Wikipedia:Village pump (technical)

 Policy Technical Proposals Idea lab WMF Miscellaneous 

• Table of contents
• First discussion
• End of page
• New post

The technical section of the village pump is used to discuss technical issues about Wikipedia. Bug reports and feature requests should be made in Phabricator (see how to report a bug). Bugs with security implications should be reported differently (see how to report security bugs).

Newcomers to the technical village pump are encouraged to read these guidelines prior to posting here. If you want to report a JavaScript error, please follow this guideline. Questions about MediaWiki in general should be posted at the MediaWiki support desk. Discussions are automatically archived after remaining inactive for five days.

« Archives, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215

wrong sort by coordinates in class "wikitable sortable" table

This is how the table should like after the first click on the "Coordinates" column: coordinates must increase from the Northern Pole to the Southern. (The second click on each column displays the reverse order; if you tried to click anything here, return to the default view via refreshing screen with F5).

Meteorological hardware on Mars

Landing order	Mission	Coordinates	From	To	sols	Device	Zone
6	Phoenix	68.2188°N 125.7492°W / 68.2188; -125.7492	25.05.2008	28.10.2008	152	MET	subarctic
1	Viking-2	47.64°N 225.71°W / 47.64; -225.71	04.09.1976	12.04.1980	1281	(NASA)	moderate
10	Zhurong	25.1°N 109.9°E / 25.1; 109.9	22.05.2021		1193	MCS	moderate
2	Viking-1	22.27°N 312.05°E / 22.27; 312.05	20.07.1976	11.11.1982	2243	(NASA)	moderate
3	Pathfinder	19°7′48″N 33°13′12″W	04.07.1997	27.09.1997	83	ASI/MET	subtropical
9	Perseverance	18.4447°N 77.4508°E / 18.4447; 77.4508	18.02.2021		1291	MEDA	subtropical
8	InSight	4.5024°N 135.6234°E / 4.5024; 135.6234	26.11.2018		2077	TWINS	equatorial
5	Opportunity	1.9462°S 354.4734°E / -1.9462; 354.4734	25.01.2004	10.06.2018	5110	solar panel	equatorial
7	Curiosity	4.5895°S 137.4417°E / -4.5895; 137.4417	06.08.2012		4318	REMS	equatorial
4	Spirit	14.5684°S 175.472636°E / -14.5684; 175.472636	04.01.2004	01.05.2009	1892	solar panel	subtropical

However, now the sort engine of "wikitable sortable" treats the contents of cells with the {{coord|47.64|...}} template either as plain figures with a float decimal or as string constants. It brings the whole idea of sorting by coordinates to an absurd, because southern and nothern latitudes are mixed disregarding the value of the hemisphere parameter (N or S)

Foreseeing the questions like 'why latitude not longitude':
sorting by longitude (E or W) doesn't arise in the tables with geographic, climatic, astronomic etc. material. However the necessity to sort by latitude arises often when the readers to group the table contents. For these reasons I kindly ask the technicians to improve the "wikitable sortable" engine with a patch providing the accurate sorting by latitude (given the values are properly put into the "{{coord|...|...}}" template).

Programmatically this patch is short and easy:
- if the parser encounters the {{coord|...|...}} template in the cell =>
=> then it must search for additional hemisphere criteria (|N| or |S| enclosed in the straight brackets) =>
=> if the hemisphere is "S", the numeric key used for sorting does not change;
while if it equals "N", then the value must be converted to negative (multiply by –1).

Thank you in advance, Cherurbino (talk) 13:08, 16 October 2021 (UTC)

You may use data-sort-value="arbitrary value" to force a sort order if you are having issues, as in e.g. | data-sort-value="68.2188" | 68.2188 N X E or | data-sort-value="-68.2188" | 68.2188 S X W. Izno (talk) 13:19, 16 October 2021 (UTC)

thank you, but this code shall not work until the values inside the {{coord|...}} shall be converted into the numeric array from -90 (Northern Pole) to +90 (Southern Pole). This array should not be explicitly visible for the readers; it should be kept within the virtual code space Cherurbino (talk) 13:26, 16 October 2021 (UTC)

Or, if the trick is in the | data-sort-value="-68.2188" | 68.2188 S X W clause, could you show it directly by editing my sample here (or in my sandbox)& Cherurbino (talk) 13:28, 16 October 2021 (UTC)

Try that. --Izno (talk) 13:34, 16 October 2021 (UTC)

Of course, you can be less precise if you have few enough values that you don't think any of them will sort wrongly. I would suggest you avoid trying to stuff the longitude into the data-sort-value in any way since several digits of precision of the latitude should be enough to separate all the values cleanly, but if you really want you could probably make it work out somehow. Izno (talk) 13:37, 16 October 2021 (UTC)

Keeping the manual opened)) I kept on trying with the header line in my sandbox. However the pipe (straight quote "|") in your argument "-68.2188" | 68.2188 S X W messed all my table. Now I see, your solution is in inserting this clause in each line of the table. Of course, it shall work - but only for one table, while there are hundreds of other tables with the "coords" templates. This is a palliative, while I seeked for the universal solution for everybody. Thank you, anyway, for your time and kind attention Cherurbino (talk) 14:00, 16 October 2021 (UTC)

A "universal" solution is probably not forthcoming. At least one dev I know has complained about the complexity of the code in this area. :^) Izno (talk) 14:54, 16 October 2021 (UTC)

Could it be a good idea to amend the WP:TABLESORT help page with the sample for the coordinates' values? Cherurbino (talk) 14:08, 16 October 2021 (UTC)

Instead of a bare pipe | try using the special five-character sequence {{!}}. --Redrose64 🌹 (talk) 14:54, 16 October 2021 (UTC)

The table sorter is client-side JavaScript. That means it cannot see the wikisource or the template-expanded wikitext. It can only see the HTML sent to your browser. In your example, {{coord|68.2188|N|125.7492|W|globe:Mars}} becomes:

<style data-mw-deduplicate="TemplateStyles:r994658806">.mw-parser-output .geo-default,.mw-parser-output .geo-dms,.mw-parser-output .geo-dec{display:inline}.mw-parser-output .geo-nondefault,.mw-parser-output .geo-multi-punct{display:none}.mw-parser-output .longitude,.mw-parser-output .latitude{white-space:nowrap}</style><span class="plainlinks nourlexpansion"><a class="external text" href="//geohack.toolforge.org/geohack.php?pagename=Wikipedia:Village_pump_(technical)&params=68.2188_N_125.7492_W_globe:Mars"><span class="geo-nondefault"><span class="geo-dms" title="Maps, aerial photos, and other data for this location"><span class="latitude">68°13′08″N</span> <span class="longitude">125°44′57″W</span></span></span><span class="geo-multi-punct"> / </span><span class="geo-default"><span class="geo-dec" title="Maps, aerial photos, and other data for this location">68.2188°N 125.7492°W</span><span style="display:none"> / <span class="geo">68.2188; -125.7492</span></span></span></a></span>

That's the output of a specific template at a specific wiki. The table sorter is used by many wikis in many languages. I don't think it would be good use of developer time to try to make this work when it's only relevant for sortable columns of coordinates on both hemispheres. However, outputting a sortkey could be an optional feature of {{coord}}. Then there could also be a choice between North-South and West-East sorting. It can be suggested on the talk page. PrimeHunter (talk) 21:47, 16 October 2021 (UTC)

Distinguishing between the hemispheres is the language-indepenent issue. Pure mathematics common to the whole world. Now sorting by coordinates returns wrong output: north is mixed with south. Solution of this problem is purely mathematical: if table sorter comes across the 'coord' template, then values for 'N' within it should be treated as negative. Cherurbino (talk) 18:45, 20 October 2021 (UTC)

Were it "purely mathematical", you would not have multiple editors tell you that what you are asking for is not going to happen as you wish. I have provided an alternative as has PrimeHunter; I suggest that you look into one or the other. Izno (talk) 18:51, 20 October 2021 (UTC)

Were this problem unsolvable, we would not see the examples, that it is already solved in some of the wiki engines. Let's take the Template:Mars map and the services putting the 'dots' on this map. This template dosn't mix N and S; it means that somebody has already written the code assigning negative values for 'N' in order to pass this argument to the proc, that calculates X and Y offsets necessary to set these 'dots'. Cherurbino (talk) 19:01, 20 October 2021 (UTC)

I didn't say this problem was unsolvable, I said you weren't going to get what you want. Please do not put words in my mouth. Izno (talk) 19:05, 20 October 2021 (UTC)

Sorry, I also didn't mean personally you constatating the unsolvability. It was a mere rhetorics when the fact 'nobody solves' is spread within the undistinguished community; nothing pesonal. Meanwhile, another example of solutions already found are infoboxes like {{Infobox crater data. User wraps the coordinates into the 'coord' template, and passes then to some underlying procedure which converts N and S values into the numeric offsets to place red 'dots' over the canvas of MOLA map. Cherurbino (talk) 19:18, 20 October 2021 (UTC)

Re: you would not have multiple editors tell you that what you are asking for is not going to happen as you wish - the arguments of PrimeHunter were not based on the subject. He told me about 'multiple wikies' while distinguising N and S is not a matter of national localization, and the long snippet he quoted does not contain the code relevant to sorting procedure. Which occurs later, maybe at geohack.toolforge.org — Cherurbino (talk) 19:26, 20 October 2021 (UTC)

Dear Izno, I feel we shall not find the answer here, so I propose to close the discussion topic. Thank you for your attention, as well as for the palliative you proposed. Best regards, Cherurbino (talk) 19:32, 20 October 2021 (UTC)

@Cherurbino: As regards whether north or south is negative, there's a simple test. Omit the hemispheres from {{coord}}, and use negative values instead - see where you end up. For example, {{coord|-22.911|-43.205|type:city|display=inline}} → 22.911°S 43.205°W / -22.911; -43.205. So there you are, negative is south and we won't change that. --Redrose64 🌹 (talk) 20:52, 20 October 2021 (UTC)

Dear Redrose64! Thanks for your sample I found the way to override the problem which didn't allow me to use the numeric +/- syntax before:

source	code	result	comments
Cherurbino	{{coord|68.2188|N|125.7492|W|globe:Mars}}	68.2188°N 125.7492°W / 68.2188; -125.7492	present state
Redrose64	{{coord|-22.911|-43.205|type:city|display=inline}}	22.911°S 43.205°W / -22.911; -43.205	sample
Cherurbino	{{coord|68.2188|-125.7492|type:city|display=inline}}	68.2188°N 125.7492°W / 68.2188; -125.7492	test OK
Cherurbino	{{coord|68.2188|-125.7492|globe:Mars|display=inline}}	68.2188°N 125.7492°W / 68.2188; -125.7492	test fails due to the

The parser does not allow to use purely numeric coordinates' values for globe:Mars parameter! OK, I shall declare Mars as a kind of 'city'))). Thank you very-very much for this lifehack... and I gotta feeling that our respected colleagues Izno and PrimeHunter shall breathe a sigh of relief after such a boring interlocutor like me shall leave the thread completely satisfied that his problem was solved. Thank you, everybody! Cherurbino (talk) Cherurbino (talk) 11:51, 21 October 2021 (UTC)

You apparently didn't understand what I wrote. Let me repeat in bold: The table sorter is client-side JavaScript. That means it cannot see the wikisource or the template-expanded wikitext. It can only see the HTML sent to your browser. "client-side" means it runs in your own browser on your own computer. It doesn't run on a Wikimedia server where the wikitext is stored. You wrote "if table sorter comes across the 'coord' template". It doesn't. It never comes across any template. It only comes across HTML. I quoted the exact HTML it comes across in your example at the English Wikipedia. In other wikis there will be different HTML, and sometimes local terms instead of N and S. Coordinates can also be formatted in several ways. It would be difficult to make the table sorter work reasonably reliably on coordinates across languages and formats. I can think of many more useful table sorter features the developers could use time on. Here are ten:

1. Descend: Option to sort a column in descending order on the first click (often best for numbers).
2. Surname: Sort by surname, e.g. sort "first names surname" by "surname, first names" when no sortkey is given.
3. Ignore: Specify words/strings to ignore in sorting, e.g "more than", "less than", "circa" in number sorting.
4. Regex: Specify a regular expression replacement to apply to each cell to make the sortkey.
5. Empty: Option to sort empty cells at the end in both ascending and descending (empty often means missing data, usually no reason to sort it at top in ascending).
6. Hide: Hide selected rows in sorting (instead of just sortbottom).
7. Colspan: Option to give a colspan cell different sort keys for different columns.
8. Transpose: Make sortable rows instead of or in addition to sortable columns.
9. Restore: Restore the original display, including rowspans and sort-bottom rows in their original position.
10. Initial: Specify a column to initially sort by before clicking anything.

Some of these can be accomplished in specific tables with sort keys but so can your coordinates. PrimeHunter (talk) 21:31, 20 October 2021 (UTC)

Dear PrimeHunter! As you see above, the problem has found its solution. I apologize for dragging you into the discussion of the specific programming side of the problem. It was not useless for those who have read it, includind myself. Now I know more than before coming here, and everybody has got an example of finding non-trivial solution for the non-trivial problem. As for you personally: you demonstrated the high level of professionalism in your field, and Wikipedia must be proud of such volunteers like you. Thank you, everybody! Cherurbino (talk) Cherurbino (talk) 11:51, 21 October 2021 (UTC)

Updating infobox templates

Scenario: 10k articles use the same infobox template. That template has code that hasn't been changed for 10 years (consider it deprecated). New template is ready. What are some advices and tips for the best workflow which causes the less amount of chaos with articles? Asking for cases outside of EnWiki. - Klein Muçi (talk) 23:57, 21 October 2021 (UTC)

Tell us more about the changes. Are there incompatible parameters? Whatamidoing (WMF) (talk) 00:38, 22 October 2021 (UTC)

@Whatamidoing (WMF), at most cases, no. But let's assume I have around 10 infoboxes suffering from the same problem and some of they may have totally new code syntax, for example having changed from traditional wiki text to Lua module invocations.

Real situation: Admin at SqWiki. Some infoboxes are outdated, we want to update them mirroring EnWiki. It's been month we've postponed that because we're worried about the chaos that my continue from that. Today I finally got the courage and updated the infobox related to musical artists, which was transcluded in +700 articles. The number of parameters and their type was more or less the same but in English so what I did was import the EnWiki version in a temporary name in SqWiki, get all the articles that it was transcluded on, feed that list on AWB together with 10 find and replace regex-es which would change the parameters on them (just translate them on English basically) and let my robot take over. Removed the old template, move the new one to the old name. Voila!

I was wondering if there are any other better ways I'm missing out on. - Klein Muçi (talk) 00:50, 22 October 2021 (UTC)

I think – and someone should please correct me if I'm wrong – that all of the good solutions inspire editors to improve the article on Vaporware, and that your current appropach is the most typical, and best, of the existing (poor) options. Whatamidoing (WMF) (talk) 16:12, 26 October 2021 (UTC)

@Whatamidoing (WMF), thanks for the answer. I was thinking I never got an answer back in this post. Good to know I'm on the right track. :) - Klein Muçi (talk) 23:18, 30 October 2021 (UTC)

Another potential option, for future use, would be to make the old template a wrapper for the new template. This ensures compatibility with the new parameters (and means you don't miss anything). Once this is done, you create a subst-only version of the old template, which then can be subst'ed away and replaced by the new template. Primefac (talk) 07:57, 31 October 2021 (UTC) (please ping on reply)

Mapframe border disappeared

Following up from here, I'm continuing to have two issues with the Mapframe at Pomona College#Campus. The first (this thread) is that the boundary line has disappeared. The odd thing here is that it worked fine for a day or so when I first added it, but then it disappeared without any edits to the article/Wikidata item/OSM item I'm aware of, and it hasn't come back. The bug is replicated above at center.

The partial workaround I've found is to change |type=line to |type=shape, as shown above at right. However, even with the fill opacity set to zero, it still means anywhere in the interior of the campus becomes clickable with the title popping up. I'd much prefer it to return to the behavior from the line type before starting the FAC; does anyone know how to get that working again? Cheers, {{u|Sdkb}} ^talk 19:15, 23 October 2021 (UTC)

It's hard to say, but generally, I don't think shapes are supposed to be available in the line service. I'd be surprised if they were. Then again, our change in the osmdb sync has brought to light multiple changes already, so maybe there is an error. I really can't say. —TheDJ (talk • contribs) 12:54, 25 October 2021 (UTC)

@TheDJ, the OSM feature from which the boundary is (I think) being derived is way 458898308, which is a line rather than a shape. I know some other institutions have a relation in addition to a way; I don't really understand what the difference is between those or if that might be related to the issue. Also, I'm not sure what you're referring to with "osmdb sync". {{u|Sdkb}} ^talk 18:57, 25 October 2021 (UTC)

@TheDJ, okay, so I found phab:T285668. I'm still not quite sure what to do about this, though. Just drawing a line on the boundary rather than an infilled shape (using "service": "geoline" rather than "service": "geoshape") seems like something that ought to be possible, and it was possible until a week or so ago but now it seems it's not. {{u|Sdkb}} ^talk 20:42, 28 October 2021 (UTC)

That's not related, that ticket is about the toolforge OSM maps, which are way more broken than wikimedia maps at this point. There isn't a ticket about this particular issue yet, but i suspect it has something to do with the new way the import/update of the OSM database is being handled now. —TheDJ (talk • contribs) 20:51, 28 October 2021 (UTC)

Okay, thanks for the info; I created a phab ticket at phab:T294598. Cheers, {{u|Sdkb}} ^talk 21:08, 28 October 2021 (UTC)

Mapframe references don't display within descriptions

Following up from here, the second issue I've been having with the Mapframe at Pomona College#Campus (first in thread directly above) is that references don't display properly within the description field. Click on the pin at right to see how <ref>{{cite news|last1=Fitz|first1=Allison|date=14 February 2020|title=New Rains Center to be completed by 2022; 5Cs prepare for construction's effects|work=[[The Student Life]]|url=https://tsl.news/new-rains-center-to-be-remodeled/|access-date=2 December 2020}}</ref> gets mangled. The same thing happens with SFNs, which I'd also like to use. I haven't been able to make progress on this since last time, so it may require a phab ticket; if that's the case, I'm mostly looking to understand how to better present the issue to increase the likelihood it gets taken up. Thanks in advance for the help, {{u|Sdkb}} ^talk 19:15, 23 October 2021 (UTC)

Courtesy pinging those who commented previously: @Izno, Ɱ, TheDJ, Tgr (WMF), and Krinkle: see also the thread immediately above. {{u|Sdkb}} ^talk 19:15, 23 October 2021 (UTC)

@Sdkb T28213 is the general class this bug belongs to. Tgr (WMF) (talk) 01:04, 24 October 2021 (UTC)

This problem is because the templates used here just dump wikitext into geojson. But geojson doesn't support wikitext or html, geojson only supports text only descriptions. —TheDJ (talk • contribs) 13:05, 25 October 2021 (UTC)

The lua/template probably can use killMarkers to remove the references from the input. —TheDJ (talk • contribs) 13:10, 25 October 2021 (UTC)

I tried that, but mw.text.killMarkers and mw.text.unstrip don't seem to work any more (only mw.text.unstripNoWiki seems still functional). Is there any downside to removing the stripmarker with gsub? —  Jts1882 | talk  15:55, 25 October 2021 (UTC)

@TheDJ and Jts1882: Mapframe does seem able to handle images, so I think that means it's able to convert at least some types of wikitext into proper output. Would getting it to do the same with references be feasible, or is that out of the question? {{u|Sdkb}} ^talk 19:00, 25 October 2021 (UTC)

@Sdkb, Jts1882, and Tgr (WMF): I think I fixed it in the sandbox ? Markers include control characters and because control characters cannot be in json, these were getting stripped from the input. By moving that logic to later in the flow and using killMarkers on content explictely meant for the json body, I think we can work around both problems. —TheDJ (talk • contribs) 20:24, 25 October 2021 (UTC)

oh great, now the static map is broken again... Guess there is still a control char somewhere.... —TheDJ (talk • contribs) 20:29, 25 October 2021 (UTC)

ah interesting.. moved the coord a fraction and it fixed itself. I think i know why.. the livepreview saved a copy of the image with a different hash, causing an incorrect group reference being generated that the static tile renderer then cannot use... This likely happens a lot and I think i might already have a patch for that particular problem somewhere.

@TheDJ: I tried adding a ref within a direct call of mapframe HTML, as you suggested. It doesn't seem to have worked, though. Am I doing something wrong? {{u|Sdkb}} ^talk 23:34, 26 October 2021 (UTC)

@Sdkb: there were just a few errors in the json. Importantly, the json values need to be escaped (esp. "), but also geojson flips the order of the latitude and longitude (apparently that's a thing in Cartesian geometry). —TheDJ (talk • contribs) 08:21, 27 October 2021 (UTC)

References

Ref parameter

For anyone following along, after some discussion with TheDJ and others on Discord (around here), we'd like to add reference parameters (e.g. |ref2=, |ref3=) to Module:Mapframe. These would add <ref>reference content</ref> into the json appended to the end of the description parameter. This would bypass a parse cycle and make it so that the ref tag wouldn't be escaped when the JSON is made, resolving the bug. If anyone who knows Lua could help to code this, I would be grateful! {{u|Sdkb}} ^talk 07:55, 1 November 2021 (UTC)

@Sdkb: Does this edit do what you want (see right)? I was puzzled why it didn't work until I saw that the stripmarkers were being stripped out in recent changes. —  Jts1882 | talk  11:53, 1 November 2021 (UTC)

References

Rollback on mobile devices

On two occasions lately I have fat-fingered rollback. On both occasions a dialog box opened that asked me if I really wanted to do that. On both occasions I answered “Cancel”, but it performed the rollback anyway. I need to disable this for myself, as I don’t tend to use rollback and obviously the option is causing more harm than good, but I thought maybe a technical-minded person could look into it. Thank you! 78.26 _{(spin me / revolutions)} 02:57, 26 October 2021 (UTC)

@78.26: before diving in any further, would you please check Special:Preferences#mw-prefsection-gadgets and confirm if your option for Require confirmation before performing rollback on mobile devices is checked? — xaosflux ^Talk 17:07, 26 October 2021 (UTC)

@Xaosflux: Yes, it is, thank you for asking. And in fact, the confirmation is popping up just like it should. The problem is that when "Cancel" is selected, the rollback is performed anyway. 78.26 _{(spin me / revolutions)} 17:10, 26 October 2021 (UTC)

@78.26: thank you, can you provide an example of the last time this occurred? I suspected it might have been Special:AbuseFilter/examine/1436794383? Can you be a bit more specific about your interface? Are you using the mobile application (ios or android), or the mobile web? If using a mobile device and web, are you using "desktop view"? — xaosflux ^Talk 17:23, 26 October 2021 (UTC)

@78.26 Which browser, device, and skin are you using? — MusikAnimal ^talk 17:25, 26 October 2021 (UTC)

@Xaosflux:, @MusikAnimal:, the last two examples are: [1] and [2]. This is on an Apple iPhone 6s, Software Version 14.6., Safari browser, "desktop view" (sorry, I really don't like mobile view). Does that help? 78.26 _{(spin me / revolutions)} 17:35, 26 October 2021 (UTC)

@78.26: using that same browser, would you mind sharing your useragent? You can fetch that from sites such as Google using this link. Feel free to redact any version numbers in there. — xaosflux ^Talk 17:52, 26 October 2021 (UTC)

Just a note, this is MediaWiki:Gadget-confirmationRollback-mobile.js. — xaosflux ^Talk 17:56, 26 October 2021 (UTC)

Mozilla/5.0 (iPhone; CPU iPhone OS 14_6 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko). 78.26 _{(spin me / revolutions)} 18:04, 26 October 2021 (UTC)

@78.26 Thanks. I unfortunately have no way of testing the behaviour on an iPhone, but I do recall someone reporting this same issue before and they were also using an iPhone. Could you try using a different browser, such as Chrome or Firefox? That will at least narrow it down to a Safari issue. The gadget itself is extremely simple and should in theory work in virtually all browsers. — MusikAnimal ^talk 20:50, 26 October 2021 (UTC)

Allright, I'll make some test edits with my alternate account and then try to "not-rollback" on both the Safari browser and Chrome. (I've never used Chrome on my mobile device, guess I'll see how I like it...) I'll ping you with the result. 78.26 _{(spin me / revolutions)} 20:56, 26 October 2021 (UTC)

@MusikAnimal: I presume this is bad news, but on both browsers (Safari, Chrome, both using Desktop view) I hit "rollback" and selected "Cancel" each time. I never selected "OK" on either browser. If you check my edit history you'll see the discouraging results. 78.26 _{(spin me / revolutions)} 21:14, 26 October 2021 (UTC)

@78.26: Yes, bad news :( That must mean it's an iPhone issue, I suppose. I will try to find a device (or virtual device) to debug this further. For now, since you say you rarely use rollback, you could hide the links entirely with User:MusikAnimal/rollbackTouch.js. — MusikAnimal ^talk 21:34, 26 October 2021 (UTC)

@MusikAnimal:Thanks! My next question was going to be if I could disable this (you can't in Preferences, already looked) because I can think of twice I've used rollback on purpose. Thank you also, Xaosflux. 78.26 _{(spin me / revolutions)} 21:38, 26 October 2021 (UTC)

If you never use rollback on desktop either, you can simply hide them by adding

.mw-rollback-link {
	display: none;
}

to your common.css. That would be simpler and faster than installing that script or enabling the gadget. Nardog (talk) 21:53, 26 October 2021 (UTC)

Yeah, I was going to suggest that, but it leaves a lingering pipe character next to "undo". In order to remove that you need to remove the outer span which doesn't have a CSS class or ID. We could hide the first span under .mw-changeslist-links, but I'm not sure that accounts for rollback links in all places, or if there are cases where rollback isn't in the first span. Finally, the script only targets mobile devices (though 78.26 made it clear they don't use rollback on any device). But anyway, yes, a CSS-only solution is always more efficient. — MusikAnimal ^talk 22:22, 26 October 2021 (UTC)

That's only true on history. On Contributions, Watchlist and RecentChanges the whole list item gets removed. It also has the problem of not taking effect after a live update. Nardog (talk) 22:27, 26 October 2021 (UTC)

Bah! Okay I've removed the .parent() (which I only added just before I commented here). A pity that the HTML isn't consistent across change lists. Thanks for pointing that out! I can do the special handling for history pages, I suppose.

Good point also about live update. This script I believe predates the live feature of Watchlist and RecentChanges, though of course other scripts/gadgets could dynamically add rollback links as well. I'll fix that, but for anyone reading this who wants to hide rollback links forever and always, like 78.26, use the CSS Nardog mentioned above :) I assume for most people the stray pipe characters on history pages is not much of a nuisance. — MusikAnimal ^talk 22:38, 26 October 2021 (UTC)

It would be simpler if you just added a class to <body> if the user agent indicates a mobile device and hid rollback links under that class in CSS. Nardog (talk) 22:51, 26 October 2021 (UTC)

@Nardog Clever! That's a pure JS solution (assuming you'll use mw.util.addCSS(), so there's still a bit of delay as opposed to pure CSS, but that seems more than acceptable here. I'll give this a go when I find the time. If you wanted to write something yourself I'll happily review it and merge it into the gadget. Just ping me :) — MusikAnimal ^talk 05:38, 27 October 2021 (UTC)

It's really weird that the rollback gets performed despite choosing Cancel, as opposed to the gadget not taking effect at all. I can totally see the gadget failing when used on Ajax-updated pages though, since it's attaching the handler to .mw-rollback-link just once, directly. I think it should either make use of event delegation as in $(document.body).on('click', '.mw-rollback-link', function (e) {... or hook it to wikipage.content and attach the handler every time the page is updated.

$(".mw-rollback-link").parent().remove() removes not just the links but the whole list items with them (i.e. .parent() is unnecessary). But I don't understand why you wouldn't just hide them in CSS. Nardog (talk) 21:53, 26 October 2021 (UTC)

Only after this lengthy discussion did I recall meta:WMDE Technical Wishes/Rollback which has been around for a while now. In Preferences > Appearance, look for "Show a confirmation prompt when clicking on a rollback link". This makes it work much like the "Thank" button which requires a second click to confirm. However there appears to be some UI issues in the Advanced Mode on mobile. But for those who only work off of the desktop UI (be it on a mobile device or otherwise), this is a nice solution to the problem. — MusikAnimal ^talk 05:47, 27 October 2021 (UTC)

@MusikAnimal: sorta - that will also force the link when not "on mobile" too. — xaosflux ^Talk 12:30, 27 October 2021 (UTC)

Difficulty in editing templates using the VisualEditor

Sorry, I'm not sure if this is the right place for this message but here goes. The VisualEditor is a more accesible way of editing and I really think we should make as many of our pages work with it as is possible. I have come across various templates such as List of listed buildings in Stonehaven, Aberdeenshire, aviation destination lists on airport pages, and multi-column templates, that make it harder or impossible to edit and add references using the VisualEditor. In frustration, I created a deletion discussion at Wikipedia:Templates_for_discussion/Log/2021_July_11#Template:Airport_destination_list but it was kept, and to be honest I felt the keep arguments were unconvincing.

I suppose I want to see if anyone has the same feelings as me here, and if an argument could be made for removing these templates on the grounds of their negative impact on accessibility. Especially in the case of templates creating tables, I just don't think it's beneficial as experienced editors still know how to edit table syntax in the source code editor, but new editors or those who find the VisualEditor easier to use are locked out from editing. Could multiple columns be created using a div and CSS, avoiding the need for a template? If we want new editors to cite their sources, we need to make it easy for them to do so. NemesisAT (talk) 11:39, 26 October 2021 (UTC)

I don't use VE, but doesn't {{Airport destination list}}'s documentation need a TemplateData section in order to make it usable with VE? I thought that was required, or at least required for ease of use. {{HS listed building row}}, which is used in the list article, is also lacking TemplateData code. – Jonesey95 (talk) 16:11, 26 October 2021 (UTC)

I'm unable to experiment with TemplateData on {{Airport destination list}} as it has been locked, however my gripe with it was that you can't add citations, or indeed any formatting, unless you know the wikicode off-by-heart at which point you may as well use the source editor anyway. In fact, I believe it is not possible to add a citation or use WYSIWYG editing for any template paramter within the VisualEditor. Regarding the listed buildings template, that article is actually one template containing many more within it. It appears to be too much for the VisualEditor to load but again even if it did load, as the entire article is located within template parameters it is impossible to use VisualEditor tools to edit it, you're stuck with using source code within the VisualEditor. Hopefully this makes sense! TemplateData is good but I don't think it would resolve the issues I'm facing here. NemesisAT (talk) 16:20, 26 October 2021 (UTC)

Somewhat perversely, TemplateData code is stored in the templates' documentation pages, which are not protected, so you should be able to add it yourself, to each of the templates in question. I don't think it makes sense to assume that the templates would not work with VE if they were properly equipped with TemplateData code; try it first before giving up. For help in adding that code, your best bet is probably Wikipedia:Help desk. – Jonesey95 (talk) 16:37, 26 October 2021 (UTC)

Ah of course, I forgot that it's on the documentation page. I'll have a look. NemesisAT (talk) 16:57, 26 October 2021 (UTC)

I have started a TemplateData for {{Airport destination list}}, but it will never be able to help you with adding a reference as the reference tools are all in the main VisualEditor toolbar, whereas the parameters of a template can only be edited in plain text. I should note that I am able to use the source code editor, in fact, I used to be opposed to the VisualEditor (both on here and on Wikia), but as it has developed, I've realised it is much easier for editing large tables and long pages with lots of references in them. NemesisAT (talk) 17:49, 26 October 2021 (UTC)

Three thoughts:

• @Johanna Strodt (WMDE) will want to see this, since WMDE is currently working on how to edit templates in the visual editor.
• Most table-oriented templates were created a long time ago. MediaWiki originally used plain old HTML to create tables (a pair of <tr> tags for each row, a pair of <td> tags for each cell, etc.) The creation of a wikitext syntax (|-for a new row, || to separate cells on the same line) for made it easier, but it's still hard, especially when the table is large, or if you want any colors or other fancy formatting. Editors created table templates because it was easier than the wikitext for large or heavily formatted tables. However, now that the visual editor handles all of "code" parts for you, it is generally easier to edit tables in the visual editor than to wrangle templates (unless you want colored text). I particularly recommend the visual editor to anyone who needs to add, remove, or rearrange the columns in an existing table.
• You can add and edit templates even if a TemplateData section doesn't exist. It's just (usually) more difficult.

Whatamidoing (WMF) (talk) 16:51, 26 October 2021 (UTC)

That's interesting, I didn't know about the history of the table syntax. NemesisAT (talk) 17:51, 26 October 2021 (UTC)

The specific WMDE link is meta:WMDE Technical Wishes/VisualEditor template dialog improvements. {{u|Sdkb}} ^talk 17:42, 27 October 2021 (UTC)

Late to the party but... sending a template to TfD on the grounds that it's difficult to edit in VE could be seen as WP:POINTy. --Redrose64 🌹 (talk) 20:17, 26 October 2021 (UTC)

Here are some scatter brain thoughts (without reviewing that TFD):

1. Templates can ensure intra-article consistency. While that is not something we necessarily need to preserve, it can be a nice benefit of table templates.
2. Templates can indicate the good 'default' information to include. (A drawback is sometimes that's an infobox with 50 parameters. ;)
3. Templates can restrict what is included, e.g. as in fan-driven wiki articles.
4. They are easier even than wiki syntax to deal with by experienced editors. Just because we know how doesn't mean we prefer to know.

They have detracting points too of course.

1. VE dislikes them. (BTW the Phab task you are looking for is phab:T52182 and related. It's old.)
2. Do we really need standardization across all such articles for all such tables?

I know in the past that some of table templates have been deleted as being generally unnecessary, but certainly not all. And those have mostly been table-start, table-row, table-end type templates, not templates that compose the entire construct.

Another Phab task of interest might be phab:T54582 for the specific template.

A specific point: Could multiple columns be created using a div and CSS Yes, but we don't want users to do that. It's fundamentally gross and requires inline CSS. In the case of reference blocks, one might consider advocating to remove {{reflist}} in the general case in lieu of <references/> since most people probably don't need what it provides anymore (columns of references), and where they do they could instead use {{div col}}, which wouldn't make VE sad. Making general columns like {{col-begin}} does is somewhere on my to do list, but I don't know when I'll get to a nice version. Izno (talk) 17:00, 27 October 2021 (UTC)

Media Search is available for all users on Wikimedia Commons

Hello! We wanted to make a gentle reminder that, since last May, there is a new default way to search files and other media on Commons. Special:MediaSearch has been developed in the past year to make use of the structured data on Commons and information from Wikidata in order to provide users with more relevant and extensive results, regardless of the language used for the query.

The new interface allows users to access some of the file information (licensing, description…) directly on the page, gives users various options to better filter results, and relies on Wikidata's concept-linking abilities and multilingual labels. Special:Search remains available for users for cases when its purposeful wikitext-based search may be more appropriate to use.

MediaSearch is just one part of the larger Structured Data Across Wikimedia project, which aims to structure content on wikitext pages, to make reading, editing, and searching easier and more accessible across projects and on the Internet. If you want to learn more about the project, please visit the project pages on MediaWiki, subscribe to the SDAW Newsletter, or get in touch with us on the project's talk page. The Structured Data team looks forward to continuing work on opening up Wikimedia content. Thanks! -- Sannita (WMF) (talk) 13:58, 27 October 2021 (UTC)

Number of editors per project

In one discussion somewhere I saw a graph of active editors in English Wikipedia compared to those of French Wikipedia. I think the graphs were a screenshot of one of the autogenerated statistics pages. Anyone know what that page might have been and could you provide a link? Regards, A Thousand Words (talk) 06:07, 28 October 2021 (UTC)

I have no idea about that (I don't remember seeing any graphs), but at List of Wikipedias#Editions overview the table there has similar information for the more-or-less current moment. No idea if that satisfies your requirements. — JohnFromPinckney (talk / edits) 08:15, 28 October 2021 (UTC)

@1Kwords: Special:ActiveUsers gives a list of all the editors who have performed an edit on the English Wikipedia in the last 30 days, you can retrieve the number of active users using the {{NUMBEROFACTIVEUSERS}} magic word. Numbers for all projects are availiable on meta at meta:List of Wikipedias. 163.1.15.238 (talk) 12:23, 28 October 2021 (UTC)

The magicword only works on the local wiki, for any wiki: {{NUMBEROF|users|fr}} = 5016049 or {{NUMBEROF|activeusers|fr}} = 16645 -- GreenC 12:31, 28 October 2021 (UTC)

Found this page with Statistics, it shows number of active users over the last 12 months. The average is also shown, about 40K users for that time period. Found it via some searching and clicking the right links. A Thousand Words (talk) 19:24, 28 October 2021 (UTC)

Sprites?

Prompted by WP:VPM#Very big pages, I took a look at 2021 in mammal paleontology. A big chunk of the load time is a pile of tiny png files of national flags, many of which are smaller than 200 bytes (i.e. the HTTP headers swamp the data). This seems like a perfect use for sprites. Does the mediawiki software support that in any way? -- RoySmith (talk) 15:35, 29 October 2021 (UTC)

No, that's not how the web works. That said, the fix is obvious: Remove the "pile of tiny png files of national flags". Izno (talk) 16:12, 29 October 2021 (UTC)

There are around 58 different flags. That's not so bad. Many articles have around 200. PrimeHunter (talk) 22:33, 29 October 2021 (UTC)

CSS image sprites is one approach to convert loading multiple little images into loading a single image. In theory, someone could make a template to encapsulate the inner workings away for something like a composite image of all the national flags. More investigation is needed to determine how much savings it might provide, and weigh it against the cost of maintaining a composite image. isaacl (talk) 05:17, 30 October 2021 (UTC)

@Isaacl: could be a possible solution when template size is exceeded on some list pages as well. — xaosflux ^Talk 12:10, 30 October 2021 (UTC)

This is doable with TemplateStyles, IIRC. We'd need to put together a standard sheet with all the flags that an article might use, though. (We couldn't practically have a separate sheet specifically for the flags used in a particular article, nor is there a system for generating them like that.) I don't know if this would make sense to do from a performance perspective. --Yair rand (talk) 17:47, 31 October 2021 (UTC)

Other admin deletion logs showing up on my watchlist

Filter options on recent changes, but the same interface can be found on the watchlist

I haven't lately checked - or unchecked - anything different on my Watchlist preferences. But my Watchlist is starting to show me the logs of Admin users. ANI suggested I uncheck the "logged actions" box. I don't see such a box, either on my Watchlist page, or on my Preferences. And it doesn't seem to be all admin logs that are showing up. Just two or three a day. Today's says "09:57 (Deletion log)‎ [Kusma‎; Favonian‎; Blablubbs‎]". And that's how it seems everyday. How do I stop this? — Maile (talk) 15:37, 29 October 2021 (UTC)

@Maile66: as far as I know, that happens when a page on your watchlist is deleted. At the top of your watchlist, there should be a box that says "Active filters". If it's hidden, click the "show" button. Then you can click the X next to "Logged actions" to stop showing them in your watchlist. I think that only lasts as long as you stay on the current watchlist page, but there's a bookmark icon to the right that allows you to save your filter preferences (mw:Help:New filters for edit review/Filtering). clpo13_(talk) 15:55, 29 October 2021 (UTC)

Something is missing here. "Active filters" is not on mine, but I can click on "Show", but the drop down does not give me the option of "Logged actions". — Maile (talk) 16:10, 29 October 2021 (UTC)

Do you have the "Use non-JavaScript interface" option checked in your watchlist preferences? That would show the old watchlist options, which doesn't include the newer, advanced filters. clpo13_(talk) 16:26, 29 October 2021 (UTC)

Aaahhhhhhhh ... I unchecked it, and the problem went away. Thanks. — Maile (talk) 16:33, 29 October 2021 (UTC)

Article in mutually contradicting categories

Rehelim is in both Category:Short description matches Wikidata and Category:Short description is different from Wikidata. Any explanation? Abductive (reasoning) 23:14, 29 October 2021 (UTC)

There is no such thing as contradicting categories from a category system point of view... — xaosflux ^Talk 23:19, 29 October 2021 (UTC)

{{short description|Israeli settlement in the West Bank}} matches Wikidata but the infobox automatically adds the different short description "Place in Judea and Samaria Area". The first short description is used. PrimeHunter (talk) 23:27, 29 October 2021 (UTC)

(edit conflict) In short - there are 2 templates that are pushing those categories, {{short description}} and {{Infobox Israel village}} - that later of which requires more patience than I have right now to unravel. — xaosflux ^Talk 23:28, 29 October 2021 (UTC)

This is a perennial topic at WT:Short description. Brilliant solutions are welcome. – Jonesey95 (talk) 23:56, 29 October 2021 (UTC)

At the very least, update the documentation for Template:Infobox Israel village to let editors know what it is they are invoking when they use it? — xaosflux ^Talk 00:04, 30 October 2021 (UTC)

 Done. Never let anyone say I didn't do the very least. – Jonesey95 (talk) 16:31, 31 October 2021 (UTC)

Color of "You have new messages" alert for IP editors

See MediaWiki talk:Common.css § Orange bar of doom now peach bar of doom. Suffusion of Yellow (talk) 00:31, 30 October 2021 (UTC)

When a revert isn't a revert

Hi! Can we all agree the software should only use the term revert to mean undoing somebody's edit to the previous state, you know like the word means? Currently it is possible to hit the undo button AND THEN MAKE ANY UNRELATED CHANGES, and the poor sob making the original change will see his changes "reverted". This is not a theoretical concern, I just got bitten by it. --Palosirkka (talk) 14:46, 31 October 2021 (UTC)

@Palosirkka: Could you explain what the issue is, because I'm not seeing the issue from that article's history. You fixed the logo size, but did so using the wrong syntax. Ahunt came along and fixed the syntax for you. You reverted his edit because ???, then Ahunt reverted your revert. The only diffs in that chain that are tagged as "undo" (i.e. having involved clicking the undo button) are your revert and Ahunt's revert of your revert. The only edits that are tagged as "reverted" Ahunt fixing the syntax and you removing Ahunt's fix. As far as I can see everything is correct? 192.76.8.77 (talk) 14:58, 31 October 2021 (UTC)

@192.76.8.77:The issue is Ahunt didnt revert my change returning the article to it's previous state but ALSO made other changes. Yet the software called it all misleadingly 'reverted'. Do you now see my point? I reverted precisely because of the word reverted was used without looking at the diff. This misleading lingo can lead to such silly reverts and even "edit wars". --Palosirkka (talk) 15:05, 31 October 2021 (UTC)

It's a partial reversion, or a "revert and do differently". However, the software does label any change made with the undo button a reversion, even if it keeps the previous edit intact and changes something unrelated. Please try not to take the "Reverted" tag personally. It doesn't necessarily mean your edit is being criticised. On this occasion, you spotted a problem and partially solved it, allowing another editor to stand on your shoulders and finish the job. Certes (talk) 15:12, 31 October 2021 (UTC)

And that should be changed. I don't take it personally at all but it causes unneccessary re-reverts. --Palosirkka (talk) 15:17, 31 October 2021 (UTC)

I reverted precisely because of the word reverted was used without looking at the diff. So you deliberately edit warred without reviewing the changes in the revision? That seems blockable. Maybe you shouldn't do that instead. Izno (talk) 15:13, 31 October 2021 (UTC)

lol, no. I reverted Ahunt's edit because the software claimed he had removed my useful change. --Palosirkka (talk) 15:17, 31 October 2021 (UTC)

I think your point is clear, but unfortunately, things aren't so simple in our world (and the "Undid revision 1052844..." text that appears in the edit summary field with "undo" is only a suggestion; editors are free to adapt that text as appropriate). In the world in which we're working, edit summaries are very often incomplete, misleading or missing altogether. It behooves us all to actually check the diffs we undo/revert/rollback; after all, we take responsibility for each and every edit we make based on content even more than ES or "Reverted" flags. I hope you can consider this one of those "lessons learned" and continue contributing to Wikipedia with this knowledge helping you. Technically, I think it'd be very hard to revise the system so that any edits before saving an "undone" edit negate the application of the "Reverted" marker in th histories. — JohnFromPinckney (talk / edits) 15:32, 31 October 2021 (UTC)

Technically, I think it'd be very hard to revise the system so that any edits before saving an "undone" edit negate the application of the "Reverted" marker in th histories. No that's already been done (phab:T259014). Nardog (talk) 15:35, 31 October 2021 (UTC)

@Palosirkka: No he didn't - It's tagged as "reverted" because you reverted it - edits which are performing a revert are tagged "undo" or "manual revert" depending upon how you do them. The first edit he made to the page [3] is not a revert in any sense of the word - he just moved the 90px into it's correct section of the infobox. It was not performed using the undo button, it is not a manual revert and the software did not tag it as such. 192.76.8.77 (talk) 15:22, 31 October 2021 (UTC)

@192.76.8.77:Good catch! So, the example is wrong. Sorry for that folks... However I still think the undo button should only remove the edit in question without allowing other changes. --Palosirkka (talk) 15:31, 31 October 2021 (UTC)

Just a note, there was a similar discussion here last week, which generally seems to indicate that such actions are perfectly acceptable, though obviously if further changes are made they are indicated in the edit summary. Primefac (talk) 16:01, 31 October 2021 (UTC)

Block new account creation only?

As a way to fight persistent creation of new socks on a /16 IP range with lots of legitimate traffic, I'm thinking the best tool might be to partial-block the /16 from some rarely-used namespace (say, "Gadget Definition Talk") and also tick the "Block account creation" box. I think that would have the effect of letting everybody, including IPs, continue to edit, yet prevent new socks from being generated. I assume that would work? Any reason not to do something like that? -- RoySmith (talk) 15:55, 31 October 2021 (UTC)

That's cute. I would use a namespace that most people can't edit such as Gadget or MediaWiki, not a talk namespace, leaning to MediaWiki because that one is guaranteed not to go anywhere. If I were to do something like this. And if it works. Izno (talk) 16:40, 31 October 2021 (UTC)

@RoySmith and Izno: you don't need that hack, you can just do what you want to do - see example at testwiki:Special:Redirect/logid/266527. — xaosflux ^Talk 16:58, 31 October 2021 (UTC)

Interesting. Hey, could you give me a mop on testwiki so I can play with this? -- RoySmith (talk) 17:14, 31 October 2021 (UTC)

Oh, I see what you did. Select "Partial", leave "Pages" and "Namespaces" both blank, and tick "Block account creation"? I just assumed you had to put something into one of Pages or Namespaces, but I guess not? -- RoySmith (talk) 17:17, 31 October 2021 (UTC)

 Done and nope - not required to actually include any "editing" blockage. — xaosflux ^Talk 17:36, 31 October 2021 (UTC)

That's pretty cool. I'll keep that in mind if I ever need to only prevent account creation from a range. Dreamy Jazz ^{talk to me | my contributions} 18:17, 31 October 2021 (UTC)

Advice regarding a template update

I am seeking advice regarding {{EB1911 poster}}. Currently the template does not function as described in the templat's documentation nor at Category:Wikipedia articles incorporating the template EB1911 poster with no parameters where it reluctantly states that failing to designate the article's title using the first unnamed parameter defaults to using {{PAGENAME}} albeit advisable to avoid. Currently, no such link is produced by such default although I have worked up a functioning solution in the templat's sandbox. Yet, because the template is protected, and I don't wish to use the TE flag inappropriately, I ask: would it be better for me to rewrite the instructions to correspond with the templat's functionality, or recode the template to correspond with the instructions? Additionally, the category's title makes it seem that the template was used without any parameters being used when in fact it only red flags and categorizes based on whether or not {{{1|}}} is used (even when other parameters are used). In this case I ask: should the category be moved to a more accurate title (like Category:Wikipedia articles incorporating the template EB1911 poster without designating the article's title for example) or here again, simply rewrite the instructions to accurately describe what triggers the categorization of member pages? Thank you for considering this request, and for any advice given. I appreciate the opportunity to learn from those regularly watching this page.--John Cline (talk) 10:19, 1 November 2021 (UTC)

Shouldn't these questions be asked at Template talk:EB1911 poster. Doesn't seem like an appropriate topic for WP:VPT.

—Trappist the monk (talk) 11:37, 1 November 2021 (UTC)

Thank you for your reply. I'll move the thread to that page. --John Cline (talk) 12:28, 1 November 2021 (UTC)

@John Cline: Yet, because the template is protected, and I don't wish to use the TE flag inappropriately... Template editor was granted to you because you are trusted to perform changes to pages protected so. Unless you have a COI of some sort (which it's hard to think would be applicable to a 110-year-old revision of Britannica), do what you think is best for the Wikipedia. Izno (talk) 12:35, 1 November 2021 (UTC)

Thank you Izno, I appreciate your reply. I've already moved the discussion to Template talk:EB1911 poster#Advice regarding an update of this template so I'll leave it open for a little while and mark this thread as resolved. Thanks again.--John Cline (talk) 13:03, 1 November 2021 (UTC)

Resolved

Indentation bug in mobile

Compare the mobile view and desktop view of an old version Wikipedia:Wikipedia Signpost/2021-10-31/Book review. For some reason, using a colon for indentation in the first para after the Signpost templates shows a large blank area in mobile view. I replaced the colon with <div style="margin-left:1em;"> which displays properly in mobile web. I have occassionally seen this type of bug in talk pages too. Any idea what's the reason behind this? ಮಲ್ನಾಡಾಚ್ ಕೊಂಕ್ಣೊ (talk) 14:33, 1 November 2021 (UTC)

A colon is not for indentation. This is a bad habit we developed on talk pages, but generally a colon is part of a definition list. We generally use the template {{hatnote}} or any of it's more specific children templates for styling this. —TheDJ (talk • contribs) 15:11, 1 November 2021 (UTC)

The content there would probably better use {{block indent}}. --Izno (talk) 15:20, 1 November 2021 (UTC)