (UTC)

Done. Works. Thanks. Barnstar. -DePiep (talk) 10:35, 12 April 2013 (UTC)

because of how hairy this whole issue is, i strongly recommend testing with several browsers (at least the big 3, safari and opera gets bonus points, and if you have access to some handheld it's a medal), and trying several screen-widths with each - table should display correctly with width of 800px or less, before declaring victory. peace - קיפודנחש (aka kipod) (talk) 13:32, 12 April 2013 (UTC)

Checked some 100+ browser × OS screenshots. Per WP:ACCESS#Resolution, I took 1024px screens to be decisive. Did not see any prohibiting failings. As for the 800px screens: tried two at home, but no issues. Same for the few mobile versions I checked. I consider this acceptable. Details and tweakings continue at WT:ELEM. -DePiep (talk) 19:13, 14 April 2013 (UTC)

IP tool links

What is going on with the links in the info box at the bottom of IP talk pages and edit histories? There is no longer a geolocate link, and the traceroute link goes to the DNSstuff site where I am requested to register and sign up for "30+ powerful tools and alerts" (which annoyingly I can't back out of because the previous page just redirects me back). Spinning Spark 23:43, 13 April 2013 (UTC)

You describe {{Anontools/ipv6}} which you get for IPv6 addresses. For IPv4 you get {{Anontools/ipv4}}. If you know better tools for IPv6 then you can make a suggestion at Template talk:Anontools. PrimeHunter (talk) 00:35, 14 April 2013 (UTC)

Ah, thanks, I hadn't realised the tools were different. Spinning Spark 09:21, 14 April 2013 (UTC)

Red link convert categories

Hello, I'm not completely sure if this is the best place to list this problem but I've seen a few problems happening lately. I'm not sure if these are supposed to be hidden categories but they keep showing up in articles that use the convert template. Here are some examples, [25] [26] [27].-- Astros4477 (Talk) 03:21, 14 April 2013 (UTC)

See #Categories need a purge it is a known issue that should resolve itself. Werieth (talk) 03:22, 14 April 2013 (UTC)

The three examples given and the two categories that I know of have now cleared. -- Alan Liefting (talk - contribs) 07:34, 14 April 2013 (UTC)

It's still showing up for the first two links I listed.-- Astros4477 (Talk) 13:58, 14 April 2013 (UTC)

It disappeared when I purged the articles. PrimeHunter (talk) 14:18, 14 April 2013 (UTC)

Redirects not working in the Module namespace

See e.g. Module:Cite. Any ideas on how to fix this? It Is Me Here ^{t / c} 12:02, 14 April 2013 (UTC)

Redirects are not implemented for the Module: namespace, because these are server-side scripts. There is no fix (nor can I imagine a use-case for redirecting a module). — Edokter (talk) — 12:10, 14 April 2013 (UTC)

Redirects don't work in the MediaWiki namespace as well, but you can transclude one MediaWiki to another MediaWiki. -- Gadget850 (Ed)^talk 13:51, 14 April 2013 (UTC)

With that said, would someone delete Module:Cite, please? —Theopolisme (talk) 13:54, 14 April 2013 (UTC)

done —TheDJ (talk • contribs) 15:19, 14 April 2013 (UTC)

Toolserver is down again

I cannot use the lovely WP:REFLINKS. -- Alan Liefting (talk - contribs) 20:27, 12 April 2013 (UTC)

Or the Edit Counts tool, or Wikistalk, or... Can someone update me on what the deal is with Toolserver? I've heard that we're going to lose it entirely, is that the case? Beyond My Ken (talk) 21:41, 12 April 2013 (UTC)

The toolserver, which is run by the German chapter of the Wikimedia Foundation (WMDE), will be shut down either late this year or early next year. In its place, the WMF is creating a new platform (Tool Labs) that is intended to have essentially the same role as the toolserver, but will have a somewhat different configuration. Existing tools will need to be converted to run on the Tool Labs platform; however, the intent is to continue to support all or most of the current population of widely used tools. Because Tool Labs will be supported / managed by the WMF directly, one of the hopes is that it will be more stable over the long-term than the existing toolserver. However, it is very early in the process (e.g. Tool Labs does not yet have database replication), so it will probably be many months before we know how well the migration will work in practice. Dragons flight (talk) 21:56, 12 April 2013 (UTC)

I went to Wikipedia:Toolserver to see what the current situation was but there is absolutely nothing about it there. Can you update that page? Cheers. -- Alan Liefting (talk - contribs) 22:25, 12 April 2013 (UTC)

Have now put a hatnote over there. -- Alan Liefting (talk - contribs) 22:55, 12 April 2013 (UTC)

It is flaky. Now I cannot use Dabsolver. -- Alan Liefting (talk - contribs) 02:36, 13 April 2013 (UTC)

Or the ever-dependable Geohack. --Slgrandson (How's my egg-throwing coleslaw?) 03:37, 13 April 2013 (UTC)

It was working fine for me just a few hours ago, but now, the whole Toolserver website is down (the main page is now only a blank page, while every other page now returns a 404 error). I'm completely aware of the financial problems that the Toolserver has had for quite some time, but with several of our projects being dependent on Toolserver (like the one mentioned at the section directly above this), and Tool Labs still under development, this could become a major problem. What now? Narutolovehinata5 ^tccsdnew 03:27, 13 April 2013 (UTC)

Dabsolver, Reflinks, Commons helper ( eg{[28]) are all flaky for me. A real shame that an important tool is now unreliable. -- Alan Liefting (talk - contribs) 10:11, 13 April 2013 (UTC)

Ditto, I haven't been able to use Reflinks for days. I'm somewhat citation-template-phobic to start with, so this is painful. --Arxiloxos (talk) 22:28, 15 April 2013 (UTC)

404 error at WP:UTRS

Hey, I'm trying to get my account exempt from IP hardblocks. After over an hour of trying to find out what I need to do, I posted a justification at the toolserver link provided at WP:UTRS, only to get a 404 error (The requested URL /~unblock/p/index.php was not found on this server.) when I submitted the form. This is seriously annoying. Sławomir Biały (talk) 02:42, 13 April 2013 (UTC)

The WP:UTRS interface runs on the toolserver and the toolserver is having problems again. See #Toolserver is down again above. It's still a bit buggy at the moment. These types of problens are usually fixed within a few hours, so checking back later is probably the best advice. 64.40.54.100 (talk) 05:08, 13 April 2013 (UTC)

toolserver / edit counter

Toolserver / edit counter seems to be down [29]. Not sure if there's anything to be done about it? NE Ent 01:33, 16 April 2013 (UTC)

Please scroll up to read the thread "Toolserver is down again" for more info. MarnetteD | Talk 02:14, 16 April 2013 (UTC)

Templates

I have this template that's for Work In Progress articles here: [30], and i'm trying to make it so that when you don't give it a proper date, it should give an error. How do i do this? RocketMaster (talk) 10:26, 13 April 2013 (UTC)

What do you mean "proper date"? What do you want the date to do? The way it currently is doesn't discriminate. Technical 13 (talk) 10:41, 13 April 2013 (UTC)

Surely it already does this: {{WIP|qwertyuiop}} should give a big red "Error: Invalid time". If you want a custom message, use {{#iferror:{{#time: m-d-Y|{{lc:{{{1|{{{date|yesterday}}}}}}}}}}|'''Error message here'''}}. — This, that and the other (talk) 12:14, 13 April 2013 (UTC)

tt-to, His template isn't on enwiki... Technical 13 (talk) 12:18, 13 April 2013 (UTC)

I can see that. What's wrong with what I posted? — This, that and the other (talk) 08:11, 15 April 2013 (UTC)

Also, I've visited his template and made some modifications for him that since there has been no further question here, I'm assuming took care of it. Technical 13 (talk) 16:06, 13 April 2013 (UTC)

Mediawiki, again

Hello from Latvian Wikipedia! I have a problem of finding the Mediawiki interface message for the message after succ. move for updating Wikidata page. Can somebody help? --Edgars2007 ^{(talk/contribs)} 13:57, 13 April 2013 (UTC)

Are you talking about MediaWiki:Movepage-moved? Or something else? —Theopolisme (talk) 13:59, 13 April 2013 (UTC)

No, that is easy founded :D I can't find message, which appears below this message that user should update the Wikidata page (if there is such). --Edgars2007 ^{(talk/contribs)} 14:03, 13 April 2013 (UTC)

Add ?uselang=qqx to the end of the URL (or &uselang=qqx if the URL already contains any parameters). That will show you the location of all the MediaWiki messages used on the page. It's an obscure but very handy feature. the wub "?!" 15:08, 14 April 2013 (UTC)

That wouldn't work here because it's a submit url with content depending on the action which lead there. The message is MediaWiki:Wikibase-after-page-move. PrimeHunter (talk) 23:41, 14 April 2013 (UTC)

Thanks, PrimeHunter! --Edgars2007 ^{(talk/contribs)} 15:47, 15 April 2013 (UTC)

Template help

I want to make a template that shows one message when given a 1, and another message when given a 0, and an error when i give it no parameters. How do i do this? Let's say, a site status template. I want it to say "This site is {{{status}}}", where status is either 0 or 1. And an error when given no parameters. How do i do this? RocketMaster (talk) 15:57, 13 April 2013 (UTC)

"This site is {{#switch:{{{status|}}}|0= Dead|1= Alive|#default= Error!}} Technical 13 (talk) 16:05, 13 April 2013 (UTC)

A minor variation to make a noticeable big red error, if I may:

...1=Alive|#default={{Error|Invalid argument "{{{status|}}}"}}}}

Also, change "status" to "{{mono|1}" if you just want it to use the first parameter value provided instead of having to use |status=. —[AlanM1(talk)]— 09:51, 15 April 2013 (UTC)

You're assuming he wants to put this template on Wikipedia Alan, as {{Error}} is a template available here. Alan's second point would be the preferred method in most cases if there is only going to be the one argument in the template as "This site is {{#switch:{{{1}}}|0= Dead|1= Alive|#default= Error!}} would work and not require |status= to be declared on every usage of the template. I only put it in there because the OP specified "This site is {{{status}}}", where status is either 0 or 1. And an error when given no parameters. Happy editing! Technical 13 (talk) 13:15, 15 April 2013 (UTC)

Clicking an image redlink brings me to the File Upload Wizard

Royal Flying Doctor Service of Australia#The service today, the second image down. I remember seeing a place that could tell me if the file was deleted and why, but I don't see that. MeekSaffron (talk) 19:56, 13 April 2013 (UTC)

It would be the deletion log although since it's an image, you need to try the one on commons as well. Anyway, the problem wasn't a deleted image, but that somebody had altered a hyphen-minus to an en-dash, thus breaking the image link, which has now been fixed. --Redrose64 (talk) 20:48, 13 April 2013 (UTC)

Given that the red link is almost useless, what's the minimum number of keystrokes to obtain the deletion log for a deleted image? I made a suggestion at WT:POPUPS last December and got nowhere. -- John of Reading (talk) 20:54, 13 April 2013 (UTC)

If you use the mouse to copy the redlinked filename, then in the left margin go for Special pages, then Logs; click in the "Target (title or User:username for user):" box and paste, then hit Go I tally a total number of keystrokes: 0.

I've found the edit which broke the file link. Quite surprised at who did that, really. --Redrose64 (talk) 22:14, 13 April 2013 (UTC)

Stuff like that will happen if you use the search-and-replace function of the toolbar to replace all hyphens. — Edokter (talk) — 22:27, 13 April 2013 (UTC)

wikEd strikes again... I'm glad I'm not the only one that does that (I try to catch and fix right away when I do). I've been meaning to go to the wikEd creators and ask for some fixes, like it should know not to add spaces after periods in URL strings and not capitalize parameter names inside templates... Just two more examples of things to look out for. Technical 13 (talk) 22:37, 13 April 2013 (UTC)

Oops! That was a bad search and replace (using an offline text editor), performed late at night when I probably shouldn't have been editing at all (I'd been up the night before with a gastro bug). Thanks for the fix, Edokter! Graham87 11:56, 14 April 2013 (UTC)

Or rather, that was the night I had just returned home from the holiday where I probably caught the bug and by the time I'd made the offending edit ... I don't even want to talk about it. Graham87 14:53, 14 April 2013 (UTC)

Note that the endashes should have been preceded by non-breaking spaces ( ), which I've fixed per WP:ENDASH (actually, that page has either been broken or it was somewhere else and, inconsistently, not there

. No wonder it's wrong in so many places. (edit)The &nbsp is mentioned at WP:MOSNUM poorly and I've asked about it at Talk...).

It would be cool if WikEd had a "hide links" mode (like the "hide refs and templates" mode) to facilitate fixes that should only apply to actual prose. It would make the global search/replace usable, or the one-at-a-time version get far fewer hits, especially when looking for bad hyphens. —[AlanM1(talk)]— 09:17, 15 April 2013 (UTC)

This has nothing to do with the type of space which goes before a dash (but since you mention it, changing spaces within filenames to non-breaking spaces would be a bad move). It's about how to access the deletion log for a file, which we determined hadn't been deleted at all but was redlinked as the result of an incorrect copyedit of a hyphen-minus to an en-dash, for which the person responsible has apologised. --Redrose64 (talk) 11:06, 15 April 2013 (UTC)

Chopping parameters from templates

{{Marriage}} just survived TFD via a no-consensus close; the delete voters were upset that it was heavily bloated with unnecessary components, while most of the keep voters (including me) wanted to keep it but thought we should remove the unnecessary components. I'd like to start removing some of the unnecessary components, but I fear damaging pages where those components are used — is there any way to figure out (automatically) which components are used on which pages? The template is transcluded on approximately 2,500 pages, so I'm not about to check all of them as long as there's any other way to do it. Nyttend (talk) 00:42, 15 April 2013 (UTC)

There are, depending on what kind of information you need, you can also add tracking categories to find what pages use which params. Werieth (talk) 00:44, 15 April 2013 (UTC)

I've used tracking categories before as Werieth suggests, and I want to let you know if you haven't done it before it is something that has to go through the job cue and "may" take a couple of days to show accurate results. If you are patient enough, it is the easiest way to do it in my opinion. Technical 13 (talk) 01:17, 15 April 2013 (UTC)

Please look at this revision of the Sandbox, plus its code — I've filled out all of the parameters, but nothing appears except the name of the person who's gotten married, the date of the marriage, the fact that it ended because of divorce, and the date of the divorce. Nothing about the location or the name of the spouse appears; is this because it's not designed to appear, or because I did something wrong? The main reason I came here is that I didn't want to remove text that this template was putting on pages, but if things like the location of the wedding ceremony aren't displayed in articles, we might as well begin discussions immediately about removing the parameters, without worrying about what pages use those parameters. Nyttend (talk) 01:31, 15 April 2013 (UTC)

okay, I'll set up the tracking categories and a control board page table to morrow. I'm in bed now and can't do it from my BlackBerry. (buttons too small and I can't see screen because of some odd bug that makes the line I am typing one line off the bottom of the screen.) :( Technical 13 (talk) 01:56, 15 April 2013 (UTC)

In you sandbox, the following large block appears in the HTML:

«start: (2000-01-02)–end+1: (2014-01-04)»"Marriage: Mrs. Booth to John Wilkes Booth" Location:11, 22, 33, place1, <a rel="nofollow" class="external text" href="//toolserver.org/~geohack/geohack.php?pagename=Wikipedia:Sandbox&params=5.000_N_5.000_E_scale:1000&title=January+2%2C+2000%3A+Marriage%3A+Mrs.+Booth+to+John+Wilkes+Booth">5°00′00″N 5°00′00″E / 5.000°N 5.000°E / 5.000; 5.000 (January 2, 2000: Marriage: Mrs. Booth to John Wilkes Booth)</a> (linkback://en.wikipedia.org/wiki/Wikipedia:Sandbox)

I assume someone wants / expects it to be there even though it is intentionally not visible to the user. As to why it exists? I don't know. I've never followed this template / issue. Dragons flight (talk) 02:09, 15 April 2013 (UTC)

Basically, the template is meant to emit some microformats, but t's not doing them properly, and nobody at the TFD appeared to want the template to continue trying to emit them. That will explain why it has those bits without displaying them — my concern was that I'd made a mistake preventing the template from understanding what I'd put in, and obviously that's not the case; thanks. Nyttend (talk) 02:19, 15 April 2013 (UTC)

Before digging into it, who says it's not doing them properly? Was it someone familiar with microformats, and was there something specific? At least one of the arguments against said that it didn't emit them at all (and was made by someone who I expected to know something about it (I don't)), which does not appear to be the case if I understand the concept correctly. If you look at the actual HTML that came out, do you see all the parameter values that you supplied, or are there some still missing? If so, they could still be conditional on other values, so we really need someone to look at the code to see what it's supposed to do. —[AlanM1(talk)]— 06:51, 15 April 2013 (UTC)

Does someone have some "pretty-printing" code for WikiML that will bring in any transclusions and then insert newlines and tabs to show the hierarchy properly? —[AlanM1(talk)]— 06:55, 15 April 2013 (UTC)

Pigsonthewing is far more familiar with microformatting than anyone else here, and at the TFD he was the one who brought up the problems with this template's microformatting. And I'm not quite sure what you mean in your second paragraph, about WikiML; could you elaborate? Nyttend (talk) 12:14, 15 April 2013 (UTC)

Indeed. The template has a load of bloat, but does nothing useful, that plain text cannot do. See my various posts on its talk page. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:39, 15 April 2013 (UTC)

You say you don't know anything about microformats, yet elsewhere argue for this retention because of the microformats to you say it emits; when it does not. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:39, 15 April 2013 (UTC)

Invitation to today's bugday (16:00 - 20:00UTC) about old bug reports

Hi everybody, please join us on the next Wikimedia Bugday on Monday, April 15th, 16:00-20:00 UTC in #wikimedia-office on Freenode IRC. We are going to take a look at old reports, trying to reproduce some plus provide feedback. Everyone is welcome to join, and no technical knowledge needed! It's a nice and easy way to get involved in the community or to give something back. Join IRC, say hello and that you're here for the bugday, and have fun. More information can be found here and for more information on Triaging in general, check this link. We look forward to seeing you! --AKlapper (WMF) (talk) 13:46, 15 April 2013 (UTC)

CSS

I'm the developer of the WP:AFC helper tool and in our actual beta version are many new improvements. Would some "designer" who is experienced in CSS style our helper tool a bit more "friendly"?

Simply import the beta version of the tool to you skin.js and click on "review" when visiting WT:AFC/sand. There is a table which needs to be sytled and button etc. There are no design guides or wishes, simply try it out and feel free to redisgn it. Thanks. mabdul 13:59, 15 April 2013 (UTC)

looking... and waiting for inspiration. —Theopolisme (talk) 15:51, 15 April 2013 (UTC)

Hey, I forwarded your request on to the design mailing list. Either WMF or volunteer designers may be able to lend you some CSS for button styles etc. Steven Walling (WMF) • talk 02:28, 16 April 2013 (UTC)

Oh great. I subscribed the list. mabdul 05:52, 16 April 2013 (UTC)

Watchlist loading time

This is a little rare for me in this area. I just want to praise the Wikimedia team as it seems recently Watchlist loading times have reduced, thereby not taking up as much memory on the computer. Unless I'm wrong. Simply south...... eating shoes for just 7 years 16:20, 15 April 2013 (UTC)

Template:Dynamic list

For some reason, Template:Dynamic list is leaving a double white space above where the template is placed on a list. See: diff. Could someone check the code to why this is happening? Thanks much, --Funandtrvl (talk) 19:10, 15 April 2013 (UTC)

I fixed it by changing the invocation of dynamic list to use the dablink class. I suspect the reason it was having a problem was because it was opening a definition list using the colon and then opening a div right after. The if-cases might not have been helping but am unsure. --Izno (talk) 19:40, 15 April 2013 (UTC)

TYVM! I did think it had something to do with the use of the colon. --Funandtrvl (talk) 19:42, 15 April 2013 (UTC)

Redlink to AfD debate

Hi, can anyone see what's going on here? A Wolf At Your Door Records, I sent it to AfD but the link in the notice at the top of the page is red, saying "page does not exist" on mouseover, if you click it it takes you to the right page. Thanks. CaptainScreebo ^Parley! 19:59, 15 April 2013 (UTC)

WP:PURGE Werieth (talk) 20:06, 15 April 2013 (UTC)

This is frequent with Twinkle nominations. See for example Wikipedia:Village pump (technical)/Archive 108#Link to AfD discussion in template incorrectly displaying as redlink. PrimeHunter (talk) 20:09, 15 April 2013 (UTC)

I had never come across this before, thanks for your replies, sorted. CaptainScreebo ^Parley! 20:48, 15 April 2013 (UTC)

Sudden drop in pageviews

The article List of best-selling books is (or was) one of our most-viewed articles, with about 5000 views per day. E.g. June 2011, January 2012, September 2012. Suddenly, late in March 2013, it dropped to about 1,000 a day, or less than 1/4th of the previous number[31][32]. The article had no huge changes at that time, and other articles don't seem to show the same drop. Anyone has any clue as to what might have caused this? No idea if it is a technical problem or not, posting it here seemed the most appropriate. Fram (talk) 08:35, 10 April 2013 (UTC)

There is not any clue in Google searh either! --Tito Dutta (contact) 08:42, 10 April 2013 (UTC)

Indeed, I can't find a single reason for this drop (it seems unlikely that this was consistently wrong for months and months, and now suddenly has been corrected, but it is a possibility of course). Fram (talk) 09:52, 10 April 2013 (UTC)

I wonder if it's either (a) something to do with Wikidata, or (b) that something else was redirecting to the page and no longer is? Black Kite (talk) 10:20, 10 April 2013 (UTC)

Spring fever? Everyone deciding to go outside to enjoy the beautiful weather inside of looking for a book to read? I wouldn't be concerned with a change in such a stat for one month. If it persists, then I would suspect something is going on and hunt for it.

from a broke bloke... Technical 13 (talk) 10:42, 10 April 2013 (UTC)

Perhaps someone deleted a link from a highly visited page, like Book. --NaBUru38 (talk) 16:03, 11 April 2013 (UTC)

It's a possibility, so far I haven't found the obvious culprit though. Fram (talk) 09:44, 13 April 2013 (UTC)

The same thing happened, at about the same time, to the Parabola article, with the daily number of reported views dropping from about 2500 to about 600 on weekdays (both fewer on weekends). Other related articles such as Hyperbola and Catenary have not been affected. I think it's highly unlikely that this drop is real, especially since the same has happened elsewhere. There's a bug somewhere. DOwenWilliams (talk) 20:15, 14 April 2013 (UTC)

Perhaps reduce Google link of https likely scaring viewers: The result in Google Search for "best selling books", although still top 1-4, where most articles have "http:" link in Google, now has a bizarre secure-server link using "https:" prefix, which requires a scary, user-browser confirmation to allow many browsers to risk seeing that webpage ("List of best-selling books") as requiring special permission. Similarly, Google has listed "Parabola" with "https" likely scaring potential readers. If users say no, then the webpage is not visited, and the stats.grok.se will not log a pageview. Although sad to see people frightened away from Wikipedia pages by "https:" confirmation messages, this is a good case to see the effects of listing a page in Google with "https:" versus the normal, non-scary "http:" protocol. I cannot remember how to rerank the http version as higher in Google-Search results, to upstage the "https" version. Just too tired tonight. -Wikid77 (talk) 03:16/03:21, 15 April 2013 (UTC)

I can confirm that searching in Google for "List of best-selling books" you get as first result a link to the https version, however people do not get a scary warning. That only happens when the certificate is invalid and that is not the case so I don't think this explains the drop in pageviews. Drdee (talk) 03:24, 15 April 2013 (UTC)

Secure-server warnings depend on browsers: Users who run the high-tech Internet Explorer (IE6 to IE10) might have builtin warnings to check the security certificate of the website, while older, low-tech versions of Firefox or Google Chrome (or other unsafe browsers) might bypass security checking and allow access by any website (such as malware). Many users of MSIE learned to always answer "no" to security warnings, because the auto-installed malware was so difficult to remove, and that is why years of scary "https" messages might be keeping WP readers away from those 2 https articles ("Parabola" and "List of best-sell..."), while no reduction in pageviews for the related, but http-protocol articles "Hyperbola" and "Catenary". However, it is ironic to use a high-security browser to read an article about the best-selling literary classics South Beach Diet and Captain Underpants. Perhaps the viewer-warnings from Internet Explorer are even smarter than we thought (just kidding!). Anyway, we should rerank the http versions in Google, as higher, rather than https. A redirect to an article will typically rank lower. -Wikid77 (talk) 04:21, 15 April 2013 (UTC)

Note that the hit counter gives you the results for a precise URL, except that it's not case-sensitive. As a result, alternate-name redirects are counted separately; if everyone decided to search for "Unites States" and didn't want to search for "United States", the hit counts on the United would plummet at the same time as hit counts for Unites were skyrocketing. Nyttend (talk) 04:42, 15 April 2013 (UTC)

Security certificate. I'd be surprised if Internet Explorer even knew what those were. I have put together plenty of web pages where IE7 complains "To help protect your security, Internet Explorer has restricted the webpage from running scripts or ActiveX controls that could access your computer". Here is an example of a page which IE7 rejects:

<HTML>
  <HEAD>
    <TITLE>Test Page</TITLE>
  </HEAD>
  <BODY>
    <H1>Test Page</H1>
    <A HREF="elsewhere.htm">Elsewhere</A>
  </BODY>
</HTML>

Now, ignoring the absence of a <!DOCTYPE>, that conforms to HTML 2.0 and all subsequent versions. It's a page I wrote myself, so I know what went into it: there are no scripts or images, it doesn't link to any other site, and uses no CSS at all. I have no idea what an "ActiveX control" is, so I don't know if that little page has one or not. So what's the problem? IE is, if anything, over-sensitive to the point where the message appears so often that users become used to seeing it so much that they ignore it - like me. --Redrose64 (talk) 09:49, 15 April 2013 (UTC)

Explanation: On March 25th, the Analytics Team removed SSL traffic from the udp2log stream of webrequests. This webrequest stream is consumed by webstatscollector, the tool that generates the data that is presented by stats.grok.se. The reason we removed SSL traffic was twofold:

Each logline is tagged with a unique number that allows us to see how much loglines we lose (aka packetloss); this numbering system was not working for SSL traffic and hence our packetloss monitoring was inadequate. You can see a nice drop in packetloss reporting as a result of this fix.
SSL traffic actually generates two hits in our log files, once when it hits the SSL server (nginx) and the second time when it hits the cache server (squid). Webstatscollector was not deduplicating these numbers and so actually the drop in pageviews that we are seeing means that we have gone back to the actual pageview count.

So removing SSL traffic from the main webrequest stream was the cause of this drop but it did not introduce a bug, it actually fixed an unknown bug of overreporting SSL generated pageviews. Thanks to Wikid77 who got me thinking about the SSL cause in the first place.

Potential Next Steps:
- WMF only recently started to enable SSL traffic so it would be interesting to see if we can find a sudden spike in pageviews for this article.
- We can try to get Google link to the non-SSL page; it should not impact the numbers anymore but WMF's infrastructure is not quite ready for handling massive volumes of SSL traffic for anonymous readers.

My name is Diederik and I am the Product Manager of the Analytics Team @ WMF. Drdee (talk) 14:49, 15 April 2013 (UTC)

- - Ah, so now we get the correct (or a more correct) number, and the earlier one was inflated? That's a bit disappointing (as I am one of the main contributors to that page :-) ), but obviously an improvement. Thanks for the explanation! Fram (talk) 15:03, 15 April 2013 (UTC)
  - I don't buy this explanation. If duplication of SSL hits was really the problem, then upon fixing it the pageviews would have dropped by AT MOST half (assuming everyone was using https). However, Fram's original post reports pageviews suddenly dropping from 5,000 to 1,000, which is a reduction of well over half. So something else is going on. jcgoble3 (talk) 17:42, 15 April 2013 (UTC)

I'm a bit sceptical, too. Diederik says that the SSL change was made on March 25, but the sudden drop in reported pageviews of the Parabola page didn't happen until March 29. And, like the page that Fran reported, the drop was more than 75% DOwenWilliams (talk) 19:21, 15 April 2013 (UTC)

As I read Drdee, all SSL secure-server pageviews are ignored in the counts, so all pageviews from Google would be omitted in the stats.grok.se counts, but Bing.com uses non-SSL "http:" prefix, so all pageviews from Bing will be counted. -Wikid77 (talk) 21:16, 15 April 2013 (UTC)

No that's not correct, SSL hits were generating two hits in our logs and we removed one of them so we would still be counting page requests but just not be overcounting. Obviously, we would not throw away request :). Second, it is likely that multiple causes are playing a role to fully explain the drop in pageview requests. I did not mean to say that the SSL explanation accounts for 100% of the drop (as it does not) but it definitely explains a big portion of the drop. Drdee (talk) 01:06, 17 April 2013 (UTC)

Well, I am relieved somewhat to find this discussion. I had developed Gone with the Wind (film) during March and put it through a GA review, only for the page views to drop from 4000 per day to 1500, much to my amazement. I was bit surprised to say the least, that my development of the article had resulted in the article losing two thirds of its audience! But I do have a question: if it were a technical glitch, wouldn't its correction have resulted in an immediate drop, since the stats for GWTW show a steady decline: [33]? Betty Logan (talk) 20:14, 15 April 2013 (UTC)

I am thinking that the change was applied to each of the related servers over a period of days, before most servers were all omitting "https:" pageviews from the udp2log stream of webrequests. BTW: Thanks for the GWTW update: that was my personal measurement for when "Wikipedia has arrived" as a quality cultural encyclopedia. We had articles about 60 Mozart symphonies, but GWTW was the canary (with tuberculosis) in the coalmine. Well done. -Wikid77 (talk) 21:16, 15 April 2013 (UTC)

That Gone with the Wind (film) page is interesting. It had that huge spike in reported views in mid-February. I wonder what could have caused that. I'm getting the feeling that we shouldn't trust these data at all. DOwenWilliams (talk) 00:54, 16 April 2013 (UTC)

Spikes on film articles often coincide with a TV airing. I couldn't say if that particular spike came from a broadcast, but the comparable Christmas Day spike definitely did. Betty Logan (talk) 03:26, 16 April 2013 (UTC)

Displaying citation comments - worth a preference option?

I just heard about the following CSS, which displays errors in references where a citation template has been used incorrectly:

.citation-comment { display: inline !important; color: red; }

I can hardly believe that I've never heard of this before, because it's extremely useful - now I'm seeing errors all over the place that need fixing. Should this be a preference option? — Hex (❝?!❞) 10:44, 15 April 2013 (UTC)

It's relatively new, and has only come in as and when each of the Citation Style 1 templates has been moved to invoke Module:citation/CS1 instead of transcluding {{citation/core}} --Redrose64 (talk) 11:32, 15 April 2013 (UTC)

Right now CS1 catches, categorizes, and reports some 19 error types. CS1 is displaying some of them now without the css. The plan is to eventually display all of the error reports at which that css will no longer be require though, no doubt, there will be editors who will want a way of turning the error messages off.

There are two versions of the css. The other lacks the color: red; line.

—Trappist the monk (talk) 12:01, 15 April 2013 (UTC)

We are turning on the errors in chunks. We are monitoring the Help Desk and elsewhere to see what sort of issues are popping up. Eventually all errors will be visible. See more at Module talk:Citation/CS1/Updates. -- Gadget850 (Ed)^talk 12:16, 15 April 2013 (UTC)

And Help:CS1 errors. —TheDJ (talk • contribs) 14:12, 15 April 2013 (UTC)

Aha, so it's a new development. Great, thanks all. — Hex (❝?!❞) 09:35, 16 April 2013 (UTC)

Showing red messages to see if fixed before auto-correction: Currently, several messages about misspelled or unknown parameters are being shown to all readers, in more than 40,000 articles, to see if people will fix them, while Bots are being modified to edit some obvious error patterns. However, we are also discussing more auto-correction of simple errors (because Lua is so fast, it can do cartwheels and handstands faster than markup processes #switch). For example, a parameter might be misspelled, such as "accessdate" by "accesdate" with one "s" but treated as if correct. Also, we might find 5,000 pages which omit "url=" from "url=http:..." and then auto-correct to treat the URL as if properly coded (with "url=" prefix), then append a short, subtle reminder: ^{"[fix cite] "} rather than show a red-error message because the auto-corrected cites would function identically to the corrected parameters. Meanwhile, monthly runs of a cite-Bot might edit-fix those types of simple problems in the 5,000 articles some weeks later. But no glaring red-error messages for a trivial auto-fixed issue. Formerly, the markup-based cites were too slow to even consider some simple auto-correction of misspelled parameters (merely ignored), as running only 14 per second, while now, the Lua-based cites run about 180 per second, or 13x times faster, even running spell-check to report over 177 misspelled parameters (by hash-name lookup) and auto-correct a few issues, such as singular "pages=34" now shows "p. 34" rather than incorrect "pp. 34". -Wikid77 (talk) 16:32, 15 April 2013 (UTC)

Exciting stuff. — Hex (❝?!❞) 09:35, 16 April 2013 (UTC)

upload.py and flickrripper.py

Hello, I often have mistakes when using these scripts.

1) Some images can't be loaded by means of upload.py

Extended content
C:\Python26\pywikipedia>upload.py No input filename given File or URL where image is now: http://epubbooks.ru/pics/cover751277.jpg Reading file http://epubbooks.ru/pics/cover751277.jpg The filename on the target wiki will default to: cover751277.jpg Enter a better name, or press enter to accept: Новая_энциклопедия_бодибилдинга.j pg The suggested description is: Do you want to change this description? ([y]es, [N]o) y Uploading file to wikipedia:ru via API.... {u'servedby': u'mw1118', u'error': {u'info': u'Permission denied', u'code': u'pe rmissiondenied'}} C:\Python26\pywikipedia>

Extended content

C:\Python26\pywikipedia>upload.py
No input filename given
File or URL where image is now: http://epubbooks.ru/pics/cover751277.jpg
Reading file http://epubbooks.ru/pics/cover751277.jpg
The filename on the target wiki will default to: cover751277.jpg
Enter a better name, or press enter to accept: Новая_энциклопедия_бодибилдинга.j
pg
The suggested description is:

Do you want to change this description? ([y]es, [N]o) y
Uploading file to wikipedia:ru via API....
{u'servedby': u'mw1118', u'error': {u'info': u'Permission denied', u'code': u'pe
rmissiondenied'}}

C:\Python26\pywikipedia>

2) When using flickrripper.py of the image are sometimes passed, and the mistake sometimes is given. I don't know, in what the reason. Photos exist and they are laid out under free licenses.

Extended content
C:\Python26\pywikipedia>flickrripper.py -user_id:85931598@N02 -removecategories -addcategory What category do you want to add? Animal rights Finished running Total photos: 0 Uploaded photos: 0

and

Extended content
C:\Python26\pywikipedia>flickrripper.py -user_id:34882043@N03 -removecategories -addcategory What category do you want to add? Animal rights 3683231774 Reading file http://farm3.staticflickr.com/2648/3683231774_31c047808d_b.jpg The suggested description is: === {{int:filedesc}} === {{Information \|Description=Uploaded by [http://www.flishr.com Flishr V2.01] \|Source=[http://www.flickr.com/photos/34882043@N03/3683231774/ KW Vegetarians Ac tion Shots] \|Date=2009-06-27 11:35 \|Author=[http://www.flickr.com/people/34882043@N03 Create For Animal Rights] \|Permission= \|other_versions= }} {{subst:unc}} [[Category:Animal rights]] Uploading file to commons:commons via API.... {u'servedby': u'mw1115', u'error': {u'info': u'The file you submitted was empty' , u'code': u'empty-file'}} 3682419489 Reading file http://farm4.staticflickr.com/3656/3682419489_77b8aa5016_z.jpg?zz=1 The suggested description is: === {{int:filedesc}} === {{Information \|Description=Uploaded by [http://www.flishr.com Flishr V2.01] \|Source=[http://www.flickr.com/photos/34882043@N03/3682419489/ KW Vegetarians Ac tion Shots] \|Date=2009-07-02 16:57 \|Author=[http://www.flickr.com/people/34882043@N03 Create For Animal Rights] \|Permission= \|other_versions= }} {{subst:unc}} [[Category:Animal rights]] Uploading file to commons:commons via API.... Upload successful. 3683230852

Extended content

C:\Python26\pywikipedia>flickrripper.py -user_id:34882043@N03 -removecategories
-addcategory
What category do you want to add? Animal rights
3683231774
Reading file http://farm3.staticflickr.com/2648/3683231774_31c047808d_b.jpg
The suggested description is:
=== {{int:filedesc}} ===

{{Information
|Description=Uploaded by [http://www.flishr.com Flishr V2.01]
|Source=[http://www.flickr.com/photos/34882043@N03/3683231774/ KW Vegetarians Ac
tion Shots]
|Date=2009-06-27 11:35
|Author=[http://www.flickr.com/people/34882043@N03 Create For Animal Rights]
|Permission=
|other_versions=
}}

{{subst:unc}}
[[Category:Animal rights]]

Uploading file to commons:commons via API....
{u'servedby': u'mw1115', u'error': {u'info': u'The file you submitted was empty'
, u'code': u'empty-file'}}
3682419489
Reading file http://farm4.staticflickr.com/3656/3682419489_77b8aa5016_z.jpg?zz=1

The suggested description is:
=== {{int:filedesc}} ===

{{Information
|Description=Uploaded by [http://www.flishr.com Flishr V2.01]
|Source=[http://www.flickr.com/photos/34882043@N03/3682419489/ KW Vegetarians Ac
tion Shots]
|Date=2009-07-02 16:57
|Author=[http://www.flickr.com/people/34882043@N03 Create For Animal Rights]
|Permission=
|other_versions=
}}

{{subst:unc}}
[[Category:Animal rights]]

Uploading file to commons:commons via API....
Upload successful.
3683230852

Somebody uses these bots? In what my mistakes or, probably, mistakes consist in a code of bots? Thanks.--Парис "Анима" надаль (talk) 16:35, 15 April 2013 (UTC)

Slow

It happens only to me or english version of wikipedia loading pretty hard? --Goldask2 (talk) 18:28, 15 April 2013 (UTC)

I've noticed it at least since this past weekend, to the point of it being a pain in the neck. Almost everything now takes longer, and my browser is continually showing the "connecting" message for a few seconds before anything happens. Only happens on Wikipedia. Until recently, I never saw that message before for anything. Windows XP, Firefox 20.0.1— Maile (talk) 11:44, 16 April 2013 (UTC)

Template:Afc decline

I need help: 29611670.x (talk · contribs) reported that AfC submissions beginning with a whitespace (.e.g Wikipedia talk:Articles for creation/ Back to the Italian) are problematic: When such submissions get declined, normally the user gets {{subst:afc decline|1= Back to the Italian|sig=yes}}, but the result is that the whitespace was removed as can be seen at [34]. So how to fiox the template? (The script adds the whitespace correctly; tested in WP:SANDBOX) mabdul 07:19, 16 April 2013 (UTC)

If you're using a script, either have it replace all spaces with _ or have it replace a leading space with {{Sp}}> Technical 13 (talk) 11:07, 16 April 2013 (UTC)

That should work. thanks. mabdul 11:49, 16 April 2013 (UTC)

An alternate approach if you're interested: Place <nowiki /> immediately after the = sign. It prevents the spaces that follow it from being stripped, and has no effect if not followed by spaces. – PartTimeGnome (talk | contribs) 21:10, 16 April 2013 (UTC)

"Content" as section header

For those who may remember this recent conversation, Wikipedia:Village pump (technical)/Archive 109#Content, I just upgraded to IE10, and the issue has not been fixed. IE10 still takes the user to the html "content" expression near the top of a page. I anchored an old header, "Content", at the newly named Encyclopedic content section on the "WP:NOT" page:

Wikipedia:What Wikipedia is not#Content

That anchor serves as a test link for this issue. – PAINE ELLSWORTH ^CLIMAX! 21:59, 16 April 2013 (UTC)

First thing I recommend doing is to tell people not to use "Content" as a section heading in WP:ACCESSIBILITY#Headings, Help:Section, and MOS:HEAD to start. Technical 13 (talk) 22:12, 16 April 2013 (UTC)
Excellent idea, Technical 13! I'm strapped for time right now, so if you want to make the changes yourself, the issue is described in detail at the VPT archive link I gave above. Thank you very much! – PAINE ELLSWORTH ^CLIMAX! 02:33, 17 April 2013 (UTC)

Template tweak needed

I want to use {{Year nav topic3}} for List of protected areas established in 2013 and the rest of the series. To make the template usable the link "subject by year" at the top of the template should start with a capital letter. If the "subject" parameter is entered as a capital letter it will work but then all the years will have the incorrect capitalisation of the form "List of Protected areas in [year]". Can the line [[{{{2|subject}}} by year]] in the template be modified so that "subject" is returned as "Subject".? -- Alan Liefting (talk - contribs) 22:52, 16 April 2013 (UTC)

Done.[35] PrimeHunter (talk) 23:16, 16 April 2013 (UTC)

I thought it might be an easy job for you byte boffins.

Thanks for that. -- Alan Liefting (talk - contribs) 23:19, 16 April 2013 (UTC)

Image displayed on black background

Tracked in Phabricator
Task T49226

File:SOAS Crest.jpg Why do the arms of SOAS display on a black background? It makes it virtually impossible to see. Other colleges of London University appear normal. Justlettersandnumbers (talk) 20:56, 14 April 2013 (UTC)

It doesn't look black to me. You don't happen to have "Use a black background with green text on the Monobook skin" checked on Special:Preferences#mw-prefsection-gadgets and are using the Monobook skin, are you? Technical 13 (talk) 21:27, 14 April 2013 (UTC)

It looks black for me as well, but solely when used with |thumb -- when I view it on the description page, everything looks fine. —Theopolisme (talk) 22:09, 14 April 2013 (UTC)

What about the thumb lower down on that same page? I too see the full-size image in its proper colours. I've asked on commons if I can upload a screenshot of this (non-free) image for discussion purposes; relieved to hear that it isn't just me that's affected. Justlettersandnumbers (talk) 22:30, 14 April 2013 (UTC)

Do any other images do this with |thumb set? What browser are you guys using? I can't make the background black for the image in Firefox 19.0.2 Technical 13 (talk) 23:06, 14 April 2013 (UTC)

Me: Safari on a Mac (Mozilla/5.0 (Macintosh; Intel Mac OS X 10_6_8) AppleWebKit/534.57.2 (KHTML, like Gecko) Version/5.1.7 Safari/534.57.2). —Theopolisme (talk) 23:08, 14 April 2013 (UTC)

Yep, it's a Safari thing (I have the same browser and OS version as Theo); loading this page in Firefox makes the image display correctly. Off to Bugzilla ... Justlettersandnumbers (talk) 23:35, 14 April 2013 (UTC)

This is a remnant of bugzilla:24854. Purging the file page should fix the old thumbnails of it. —TheDJ (talk • contribs) 09:27, 15 April 2013 (UTC)

Thank you. I've purged both the file page and the article page, first by making null edits, then with ?action=purge. The thumbnail still displays on a black ground on the SOAS article page. The image file has been edited by AnonMoos, for which many thanks, and the thumbnail of the new version displays correctly on the file page. 24854? I wouldn't know. But that is fixed, and this is not. Justlettersandnumbers (talk) 21:56, 15 April 2013 (UTC)

Your browser may still have the incorrect image cached, in which case you will need to clear your browser's cache. --Redrose64 (talk) 10:39, 16 April 2013 (UTC)

Yes, I should have mentioned that I did that several times also. Anyway, thanks to the work of AnonMoos, the problem is now resolved on the article page. I'll leave the bug report open as the thumbnail of the old version of the image still displays incorrectly on the file page, and maybe someone cleverer than I can learn something from that. Thanks to all who helped to resolve this, Justlettersandnumbers (talk) 21:44, 17 April 2013 (UTC)

Small bit of code being added to a few articles

A new user (User:Sgccgs, but also as an IP before they registered) has started adding a bit of code to some pages, centered around the "printability" of templates and wikitables. They have been reverted several times, their edit summaries are a little insulting and borderline incomprehensible to those of us who don't write markup code("WARNING: Article could not be rendered - ouputting plain text. - When You put a table.. learn how to do it, and try save as PDF after you can save that thank you! PleaseDon'tUseUndoIf YouDon't Know What IS The Propouse Of That Div!") and they haven't yet gotten the hang of using the talkpages. If the issue they raise has merit, shouldn't this be a site wide fix to our wiki code and not just adding a small bit of code to every article that uses tables and templates? I wasn't sure where else to bring this. If there is somewhere more appropriate, please point me in the right direction. He iro 19:44, 15 April 2013 (UTC)

Correct. The user should be informed to stop and discussion should continue here (since you've raised it here). If the user in question does not stop, then that is grounds for a block probably. --Izno (talk) 19:56, 15 April 2013 (UTC)

I've instructed them to come here already, but they haven't been all that communicative so far, and have simply reverted anyone who removes their addition. They have added this to around 10 articles so far. Something to do with exportibng pdfs so they can use it for a book. He iro 19:59, 15 April 2013 (UTC)

I've reverted the 4 most recent ones (and one back when they were just an IP) and left a warning on there talk page followed by a THinvite. I hope they do go somewhere to discuss it with someone. Technical 13 (talk) 20:11, 15 April 2013 (UTC)

They have posted several times on my talk page. See User talk:Rmhermen#Broken tables. I don't seem to be explaining well enough, though. Rmhermen (talk) 00:50, 17 April 2013 (UTC)

You can't understand... because is very hard... Firts problem: when the table is too wide you can't put that in a paper dimension (A4) Or (A5) or printing with pdf because the soft (mwlib toolkit) can't render the article with that table. Second Problem: When the code for the table is too large, don't work because the render don't work with large code for tables. Here is and example: if someone... are creating a book with this category ( http://en.wikipedia.org/wiki/Category:Ancient_peoples) and inside are this article (http://en.wikipedia.org/wiki/List_of_archaeological_periods_(North_America) they can't. Because that article have a very large table. And They never know why they can't create a book with this category. Please try make a book with the all articles, and try create a book without the article. In this case you know what article have the error. But if you don't know you have to search one by one and find the article with error. And anyway... you can view the article online.... and still view the article offline in pdf.. without the table. What is better have an article with errors .. or have a article without errors in pdf?--Sgccgs (talk) 00:28, 17 April 2013 (UTC)

Group (periodic table) which contains one of the most complex tables I know of on Wikipedia works fine as a PDF so I don't think the issues is merely a large or complicated table. Rmhermen (talk) 17:30, 17 April 2013 (UTC)

I suspect it has to do with poorly formed tables. Tables that use (row|col)span and don't make sure to account for all of the cells in the table. I've been researching it since this morning, and am sure it will take me quite a while longer. Technical 13 (talk) 19:24, 17 April 2013 (UTC)

And they are back at it again, adding their bit of code to a template. He iro 21:18, 17 April 2013 (UTC)

Menus in Wikipedia app

Hi all,

The Wikipedia app should have more menus, specifically the top menu (read, edit, history, view talk, search) and the nav menu on the left side. That way it's like viewing the desktop site except on phone. All of these menus would be collapsible, especially the left menu, which would pop out with a gesture to the right as in the Facebook app. The top menu would be scrollable.

You should also be able to open editing for a section of a page by tapping and holding.

Thanks, 68.173.113.106 (talk) 20:19, 16 April 2013 (UTC)

The web only version of the website already behaves as such. It is reasonable to expect that the App will follow this same approach in the future, but there are only a few people working on the native apps, so it might take some time. —TheDJ (talk • contribs) 09:46, 17 April 2013 (UTC)

blocking sockpuppets

When a user is blocked, later shows up as a sockpuppet, which gets blocked, shows up as another sockpuppet, gets blocked, etc - is there a way to stop that cycle? Bubba73 ^{You talkin' to me?} 04:00, 17 April 2013 (UTC)

Eventually, in rare cases where the problem really gets serious enough, the IP address of the user can and will be forced out and blocked by a CheckUser. -- King of ♥ ♦ ♣ ♠ 04:02, 17 April 2013 (UTC)

In the meantime, they can create another account in second whereas it takes good editors perhaps an hour or more of combined effort to get it blocked. :-( Bubba73 ^{You talkin' to me?} 22:25, 17 April 2013 (UTC)

UTRS oddness

In the past couple of days, UTRS has received a large number of IP unblock requests (23, at the last count) which are listed as coming from User:185.15.59.204. They appear, from context, to come from a wide range of different users in different locations, and at least one of the users affected was presented with an autoblock message for a range which 185.15.59.204 does not fall within. Given that 185.15.59.204 is registered to the Wikimedia Foundation, I suspect that some sort of glitch has arisen which is causing UTRS tickets to list the WMF IP instead of their actual IP (as far as I can tell, there've been no recent anon UTRS requests that haven't originated from 185.15.59.204). I know the toolserver's been a bit squiffy this week, which may be related. Anyone with UTRS/toolserver access and a technical brain fancy taking a look? Yunshui 雲‍水 09:40, 17 April 2013 (UTC)

I also experienced this issue, which I reported at Wikipedia:Administrators' noticeboard/IncidentArchive792#185.15.59.204, but no one was able to come up with a good answer. I'm merely a tool admin (which gives me a few extra buttons), not a dev, so I can't see the inner workings of the tool. -- King of ♥ ♦ ♣ ♠ 09:54, 17 April 2013 (UTC)

By the way, I've created the template "Need Block Information (Corrupt IP)" for asking requesters from IP addresses to restate their IP address. -- King of ♥ ♦ ♣ ♠ 09:55, 17 April 2013 (UTC)

DeltaQuad knows more about this, but it's something about the toolserver is behind some kind of proxy that forwards the packets. We have code in there that is supposed to recognize it and use the HTTP X FORWARDED FOR address instead, but perhaps it has recently changed and we need to update it. I'll check github to see what's in there.--v/r - T P 12:29, 17 April 2013 (UTC)

I have updated it with the new IP address on Github with this diff. I'll ask User:Crazycomputers to pull down to the live site immediately.--v/r - T P 12:36, 17 April 2013 (UTC)

Done. I cherry-picked your change onto the live branch and redeployed it. Let me know if anything else needs to be done. Thanks for catching this. --Chris (talk) 13:15, 17 April 2013 (UTC)

Favicon W

As some of you may have noticed, the favicon was recently updated - rounded corners, high-res varient, general rerender of the W itself...

Unfortunately I'm not entirely sure the W was rendered from the correct wordmark - there are a few different variants floating around Commons, many of which have been used in the past. Would anyone happen to know which would be be the correct one at present so I can check the favicon and if need be fix it? -— Isarra ༆ 18:25, 17 April 2013 (UTC)

Nevermind, looks like it was the right W. -— Isarra ༆ 02:43, 18 April 2013 (UTC)

Errors with pipes in titles

I'm seeing a number of errors caused, where the title attribute of a web page which includes pipe characters ("|") has been automatically copied to the |title= of a citation template; for example:

|title=Albanian celebrations leave Serbs defiant | World news | The Guardian

from http://www.guardian.co.uk/world/2008/feb/18/kosovo.serbia1

It would be good to find out which of our citation tools are doing this (RefToolbar seems to be one of the culprits); and have them remove, substitute or escape the characters concerned, if not (prompt users to) strip out superfluous text. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:21, 15 April 2013 (UTC)

I don't know what is causing this. There is no way for the Lua templates to figure out that these should be converted to something else. But, they will show an error since there are no parameters and will be placed in a category.

Albanian celebrations leave Serbs defiant. {{cite book}}: Text "The Guardian" ignored (help); Text "World news" ignored (help)

-- Gadget850 (Ed)^talk 20:37, 15 April 2013 (UTC)

Yes, the category is how I'm finding them. You can see this caused yourself, using RefToolbar. Our tools should not be routinely causing errors which require human rectification. The fix is required before the "poisoned" data reaches a template, Lua or otherwise. Zotero, for instance, manages to avoid this problem, when exporting Wikipedia refences. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 21:10, 15 April 2013 (UTC)

Reflinks can cause this problem. Graham87 05:33, 16 April 2013 (UTC)

Thank you. I've asked its author to join us here (as I have for RefToolbar). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:51, 16 April 2013 (UTC)

My apologies; it turns out that it uses the HTML entity instead of the raw vertical bar character. Graham87 02:34, 19 April 2013 (UTC)

I sympathise, Andy; but whenever I detect that a tool is the cause of bad content - whether it be AWB, Reflinks or something else - those writing/maintaining the tool tend to say something like "it's the user's responsibility"; "users should check and clean up before they save"; "the user might want to do that"; or similar. Too many users of automated tools use them in good faith "the tool said to do it, so it must be right". Which is why I do all my editing manually, and my only tool is an offline database of preconstructed {{cite book}} templates, for my personal library. --Redrose64 (talk) 07:00, 16 April 2013 (UTC) amended Redrose64 (talk) 13:41, 16 April 2013 (UTC)

I trust that that won't be the case here - even a text warning to their users will help - but more generally, we as a community can reject such tools, blocking them (or their recalcitrant authors) until fixed if required. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:51, 16 April 2013 (UTC)

Don't forget ProveIt GT. -- Gadget850 (Ed)^talk 14:59, 16 April 2013 (UTC)

Does that cause the error, too? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:09, 17 April 2013 (UTC)

Error messages

In this case, the new template is pointing out errors that were previously not visible to the reader, and therefore not significant enough to fix. Are we just creating busywork for editors by highlighting these issues to readers? GoingBatty (talk) 03:10, 18 April 2013 (UTC)

The errors discussed above are caused by software tools. If we can stop them being caused, we should, and it need not trouble editors. The error messages are a separate issue, so I've made this a sub-section. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:59, 18 April 2013 (UTC)

Discussing auto-correction of simple errors: We have been discussing these extra error messages (at Module_talk:Citation/CS1#Perhaps consider fix-cite superscripts), to instead just append the extra text at the end of the citation with a quiet superscript note, such as:

"<citation here>. World news. The Guardian.^{[fix cite]}

Because all those red-error messages are new (previously such bar "|text" was ignored) and years have passed since some extra bar text was added ("|World news") then it might be excessive to log red-error messages into 16,000 pages, as if the extra cite text was a fatal problem that demanded immediate editing to correct (it's been there for years). Instead, it would be simple to just append the extra text, to each cite, with a small superscript note. -Wikid77 (talk) 11:28, 18 April 2013 (UTC)

Thanks - I'll reply there. GoingBatty (talk) 22:45, 18 April 2013 (UTC)

When I looked this morning, across all namespaces the Citation Style 1 engine had detected 170,863 errors in 29,968,922 pages (0.57%). That's a pretty small percentage so I don't think we're creating busy work. And, editors who don't want to see any of the error messages can turn them off. Some error messages are still hidden but can be made visible. Instructions for both at Help:CS1 errors.

—Trappist the monk (talk) 12:38, 18 April 2013 (UTC)

While it's a small percentage, it's still 170,863 more times that readers who don't understand the inner workings of Wikipedia will see errors that they might not understand, and may not be saavy enough to know how to turn off. I'm worried that this will undermine the perception that Wikipedia is a quality encyclopedia. GoingBatty (talk) 22:45, 18 April 2013 (UTC)

I don't share your worry. Since I first experienced Wikipedia, I have noticed that the widely variable quality of article reference sections has always been a weak point. Surely I'm not the only one who sees this. If you want to ensure that readers' perceptions of Wikipedia as a quality encyclopedia are good, do everything you can think of to improve the quality of article reference sections. The new CS1 error messaging is one tool in the box for editors to use to make those improvements. The developers of the new CS1 Lua module have done a terrific job for which they are sorely under-appreciated.

—Trappist the monk (talk) 03:09, 19 April 2013 (UTC)

Why is ogv video playback on Wikipedia happening in Flash?

I was looking at the Boston Marathon bombings article which contains File:President Obama speaks on explosions in Boston (2013-04-15).ogv. Why is there a big Adobe Flash button? I'm running Firefox 20.0.1, it has HTML5 video support. I'm fairly sure that videos on Wikipedia did not require Flash, so why is it happening now? - hahnch e n 10:04, 17 April 2013 (UTC)

Perhaps you can make a screenshot, so that we can see it as well ? —TheDJ (talk • contribs) 12:17, 17 April 2013 (UTC)

Nothing really worth taking an image of. In Firefox, the playback is done in Flash. In Chrome, it's not. - hahnch e n 12:28, 17 April 2013 (UTC)

There's some weird fallback stuff going on. If I have Flashblock ON, it defaults to using Flash video. If I turn Flashblock off, it uses Native. Seems like a bug. - hahnch e n 13:10, 17 April 2013 (UTC)

Sounds like Flashblock might have a bug with HTML5 video or something. —TheDJ (talk • contribs) 09:47, 18 April 2013 (UTC)

Yeah, it was a Flashblock issue. Got confused with the blocking options, if you block HTML5 video - and then unblock it - it gives you the default Firefox player - and not the fancy Wikimedia one. - hahnch e n 16:14, 18 April 2013 (UTC)

Possible way to cut down on the number of new editors submitting articles without references at the Afc

Dear technical folks: I posted this in the Proposals section, but didn't get what I needed, so I am reposting here.

We Afc reviewers waste a lot of time declining articles that have no references at all. Several of us have been discussing ways to discourage users from submitting such articles. I made up a proposal for an intermediate step in the submission process. I posted it at User:Anne Delong/AfcBox and requested input from other reviewers at User talk:Anne Delong/AfcBox so that we could agree on a proposal to bring to those who implement these types of things.

Another editor, Techncal 13 has made a counter proposal, and his ideas are also on the talk page above.

I would appreciate a technical assessment from someone who is familiar with the technical ins and outs of Wikipedia about the practicality of both ideas. Each proposal seems to have its own strengths and weaknesses. Or, perhaps neither is practical and we should continue manually declining the articles (sigh). Can someone in the know please read the proposal at User:Anne Delong/AfcBox and then add technical comments to the discussion at: User talk:Anne Delong/AfcBox ? Thanks. —Anne Delong (talk) 17:06, 17 April 2013 (UTC)

Both have merits, both can be done. Personally, I prefer 'post review' over 'gatekeeping'. Your real problem is 'who is gonna write it'. —TheDJ (talk • contribs) 09:40, 18 April 2013 (UTC)

Live Pending Changes reviewer gadget

Now the average time for reviewing is more than 2 hrs. And reviewers and admins have to go to Special:PendingChanges frequently to check them. I just saw that we can transclude that special page {{Special:PendingChanges}}.

Articles with edits awaiting review (click show)

There are currently no pages meeting these criteria.

So is it possible to build a gadget for transcluding this at the top (or somewhere) of every article a user visits? If it is developed, the delay would get reduced from this two hours to a few minutes and reviewers will no longer be needed to go to Special:PendingChanges to see the list. Thanks for your kind help.···Vanischenu^「m/_Talk」 20:11, 17 April 2013 (UTC)

Wait, what?! So is it possible to build a gadget for transcluding this at the top (or somewhere) of every article a user visits? Uh, yeah, it's possible, but why on earth would want to clutter up every article with this? :p Plus, it still requires one to click [show]...making the whole thing rather pointless indeed. A better solution would be an IRC feed or something like that, which you could then join. In fact...has anyone done that already? —Theopolisme (talk) 00:54, 18 April 2013 (UTC)

legobot2 in #wikipedia-en-pc ^connect does the same every ~5 minutes. :) gwickwire^talk_editing 00:56, 18 April 2013 (UTC)

Ah, the magic of the internet. Thanks. —Theopolisme (talk) 01:01, 18 April 2013 (UTC)

Theopolisme: I think the idea is it'd be something to opt-into if you wanted to be an active pending changes reviewer and have the list handy.--v/r - T P 16:13, 18 April 2013 (UTC)

we have a similar opt-in gadget in hewiki for patrollers (we use the "patrol" system that's not in use in enwiki except for new pages, and we do not use the "review" system). after several iterations and users' feedback, we found that a better place for this gadget, rather than adding a box at the top of every page the reviewer visits, is to add a new menu item to the toolbox, which also servers as a link. the gadget itself is very simple: it executes an API call every <some reasonable interval>, and if it finds the list of articles waiting for review not empty, it adds the link to the toolbox. once the link is added, polling stops of course. the whole thing looks approximately so (100% untested):

(function doit() {
    mw.loader.using( 'mediawiki.api', function() {
        var api = new mw.Api();
        api.get( { list: 'oldreviewedpages', orlimit: 1 }, function( data ) {
            if ( data && data.query && data.query.oldreviewedpages && data.query.oldreviewedpages.length )
                mw.util.addPortletLink( 'p-tb', '/wiki/Special:PendingChanges', 'Pending changes' );
            else
                setTimeout( doit, 10000 );
        } );
    } )
})();

the idea is that people with reviewer rights (i.e., those that opted in), can see in a glance at the toolbox that there are articles awaiting review, and the indication is a link to Special:PendingChanges.

it's possible to place thr indication/link elsewhere (i.e., not in toolbox) by using something other than "p-tb".

peace - קיפודנחש (aka kipod) (talk) 17:53, 18 April 2013 (UTC)

Thanks to all of you. It would be great if we have a code for transcluding it to each page. Just the code to transclude it, do not hide or do any modification. e.g., I want to transclude {{User:Vanischenu/transcludeme}} to all pages. Now depending upon my needs, I can change the contents to be transcluded by editing that page. (If you want to hide, then the markup for hiding can be provided inside the page to be transcluded.)···Vanischenu^「m/_Talk」 21:30, 18 April 2013 (UTC)

Sortable annotated tables?

Is there a way to make annotated tables sortable? I'm talking about the tables that, for each entry, have a first row of columnar data with a second row of table-width text giving annotations, such as the standard episode list here. The goal is to have the columns sort only by the values in the data row while each entry's nonsorting annotation row remains directly beneath the data row.

There at least used to be a workable hack, but sometime within the last year or two the underlying table code was changed (whenever the placement of the sort button in headers also became fixed) and it no longer works properly as far as I can tell. postdlf (talk) 00:15, 18 April 2013 (UTC)

I don't think this is possible. The problem is that there is no semantic relation between row 1 and row 2 of what the 'visual' users consider to be one row. As such, you cannot explain this relation to an algorithm. The 'proper' way to do this is to create a 'subtable' inside each row to hold these elements, but then you have the problem that the column widths won't align anymore (without setting fixed widths, which is not really desirable either). —TheDJ (talk • contribs) 09:29, 18 April 2013 (UTC)

The larger reason this doesn't work is because these tables are being used not as data tables (which you could say that they are) but as presentation tables (with certain elements of data). A reformat of them as below would fix it, but that would upend years of having a particular view of these tables and which may not look very pretty on mobile. :)

No.	Title	Directed by	Written by	Original air date	Production code	U.S. viewers (million)	Plot
1	"Pilot"	R.J. Cutler	Callie Khouri	October 10, 2012	101	8.93[7]	Rayna Jaymes is a famous country singer, whose tour is not selling as quickly as she had expected. In order to save money, her record company wants her to be an opening act for up-and-coming singer Juliette Barnes. Rayna's company tells her to either go on tour with Juliette or leave the label. Juliette is revealed to constantly be called on the phone by her mother, who is a drug addict and wants money to feed her drug needs. Juliette records three songs Rayna turned down for her album. Rayna eventually wants them back but it's too late.

--Izno (talk) 16:37, 18 April 2013 (UTC)

The hack that used to work basically created dummy table cells for the second row, which provided proper sort keys that did not display, and then only the last cell would be visible, but manually positioned (by div code, I believe) so that it was superimposed beginning at the far left of the table and then stretched across its length over the non-visible cells. Given that most tables are constructed with separate templates for each multi-line entry, there might still be a way to set it up so that the table treats the two (or more) lines as one sorting unit but still sorts by the right cell information.

But I think this functionality is worth adding even if it requires developer effort, because these kinds of tables are widespread across many subjects, with entries that are largely sortable data apart from the annotation. Anything that adds to their functionality and flexibility benefits both readers and editors, and there's really a lot of room for innovation in organization and presentation here if the software would just support it instead of narrowing our options. postdlf (talk) 18:16, 18 April 2013 (UTC)

Need a tool to connect article to Wikidata on creation

Can anybody make a gadget which near editing window of page being newly created will display a field or two where one can specify either Qxxx ID, or langcode and pagename to find one, to which the article will immediately be linked after creation? It's difficult even for experienced user to go on WD and do there several clicks each time the page is created (to say the true, a great patience is needed to wait until WD page will be rendered, and it would be better not to load the full UI an extra time). For newbie, it can be scaring out from establishing any iw's whenever. It seems to me that we should replace the #wbSave input type=submit on input type=button, which sends a POST query to save the page without loading it via AJAX. In on-success method of the query we send another request (cross-domain one, which ones I myself understand poor) to Wikidata to add a sitelink; after this query is executed, the created page is loaded, or if it returns error, the page is however loaded after error message. Maybe also a new WD item should be created for new page (at least, if user checks some box to confirm it), but seems like we can leave it to bots. Ignatus (talk) 11:39, 18 April 2013 (UTC)

Just insert interwiki links at page bottom to override Wikidata: All editors can continue to use the old-style interwiki links at page bottom. If the current Wikidata entry seems incorrect, then just edit the page and insert the corrected other-language page as an old-style interwiki link, such as "[[dk:xxxx]]" to reset the article name as being "xxxx" for Danish Wikipedia. -Wikid77 (talk) 14:13, 18 April 2013 (UTC)
Please don't do what Wikid suggests.
To respond directly to that question, it is planned to develop this in the core software (at some point). I'm not sure if there's a bug for it yet in bugzilla. --Izno (talk) 16:44, 18 April 2013 (UTC)

This morning I looked at WP:Wikidata, Wikidata, Create a new item, and Item by title and found it all to be completely inscrutable. What documentation there is, appears to have been written by the engineers who developed it. Clearly they knew what they were writing about but that knowledge isn't conveyed to those of us who had no part in Wikidata's creation. A lot of work is needed to make it ready for non-technical editors. Until such time as a novice editor can use the Wikidata UI, adding old-style interwiki links and letting a robot sort it all out makes a lot of sense.

—Trappist the monk (talk) 17:38, 18 April 2013 (UTC)

You don't need to know the Q number on Wikidata. For any page on English Wikipedia, the Wikidata is held at [[d:Special:ItemByTitle/enwiki/fullpagename]] hence the Wikidata for this page is at d:Special:ItemByTitle/enwiki/Wikipedia:Village pump (technical); the Wikidata for England is at d:Special:ItemByTitle/enwiki/England. --Redrose64 (talk) 20:26, 18 April 2013 (UTC)

We are trying to deploy the widget for this on Monday. --Lydia Pintscher (WMDE) (talk) 21:32, 18 April 2013 (UTC)

My talk page posts deleting a more or less simultaneous post from another editor

At this edit my post deleted a more or less simultaneous post by another editor – were both replying to a third editor's post. There was no edit conflict warning. This has happened to me three times now, all on this same page and, as far as I know, only when the other's post is to the same talk page section. Anyone know what's going on?

—Trappist the monk (talk) 15:54, 13 April 2013 (UTC)

The problem you are reporting sounds like a potential issue in the code of the MediaWiki software or the server configuration. If the problem is reproducible, it would be nice if somebody who has this issue could send the software bug to the 'Bugzilla' bug tracker by following the instructions How to report a bug. This is to make developers of the software aware of the issue. If you have done so, please paste the number of the bug report (or the link) here, so others can also inform themselves about the bug's status. Thanks in advance! --AKlapper (WMF) (talk) 16:19, 14 April 2013 (UTC)

Hasn't happened since the post I referenced, though I've been extra careful to make sure that other posts haven't occurred while I've been writing a new post. I'll go back to just hacking away and posting and see if I can get it to happen again.

—Trappist the monk (talk) 17:15, 14 April 2013 (UTC)

Just happened in a slightly different way. I was replying to a post by Editor Technical 13 at Can category presence be done by a template?. Somehow, Editor WOSlinker also replied to Editor Technical 13 but I didn't get an edit conflict notification when I saved my post. Fortunately, this time, Editor WOSlinker's post wasn't deleted.

—Trappist the monk (talk) 11:58, 16 April 2013 (UTC)

That’s normal. The system can resolve edit conflicts on its own in simple situations.—Emil J. 12:27, 16 April 2013 (UTC)

I just commented about this under "invisible glitch" below. It seems to be happening a lot. Wnt (talk) 04:05, 19 April 2013 (UTC)

@Emil J.: So you're saying that it's normal for the system to delete or replace an intervening edit and not give an edit conflict warning? Really? If that's true then something is badly broken.

—Trappist the monk (talk) 10:49, 19 April 2013 (UTC)

Emil J. is responding to your comment saying that the "post wasn't deleted". I.e. properly merging the changes without deleting content is normal. – PartTimeGnome (talk | contribs) 21:27, 19 April 2013 (UTC)

I have a strange problem

Tracked in Phabricator
Task T49291

When I call up Bantayan Island it is fine. But if I click on 'edit' then straight away it displays a whole lot of red Cite errors. They're in pairs, one says missing reflist, the other says missing reflist|group=lower-alpha.

I get one pair immediately after heading, below coordinates but above hatnote about copyright. Then the page body is OK, but at the end there is one pair in the edit summary blurb, then a gap for the summary box; two pairs between 'minor edit' and 'By clicking'; one pair after the three buttons, one last pair before 'View templates'.

I just tried rebooting and it made no difference. I've been editing all day (now 23:00) no problem. However Bantayan Island is the page with which I associate myself most, and which I have worked on plenty. Although nothing for more than a week.

One other thing - when I went to my sandbox this evening, first time this year, it looked nothing like I left it before. It had lots of code starting with /unlink and displayed weird things. However nothing in history.

Any ideas?

John of Cromer in Philippines (talk) mytime= Tue 23:08, wikitime= 15:08, 16 April 2013 (UTC)

Are you trying to edit sections or the whole page when you click [edit]? If you are editing sections, then that is totally normal because the section you are editing doesn't include the rest of the page where the {{Reflist}} would be. If you are editing the whole page (which I just tried and saw no errors), then it may have something to do with other lua module citation errors that have been happening lately while things are being moved around, but someone else would have to offer more detail if that is the case. Technical 13 (talk) 15:25, 16 April 2013 (UTC)

I can confirm that this also happens when editing the whole page and hitting Show preview. -- Toshio Yamaguchi 15:29, 16 April 2013 (UTC)

Then I deffer any knowledge of the issue to Help:CS1 errors to hopefully be able to answer your question or at least offer some ideas that you can point to and say, tell me more about ____ Technical 13 (talk) 15:45, 16 April 2013 (UTC)

Yeah, edit button at the top. I'd take it to CS1 but that Help:CS1 errors is not an error-reporting page - it just tells you what the new error messages mean.

It happened straight away with me, before I did anything, no need to 'show preview'. Actually now I think about it, I have my preferences set to show preview on first edit. So maybe it's the preview which is cocking up.

John of Cromer in Philippines (talk) mytime= Wed 00:05, wikitime= 16:05, 16 April 2013 (UTC)

Did the edit I just made correct the errors? If not, I'm thinking it is the use of <ref>...</ref> used inside of templates and I know how to fix that. The error doesn't show on my screen, but I have a lot of gadgets running. Technical 13 (talk) 16:31, 16 April 2013 (UTC)

I have auto preview enabled. When I edit and preview, before the Preview I see:

Cite error: There are <ref> tags on this page, but the references will not show without a {{Reflist}} template or a <references /> tag; see the .
Cite error: There are <ref> tags on this page, but the references will not show without a {{Reflist|group=lower-alpha}} template or a <references group="lower-alpha"/> tag; see the .

Content that violates any copyrights will be deleted. Encyclopedic content must be verifiable. Work submitted to Wikipedia can be edited, used, and redistributed—by anyone—subject to certain terms and conditions.

And yes, the cite errors are missing the help page link. Some of the fonts in the References section are odd even out of edit mode. -- Gadget850 (Ed)^talk 17:13, 16 April 2013 (UTC)

This post tells me that it's nothing to do with CS1. You would get exactly the same problems with refs of the form <ref>plain text without templates</ref> --Redrose64 (talk) 19:18, 16 April 2013 (UTC)

OK- this uses List-defined references and has citation in the References section like:

<ref name="blair07">{{:Wikipedia:Tambayan Philippines/History of the Philippines (citations)|transcludesection=blair-tpi-07}}</ref>

-- Gadget850 (Ed)^talk 17:17, 16 April 2013 (UTC)

No, it's no better. In fact now it also shows 'PH-07' before coordinates John of Cromer in Philippines (talk) mytime= Wed 01:21, wikitime= 17:21, 16 April 2013 (UTC)

I need to sleep now - it's 1:00 am. btw I use W7-starter and MSIE 10.0.9.9200.16540

I just tried Opera (not logged in) - it's ok

Safari - OK

Firefox - OK

No chrome on my computer thank you.

John of Cromer in Philippines (talk) mytime= Wed 01:21, wikitime= 17:21, 16 April 2013 (UTC)

As best I see it: The article is transcluding references from Wikipedia:Tambayan Philippines/History of the Philippines (citations). The problem is that in edit mode these are not transcluding, thus the errors. If this were a proper template, then these sections would have trascluded as expected. -- Gadget850 (Ed)^talk 17:23, 16 April 2013 (UTC)

I can't reproduce the error with Firefox or IE7, but I've fixed the PH-07 and made another change that hopefully fixes the ref issue. Let me know. Thank you. Technical 13 (talk) 17:45, 16 April 2013 (UTC)

I still see it with FF20. BTW: {{refn}}. -- Gadget850 (Ed)^talk 19:12, 16 April 2013 (UTC)

I wonder how many resources this eats up. There are 322 citations, each wrapped in <onlyinclude> which includes an #ifeq to check for the name of the citation. If I am interpreting this correctly, all 322 citations are transcluded on each use, but only the one that matches in the #ifeq is shown. -- Gadget850 (Ed)^talk 19:21, 16 April 2013 (UTC)

The cause appears to be a broken image inside a reference, specifically File:Daanbantayan Map.jpg. A minimal test case is something like <ref>[[File:X]]</ref><references/>. Why that's blowing things up, I have no idea. BJorsch (WMF) (talk) 19:22, 16 April 2013 (UTC)

I've taken my last crack at it, I'm out of ideas, if my most recent edit didn't take care of it, then it is out of my realm of experience. Technical 13 (talk) 19:27, 16 April 2013 (UTC)

Ok, I have an idea now. T49291 explains it. BJorsch (WMF) (talk) 19:54, 16 April 2013 (UTC)

Avoid including cites from essays written as onlyinclude cite-speak: This is a prime example of massive over-complification in using cites. Many of the cite templates are being transcluded from within the large, complex essay page, WP:Tambayan Philippines/History of the Philippines (citations), but not visible there (for proofreading) because those cite templates are set as "<onlyinclude>" where they can only be seen after editing (and saving) the essay page, then edit-preview of the main article, then re-edit/save the essay to make further updates. Okay, we already know that people can write articles which invoke essays which invoke templates which invoke other templates which invoke wp:parser functions, where the whole contraption is so convoluted, so needlessly over-complicated, and we do not need more examples. This is an excellent case where technical complexities should drive the need for a "keep it simple" policy, so that new users will always have direct view of text which generates citations, rather than change the cite templates in one page (essay), then have to edit-preview another page to see the results of having changed those cite templates in the essay page. We need some technical-usability policies to deter these excessive convoluted over-complifications of ultra-contorted complexities. No transclusions allowed from essays of only-include cite-speak. I hope people can better understand, now, why the U.S. Federal Government has spent many thousands of dollars to write keyboard standards:

The '1' key shall be used to insert the numeral "1". The 'delete' key shall be used to delete one character.
The '2' key shall be used to insert the numeral "2". The unshifted 'A' key shall be used to insert the lowercase letter 'a', etc.

There need to be policies based on technical usability. -Wikid77 (talk) 19:50, 16 April 2013 (UTC)

OK I'm back. I don't really understand what this piece is saying. However there are a couple of things you seem to have lost sight of:

This is not a new page, but an old one. It worked perfectly fine up until (and including) its last edit a week ago. So any fault must be arising from an external change.
If you look closer at Wikipedia:Tambayan Philippines/History of the Philippines (citations) you will see that all {{cite book (etc)}} occurrences are inside comments. Thus nothing gets parsed directly, but only after transclusion. Transclusion works perfectly well inside comments, which is how I would expect it to work: basically the include operation causes it to read down the object until it matches an <onlyinclude>, then copy until it hits a(n) </onlyinclude>. In fact as expected, it reads the whole object file, so it is possible to transclude several chunks with the same key. NB the transclusion object file is not parsed by the transclusion process, and it is possible to read unparsable pieces.
Transcluding citations like this continues to work perfectly well on other pages - Emilio Aguinaldo for instance.
If I go back and try to edit an old version of the file, the error still occurs.
Second Law of Debugging - if you can't see it, you're looking in the wrong place.

My belief continues that this problem is outside the Bantayan Island page and arises from some system change within the last seven days. Don't tell me there haven't been any! I tried removing (commenting out) all the transcluded citations, and it made no difference, as I would expect.

I am not yet convinced that the file has not been damaged in some way, perhaps maliciously.

John of Cromer in Philippines (talk) mytime= Wed 08:55, wikitime= 00:56, 17 April 2013 (UTC)

Last page change was 21 March. I have now restored that, because none of the test changes here made any difference. The solution lies outside the page. Ironically, I want to edit it to replace some of the transcluded citations with {{cite isbn}} (for later works which actually have an isbn).

I don't know who deleted Daantayan Map.jpg nor when.

John of Cromer in Philippines (talk) mytime= Wed 11:02, wikitime= 03:02, 17 April 2013 (UTC)

OK here's the current situation:

I changed my preference so there is no preview on first edit.
I edited the text how I wanted, then saved without preview. That was fine, and revised page shows properly
I tried to edit again, without initial preview. That was fine, but when I clicked on 'show preview' the problem reappeared.

Which leads me to think the problem is in the preview action – either not doing something it should, or doing something it shouldn't. I can't explain though why it is just this one page out of the hundreds I have edited recently. However I continue to think the problem is outside the page. John of Cromer in Philippines (talk) mytime= Wed 12:59, wikitime= 04:59, 17 April 2013 (UTC)

Complex problems seem to depend on which server reformats, not preview: I recently saw the prior top-revision (cache copy) of "Bantayan Island" show those red-error messages repeated (at the top), even without edit-preview (for the 100-200 pageviews per day). It showed:

Cite error: There are <ref> tags on this page, but the references will not show without a {{Reflist}} template or a <references /> tag;...

I again displayed the top-version, and saw the same red-error messages again. In alarm mode, I edited the article to start simplifying the cites, and the red-error message went away temporarily, but returned after the 3rd edit-save, so I suppressed the messages. However, upon redisplaying the old revisions, the red-error messages were gone now, both from my red-msg copy and the original "top-revision" red-msg copy, as reformatted by other servers. We have seen this before with Lua errors (Script error) which get stored into the top-revision and always repeated in each cache-copy viewed. If the red-error messages appear again, then try a null-edit to reformat the top-revision, as then perhaps that next server will not regenerate the red-error messages ("Cite error: There are <ref> tags on this page, but the references will not show...") as the cached form of the top-revision. Meanwhile, among all the complexity, I have discovered the cites of books (in article "Bantayan Island") did not have page numbers, and the cite-speak essay-template did not reveal which cite is which, all scrambled in redisplayed scrambled order, so then I set to {|safesubst:} the 322 citations in the essay, but then reftags do not allow wp:subst, so then I hard-coded some of the book cites, so then I confirmed that transcluding cites from a 322-entry essay is very slow, and so then rewriting the cites as simple text into the article will reformat the article 3x faster, and allow new editors to see each cite as markup (rather than essay-based cite-speak), and they can add the needed page numbers for each book (which I noted ^{[need page]}) and that will begin to resolve some of the complexity, to allow proper book cites with page numbers. After all cites are recoded, then perhaps the red-msg problem will also go away, even though only some servers seem to regenerate the red-msg when formatting the top-revision of that article, as repeated in each cache-copy which the various users view. Gotta run. -Wikid77 (talk) 14:07, 17 April 2013 (UTC)

So basically you are saying the problem is external to the page?

Meanwhile, after all your tweakings, the citations are broken, and the problem is still there. If you look at each volume of Blair and Robertson, you will see they show the same date (1599–1602) whereas the date varies by volume. The date shown corresponds to volume 11. If you look at the external link, which is where you substituted vol 3, you can see correct information. Also titles are wrong – this changed after volume 5, so the title for vol 3 should be 1493–1803 which is how it is in the previously subst:ed one. You will also see the subst:ed one includes a whole lot more info, now missing from copied: wikilink of authors, publisher name and location, others, asin, url link for work, quote. I note that format is now in plain font. I am in the process of adding oclc, which also doesn't show. Updates to base material is a strong argument in favour of transclusion rather than substitution. As is adherence to 3rd normal form. I doubt many pages would contain as many as ten transclusions, which mean timing considerations are immaterial.

When I experimentally reverted volume 7 back to transclusion, all the material reappeared.

Why are you making experimental changes to a live page? Isn't that what sandbox is for?

NB1 it's not a 322-cite essay. It's a null-cite essay, mostly comment. But it is a 300,000 byte file to be scanned each transclusion.

NB2 "pages needed" is not a cite error per se.

If it is some sort of lua scramble, is there any possibility of (a) copy text to a new page (b) delete current page (c) rename copied page to original (d) somehow preserve history? This would shed any swarf in file headers etc

John of Cromer in Philippines (talk) mytime= Thu 09:46, wikitime= 01:46, 18 April 2013 (UTC)

Both internal and external problems with cites: I can understand the frustration with handling all the problems in article "Bantayan Island" but when citing several large books (and volumes) as references, then the page numbers should be listed for each. Also, I was not making "experimental changes" but rather the citations were too complicated, and I simplified each cite to allow adding page numbers where needed. Also, cites of 6 volumes of the encyclopedia had repeated the same 51-word quotation, so I cut that to just one, and removed the repeated 255 words from the other 5 cites. I apologize that I incorrectly changed the years in 5 volume names, and I have since corrected them. However, to edit an article-sandbox copy, use:

Talk:Bantayan_Island/sandbox

In that copy, then all complex cites can be bypassed to see if the overall cite-error problem stops. -Wikid77 (talk) 12:56, 18 April 2013 (UTC)

Fixed As BJorsch noted and reported, there is an {{efn}} that included the non-existant File:Daanbantayan Map.jpg. In addition, the {{efn}} was missing the closing braces. I deleted the entire sequence and the page no longer shows the errors when editing. The page probably needs some cleanup from other attempts at fixing it. -- Gadget850 (Ed)^talk 14:25, 18 April 2013 (UTC)

Not everyone is convinced of "Fixed" yet: I haven't seen any changes which would impact the <references> section, so we can just wait for more problems. That {efn} template was closed at "{{clear}} }}". Also, a redlinked image is no problem: trying <ref>[[File:X]]</ref><references/> will still format the one reference correctly. Meanwhile, I commented, at Talk:Bantayan_Island, that some of the cited books need the page numbers to be specified in the cites. I would not delete any more text, as that is likely to just further anger the author(s) of the article. If there are cite-check problems in some servers, then perhaps the cite-error messages will reappear. -Wikid77 18:38, 18 April 2013 (UTC)

I think it is now clear. As you say, it wasn't an unclosed {{{efn}}, it was properly closed. What was causing the grief though was the file missing. Removing the efn removed the entire problem. However I have just restored the missing file to commons, or one like it, and the old versions of the page work properly. I don't know when the file was deleted, nor by whom, because I can't find any entry in delete logs but it must have been some time after March 21st.

Seeing as how the citations are transcluded (or used to be) in several articles, the full citation including quote ought to be appear in each. Blair & Robertson for instance is a notable work within its sphere, and cited in many places, but not all volumes at once. Similarly, page references should appear when the citation is first referenced, i.e. within the body text, not tagged onto the citation itself, unless it is specific.

Summarising, for future reference, the fault was caused by a missing file inside an {{efn}}.

John of Cromer in Philippines (talk) mytime= Fri 08:08, wikitime= 00:08, 19 April 2013 (UTC)

If you want to do the transcluded references properly, you might as well use the new <section>. -- Gadget850 (Ed)^talk 02:15, 19 April 2013 (UTC)

Clarifying as sporadic error for some browsers: I have re-inserted the prior note [k], now with the restored map image, into article "Bantayan Island" but in all viewings of older revisions with Firefox browser, there were no missing-references messages. With older versions of MSIE, those messages only appeared sometimes (even in the formatted article), and apparently, in MSIE 10.0.9.9200.16540, the missing-references error appeared often. Again, Firefox browser had no problem with a redlinked image inside an {{efn}} footnote template. I guess the experiment should be to misspell a file name inside an {efn}, and see which browsers get error messages. -Wikid77 (talk) 02:36, 19 April 2013 (UTC)

I think it's a bit more obscure than that. Basically the file existed at the time the {{efn}} was written. It was only later when the object was deleted it went ape. So the test should be to rename (or delete) an in-use file and see how previous uses of it behave.

I'll check out <section> but right now I'm looking for oclc numbers for all cites in the hope that a {{cite oclc}} (a la {{cite isbn}}) will come along. Or do it myself.

John of Cromer in Philippines (talk) mytime= Fri 18:07, wikitime= 10:07, 19 April 2013 (UTC)

The animation of thumbnails gifs

UPDATE: Null-edit of image restarts animation, for some browsers. -Wikid77 23:26, 18 April 2013 (UTC)

So if I'm understanding this correctly if this image is re-sized to 401px the animation will work. Hmmmm
Chrome: Doesn't seem to be working
IE9:Doesn't work

Almost three years ago thumbnail animations were turned off because developers were worried about phones like the Sony Ericsson phone browser or Safari 3 didn't handle thumbnail gifs very well. Well its been 3 years and phone technology has progressed leaps and bounds. Is there any way we can restart thumbnail animations? -- Esemono (talk) 12:04, 18 April 2013 (UTC)

Could have policy requiring click for animation: I agree that the GIF animations should be reactivated, but perhaps we could write a policy to require viewers to click a photo to see the animated sub-page. In particular, all the animated horse photos from 2006 now make Wikipedia seem primitive and low-tech by comparison. It is not easy to explain a "canter" as shown by 16 frames in the (formerly) animated GIF below:

File:Gallop_animated.gif
See article "Canter" where 2nd image is animated for Firefox and other browsers.

There could have been a note as "(right-click to animate)" in the photo caption, to avoid downloading the animation with every pageview. -Wikid77 (talk) 16:17, 18 April 2013 (UTC)

Yeah, I was actually writing this 3 years ago. Didn't feel really inspired by some of the comments of the community though, so never finished it and took a 6 month break instead. —TheDJ (talk • contribs) 21:34, 18 April 2013 (UTC)

In my browser at least, the horse gif is displaying as animated, but the map is not. postdlf (talk) 18:17, 18 April 2013 (UTC)

I also see the horse animated in Firefox, so I changed the above image to link article "Canter" to avoid 750 kb image download here. -Wikid77 (talk) 19:11, 18 April 2013 (UTC)

first thing, this whole discussion has nothing to do with WP:VPT on enwiki. it has to do with WMF and not with enwiki.

second, ttbomk ^{(the rest of this schrift is all "to the best of my knowledge" - it can be completely untrue. if anyone knows better, please correct me)}, there was a time where the imageservers that the foundation uses had some bugs or limitations, and they did not import the GIF animation when resizing images. those issues were solved at some unknown point in time (at least, i do not know when), and all is well now. what we see with the non-animating thumbs is not a policy - it's a bug that makes it very difficult/impossible to purge old resized images from the image servers (this bug was discussed on WP:VPT several times in recent months, with each of those discussions begin as if previous discussions never happened). to "prove", or test this point, i changed the size of the displayed "CSA_states_evolution.gif" thumb from 400px to 401px, which forced the image server to create a fresh copy of the thumb, and lo and behold - it's animated (just to clarify - changing it to 399px would have the same effect).

peace - קיפודנחש (aka kipod) (talk) 20:29, 18 April 2013 (UTC)

(later addition): i was too lazy, and did not read the link Esemono posted in his (or her?) first post in this thread. the link is to an announcement from april 2010, which actually tells us that thumb animation on WMF iageservers was turned back ON, not "turned off" is he erroneously read it. in short - all is well (well, almost all: we still have the annoying image-purging issue). peace - קיפודנחש (aka kipod) (talk) 20:40, 18 April 2013 (UTC)

Tracked in Phabricator
Task T43130

I think the stuck thumb is another case of bugzilla:41130. Really old Last-Modified date on the HTTP response, even though it clearly was changed since then (also tried thumb.php + purge). —TheDJ (talk • contribs) 21:52, 18 April 2013 (UTC)

update: i performed a "null edit" (i.e., press "edit" and then "save", with no changes) on the image page, and it seems to be enough - i returned the thumb of "CSA_states_evolution.gif" image at top of page to size 400, and it still animate. i do not know if this remedy will work for every problematic image, but i'm going to try it from now on. peace - קיפודנחש (aka kipod) (talk) 22:29, 18 April 2013 (UTC)

Still stuck on frame 0 for me. Possibly it's only on some cache servers ? This is my response from .nl —TheDJ (talk • contribs) 08:13, 19 April 2013 (UTC)

Date: 19 Apr 2013 07:59:28 GMT

Via: 1.1 varnish, 1.1 varnish, 1.1 varnish

Age: 79456

X-Cache: cp1035 hit (92), cp3009 hit (16), cp3007 frontend miss (0)

Connection: keep-alive

Content-Length: 24911

Last-Modified: Wed, 11 Apr 2012 18:02:46 GMT

Etag: af188a5ca5f2b4b10222f7d961bc9493

X-Varnish: 1361273040 1326902984, 766261359 753202427, 1040169128

Access-Control-Allow-Origin: *

X-Timestamp: 1334167366.43913

Accept-Ranges: bytes

Content-Type: image/gif

He, it suddenly started animating.

Date: Fri, 19 Apr 2013 10:38:07 GMT

Via: 1.1 varnish, 1.1 varnish, 1.1 varnish

Age: 555

X-Cache: cp1035 hit (6), cp3009 miss (0), cp3010 frontend miss (0)

Content-Disposition: inline;filename*=UTF-8CSA_states_evolution.gif

Connection: keep-alive

Content-Length: 168548

Last-Modified: Fri, 19 Apr 2013 10:28:51 GMT

Server: nginx/1.1.19

X-Varnish: 1372972496 1372781083, 781540829, 1522202756

Access-Control-Allow-Origin: *

Accept-Ranges: bytes

Content-Type: image/gif

Thanks. Those are excellent ideas, and we can explain in some help files. -Wikid77 23:26, 18 April 2013 (UTC)
Sorry non-technical user here. Where does this stand. The image appears blank to me (This is what it should look like BTW). Am I doing something wrong?-- Esemono (talk) 06:37, 19 April 2013 (UTC)
- This could be a completely different issue. It seems the generation of thumb at that size is completely failing during scaling. This means that the resize software doesn't consider this to be a valid file. —TheDJ (talk • contribs) 08:13, 19 April 2013 (UTC)

Tracked in Phabricator
Task T49409

Thanks for reporting this! For the 400px thumbnail at https://upload.wikimedia.org/wikipedia/commons/thumb/1/1d/CSA_states_evolution.gif/400px-CSA_states_evolution.gif , purging helped and it works now (thanks to whoever did that). The 120px version at https://upload.wikimedia.org/wikipedia/commons/thumb/archive/1/1d/20120411151857!CSA_states_evolution.gif/120px-CSA_states_evolution.gif was not animated for me. So I first went to https://upload.wikimedia.org/wikipedia/commons/thumb/archive/1/1d/20120411151857!CSA_states_evolution.gif/120px-CSA_states_evolution.gif?RandomNumberHere1234 and then to https://en.wikipedia.org/wiki/File:CSA_states_evolution.gif?action=purge (note the URL suffixes). This fixed it for me. But https://en.wikipedia.org/wiki/File:Non-Native-American-Nations-Territorial-Claims-over-NAFTA-countries-1750-2008.gif is a different issue. Here the thumbnail is already missing, and trying *any* different thumbnail size only creates a generic "Error generating thumbnail". The image was uploaded more than three years ago and purging does not help here. Maybe the 97 frames are too many for the image scaler or whatever. :) Worth to investigate, hence I filed a bug report at bugzilla:47409 at so the developers can get aware of it and investigate. --AKlapper (WMF) (talk) 13:24, 19 April 2013 (UTC)

AWB question

I wanted to try AWB to do some file cleanup, but I am not very experienced with AWB

For example, in File:Herbert Zwiebel or Herb Bell founder of PackardBell 1933-1971.jpg, after removing the older version of the file, I need to remove the template: {{non-free reduced|date=00:43, 2 April 2013 (UTC)}}

Because the time stamp will be different for each one, I cannot use a simple find and replace. I though maybe I could use a wildcard. but that doesn't seem to be working. Alternatively, I see that AWB has a Template Substitution option, but I simply want to remove it, not substitute, and I don't see how to specify the template. Anyone with some suggestions? --SPhilbrick (Talk) 13:46, 19 April 2013 (UTC)

Use \{\{non-free reduced\|date=.*?\}\} in the find and replace and mark it as a regex (regular expression) Werieth (talk) 14:03, 19 April 2013 (UTC)

Sorry, I'm probably doing something stupid, but it isn't finding a match.

See this screenshot

I get that the backslash is needed to specify that some characters should be interpreted as characters. I thought * was the wildcard, but now I see that a dot is. However, it doesn't seem to find the line.--SPhilbrick (Talk) 14:24, 19 April 2013 (UTC)

Update, I tried it without the "?" and it seems to work. Am I doing something dangerous? And thanks for your help.--SPhilbrick (Talk) 14:34, 19 April 2013 (UTC)

I believe the syntax is similar to Regular expression#POSIX Basic Regular Expressions. If so, a dot is a wildcard for a single character, and an asterisk means "match 0 or more instances of the preceding element". Thus a dot followed by an asterisk means "any quantity of any character". --Redrose64 (talk) 14:56, 19 April 2013 (UTC)

Thanks. So I didn't get why the question mark was in there. It seemed to fail with it, and work without it.--SPhilbrick (Talk) 15:27, 19 April 2013 (UTC)

The asterisk will normally match as many characters as possible; this is called being "greedy". The question mark makes the asterisk quantifier "non-greedy" or "lazy", so that it matches as few characters as possible while still creating an overall match. Thus without the "?", if the page contains something like

{{non-free reduced|date=00:43, 2 April 2013 (UTC)}}{{another template on the same line}}

then the .* in the regex will match 00:43, 2 April 2013 (UTC)}}{{another template on the same line instead of only the timestamp, causing the other template to be removed as well. So the question mark is desirable here to prevent additional templates from being removed. I'm not sure why it isn't working with the question mark; maybe try a plus sign in place of the asterisk? jcgoble3 (talk) 19:42, 19 April 2013 (UTC)

Converting Kindle locations to page numbers in sourced material

Is there a simple way to convert page numbers to Kindle locs (and vice versa)? This website offers a convoluted means: http://www.bookmonk.com/labs/numbers.php 36hourblock (talk) 19:20, 19 April 2013 (UTC)

Some Kindle e-books contain page numbers based on a paper copy, they can be seen at the bottom of the screen when the Menu button is pressed (Amazon documentation). the wub "?!" 22:38, 19 April 2013 (UTC)

As I recall, only newer Kindles or some older Kindles with upgraded software support page numbers and only with newer eBooks. -- Gadget850 (Ed)^talk 23:57, 19 April 2013 (UTC)

Mobile CentralNotice Banner (on en.m.wp.o)

The mobile team in conjunction with the fundraising team is going to test out the new integration between the MobileFrontEnd and CentralNotice extensions. We will be doing this the next week (Apr 22-28th). What we will be putting up is a banner recommending users try out the new commons app (Banner). Specifically this banner will be targeted at logged in alpha/beta mobile users of en.m.wikipedia.org browsing in english and having a handset that we identify as an android device. If you have any problems please report them to meta:Help_talk:CentralNotice/Mobile. Mwalker (WMF) (talk) 22:24, 19 April 2013 (UTC)

invisible glitch

Tracked in Phabricator
Task T49438

Theres a serious glitch here with sections going missing. A few days ago some categories had entries mysteriously missing. Now a post by an admin on AN/I is only visible during a diff, but not without the diff. Is this only happening to me? What can i do? Its like shit is mysteriously going missing. Pass a Method talk 03:35, 19 April 2013 (UTC)

Links would help figure out what is going on. EVula // talk // ☯ // 03:45, 19 April 2013 (UTC)

~~Have you tried purging either your own computer's cache, or the page itself?~~ Someguy1221 (talk) 03:46, 19 April 2013 (UTC)

Never mind, I found it. Masem accidentally deleted my comment. No glitch - just an editor's error. Someguy1221 (talk) 03:49, 19 April 2013 (UTC)

I had my own comments on that article talk page (not the ANI) deleted twice apparently by accident also. And it seems like I'm constantly having edit conflicts even though the others aren't editing that particular section. There isn't something gone wrong recently with how the edit conflict code is processed, is there? (I never use it myself - I back-arrow, copy the text, hit the Editing section.. version in my browser history, and paste into that.) Wnt (talk) 04:01, 19 April 2013 (UTC)

Wait a minute: "new section"? And it overwrote the old material? This should not be allowed to happen. Let's say someone is editing a section. After he presses "Save page," this is what should happen:

Acquire a lock on the entire page. If acquisition fails, repeat.
Check if anyone else has saved changes to the section being edited. If so, return an edit conflict and release lock. If not, proceed.
Combine the current section with all the other sections to fill in the content of the page.
Save the page.
Release the lock.

Is this not what happens? -- King of ♥ ♦ ♣ ♠ 04:53, 19 April 2013 (UTC)

Not quite right. Firstly, the edited section is combined with the rest of the page before checking for conflicts. MediaWiki determines what you changed by comparing against the previous revision, then applies those changes to what is now the current revision. You only get an edit conflict if MediaWiki is unsure how to apply your changes. This typically happens if another editor changed the same line as you, or changed a line immediately above or below one you had changed.

There is a bit of a difference when using "new section", though. In that case, the section you are adding just gets tagged to the end of the current page. There should be no possibility for an edit conflict or for anything to be removed.

Of course, that's the theory. I've seen several cases of someone accidentally removing a previous comment; I suspect something goes wrong sometimes. I don't think it's anything recent; I've seen it several times over the last few months. – PartTimeGnome (talk | contribs) 21:50, 19 April 2013 (UTC)

But I am not talking about the part that concerns opening up a section to edit, which can take a long time and is what edit conflicts are for. I'm talking about the actual save, which is ideally instantaneous but not so in practice. So what could happen is that two people are editing the last section and creating a new section, respectively, and both save at the same time. The software sees that there are no issues, and what happens is that the second person's revision overwrites the first person's revision. -- King of ♥ ♦ ♣ ♠ 00:16, 20 April 2013 (UTC)

I.e. a lost update problem (though the previous edit isn't really lost in our case, since it can still be retrieved through history). I suspect some problem with locking. I'm interested to see what the devs find on the bug you filed... – PartTimeGnome (talk | contribs) 14:17, 21 April 2013 (UTC)

Different display for readers vs. editors

Is there any existing mechanism, via CSS or otherwise, to vary what content is displayed to logged-in editors vs. general readers? It occurs to me that some of our message boxes and error messages can appear fairly mysterious to the general reader. I'm wondering whether it might sometimes be useful to display a more general information statement to readers while having a different and more focused message shown to editors that explained how to fix the problem. However, I don't think we presently differentiate any content for editors / readers, so I'm unsure if there is any natural way to accomplish such a thing. Of course one would also want to get the community's approval on whether such differences are a good idea in any particular application, but I'm wondering whether they are presently possible in principle or not. Dragons flight (talk) 01:06, 20 April 2013 (UTC)

My first impulse: a hacky way to do it currently is a by-default-enabled CSS/JS gadget. —Theopolisme (talk) 02:08, 20 April 2013 (UTC)

MediaWiki supports the ability to add CSS and JS that is only seen by users in a certain user group, and we use this for admins (see MediaWiki:Group-sysop.js and particularly MediaWiki:Group-sysop.css). The latter of those two shows CSS that is used to show information to admins only (MediaWiki:Common.css sets that class to "display:none" by default, which is overridden by MediaWiki:Group-sysop.css. However, this appears to only be supported for certain user groups, as this list does not show CSS/JS messages for all user groups (interestingly, the list does not seem to recognize MediaWiki:Group-accountcreator.css as a valid MediaWiki message). Autoconfirmed is supported, but not simply user. However, it would likely be simple (I'm guessing) for the devs to set it up for the "user" user group. jcgoble3 (talk) 02:58, 20 April 2013 (UTC)

Maybe Mediawiki:Group-editor.css from your list is meant to be all logged in users? Dragons flight (talk) 06:00, 20 April 2013 (UTC)

Editor is a special group used by Wikinews, but not enwiki. There's many groups on that list that enwiki does not use. Per MediaWiki:Grouppage-user and MediaWiki:Group-user, I believe simply "user" represents all logged-in users, and this is confirmed by going to my preferences, looking at the list of groups I'm in near the top, and clicking on "Users", which takes me to Wikipedia:Why create an account?. And the "user" user group, as noted above, does not seem to have its own JS/CSS pages. jcgoble3 (talk) 06:58, 20 April 2013 (UTC)

It seems that the group 'user' is explicitly excluded from the list of groups to add specific CSS and JS for. I'm not sure if there is a proper reason for this, but that is how it is currently coded. —TheDJ (talk • contribs) 11:28, 21 April 2013 (UTC)

Unnoticed editing

How is it that the preloaded preview template was edited by a new user, and went unnoticed for a good two months? This was causing that user's edits to show up every time you made a new PR. Very odd that no one noticed in TWO MONTHS. Ten Pound Hammer • ^{(What did I screw up now?)} 17:14, 20 April 2013 (UTC)

Because it doesn't have any watchers, which implies that you're not even watching it yourself. No watchers=nobody who cares. --Redrose64 (talk) 17:41, 20 April 2013 (UTC)

But still, the fact that someone else's review comes up whenever you make a new PR… Ten Pound Hammer • ^{(What did I screw up now?)} 18:09, 20 April 2013 (UTC)

Here I come....to watch the page! —Theopolisme (talk) 02:03, 21 April 2013 (UTC)

Little help with wikitable

Does anyone know of a more efficient way of centering all the content after Row2-Column3 in this table? (instead of individually centering each cell?) Thanks in advance. :) Reh man 06:16, 21 April 2013 (UTC)

:I'm answering this for two reasons, first is of course to answer your question, and the second is to wonder if anyone else is seeing the extended content that appears to be part of this post but isn't. It seems to have happened at this edit by User:MiszaBot II. Fixed by Editor This, that and the other.

I re-fixed the fix. A section contained three chunks of example output from some external process, each chunk was enclosed in {{cot}}/{{cob}}. One of these chunks contained == {{int:filedesc}} == which MiszaBot II assumed was the start of a new section, so it archived that off and in so doing took the third {{cob}} with it. An unclosed {{cot}} always moves to the bottom of the displayed page. --Redrose64 (talk) 11:59, 21 April 2013 (UTC)

I don't think there is a way to center the content of a single column because the wiki table markup is just an abstraction of the HTML markup which is all row-centric: table row tags <tr>...</tr> hold multiple table data tags <td>...</td> that hold the row's column data. Styling is applied to <td> so for a particular column must be applied individually at each row's column. There is an obvious need, I wonder why it hasn't been addressed.

—Trappist the monk (talk) 10:25, 21 April 2013 (UTC)

If you want to center the majority of all cells in a table then you can start the table with {|class="wikitable" style="text-align:center;", and then left or right align the cells you don't want centered. PrimeHunter (talk) 10:45, 21 April 2013 (UTC)

Thanks Prime. That should do for now I guess. Reh man 11:56, 21 April 2013 (UTC)

Datasift/Tweetmeme crap: heads up for tool maintainers

46.236.7.249 - - [21/Apr/2013:06:28:57 -0700] "GET /linksearch.jsp?link=htm2pdf.co.uk HTTP/1.1" 403 300 - "Mozilla/5.0 (compatible; TweetmemeBot/3.0; +http://tweetmeme.com/)" "wikipediatools.appspot.com" ms=297 cpu_ms=23 cpm_usd=0.000034 app_engine_release=1.7.7 instance=00c61b117c992c970338b324421579b708512e
...
46.236.24.47 - - [21/Apr/2013:06:21:15 -0700] "GET /linksearch.jsp?link=loadmastertrailerco.com HTTP/1.1" 403 300 - "Mozilla/5.0 (compatible; TweetmemeBot/3.0; +http://tweetmeme.com/)" "wikipediatools.appspot.com" ms=245 cpu_ms=23 cpm_usd=0.000034 app_engine_release=1.7.7 instance=00c61b117c992c970338b324421579b708512e

$ whois 46.236.7.249
inetnum:        46.236.7.0 - 46.236.7.254
netname:        DEDIPOWER-DEDICATED-O1FL
descr:          MediaSift Limited

Just a heads up: I've had 1000+ hits from these guys in the last 37 minutes (that's about 1 every two seconds). robots.txt is ineffective, so you will have to hard blacklist them. They're also not paying attention to Wikipedia's robots.txt. This is a social media monitoring company, so there is no reason whatsoever they should be accessing tools (or Wikipedia, for that matter). MER-C 13:58, 21 April 2013 (UTC)

Template:Countyrow error

(Note:The below message was originally posted at Wikipedia talk:WikiProject United States, but the answer might be more likely found here at the Pump. — Maile (talk) 14:16, 21 April 2013 (UTC))

I'd like to draw attention to the bug in List of counties in Nevada for the Carson City entry. Carson City is not a part of any county, so the syntax given is different. However, the way the syntax is right now, "class=Nickname" appears inside the link target, causing the words "Carson City" to link to Carson Cityclass="Nickname", Nevada. The problem may be with Template:Countyrow or with the syntax for its usage on the page. Either way, I'd prefer if someone with a bit more wikicode knowledge would take a look at this; I'm not able to find the problem. Someone the Person (talk) 14:08, 21 April 2013 (UTC)

I fixed it by reverting these two edits. --Redrose64 (talk) 15:06, 21 April 2013 (UTC)

Thank you for taking care of this. — Maile (talk) 19:14, 21 April 2013 (UTC)

Deleted diffs and Fullurl

If we take e.g. [36] in basic, non-Fullurl notation, when I try to represent it in Fullurl as [37], clicking on that second link it seems to drop the &timestamp= parameter. Am I doing something wrong, or is it something to do with the Fullurl mechanism itself? It Is Me Here ^{t / c} 17:50, 21 April 2013 (UTC)

You don't need to separate all the query elements. It should be like this: [38] —TheDJ (talk • contribs) 17:58, 21 April 2013 (UTC)

I see. So, in general, how do you know when to change an & to a |, and when not to? It Is Me Here ^{t / c} 18:01, 21 April 2013 (UTC)

The fullurl syntax is at mw:Help:Magic words#URL data: {{fullurl:page name|query_string}}. Never split the query string. fullurl is a parser function and not a template. It doesn't have named parameters. target=%40field+%28video+game%29&timestamp=20120905222706 in {{fullurl:Special:Undelete|target=%40field+%28video+game%29&timestamp=20120905222706}} is one long unnamed parameter and not an assignment to a fullurl parameter called target. PrimeHunter (talk) 20:50, 21 April 2013 (UTC)

Twinkle question

I have rollbacker rights, but I did not ever download Twinkle. Due to a recent edit of mine, I need clarification. I'm being accused of abuse, I think. Please see Talk:Minnie_Fisher_Cunningham. First of all, I was unaware Rollbacker was operated by Twinkle. Secondly, is this true that it shouldn't be used to revert a good-faith edit? Good faith is assumed unless it's vandalism, but the assumption does not mean the reverted edit was correct. I did not accuse this editor of vandalism. If we can't use the Rollbacker rights to revert good-faith edits, then why isn't it limited to strictly vandalism? Please advise. — Maile (talk) 21:07, 21 April 2013 (UTC)

Technically, rollback (native, the button that says [rollback (X edits)] near it) should only be used for pure vandalism. However, WP:TW, which is a gadget that you don't even download and is enabled in your preferences under "gadgets", offers the vandalism rollback (for those without native rollback) and also has a "custom" rollback and an "AGF" rollback. That way, along with a message to the user, you can still easily undo non-constructive edits without calling them outright vandalism. You in fact do have Twinkle installed, and you didn't abuse it at all as far as I'm concerned. It's the other person's problem if they don't want to discuss with you. They did the right thing in piping the link to display differently. I'm not sure if there's an actual problem here now that it's all taken care of. Maybe everyone just needs some tea :) gwickwire^talk_editing 21:15, 21 April 2013 (UTC)

Thank you for the clarification. That particular editor has only been editing since March 10, 2013. But even us seasoned editors sometimes need clarification on how we handle edits. Much appreciated. — Maile (talk) 21:20, 21 April 2013 (UTC)

Title blacklist false positive

I'm trying to move Spring Break…Here to Party (with an ellipsis) to Spring Break...Here to Party (with three periods) per WP:ELLIPSIS, which discourages the use of the ellipsis character. But for some reason, the latter title is flagging the blacklist. However, I can create a page at Spring Break...Here to Party no problem. Any idea why this false positive is popping up? And can someone move the page for me? Ten Pound Hammer • ^{(What did I screw up now?)} 21:29, 21 April 2013 (UTC)

Moved. No idea why the title blacklist is catching it, though. —Theopolisme (talk) 22:02, 21 April 2013 (UTC)

Wait... how were you able to do it while TPH was not? It doesn't seem to me that reviewer, account creator, rollbacker, or course online volunteer would give a user extra ability to bypass the title blacklist. -- King of ♥ ♦ ♣ ♠ 22:13, 21 April 2013 (UTC)

Account creator has the ability to bypass the title blacklist, for technical reasons. With that said, could you actually remove the userright, since I'm not active with ACC anymore? —Theopolisme (talk) 22:17, 21 April 2013 (UTC)

Well the line that's catching it is .*\.\.\.H.* <moveonly> which was added by User:NawlinWiki in this edit. they're still active so you could ask them why it's there if your that bothered. Dpmuk (talk) 22:38, 21 April 2013 (UTC)

And right removed - somehow managed to hit save before putting in a reason so if that bothers you let me know and I'll add it and remove it again with a proper reason. Dpmuk (talk) 22:41, 21 April 2013 (UTC)

A "proper reason" would be great, if only to keep everything understandable. Thanks! —Theopolisme (talk) 01:09, 22 April 2013 (UTC)

Done King of ♥ ♦ ♣ ♠ 05:41, 22 April 2013 (UTC)

Can category presence be done by a template?

There are some really clever things that happen when executing templates. Can someone tell me if the presence of [[Category:Category name]] be detected and changed to [[:Category:Category name]]? (note the colon) I want to have such a function on {{AFC submission}} to stop AFCs from polluting content categories. -- Alan Liefting (talk - contribs) 07:54, 16 April 2013 (UTC)

Detection of things that aren't parameters in the template requires JavaScript. Ping me and we'll discuss it. Technical 13 (talk) 11:09, 16 April 2013 (UTC)

Could be detected with a LUA template. -- WOSlinker (talk) 11:41, 16 April 2013 (UTC)

It's "Lua", not "LUA" (cf. <http://www.lua.org/about.html#name>). --MZMcBride (talk) 21:27, 22 April 2013 (UTC)

[edited] I think I disagree. Not about Lua, but that Javascript is required.

{{#ifexist: Category:Non-talk pages that are automatically signed | Category:Non-talk pages that are automatically signed | Category:Non-talk pages that are automatically signed }}

→ Category:Non-talk pages that are automatically signed exists

{{#ifexist: Category:Category name | Category:Category name exists | Category:Category name doesn't exist }}

→ Category:Category name doesn't exist

See mw:Help:Extension:ParserFunctions.

—Trappist the monk (talk) 11:46, 16 April 2013 (UTC)

I'm not sure what you are getting at Trappist. The problem is that when new users are creating AfC articles, they are adding categorization to the pages. We have NO idea what categories they may add, but don't want any of the pages in content categories. It doesn't matter if the categories exist or not. What is wanted is something like:

if ((mw.config.get('wgNamespaceNumber') == 4 || mw.config.get('wgNamespaceNumber') == 5) && mw.config.get('wgTitle').match(/Articles for creation\//){
  if ($('div#catlinks')){
    /* Edit the page and replace /\[\[Category:(.*?)(\|?.*?)?\]\]/gi with [[:Category:$1$2]] */
  }
}

Although I would have to learn a lot more about writing JavaScript for MediaWiki to do that. Technical 13 (talk) 12:00, 16 April 2013 (UTC)

I've written Module:AFC submission catcheck and added it to Template:AFC submission. If any pages are found, they will be added to Category:AfC submissions with categories -- WOSlinker (talk) 12:14, 16 April 2013 (UTC)

I've tried it with Wikipedia talk:Articles for creation/Alexander Karasyov: while the obvious mainspace cats are now commented out, the page remains in the new "with categories" category you created. Perhaps further checks which cats are acceptable and which warrant an inclusion in that new cat are necessary. Fram (talk) 12:26, 16 April 2013 (UTC)

Unfortunatly, because of the way the templates are processed, there is a lag in detection and you need to edit twice to get it to notice the change. -- WOSlinker (talk) 12:33, 16 April 2013 (UTC)

Thanks. An additional null edit did the trick indeed. Useful cat, I used to search these through AWB but this makes life easier. Fram (talk) 12:59, 16 April 2013 (UTC)

WOSlinker, I would rather they never get into content categories at all (and I would hope everyone agrees on this!). We now have a tracking category sitting there with work to be done that may not get done (seen the WP:BACKLOG lately?). BTW, good effort tho. -- Alan Liefting (talk - contribs) 21:31, 16 April 2013 (UTC)

What we've got here is failure to communicate. As I consider Editor Alan Liefting's post a bit more, I wonder if there is a solution. If the goal is to prevent an article in the AFC process from being included in any categories until after the approval process has run its course, then the method I outlined will work if an editor implements it. I'm guessing that the same it true for Editor WOSlinker's solution and for your solution. The problem with all of these is that an editor must act.

If I understand how categories work, and I'm far from expert on them, articles will end up in categories whether anyone reads the article or not. As long as there is a [[Category:Category name]] wikilink in the article source, the article will be categorized. So isn't the question: How do you prevent an article in the AFC process from being categorized? Whatever the mechanism is, it must actually edit the article source to change [[Category:Category name]] to [[:Category:Category name]]. A robot that knows about {{AFC submission}}s? A special quarantine area for AFC articles that inhibits categorization?

—Trappist the monk (talk) 12:55, 16 April 2013 (UTC)

I don't care as to the mechanism. As long as AFCs don't get into content cats I am happy. Having a bot chase after AFCs are submitted that have had categories added is "messy". -- Alan Liefting (talk - contribs) 20:22, 16 April 2013 (UTC)

I think that's not possible without using a bot, and not too messy with a bot, even before submission, since AFC is a well-defined area. The more challenging task would be to de-activate all content category tags on articles developed in user space. — HHHIPPO 21:33, 16 April 2013 (UTC)

Unmm, hang on. Once an AFC is submitted and it contains content categories it is immediately visible by readers perusing categories since it will take a finite and possible lengthy amount of time to remove the categories. The issue of categories in user namespace is quite separate to the AFCs. The polluted category database reports help out with that problem. -- Alan Liefting (talk - contribs) 21:50, 16 April 2013 (UTC)

Incidentally, we have no way of differentiating content categories form maint categories. -- Alan Liefting (talk - contribs) 21:31, 16 April 2013 (UTC)

You're right. Identifying content categories is tricky. It can be done on a WikiProject level, but for all of Wikipedia it would be a nightmare. So you're suggesting to disallow all categories on AFCs? That would probably require an new namespace, probably also software changes. How about making another pollution report for categories that contain both main and project space pages? — HHHIPPO 06:27, 17 April 2013 (UTC)

Maintenance categories should all sit somewhere below Category:Wikipedia administration. -- John of Reading (talk) 07:06, 18 April 2013 (UTC)

Reflist script error?

This revision [39] of Patriots' Day looks fine. Then in this edit [40], two words, "of Hawaii", not even inside ref tags, got removed, and now the article [41] shows "Script error" for all the entries under the "References" heading. What's going on? —Lowellian (reply) 08:33, 16 April 2013 (UTC)

I don't see any errors (in refs or anywhere else, other than a redlink) in that or the version after it (which is current). —[AlanM1(talk)]— 09:08, 16 April 2013 (UTC)

Must have been some sort of transient server error, because the links no longer show the errors that were there earlier. —Lowellian (reply) 09:27, 16 April 2013 (UTC)

~~Check the history. I fixed it by using mdy dates... as it seems.~~ mabdul 11:07, 16 April 2013 (UTC)

strike that, seems to be an server error. mabdul 11:08, 16 April 2013 (UTC)

Neither {{reflist}} nor Cite use "Script error". -- Gadget850 (Ed)^talk 13:22, 16 April 2013 (UTC)

I saw the message: it was a LUA error message with extended error message when clicking it. Sadly I didn't copy and pasted it. mabdul 13:34, 16 April 2013 (UTC)

It's "Lua", not "LUA" (cf. <http://www.lua.org/about.html#name>). --MZMcBride (talk) 21:26, 22 April 2013 (UTC)

Extremely busy servers might hit Lua timeout to show Script error: We have been trying to get the Lua timeout-limit raised from 10 seconds to perhaps 30 seconds, because in very rare cases, when the servers get extremely busy for over 10 seconds during Lua processing, then the Lua clock is charged those 10 seconds, and the next Lua-based template (or #invoke of a Lua function) will detect a past-due timeout, such as 11.3 Lua seconds or such, and then all subsequent Lua functions might show "Script error" as the text stored into the cached form of the top-revision of a page. Unless the page cache is reformatted, such as by changing an embedded template/module, then the red "Script error" text will remain in the cached form (for top revision) and get pulled into each cache-copy which overlays a user's username and skin menus to generate the rendered webpage. In cases where the top-revision shows the Script-error text, try a null-edit. However, in edit-preview, just rerun as another edit-preview, because the occurrence is so rare, for an extremely busy server to be delayed nearly 10 seconds, during a Lua function. In fact, it is so rare that some people might insist "no one" really saw "Script error" in a page that typically uses "0.03" seconds of Lua time, when they really did. This is just a matter of getting the developers to understand how the Lua-clock elapsed time is really affected, can really stretch beyond 10-second timeout, on busy servers. I have two degrees in computer science, and it was a total shock to me as well, but the evidence keeps recurring. -Wikid77 (talk) 18:56, 17 April 2013 (UTC)

Subscript spacing

Has the vertical spacing of subscripts been adjusted recently? I have started to notice in Firefox that subscripts are very low down causing the line spacing to open up on lines that have subscripts. Possibly this is a Firefox rendering problem, but I'm sure it never used to be like this. Internet Explorer doesn't have this problem, the subscripts are almost aligned with the bottom of normal text. Spinning Spark 15:01, 17 April 2013 (UTC)

Perhaps depends on browser version: What version of Firefox browser are you running? Compare results and try different skins (&useskin=monobook):

For years, the height of superscripts^x seemed excessive, as sitting at the top of lowercase 'n'^x, compared to subscripts(x), which were only a half-letter lower (5px?) than flat text, as here: (joggy). Compared to subscript(Texting), the excessive height of some superscripts, perhaps caps^ABC would even chop the tops off letters in some skins, such as &useskin=monobook.

For years, the height of superscripts^x seemed excessive, as sitting at the top of lowercase 'n'^x, compared to subscripts _x, which were only a half-letter lower (5px?) than flat text, as here: _joggy. Compared to subscript _Texting, the excessive height of some superscripts, perhaps caps^ABC would even chop the tops off letters in some skins, such as &useskin=monobook.

Perhaps newer versions of Firefox lower the subscripts, whereas older versions of Firefox keep the same line-height, regardless of subscripts, and only superscripts had caused Firefox to lower the line level. What browser skins have you compared? -Wikid77 (talk) 17:36, 17 April 2013 (UTC)

I use the Monobook skin and am running Firefox 20.0.1. I have just tried Vector and Classic as well with the same result, that is good in IE (10.0.4) but too low in Firefox. Spinning Spark 22:34, 17 April 2013 (UTC)

Just checked on my notebook which is still running Firefox 19.0.2 (until I restart it). The subscripts look normal in that. Spinning Spark 22:40, 17 April 2013 (UTC)

I see no difference on MacOS X Mountain Lion with FF 19 vs FF 20. Perhaps it is a Windows only issue ? It can also be an issue with the font (or font selection) perhaps. —TheDJ (talk • contribs) 09:45, 18 April 2013 (UTC)

I have now let my notebook upgrade to FF20 so it is on the exact same version and there is no problem there. I have checked the advanced Font settings and they appear to be identical. The only significant difference is that the computer with a problem is Windows 7 and the notebook is XP. But if it's an issue with Windows 7, why is it ok in IE? Spinning Spark 10:55, 18 April 2013 (UTC)

Firefox uses a different method of hardware acceleration on Windows 7 vs Windows XP. On 7, Firefox uses DirectWrite to render fonts. DirectWrite is only available in DirectX 10 or greater, and that version requires at least Windows Vista. On XP, Firefox uses the older GDI technology to render fonts, which provides little to no acceleration.

As Internet Exlporer version 9 and greater also use DirectWrite, it is unlikely the bug is with Windows. It is most likely that there is a problem with Firefox's DirectWrite implementation, so you should probably file a bug at the Firefox Bugzilla. It is possible that Internet Explorer either uses DirectWrite differently or works around a bug in the software. However, that's something for the Firefox developers to figure out.
— trlkly 20:26, 22 April 2013 (UTC)

Why are new edits introducing seemingly random errors into previous page content?

Another user's recent edits to my user talk page have resulted in some peculiar changes to the existing content on that page, for example: diff. This is just related to that one user's edits. Can someone explain what is going on -- and can I/we prevent this from happening in the future? --Orlady (talk) 02:54, 18 April 2013 (UTC)

As a first step, I've posted at User talk:Gregbard#Strange damage to text to ask if there's a simple explanation. -- John of Reading (talk) 07:18, 18 April 2013 (UTC)

I am having intermittent problems with my computer. It is fairly rare, but every once in a while it happens. My apologies for this instance. Greg Bard (talk) 07:21, 18 April 2013 (UTC)

Thanks for looking into this. It appears that this could be some sort of a software problem -- systematically replacing certain characters or strings with something else. I thought it might be related to a strange incompatibility of user font preferences on Wikipedia, but it does sound like the issue was on the user's computer. Since it happened multiple times on my talk page and took an inordinate amount of effort to repair (I couldn't "undo" due to subsequent changes on the page), I wanted to make sure it wouldn't be happening again. I hope that's the case! --Orlady (talk) 14:28, 19 April 2013 (UTC)

The nature of the errors makes me think you likely have malware on your computer, Greg Bard. I am not sure if there is a Wikipedia essay on removing Malware, so I will point you to a guide I've seen multiple people use successfully: [42]. Please follow the instructions there to help make sure that this sort of thing doesn't happen again. Cheers. — trlkly 20:35, 22 April 2013 (UTC)

New deployment date for Wikidata phase 2 (infoboxes)

Hey :)

As previously announced Wikidata phase 2 will be deployed on English Wikipedia next. 11 Wikipedias have already been using it successfully for a few weeks. All other Wikipedias are scheduled to follow 2 days after English Wikipedia. There were some technical issues we had to fix that delayed the last deployment. We now have a new date for it: Monday next week, so 22 of April.

This version is basic and only the start of phase 2 here and will allow using all data from Wikidata, not just the language links as done so far here. This phase will not do any automatic changes to existing articles or infoboxes. It will only be used if editors here change their article accordingly.

As a reminder: We have prepared an FAQ for this deployment and are looking forward to your testing and feedback. You can already test it on test2.

And some more details about how this will work: There are two ways to access the data:

Use a parser function like {{#property:p169}} in the wiki text of the article on Yahoo!. This will return “Marissa Mayer” as she is the chief executive officer of the company.
For more complicated things you can use Lua. The documentation for this is here.

We are working on expanding the parser function so you can for example use {{#property:chief executive officer}} instead of {{#property:p169}}. The complete plan for this is here.

Thank you so much for helping us bring Wikidata to all Wikipedias!

Cheers --Lydia Pintscher (WMDE) (talk) 12:55, 19 April 2013 (UTC)

=) ·Add§hore· ^{Talk To Me!} 13:13, 19 April 2013 (UTC)

We really need some automated tools otherwise this will be a complete disaster. I can foresee a huge flood of e-mails to OTRS about this.--ukexpat (talk) 13:55, 19 April 2013 (UTC)

Because of the nature of this change, which if implemented would mean the migration of article content to another site with potentially differing policy on sourcing or inclusion, and which many editors will not know how to edit. Any and All suggested changes of infoboxes to this standard will need a very strong consensus.Nigel Ish (talk) 14:06, 19 April 2013 (UTC)

If the documentation for this phase is as inscrutable as the documentation for Phase 1, then Phase 2 is not ready and should not be released. See WP:Wikidata, Create a new item, and Item by title. From those Phase 1 pages it is clear that Phase 1 documentation is not done (not started) yet.

Wikidata is a complex subject. Non-technical editors will need to be led through the process in a way that they can understand. The documentation as it currently exists may be quite sufficient for those who developed Wikidata, but it is woefully inadequate for the non-technical editor.

At the risk of being overly redundant: Phase 1 isn't ready for release so even thinking about releasing Phase 2 is premature. Don't do it.

—Trappist the monk (talk) 15:14, 19 April 2013 (UTC)

Frankly, documentation is never going to be marked "done" - and actual, practial use of wikidata is likely to substantially change what the documentation needs to cover, as opposed to what it is (in advance) perceived to cover. Wikidata is a wiki - let Phase II be launched, let users flood in, and if they can't find how to do something then, when they work it out, they can add it to the manual. Ironholds (talk) 18:29, 20 April 2013 (UTC)

While I'm excited at the prospect of a central repository for data that is used in multiple articles (like properties of countries, people, companies, public offices, etc.), I, too, am concerned that articles will start to include {{#property:p169}} instead of Marissa Mayer because some small number of people know how to do it, much like the way ~~many~~ [some] templates are now maintainable only by those who speak Lua, which (I'm guessing), is an awful lot smaller number than those who speak WikiML. Until there is a simple (e.g. Ctrl-right-click on {{#property:p169}}) procedure to create/edit/find these values, extreme caution is necessary. —[AlanM1(talk)]— 23:32, 20 April 2013 (UTC)

There's a request for comment concerning how we should use Wikidata phase 2 at Wikipedia:Requests for comment/Wikidata Phase 2. Please leave your opinions there. MER-C 12:00, 22 April 2013 (UTC)

HTML5 video notice displaying in Firefox

On the video at Flag of Austria, the notice "For a better video playback experience we recommend a [ html5 video browser]." is being displayed, and the playback button links to the video file. File:Flag-of-Austria-320x240.ogg is Theora, which has been fully supported in Firefox for many years (this is backed up by the compatibility table at HTML5 video#Browser support). I am using the April 13 nightly build of 64-bit Firefox 23.0, which displays the browser's inbuilt Theora player when the video file is opened.

Where is this notice coming from? I'm slightly concerned about the placeholder "[ html5 video browser]" bit (complete with space inside bracket), as well as the fact that it is triggering itself on a browser that supports Theora natively. I've disabled Greasemonkey and do not have AdBlock or NoScript installed, so I don't believe it's an add-on issue. — TORTOISE WRATH 23:25, 21 April 2013 (UTC)

This message comes from mw:Extension:TimedMediaHandler, which is a MediaWiki extension. (In case you don´t kow what MediaWiki is, then it is a software that powers Wikipedia) The message is at MediaWiki:Mwe-embedplayer-for best experience.--Snaevar (talk) 15:12, 22 April 2013 (UTC)

Toolserver funkiness

Is Toolserver having fits? I had this same error some days ago, and then it went away before I had a chance to post over here. This only happens to me when I'm on Page History and click "External tools: Contributors". No problem with the other external tools there. But the error message it brings up is

"Database Error: User has exceeded the 'max_user_connections' resource (current value: 15) (sql-toolserver) on sql-toolserver/toolserver- failed to connect to WikiList database - failed to connect to WikiList database"

Doesn't matter what page I try that on. I started with Loyal Valley, Texas and got this, and then tried it on my own user page and got the same message. I get this on every page right now, if I click on "Contributors". Chances are, this will resolve itself eventually, but why is it happening? Could this be a glitch with Firefox 20.0.1? — Maile (talk) 23:27, 21 April 2013 (UTC)

Don't know what it would be (presumably some sort of issue), but it certainly looks like a server issue and not an issue with the Fx patch. — TORTOISE WRATH 23:35, 21 April 2013 (UTC)

Yeah, and the issue cleared itself right now. Wonder what's going on at Toolserver. — Maile (talk) 23:39, 21 April 2013 (UTC)

What's going on at Toolserver is a lack of funds. It's hosted by WMDE who have previously been happy to pay for it, but they don't want to carry on doing that. Plenty on this in the archives for this page - mainly in the last nine months. --Redrose64 (talk) 09:50, 22 April 2013 (UTC)

The Toolserver is bound to disappear during 2013-2014. All tools will move to the Tool Labs which should be much more stable. See mw:Tool Labs/Roadmap en. Darkdadaah (talk) 12:27, 22 April 2013 (UTC)

It's been completely down for the last few hours.--Gilderien Chat|List of good deeds 21:52, 22 April 2013 (UTC)

Userspace "template" border color issue

I am trying to create a "template" in my userspace to use as an award and I want to make the border gold. As you can see here, it is not working. Could somebody please explain what the problem is? AutomaticStrikeout (T • C • Sign AAPT) 01:09, 22 April 2013 (UTC)

border: 3px #FFFDF5 uses a default of "none" for the border-style, since you've left it unspecified. Try border: 3px solid #FFFDF5 instead. Anomie ⚔ 01:32, 22 April 2013 (UTC)

That actually didn't fix it, but I used border: 3px solid gold instead and that worked. AutomaticStrikeout (T • C • Sign AAPT) 01:39, 22 April 2013 (UTC)

Curious that it didn't work like that - the CSS 2.1 spec shows that colours may be specified in a number of ways, which include a six-figure hexadecimal RGB value, if that value is prefixed by the # sign, so #FFFDF5 should have worked. Anyway, in relation to that, #white is invalid - to specify the colour white, you have a number of options: these include white and #ffffff Others are available, such as #fff rgb(255,255,255) and rgb(100%,100%,100%). CSS 3.0 (which is not yet supported by all current browsers) also allows several other ways of specifying white, such as hsla(0,0,100%,1). --Redrose64 (talk) 09:47, 22 April 2013 (UTC)

Looking at the history, it did work. Except that AutomaticStrikeout was looking for something like #FFD700 ( ) rather than the very subtle #FFFDF5 ( ). Anomie ⚔ 10:50, 22 April 2013 (UTC)

CSS help

I need some help with Chrome and position: relative. See Template talk:Anchor#Positioning. -- Gadget850 (Ed)^talk 12:10, 22 April 2013 (UTC)

CAPTCHA

I just added a {{reflist}} to an article and had to enter a CAPTCHA. I did not add any EL to the article, but there was an EL in one of the references that appeared due to the {{reflist}}. I don't think that it is necessary to enter a CAPTCHA unless I also add an EL. If it is not a major expense, can this be fixed? Rgrds. --64.85.215.205 (talk) 21:25, 22 April 2013 (UTC)

The external link detector just looks at the rendered version of a page. This avoids people getting around being detected as adding a link if the link is embedded in a template, or partially embedded in a template, or broken up with <includeonly>, or whatever. But that also means you'll be seen as adding a link if you add a {{reflist}}, or happen to be the first to edit a page in the new year when some template generates links with {{CURRENTYEAR}}, or certain other edge cases like that. Anomie ⚔ 01:22, 23 April 2013 (UTC)

Too difficult to create username

(I will just copy this from another page):

You must change the policy. It is way too easy to trigger "username is too similar to existing". It happens when it is not that similar and isn't going to confuse anyone. People are running out of names. I tried 5 different names already. 174.236.79.80 (talk) 1:09 pm, Today (UTC−6)

I think name changes take a couple of weeks. You could create a temporary name and if you like Canoe1966 discuss it with Canoe1967. If they don't mind a name similar to theirs, then have admin change your temp one.--Canoe1967 (talk) 1:44 pm, Today (UTC−6)

The inconvenience discourages people from creating an account. 174.236.79.80 (talk) 2:02 pm, Today (UTC−6)

I see your point and sorry for just providing a work around. I don't know if the filter is easily adjustable or whether it needs a policy change. Someone on this page may know or at Wikipedia:Village pump (technical).--Canoe1967 (talk) 2:09 pm, Today (UTC−6)

(end of paste)

IIRC, account creators can override that limitation and create accounts with similar names, although I'd imagine they wouldn't create anything too similar. You can request an account at Wikipedia:Request an account. Chamal ^T•C 22:29, 20 April 2013 (UTC)

Thanks for the quick response. I think the OP feels that the filter may be causing us to lose editors though and may consider that as a work around. Is it adjustable or a policy thing? Could we add a message and link to Wikipedia:Request an account when IPs get the above message?--Canoe1967 (talk) 22:44, 20 April 2013 (UTC)

As an new ACC, I'm not even sure whether Canoe1966 is too similar or not. It's certainly reasonable to assume people will use different numbers after the same name to DAb them, but how similar is too similar (to 1967)? 1966? 1963? 1956? 19xx? 19? It does make more sense if, when told the username is too similar, to follow the ACC process if you feel the name is not too similar (versus creating another name and getting it changed later). It requires a lesser-privileged user, and is likely faster now that the ACC backlog is cleared. —[AlanM1(talk)]— 23:14, 20 April 2013 (UTC)

I just used Canoe1966 as an example in my work around before I knew Wikipedia:Request an account existed. We may wish to either put that link on the error page or loosen the filter. It seems the OP feels we are losing editors because the filter is too tight.--Canoe1967 (talk) 23:49, 20 April 2013 (UTC)

Balancing the caution due to malicious intent against the accessibility for those with beneficial intent is not that easy. Some people do try to create accounts that spoof an admin who blocked them or insult someone else. There is also the concern that people who are active contributors who have nigh the same user name could be confused for each other. Canoe1966 should technically be possible. If you were to attempt Canoel967 (using a lower-case letter L in place of the digit one) it should tell you it is too similar.
Without knowing what 174.236.79.80 (talk) was attempting to register it is impossible to be certain why 5 'too similar' notices were the results of 5 attempts to register an account.
Those with the admin or account creator user right can over-ride the spoofing checks and the title blacklist. (The latter because some proper names in other languages are profane slang in English and other like reasons.) The link to wikipedia:request an account is there but it is presented as an aid for those who have problems with the captcha. Surely it could be presented more clearly as an option for those having issue with the captcha or who believe they are getting a false positive on the spoofing check. delirious & lost ☯ ^~hugs~ 00:14, 22 April 2013 (UTC)

I see the point of the tight filter now. I know User:Jimbo online freaked me out when I first ran into him. Would it be an easy fix to add a more obvious link to wikipedia:request an account when a new name is rejected? I can't see it being controversial and it would be interesting to see if the number of requests increases after the change. This may indicate that we may have lost editors or have some named different than their first choice.--Canoe1967 (talk) 13:30, 23 April 2013 (UTC)

I took a screen shot of a failed account: File:Account rejected screen 2013.jpg. Can we add the link to wikipedia:request an account inside the pink box? Also what is the 'reason' box for? Does that need to be clarified?--Canoe1967 (talk) 14:48, 23 April 2013 (UTC)

I am going to move this to the bottom as a new issue.--Canoe1967 (talk) 17:16, 23 April 2013 (UTC)

Done here. Moved below.--Canoe1967 (talk) 17:16, 23 April 2013 (UTC)

Bug in undeleting 2 pages with many edits

Tracked in Phabricator
Task T45911

Yesterday I history-merged the pages RZA and GZA (not with each other!). In each, at the end I had to undelete a long list of edits, omitting a few of those edits because they are tramp redirect edits belonging to another editing thread. But whenever I tried that selective undelete, the system acted as if I had made a search among deleted pages, instead of undeleting. But an unselective undelete of all edits under that name, worked. This happened again just now. Anthony Appleyard (talk) 17:19, 21 April 2013 (UTC)
- I've experienced this problem as well, but I thought it was just a quirk of my screen reader and/or browsers (it happens in Internet Explorer 9, and something similar happens in Firefox (I can't remember the exact details)). Graham87 10:06, 22 April 2013 (UTC)
  - Move from ANI. Graham87 10:09, 22 April 2013 (UTC)
    - The problem you are reporting sounds like a potential issue in the code of the MediaWiki software or the server configuration. If the problem is reproducible, it would be nice if somebody who has this issue could send the software bug to the 'Bugzilla' bug tracker by following the instructions How to report a bug. This is to make developers of the software aware of the issue. If you have done so, please paste the number of the bug report (or the link) here, so others can also inform themselves about the bug's status. Thanks in advance! --AKlapper (WMF) (talk) 14:31, 22 April 2013 (UTC)
      - Yeah, I should've thought to serch Bugzilla first ... I thought the "invert selection" feature was specific to the English Wikipedia, and the problem could have therefore been fixed here. Obviously not; it's already been reported as bug 43911. Graham87 10:59, 23 April 2013 (UTC)

Hotcats and infoboxes

Apologies if this has been queried before, I couldn't see it on a quick scan through archived discussions... I note that HotCat doesn't seem to work with my browser when there is more than one navigation box template at the bottom of an article (and sometimes not when there is just one). Is this a known bug? I realise my browser is not the newest (Safari 5.1.7) but I would expect it's new enough that others will be facing the same problem. Thanks in advance - Grutness...wha? 12:38, 22 April 2013 (UTC)

You may wish to post your issue on Wikipedia talk:HotCat. Good luck! GoingBatty (talk) 23:57, 23 April 2013 (UTC)

Done - thanks. Grutness...wha?

Wikidata phase 2 is now live

Heya folks :)

It took a while but phase 2 (infoboxes) of Wikidata is now live here. This enables you to include data from Wikidata in infoboxes for example. For more details about this please see this blog entry. A request for comments is currently ongoing about how to make use of phase 2 on this Wikipedia exactly. Please follow it.

At the same time we have enabled a new feature that lets you add another language link here without having to go to Wikidata if there is no other language listed on Wikidata yet.

Please let me know if you see any issues related to this deployment or have further questions. --Lydia Pintscher (WMDE) (talk) 19:23, 22 April 2013 (UTC)

Where is the new feature please?--ukexpat (talk) 16:41, 23 April 2013 (UTC)

This was partially broken until some hours back, but we fixed it now. You can find it on any page without langlinks on the lower left as "Add links" (eg. on Kebithigollewa Divisional Secretariat). Please notice that it's only available for logged-in users at the moment - Hoo man (talk) 21:46, 23 April 2013 (UTC)

OK so that works for ILLs, is there anything for infobox fields? Thanks.--ukexpat (talk) 00:46, 24 April 2013 (UTC)

Mobile CentralNotice banner going out today to en.m.wp

Reminder that the mobile and fundraising teams are working together on a banner test this week on the English Wikipedia mobile site (en.m.wikipedia.org). The banner will only appear for mobile users who have opted into beta mode, are logged in, and are on an Android phone. In addition to testing out how CentralNotice code runs on mobile, we're using this as an opportunity to market the new Wikimedia Commons app for Android to logged in mobile Android users. You can see the banner here.

If you have any problems or you see the banner show up on the desktop site (this should not happen) please report them to meta:Help_talk:CentralNotice/Mobile. Thanks! Maryana (WMF) (talk) 18:49, 23 April 2013 (UTC)

Mobile CentralNotice Banner (on en.m.wp.o) -- Now Active

As stated above in #Mobile_CentralNotice_Banner_.28on_en.m.wp.o.29 -- banners are now live on the mobile site. If you have any problems please report them to meta:Help_talk:CentralNotice/Mobile. Mwalker (WMF) (talk) 23:40, 23 April 2013 (UTC)

Actually I lied -- we ran into some technical issues so have disabled the code again. Mwalker (WMF) (talk) 23:56, 23 April 2013 (UTC)

Visual editor

I had the visual editor enabled in my preferences for testing purposes, and it just changed from giving me an edit tab and a visual editor tab to an edit tab (that is the visual editor) and an edit source tab. Can I switch that back somehow? Ryan Vesey 05:02, 19 April 2013 (UTC)

https://en.wikipedia.org/wiki/Wikipedia:VisualEditor/Feedback might be a good place to ask, too. --AKlapper (WMF) (talk) 07:48, 19 April 2013 (UTC)

Tracked in Phabricator
Task T49396

I doubt you can change it back, this is probably in preparation of the full rollout of the editor. I notice that my e-shortcut key no longer works now, that's slightly annoying. And I wonder if we can find a better title than 'edit source'. —TheDJ (talk • contribs) 07:51, 19 April 2013 (UTC)

Bug 47396 concerns the shortcut issue.--Eloquence * 02:32, 20 April 2013 (UTC)

Is there any CSS code that will swap the functionality of "edit" and "edit source", so that "edit" functions as before and "edit source" is renamed "vis ed" and invokes the visual editor? Thanks--ukexpat (talk) 13:52, 19 April 2013 (UTC)

That's actually what I was looking for, is a bit of CSS code. Ryan Vesey 15:20, 19 April 2013 (UTC)

I've seen this also, and found it very confusing. The tab added should be "visual editor", which is explicit for what is after all not yet our usual editor, and does not yet have some key functionality, such as the ability to edit references. I don't want to have to add code to do this: it should be the default behavior. The way I've dealt with it is to remove the preference; I suspect others will do likewise, if the intent was to get more people to use it, it may have the opposite effect. Haven't we learned yet to stop making unannounced changes in the interface, including unannounced changes to widely used test features. — Preceding unsigned comment added by DGG (talk • contribs) 21:45, 19 April 2013 (UTC)

DGG, remember that this is an opt-in only alpha where you're specifically warned that it may be buggy or change. It's not really the same as making an unannounced change to a production interface that everybody depends on or has to opt out of to avoid. Steven Walling (WMF) • talk 19:23, 24 April 2013 (UTC)

The bug status says resolved, but it isn't for us dinosaurs who are still using Monobook - I still see the former "edit" and "edit source" tabs.--ukexpat (talk) 00:53, 24 April 2013 (UTC)

Two reasons for that: first, the last comment on the bug says that the change will not go live until Thursday, and second, that bug only concerns the lack of a tooltip or access key on the edit tabs, not the actual naming of the tabs. jcgoble3 (talk) 01:03, 24 April 2013 (UTC)

OK so back to the previous question: is there some CSS code that can do the "swap" as I asked above? Thanks.--ukexpat (talk) 13:24, 24 April 2013 (UTC)

Probably JavaScript actually. My best guess is:

( function ( $, undefined ) {
 jQuery(document).ready( function(){
$( '#ca-edit' ).find( 'a' ).text('VisualEditor');
$( '#ca-editsource' ).insertBefore($( '#ca-edit' )).find( 'a' ).text('Edit');
} );
} ( jQuery )); // End wrap with anonymous function

Unfortunately I think it gets into a race condition with the creation of the edit source tab so it only renames one of the tabs. There's probably an easy fix but I don't really have the time now, sorry. Hope that helps, - Jarry1250 ^{[Vacation needed]} 13:42, 24 April 2013 (UTC)

Brilliant, works for me. Thanks so much!--ukexpat (talk) 19:12, 24 April 2013 (UTC)

Are you sure you want to do this, Dave?

I'm getting "the page is asking you if you want to leave" notices now (using Firefox 5.0) when clicking 'back', or on a link, from an active (i.e. typed-in) edit window. I don't remember this happening before, was something changed? - The Bushranger _{One ping only} 20:38, 22 April 2013 (UTC)

I'm pretty sure this is a feature of your browser and not Wikipedia. Also, if you type in "Daisy" it sings a creepy song. Zad68 20:42, 22 April 2013 (UTC)

It's full of stars! But it hasn't previously done this on Wikipedia, and I haven't changed anything...odd. - The Bushranger _{One ping only} 20:47, 22 April 2013 (UTC)

I found on IE 6.0.2900 SP3, IE 8.0.6001, Chrome 26.0.1410 and FireFox 19.0.2 they all do it, you get the "Are you sure?" when you try to back out of an edited window... and I think you're right it didn't used to do that, but I've seen it doing it for a while, for what feels like at least weeks or months. Not sure now. Zad68 21:08, 22 April 2013 (UTC)

It's definitely done this for at least the past 5 months, and if my memory isn't shot then longer. I remember because there's tons of times I have hit my trackpad outside the editing window, and accidentally hit backspace and I get the saving grace of a warning first... I'm not sure that it's a big deal, but there's probably a JS/gadget to remove it, or if not I'm sure it wouldn't be too hard to make. gwickwire^talk_editing 21:20, 22 April 2013 (UTC)

That's very odd, because I now recall for a fact it wasn't doing it for me yesterday - accidentally hitting "preferences" instead of another tab (just above that) yesterday had me have to scramble to hit 'stop', but today gave me the Hal 9000 act. - The Bushranger _{One ping only} 21:29, 22 April 2013 (UTC)

With Safari on the Mac, I've been getting the "are you sure" at least since June 2012. Seems to just be a browser thing, although perhaps mediawiki has some influence on it (??). —Theopolisme (talk) 21:32, 22 April 2013 (UTC)

Yeah, this has been happening to me lately as well. There's an associated user preference at Special:Preferences#mw-prefsection-editing. I think it may have somehow gotten reset with a recent MediaWiki deployment. --MZMcBride (talk) 21:32, 22 April 2013 (UTC)

The warning requires Javascript to be active, if that's of interest. Johnuniq (talk) 23:26, 22 April 2013 (UTC)

Thanks for finding that. - The Bushranger _{One ping only} 02:18, 23 April 2013 (UTC)

As far as I know, this is a feature that is part of the Vector skin and has been around for as long as Vector. I know it's been around since at least 2010 when I began editing anonymously. jcgoble3 (talk) 01:44, 23 April 2013 (UTC)

Therein lies the rub - I use MonoBook. - The Bushranger _{One ping only} 02:18, 23 April 2013 (UTC)

personally, i love this feature and would not dream to turn it off, but if it bothers you, you can, under "Preferences => Editing => Warn me when I leave an edit page with unsaved changes". as to what happened recently - we can all guess that someone ported this feature from "Vector" to "Core", though this is just speculation, of course. peace - קיפודנחש (aka kipod) (talk) 05:36, 23 April 2013 (UTC)

It was introduced (in Spring 2010) with the Vector skin, where it has always been "opt-out" (as Kipod says, it's at Preferences → Editing → Warn me when I leave an edit page with unsaved changes); it became available in the Monobook skin at the same time, but was "opt-in". --Redrose64 (talk) 08:55, 23 April 2013 (UTC)

I'm suddenly getting it in Monobook on Firefox 18.0.2, which I have been on for some time. I have definitely NOT opted in, and I definitely do not want it. I associate this sort of pop-up with rather more sleazy sites where they want you to stay so they can advertise at you, not with reputable sites where the user is assumed to have at least half a brain. I thank you for your advice, and have opted out again - but would like to deliver a very wet trout to whoever changed the setting. Peridon (talk) 11:01, 23 April 2013 (UTC)

If you used to edit Wikipedia with scripting disabled, you would not have seen the warning. If you have recently enabled scripting, you will now see the warning (if the preference "Warn me when I leave an edit page with unsaved changes" is set, which presumably it is, as you are seeing the warning). Johnuniq (talk) 11:48, 23 April 2013 (UTC)

No, I haven't changed anything with regard to scripting. It was on before. Peridon (talk) 14:06, 23 April 2013 (UTC)

The feature was never "opt-in" for Monobook. The preference was there and checked by default, it just had no effect in skins other than Vector. Gerrit change 56881 moved the feature from the Vector extension to core, and at the same time made effective for all skins rather than just Vector; this change was deployed here as part of 1.22wmf2. Other Vector features, such as the collapsing of the "templates used in this page" and hidden categories lists at the bottom of the edit form, will likely get the same treatment at some point. Anomie ⚔ 12:36, 23 April 2013 (UTC)

Why are things that Vector has being put into skins like Monobook? Monobook users use it because they don't like Vector and the way it works. I at least object to having changes like this brought in sneakily to get Monobook to be like Vector. If I wanted Vector. I'd use Vector. I've no objection to Nanny's toys being available to people - just don't force them on the rest of us, those of us capable of deciding to leave a page all by ourselves. Peridon (talk) 14:06, 23 April 2013 (UTC)

forgive my french, but this is just silly. this feature is something that users come to expect from every editor: if i try to close my editor, and it does not matter if it's word, notepad++, kate, or my software development IDE, they will all try to warn me if i have unsaved changes that i'm about to lose by closing it.

this is such an elementary feature that i can't see why someone would whine about it when WP adds it to the functionality. the fact that you can opt out of it makes this whining even more bizarre. the fact that this feature did not work until now in monobook can only be classified as a bug, and whining about mw fixing a bug is not nice. peace - קיפודנחש (aka kipod) (talk) 15:48, 23 April 2013 (UTC)

That's not exactly the same: if you leave a WP edit page without saving the changes you can come back to that page and recover your unsaved changes through the browser history. That doesn't work for typical standalone editors. It even works if you close the browser tab with the edit page, but in this case you don't get a confirmation dialog (should we call this a leftover bug?). Anyway, I don't like it, but since it's opt-out I don't see a reason to call for any changes. (Firefox 20.0 on Ubuntu 12.4, Monobook) — HHHIPPO 19:06, 23 April 2013 (UTC)

if you use either chrome or ff on windoze, linux or android, you definitely get a warning when trying to close the tab (or, for that matter, close the browser). i will be very surprised to learn that it behaves differently for mac, opera or IE. as to the ability of your browser to recover lost work: this is very flimsy, and depends on browser, browser settings, and browsing mode. for instance, if you edit using "private browsing" (or "incognito browsing" in chrome lingo) this functionality is not there. קיפודנחש (aka kipod) (talk) 19:49, 23 April 2013 (UTC)

Sure, the browser history method is not safe enough to justify removing the feature or making it opt-in, it's just good enough for me to opt out. I tried it again with closing tabs, works indeed. What tricked me was that you don't get additional confirmation dialogs if you leave and return several times without making changes in the editor. I guess that's fine, so no leftover bug. — HHHIPPO 20:34, 23 April 2013 (UTC)

I've been getting this on Modern skin. I didn't opt-in to anything, but sure would like to know how to opt-out. It started a week or so ago on the tabs. Firefox 20.0.1. OK, it was just letting me know I was trying to close out of all the tabs at once. Today, it's been happening a great deal, just switching pages. Annoying to have it happen every time I want to leave a Wikipedia page. Please tell me how to opt-out on Modern for good. — Maile (talk) 15:01, 24 April 2013 (UTC) OK, i just went into "Preferences => Editing =>and unchecked "Warn me when I leave an edit page with unsaved changes". That that still doesn't explain why it started happening out of the blue on my browser. I didn't check the "warn me" opt-in recently (altho, admittedly, could have long ago) nor does it explain why it warns me about the tabs or, as of today, warns me when I just want to change pages without being in the edit window. What has Wikimedia done to set this in motion? — Maile (talk) 15:10, 24 April 2013 (UTC)

Adding to those reporting this as an annoying change. Some time ago, the behavior of preview was changed such that when one backed up out of preview to make further changes, the software did not remove the previously made changes, as it had sometimes often did; from my browser history I saw that it was substituting a moving forward to a new version as if I had made the new changes in the edit window below the preview before hitting preview again. That was very welcome and secured previously made changes against absent-mindedness. This gives at least the appearance of having negated that useful change. If it has, I'd like it undone, please. A scary message that has to be clicked on is in no way a good substitute for not undoing previously performed actions when someone previews - which is something we want editors to do. Yngvadottir (talk) 18:13, 24 April 2013 (UTC)

Preferences → Editing → Warn me when I leave an edit page with unsaved changes -- Gadget850^talk 18:22, 24 April 2013 (UTC)

Only 30 days per login AGAIN?

Can someone tell me why the maximum amount of time a user can remain logged in was reduced back to 30 days from the previous 180 days? Personally, I'd like to never be logged out (as it changes my skin and adds an extra two pages before I can edit), but 180 days was a nice compromise. Did some abuse happen that caused the duration to be reduced again?

I know it is technically possible, as I can still use a cookie editor to keep the login cookie from expiring. However, I had stopped having to do this when I was able to stay logged in for 180 days. I would like the cookie to be allowed to last at least that long. Heck, even 90 days would be superior to the current situation. Needing to log in once a month is far too frequent.

I note that Google also uses a 180 day limit, and the problems that can happen if someone steals your Google account information is much more extensive than happens if someone grabs your Wikipedia account. (Not that either has a remote chance of happening in my case.)

Thanks for looking into this. — trlkly 20:54, 22 April 2013 (UTC)

Hi. Yeah, I noticed this change as well and your post made me curious enough to investigate. The number of days is controlled by a global configuration variable called $wgCookieExpiration. The default value for this variable is 180 * 86400 in MediaWiki core. Wikimedia wikis are (now) overriding this value (cf. <https://noc.wikimedia.org/conf/CommonSettings.php.txt>: "$wgCookieExpiration = 30 * 86400;"). I pinpointed the change to gerrit:54505. I left a comment on that changeset asking Sam why this change was made, as there's no referenced associated bug report or comment explaining the change. --MZMcBride (talk) 21:56, 22 April 2013 (UTC)

Hmmm. I'm surprised it's 30 days, which is much longer than every other site I've ever logged into. Possibly a standard security measure? Actually, logic would suggest that the length of login should be linked to level of permission; so, "regular" editor xxx days, admin 24 hours, checkuser/oversighter/steward even shorter. Risker (talk) 22:11, 22 April 2013 (UTC)

It's set this way due to the privacy policy specifically saying that cookies will be "saved for up to 30 days". The previous period of login persistence of 180 days was unintentionally out of step with the policy. I know the legal team is slowly but surely working toward a draft of a new revision to the policy that they will unveil for a lengthy community discussion period, so keep an eye out for that and advocate for a change if you want something longer than 30 days. Steven Walling (WMF) • talk 22:30, 22 April 2013 (UTC)

Looking into the history a bit, it appears the default setting was 30 days when the current Privacy Policy was adopted in 2008. In August 2011, the default setting for Mediawiki was increased to 180 days (while citing an argument that 30 days was unusually short for comparable sites, though I haven't seen the original discussion). In December 2012, apparently someone noticed that the Privacy Policy and our cookie setting had become out of step, and since changing a legal document like the Privacy Policy is a pain in the ass, the solution now enacted was to restore the 30 days expiration for all Wikimedia sites. Dragons flight (talk) 22:50, 22 April 2013 (UTC)

Yes, that's all correct. We'll take this issue into account in the pending privacy policy revisions, but right now we have to deal with the limitation stated in the current policy -- the language unfortunately doesn't give us much flexibility here. :( --Eloquence * 00:20, 23 April 2013 (UTC)

I'm not sure where "December 2012" came from. Isn't gerrit:54505 the relevant change (from March 2013)? --MZMcBride (talk) 02:35, 23 April 2013 (UTC)

The discrepancy was mentioned on wikitech-l in December, including a comment that it was referred to legal. I don't know why it wasn't resolved till now. Dragons flight (talk) 02:59, 23 April 2013 (UTC)

Ah, I see now. Thanks. Relevant posts: [43] [44]. --MZMcBride (talk) 03:31, 23 April 2013 (UTC)

Thanks for the information. I hope the new privacy policy takes into account that people sometimes voluntarily give up a bit of privacy for convenience. — trlkly 12:47, 24 April 2013 (UTC)

CS errors

I have an impression that citation templates have been screwed up recently. For example older articles, particulalrly those without Harvard refs, like Liverpool F.C. now have citation style errors like Missing or empty |title= (while Liverpool F.C. for example does mention the related books, but in a separate ref section). Is there a way to fix it or all such refs should be reformatted? Brandmeister^talk 08:02, 23 April 2013 (UTC)

The new Citation templates have more advanced validation, and this has now been enabled for some of it's elements. This page is using shortened footnotes for some of the references, but without actually using the appropriate templates {{sfn}} for that style. If you follow the help of shortened footnotes, the errors will disappear. —TheDJ (talk • contribs) 08:13, 23 April 2013 (UTC)

PS, that was even explained in the 'help' link that follows the error. People really did put a lot of time in that help system, I think it will be a great new asset for our editors. —TheDJ (talk • contribs) 08:16, 23 April 2013 (UTC)

Note that {{sfn}} is not necessary for shortened footnotes, and gains you little over formatting them manually. And makes it less clear that there's a reference there, since the <ref> tags are hidden. Anomie ⚔ 12:42, 23 April 2013 (UTC)

Consider a ref that is presently formatted as <ref>{{cite book |last=Graham |page=14 |year=1985 }}</ref> - that is not how {{cite book}} was intended to be used. Instead this can be rewritten either as <ref>{{harvnb|Graham|1985|p=14}}</ref> or as {{sfn|Graham|1985|p=14}} --Redrose64 (talk) 10:17, 23 April 2013 (UTC)

<- just manually types it in when he uses shortened footnotes, per Anomie. --Izno (talk) 15:46, 23 April 2013 (UTC)

Shortened footnotes explains the various methods you can use, from manual to templates. -- Gadget850^talk 08:21, 24 April 2013 (UTC)

I think the templates include some redundant, moot parameters like penciller (in the refs for Terminator (franchise)), etc. Brandmeister^talk 22:06, 24 April 2013 (UTC)

First letter capitalization in article titles

According to Wikipedia:Naming conventions (technical restrictions)#Lower case first letter: "The MediaWiki software is configured so that a page title (as stored in the database) cannot begin with a lower-case letter." My question is simple ... Why?... Why is the software configured this way? Blueboar (talk) 12:21, 23 April 2013 (UTC)

As a guess I would say to prevent typos of lower case from showing up. I will phone Mr. Wales and see if there is a hidden agenda on it.--Canoe1967 (talk) 12:25, 23 April 2013 (UTC)

Well... Typos can be dealt with by simply moving the page to a corrected title. But in a few cases (the two most prominent being eBay and iPod) a lower case first letter isn't a typo. (and yes, I know that we can deal with such cases by adding the {{lowercase}} template - which makes the title display properly).

I know that having a lowercase first letter does not interfere with linking to or searching for articles... so I assume it must interfere with something else in the way the MediaWiki software works. Is my assumption correct? If so, what is it?... If not, is it be possible to change the software to allow lowercase first letters (where appropriate)? (note... I am not arguing that we should do this... I am asking if it is technically possible to do it). Blueboar (talk) 12:55, 23 April 2013 (UTC)

A lowercase first letter does not interfere with linking with the current configuration where lower case is automatically changed to upper case. Wiktionary allows both upper and lower case first letter (with lower being far more frequent). There you have to use the right case in wikilinks. For example, [[Dog]] makes a red link in Wiktionary while [[dog]] is blue. It would be very annoying and create loads of errors if Wikipedia had to pipe words like [[Dog|dog]]. The url http://en.wiktionary.org/wiki/Dog does redirect to http://en.wiktionary.org/wiki/dog with a delay. I don't know whether it's possible to configure MediaWiki so [[dog]] automatically goes to Dog if "Dog" exists and "dog" doesn't exist, but allowing that would create inconsistencies. Even if this configuration is possible, I wouldn't support it . PrimeHunter (talk) 13:40, 23 April 2013 (UTC)

Because 99.9% of "titles" of articles should be capitalized on the first letter, and thus removes the erroneous creation of a) wrongly-titled articles and/or b) duplicate titles? (✉→BWilkins←✎) 13:10, 23 April 2013 (UTC)

Page titles are configured to start with an uppercase letter in $wgCapitalLinks. Turning that option off would make page titles case-insensitive and it would also break case-sensitive links.--Snaevar (talk) 13:59, 23 April 2013 (UTC)

Ah... now we are getting somewhere... let me see if I understand. The only way to allow an initial lower case letter would be to turn off $wgCapitalLinks ... but that would mean making titles and links completely case-insensitive. While we would gain the ability to format eBay directly (using a lower case "e")... we would lose the ability to use capitalization as a disambiguation (the software would lose the ability to distinguish between Red Meat and Red meat... the software would treat both as "red meat"). Is this correct? Blueboar (talk) 15:15, 23 April 2013 (UTC)

No, it’s the other way round. It would make titles and links completely case-sensitive, so that we would have to pipe all links needing a different capitalization than in the page title (e.g., at the beginnings of sentences).—Emil J. 15:24, 23 April 2013 (UTC)

OK... I am confused again... The software currently distinguishes between Red meat and Red Meat... but not between EBay and eBay. So what is going on? Blueboar (talk) 15:28, 23 April 2013 (UTC)

Right; the case of the first letter in a title doesn't matter, but the case of any letter after that does matter. What we would "gain" by changing the flag is the ability to distinguish between eBay and EBay. Right now, because of that convention, those two article titles are really one and the same; they always go to the same page without a need for a redirect or anything, and they can't be separated. (On the eBay page, it would appear that the page is called "eBay" without the caps, but this is just a cosmetic change; internally, the software still refers to it as EBay.) If we turn that flag off, then "eBay" and "EBay" become separate pages. After that first letter, case always matters; EBay and Ebay will always be separate pages, whether that flag is enabled or not. We have Ebay currently redirecting to EBay, but that could be changed by simply editing the Ebay page. Right now, that flag is making the software treat the first letter differently from the rest of the letters; turning that flag off would make the software treat the first letter the same as the rest of the letters.

The question is: do we actually want the ability to distinguish the pages eBay and EBay? All it would mean, for the vast majority of articles, is that we would either have to create a redirect for every page, pointing from the uncapitalized title to the capitalized title or vice versa, or live with annoying piped links like BWilkins's example of [[Dog|dog]] (or [[dog|Dog]]). Plus, we already have a system to disambiguate those kinds of things. So, it's probably much more trouble than it's worth. Does that make sense? Writ Keeper ⚇♔ 15:32, 23 April 2013 (UTC)

No... my question is not "do we want to"... that's a style issue and not a technical issue. My question is: Is there something in the software that prevents us from doing so (whether we want to do so or not)? I am trying to understand the technical reasons why the software can distinguish between Red Meat and Red meat, but can not distinguish between iPad and IPad? Blueboar (talk) 15:43, 23 April 2013 (UTC)

Well then, the answer to your question is no. There is no strictly technical reason that the software cannot distinguish between iPad and IPad in the same way that it does Red Meat and Red meat. But, for stylistic reasons, we've told the software to ignore the difference between iPad and IPad, and so it does. We've told it this through the $wgCapitalLinks flag. So, because we've told it to do this, the software will now always act as though there is no difference between things like iPad and IPad, even though it would otherwise be possible for it to tell the difference (as wiktionary shows). Writ Keeper ⚇♔ 15:50, 23 April 2013 (UTC)

The parameter noted above by Snaevar prevents us from doing so. It makes only the first letter of a title case insensitive, as explained above. To change it would make the first letter case sensitive, but that would break hundreds upon hundreds of links probably. --Izno (talk) 15:51, 23 April 2013 (UTC)

The software itself doesn't distinguish between Red Meat and Red meat; what distinguishes them is that there are two different articles with those titles. It's the same reason that REd meat, ReD meat, Red mEat, Red meAt, and Red meaT are all redlinks. EVula // talk // ☯ // 16:27, 23 April 2013 (UTC)

I think I got it now... thanks. Blueboar (talk) 17:44, 23 April 2013 (UTC)

it's worth noting that the behavior is not always 100% clear or 100% obvious, and not always all part of the system work in total harmony. for instance, this "automatic first letter capitalization" does not work the way you may think it works for subpages: for instance, Wikipedia:Sandbox and Wikipedia:sandbox are the same page (or rather, the second is an "alias" for the first), but Special:MyPage/sandbox and Special:MyPage/Sandbox are not. This can create interesting hitches with some things: for instance, the "template sandbox" extension allows you to substitute sub-pages under the "sandbox prefix" for pages in other namespaces (e.g., if your "sandbox prefix" is "Special:MyPage/sandbox", the extension will substitute "Special:MyPage/sandbox/Module:Blah" for "Module:Blah"). however, even though first-letter-auto-capitalization is performed for "Module:Blah" on both page-name side and invocatin side (i.e., you can invoke it as {{#invoke:blah _or_ as {{#invoke:Blah, and both will work), same auto-capitalization is *not* performed on the subpages: if the subpage you create is "Special:MyPage/sandbox:Module:blah", it will not be substituted, because the system will auto-capitalize the module name on the invocation side, but will not auto-capitalize the subpage name itself (same failure mode for "Special:MyPage/sandbox:module:Blah"). hope this awkwardly-phrased caveat was not too obscure to actually understand what i meant... peace - קיפודנחש (aka kipod) (talk) 15:28, 24 April 2013 (UTC)

Correction: the "automatic first letter capitalization" does not worth the way you may think it works for subpages. :P The subpage name itself isn't the first letter of the article, so it of course (to me, at least) it would matter how it is capitalized. It's the same for websites: apple.com/iphone is the correct URL, but Apple.com/iphone works (because the top-level domain capitalization is irrelevant; aPPle.com/iphone works too), but apple.com/iPhone does not. Replace "top level domain" with "article title" and voila, it behaves exactly the same way. EVula // talk // ☯ // 00:07, 25 April 2013 (UTC)

Add link

I moved this plan from the "Too difficult to create username" section above.

I took a screen shot of a failed account: File:Account rejected screen 2013.jpg. Can we add the link to wikipedia:request an account inside the pink box? An IP editor claims to have had 5 names rejected and thinks we may lose editors. I don't think they knew that wikipedia:request an account existed. Can we claim 'this method usually takes just a few minutes' and a link to a 'Why similar names are rejected' page? Also what is the 'Reason' box for? Does that need to be clarified?--Canoe1967 (talk) 17:31, 23 April 2013 (UTC)

On the last question, the reason field is only shown to users who are creating an account through that form while logged into another account. The purpose is so that users can state why they are creating multiple accounts (i.e. what part of WP:SOCK#LEGIT does the new account fall under?). Users who are logged out or creating an account for the first time do not see the reason field. jcgoble3 (talk) 19:45, 23 April 2013 (UTC)

If you mean the time it takes to get an account created manually, "just a few minutes" isn't accurate, though it's better than it was. Probably depends on the time of day and other unpredictable factors. The current oldest non-handled request is about 13-hours old. —[AlanM1(talk)]— 20:33, 24 April 2013 (UTC)

I uploaded a new version that shows a screen while I was logged out (cache is slow and may show older version). Could the wording be changed in the pink box and a link to wikipedia:request an account in the pink box as well. It seems they need to fill all the fields out again as well as those captchas that many find difficult. We could also add hours/less than a day type statement. "Acoount creation usually averages less than 12 hours."--Canoe1967 (talk) 23:52, 24 April 2013 (UTC)

Not staying logged-in again

I reported at Wikipedia:Village pump (technical)/Archive 110 § Change in logged-in status that a change appeared to have occurred, allowing me to remain logged in for weeks at a time with a browser session that I didn't close (only hibernated when not in use). As of a few of days ago, it has reverted to its old behavior, requiring me to log in again every day or so. I did perform the latest Win7 and various add-on updates. I updated to FFox 20, though, before the behavior change. Has there been a server change to explain this? —[AlanM1(talk)]— 02:05, 24 April 2013 (UTC)

See above. --Redrose64 (talk) 07:03, 24 April 2013 (UTC)

I'm not sure that the "See above" thread is relevant, since AlanM1 is having to log in again "every day or so". -- John of Reading (talk) 08:02, 24 April 2013 (UTC)

Right. Can someone describe how the "stay-logged-in" process works? Which cookie is it, and how is it used? —[AlanM1(talk)]— 20:17, 24 April 2013 (UTC)

Mobile wikipedia: "read in other languages" links disappeared?

Resolved

Tracked in Phabricator
Task T49597

Is it intentional? --fireattack (talk) 07:02, 24 April 2013 (UTC)

I doubt it, I reported it in bugzilla. —TheDJ (talk • contribs) 07:33, 24 April 2013 (UTC)

Fixed and deployed. Max Semenik (talk) 15:12, 24 April 2013 (UTC)

Regex wizards wanted

Hey all. Trying to do a little text insertion and, since I'm still a relative newbie when it comes to regex, wondered if you could help. Say I have the following Wikicode:


== *DYNAMIC SECTION TITLE THAT WILL BE PASSED AS A VARIABLE* ==

{{TAFI nom|article=Mario Kart|class=C|pic=y}}
* Subset of the Mario franchise, and a hugely successful series of games in its own right.--[[User:Coin945|Coin945]] ([[User talk:Coin945|talk]]) 15:35, 1 February 2013 (UTC)
# '''Support''' --[[User:NickPenguin|<font color="darkgreen">Nick</font>]][[User talk:NickPenguin|<font color="darkblue">Penguin</font>]]<sub>([[Special:Contributions/NickPenguin|<font color="blue">'''contribs'''</font>]])</sub> 06:03, 7 February 2013 (UTC)
# '''Support''' --[[User:Horai 551|Horai]] [[User talk:Horai 551|551]] 08:55, 7 February 2013 (UTC)
# '''Support''' '''<span style="text-shadow:#808080 0.2em 0.2em 0.2em">[[User:ZappaOMati|<font color="#0000FF">Zappa</font>]][[Special:Contributions/ZappaOMati|<font color="#00FF00">O</font>]][[User talk:ZappaOMati|<font color="#FF0000">Mati</font>]]</span>''' 04:33, 8 February 2013 (UTC)

**I'D LIKE THE TEXT TO BE INSERTED HERE**

==*ANOTHER SECTION TITLE THAT WILL NOT BE PASSED AS A VARIABLE*

I imagine it has to do with some fancy lookahead-behind-and who-knows-where, but I don't really know where to begin. Any help would be appreciated! —Theopolisme (talk) 10:52, 24 April 2013 (UTC)

msg = "text i want to add"
header = "the header thing"
text = page.get()
regex = re.compile("""==\s*?{0}\s*?==(.*?)\n==""".format(header), flags=re.DOTALL)
newtext = regex.sub("""== {0} ==\1 {1}\n==""".format(header, msg), text)

Try that. Legoktm (talk) 11:24, 24 April 2013 (UTC)

In which programming language do you need the code? mabdul 06:23, 25 April 2013 (UTC)

Pending release of Echo (notifications)

Hey all :). Tomorrow, if things go according to plan (or next Tuesday if they don't!) we'll be launching Notifications, or Echo, which does what it says on the tin - offers a notifications system for Wikimedia projects. You can read more about what's in the release and what we've got planned on the project page. Once things are launched, we'll be gathering feedback on any bugs, annoyances or features people would like to see - if you'd like to get involved in that (and I sincerely hope you all will!) the best venue is the talkpage.

Thanks! Okeyes (WMF) (talk) 21:43, 24 April 2013 (UTC)

Super! Bennylin (talk) 21:59, 24 April 2013 (UTC)

It's been deferred to Tuesday 4/30.--Eloquence * 02:32, 25 April 2013 (UTC)

"Insert citation" in toolbar

Hi, I want to copy the code to insert citation via toolbar. I noticed it's a special feature not found in my wiki. I can't find it on the usual places (Common.js), anyone know where should I look? Bennylin (talk) 21:58, 24 April 2013 (UTC)

WP:RefToolbar -- Gadget850^talk 23:52, 24 April 2013 (UTC)

Help on creating category page

Hi all if someone could take a look at Category:Rollins College Wikipedians I'm running into 2 problems:

1 I inserted a table using the drop down but the userboxes are spreading out horizontally, again would like it to look like a 2 rowed table.
2 What is the wiki code for deactivating code, I would like to put the {{ and }} in the code at the top row of the table.

Thanks again. ^{Market St.⧏}_{⧐ Diamond Way} 02:42, 25 April 2013 (UTC)

This edit seems to have fixed it, though I confess that I don't know why. To stop the parser from interpreting wikicode, surround the text in question with <nowiki>...</nowiki>, as I have done to fix your post above. jcgoble3 (talk) 03:08, 25 April 2013 (UTC)

Thanks jcgoble3! This fixes it.^{Market St.⧏}_{⧐ Diamond Way} 03:38, 25 April 2013 (UTC)

Articles load improperly on Kindle

Tracked in Phabricator
Task T49499

Recently, I've noticed that articles no longer load properly on the older "paperwhite" Kindle that I often use. Just as an article completes loading the page disappears and only the search box with a left-arrow icon remain on screen. If the arrow is clicked the article will then reappear. This happens on every article every time. Have there been some recent global change to the underlying article markup or style sheet? TenebrisCaelo (talk) 18:59, 21 April 2013 (UTC)

Yes there have been changes, thank you for reporting this. Can you get more specific version information on your Kindle ? That would be helpful for the devs. —TheDJ (talk • contribs) 08:03, 22 April 2013 (UTC)

Picked a random article, and Bass guitar looks good on both the Kindle 3 and Kindle Fire. GoingBatty (talk) 23:52, 23 April 2013 (UTC)

It is called "Kindle 3" I believe, version 3.2.1; although in reviewing the bugzilla discussion, it appears that the problem has been found and corrected. This matches the behavior I noticed when I checked earlier today, every article I visited loaded properly. TenebrisCaelo (talk) 01:13, 26 April 2013 (UTC)

Twinkle not installed?

I can't seem to be able to use Twinkle on my Wikipedia account. I checked the box that says "Twinkle" in My Prefrences under Gadgets but there still is no TW at the top of articles. The page User:Missionedit/twinkleoptions.js does not exist and WP:Twinkle/Preferences says that Twinkle must be installed in order for me to change my Twinkle prefrences. I tryed purging my catche with no effect. I am using Internet Explorer version 8. Can you help? ~ Anastasia (talk) 19:57, 23 April 2013 (UTC)

And there's your problem, Twinkle doesn't work with IE8 or earlier. If you want to use it with IE you need to upgrade to IE9 or 10 or switch to another browser. NtheP (talk) 20:04, 23 April 2013 (UTC)

Oh, well there you have it. Thanks so much! ~ Anastasia (talk) 00:23, 24 April 2013 (UTC)

Can the preferences page be made sensitive to this to gray out and add an explanatory comment to the enabling checkbox when the user isn't running a compatible browser? —[AlanM1(talk)]— 20:26, 24 April 2013 (UTC)

WP:TW already states that Twinkle doesn't work in IE8. To get to WP:TWPREFS, you have to go via WP:TW, so this seems like it should be enough. — This, that and the other (talk) 06:34, 26 April 2013 (UTC)

Deleted special page and skins

Hello,

I have noticed over the past few days that this page does not exist. I managed to view it recently, and it can still be viewed on some smaller wikipedias.

I also have noticed that a few skins, such as nostalgia and amethyst, have been deleted too.

Why were they deleted?

Thank you! --The Million Rabbit 02:54, 24 April 2013 (UTC)

I believe I read that the Active Users page was too demanding on server resources. Those skins were eliminated so that fewer skins have to be supported when it comes to new features. Chris857 (talk) 03:48, 24 April 2013 (UTC)

Re Special:ActiveUsers - I can't find the discussion, but I definitely saw one, and it's basically what Chris857 said. Re skins - see also Wikipedia:Village pump (technical)/Archive 110#Classic skin and CSS and MediaWiki talk:Common.css#Obsolete skin files; I think there were others. --Redrose64 (talk) 07:09, 24 April 2013 (UTC)

For Special:ActiveUsers see also bugzilla:41078. --AKlapper (WMF) (talk) 10:11, 24 April 2013 (UTC)

Thank you for your answers. Should the skins be deleted from other wikipedias? --The Million Rabbit 02:58, 26 April 2013 (UTC)

Book deleter?

I've been using the book creator, & in the last week or so, I've had it wipe out my selections five times, for no apparent reason. If the page selection is interrupted, or the rendering is interrupted, it seems to reset & "lose" the rendering, & forces me to delete & add every page all over again. Last time, I left the rendering in progress overnight, & when I came back in the morning, the rendering was gone, & so was the list of pages. (Before that, at least the list survived.) The only connection I can find is my screensaver going on. Clearly, this system is buggy... TREKphiler ^{any time you're ready, Uhura} 22:30, 24 April 2013 (UTC)

Could you provide a specific example, and steps to reproduce for the problem? --AKlapper (WMF) (talk) 09:04, 25 April 2013 (UTC)

broken template

Could someone please fix {{repetition-inline}}; there is a trailing pipe "%7c" breaking the link. I couldn't fix it myself. Thanks. Fgnievinski (talk) 07:40, 25 April 2013 (UTC)

I've made an edit, though there may well be a better target for the "link" parameter. -- John of Reading (talk) 08:06, 25 April 2013 (UTC)

Technical proposal idea for the watchlist

While I was in the process of scanning for vandalism in my watchlist, I realized that I was having difficulty looking through my watchlist for vandalism due to the fact that all of the talk pages that I have on my watchlist are showing up as well. In effect, I had a thought, a thought I would like to run through this forum before starting some sort of "official" proposal, since I am under the assumption that this request might require the filing of a bug of some sort...

How feasible would it be to add a "Hide talk page edits" option to the watchlist? Steel1943 (talk) 05:14, 21 April 2013 (UTC)

Well, you can change it to look for specific namespaces; the default is "All". Just set it to "(Article)" and that'd be what you're asking for, I believe. EVula // talk // ☯ // 05:16, 21 April 2013 (UTC)

Not completely. That option only allows viewing of the pages in the "Article" space. I'm looking for an option that allows viewing of all spaces that do not have "talk" in their names all at once. Steel1943 (talk) 05:35, 21 April 2013 (UTC)

A smarter editor than I would be able to tweak User:Markhurd/hidetopcontrib.js, I bet. ~ Amory (u • t • c) 03:47, 23 April 2013 (UTC)

You can select a namespace and then invert the selection. This may help; i.e., it allows you to exclude all pages in the Talk namespace. --MZMcBride (talk) 21:33, 26 April 2013 (UTC)

A "refreshing" change to the CAPTCHA

Hi all,

Forgive my punning. This an announcement about a new feature associated with the CAPTCHA, namely a small link to refresh the CAPTCHA image without having to fully refresh the page. It looks like the screenshots to the right, and will appear everywhere that the CAPTCHA happens: account creation, the login form (if you fail too many times), and on the edit form given certain cases (like submitting an external link as an IP).

We anticipate that we'll deploy this update Thursday the 25th, barring any last minute technical issues. This is a global change affecting all wikis, but I wanted to drop a special note here because as far as I know, English Wikipedia has the highest rate of account creation requests, of which no small amount are due to an unreadable CAPTCHA. I am sincerely hoping that giving users to cycle through images at a reasonable rate without needing to submit their entire signup form again and again will reduce frustration, and perhaps even the queue for account creators. You can see even from the preceding VPT thread that people find our CAPTCHA annoying...

Many thanks, Steven Walling (WMF) • talk 21:31, 22 April 2013 (UTC)

Nice to finally see this. I'm wondering about the usability aspect of it however. 'Refresh' is rather vague (refresh what?). This is made worse because the element you are refreshing is actually to the left and above the refresh control, which is positioned in line with the inputbox, giving the user the impression it is part of the input box context rather than the 'captcha image' context. something to consider ? —TheDJ (talk • contribs) 21:40, 22 April 2013 (UTC)

Refresh, plus the icon, is fairly universally understood term. I agree the placement of the button isn't perfect, but it's very close, and the reason it's not closer to the image is because of the rather stupid way that FancyCaptcha spits out HTML for the whole thing. The number one place the CAPTCHA is seen is during account creation (so far as I can tell). In our upcoming redesign of that (mockup), we've improved the placement using borders and background color to more clearly associate the refresh with the CAPTCHA. Steven Walling (WMF) • talk 21:47, 22 April 2013 (UTC)

If the mockup is where we are going, then I'm good ! —TheDJ (talk • contribs) 07:52, 23 April 2013 (UTC)

I filed bugzilla:47713 for the placement point TheDJ raised above. Superm401 - Talk 08:43, 26 April 2013 (UTC)

Will there be some reasonable limit to the number of times the "refresh" link can be used, to prevent this from being a vector for brute-forcing (ie for a bot to keep on "refreshing" the CAPTCHA over and over until it finds one it can OCR, thus escaping the usual login attempt throttling)? -- The Anome (talk) 08:02, 23 April 2013 (UTC)

A better question would be: is there any reason to believe these CAPTCHAs are effective at preventing spam? Unlike reCAPTCHA and other implementations, the CAPTCHA used on Wikimedia wikis involves concatenating two common English words using a small, publicly accessible dictionary. The bots are undoubtedly trained to submit a best guess. If the answer is wrong, the CAPTCHA is automatically refreshed post-submit. This presumes that humans aren't being employed to solve the CAPTCHA. If that's the case, it becomes even more useless (arguably actively harmful if it's then only preventing the blind and having no impact on spammers). --MZMcBride (talk) 15:35, 23 April 2013 (UTC)

Anome: we got a review pre-merge of this change from Chris Steipp, the security engineer at WMF, and we agreed with that the throttling should be added outside/regardless of the refresh action, so it has the same effect on manual page refresh. As MZ points out, we have good reasons to suspect our CAPTCHAs are already a pretty weak first line of defense that cause a lot of people grief, and in fact we should probably test our assumptions in this matter at some point. But regardless of the current state of affairs regarding how effective the CAPTCHA is, this enhancement is for humans, many of whom find our CAPTCHAs incredibly frustrating. Steven Walling (WMF) • talk 16:51, 23 April 2013 (UTC)

Done This is deployed and live now. All seems kosher, but speak up if you catch any errors and we'll file the necessary bugs. Steven Walling (WMF) • talk 22:40, 25 April 2013 (UTC)

Colorized recent changes

I'm trying to interpret my user contributions page at Wikibooks. I can't find anything on the Wikibooks, Wikipedia, MediaWiki, or Meta projects, about this colorized recent changes scheme. The closest I got was some JavaScript mention...

Recent changes that are pending changes of history pages have a certain color to that item on the list. OK. Does the color have a general meaning as regards recent changes? Wikibooks user contributions pages have a color scheme that has the "pending" color, and it has other colors. — Cpiral Cpiral 01:20, 25 April 2013 (UTC)

Wikibooks has customized the recent changes to have tag markers coloured in pink. Is this what you mean, or are you referring to some other colour? This is a customization that the wikibooks folks did themselves, and is not a default part of the software. These tags are added by the abuse filter to mark edits which meet certain patterns. See Wikipedia:Tags for info on revision tagging. Bawolff (talk) 16:20, 25 April 2013 (UTC)

Wikibooks b:Special:Contributions/Cpiral reports green and tan today. (The tan looked to be near the color as mw:Special:PendingChanges when I first saw it.)

You said the colors are indicated in custom CSS, and I say "those criteria" that colorize the contributed items are not user-documented in such a way that an experienced editor can search to find within a reasonable amt of time. Is it possible to conclude here that Wikipedias Help:User contributions is missing a mention? It would be if there was an agreement of colors at meta. (As you say, such an agreement is surely not a concern of MediaWiki.) If there is not an agreement at meta, then Wikibooks needs to import our page, and update it with the color scheme. Its just wrong to give a requested report without an explanation of what it means. This is just a question of what work we might agree needs to be done to remedy that glaring omission. — Cpiral Cpiral 16:35, 26 April 2013 (UTC) — Cpiral Cpiral 18:28, 27 April 2013 (UTC)

Weird...

Was updating the centiJimbo userbox and stumbled on this in Jimbo's userpage history. The "older" edit was indeed to the userpage [45], but 2 years later. Any clue what's up with it? Just a weird ghost in the machine from all the software transitions back then? Screwed-up histmerge? — PinkAmpers &^{(Je vous invite à me parler)} 12:02, 25 April 2013 (UTC)

A known side effect of history merging. —TheDJ (talk • contribs) 13:50, 25 April 2013 (UTC)

General MediaWiki database corruption from 2004: There are other pages which show corrupted revision links from 2004. In this case, with page "User:Jimbo_Wales" the original entries were from January 2002, but it seems 81 of those oldid numbers, in range 7476700-7476782, were reassigned to later transactions of edits made during September/November 2004, which should have been higher id numbers. Compare some of those revisions around id=7476700:

dif670 dif680 dif690 dif700 dif710 dif720 dif730 dif740
dif750 dif760 dif770 dif780 dif784 dif790 dif800 dif800 dif800

I guess someone could hand-update those database records, to have the correct next/previous oldid links, but there are over 110 bad links in those database id numbers already. -Wikid77 (talk) 14:55, 25 April 2013 (UTC)

@TheDJ, it's not a side effect of history merging anymore, but it was before MediaWiki 1.5 was released in 2005 (see my musings on the subject). The history merge that ddid this appears to be this one; see the relevant deletion log (warning: ginormous page!) and search for the username "Fennec" without the quotes. Graham87 08:16, 26 April 2013 (UTC)