Automatically synchronize content with upstream via magit

Autosync-Magit provides a minor mode to automatically synchronise a local git repository branch with its upstream. It is intended to be used exceptionally: when git is used solely to synchronise private content between devices. With this use case, there is typically no need to create branches, and all changes can be pushed to the remote as soon as they are committed. The author created it to synchronise his personal notes between different devices.

Autosync-Magit should never be used for other use cases and especially not for team settings.

To configure a repository to automatically synchronise, turn on autosync-magit-mode in a buffer, and set the package variables accordingly. Settings can be made permanent by adding .dir-locals.el in repositories you want to synchronise. Example:

((nil . ((autosync-magit-commit-message . "My commit message")
         (autosync-magit-pull-timer . 300)
         (mode . autosync-magit))))

The configuration above turns on the minor mode for any file visited in the same directory as .dir-locals.el or in its sub-directories. The autosync-magit-commit-message is used as the commit message for each commit. The autosync-magit-pull-timer controls the period between background pull attempts, in seconds. See the documentation of each variable for more details.

This is a simple package, that lends much of its functionality to magit that does all the work asynchronously under the hood.

With straight.el and use-package.el, add this to your ~/.emacs.d/init.el:

(use-package autosync-magit
  :straight (:host github
             :repo "sbougerel/autosync-magit"
             :files ("*.el")))

And restart Emacs. If you're using Doom Emacs, add this to your ~/.doom.d/packages.el:

(package! autosync-magit
  :recipe (:host github
           :repo "sbougerel/autosync-magit"
           :files ("*.el")))

Then add the following to ~/.doom.d/config.el:

(use-package! autosync-magit)

Then run `doom sync' to install it.

0.4.0 - Introduces a background timer for periodic pull. This is superior to the previous pull-on-events model, which does not work fast enough in a variety of use cases. Add autosync-magit-pull-when-visiting and autosync-magit-pull-timer for background periodic pull. Users are advised to switch from setting autosync-magit-pull-interval to setting autosync-magit-pull-timer in directory-local variables. Additionally, the deprecated variable autosync-magit-dirs was removed. For users that wish to start synchronisation as soon as Emacs starts, they may simply visit the directory in a temporary buffer during initialisation.

0.3.0 - Merges are synchronous, all other operations are asynchronous. This prevents any possible concurrency issues with find-file-hook functions.

0.2.0 - Use per-directory local variables. Deprecation of autosync-magit-dirs in favor of .dir-locals.el.

0.1.0 - initial release

Minimum interval between any pull attempts, in seconds.

autosync-magit starts pulls updates either via a timer and/or when visiting a file if autosync-magit-pull-when-visiting is t for that buffer.

This variable sets the minimum interval between any two pull attempts, it is always enforced. This is to ensure that autosync-magit-pull-timer or autosync-magit-pull-when-visiting will never run too close to one another.

Interval between background pull attempts, in seconds.

autosync-magit start pulling updates from remotes periodically via a background timer runing as soon as a buffer with autosync-magit-mode visits a file in a repository. This variable sets or updates the period of the background timer. Updates are not guaranteed to be pulled from the remote on the timer expiring; see autosync-magit-pull-interval.

It is recommended to use directory-local variables (in .dir-locals.el) to set this variable value. autosync-magit keeps a single copy of this value per repository. When autosync-magit-mode is turned on in a buffer, the variable value is copied to the per-repository setting, overriding any previous value.

When non-nil, find-file triggers autosync-magit-pull.

When autosync-magit-mode is enabled for the buffer, visiting a file also triggers a pull.

Do note that pulls are done asynchronously, even if merges are synchronous. All hooks running on the buffer will likely have completed before git merge. This setting is best used when paired with Auto-Revert Mode.

Since it was the default behaviour in 0.3.0, before this tunable was added; it is kept as t by default. It will likely become nil in future versions.

It is recommended to use directory-local variables (in .dir-locals.el) to set this variable value.

Default duration in seconds that must elapse before the next push.

When you save a buffer, wait for autosync-magit-push-debounce to elapse before pushing to the remote (again). This ensures that multiple file saves in a short period of time do not result in multiple pushes.

It is recommended to use directory-local variables (in .dir-locals.el) to set this variable value.

Commit message to use for each commit.

This variable is buffer-local. Since the variable is buffer-local, and commits & pushes are triggered from write-file-functions, each file can have its custom commit message. Caveat: when multiple file saves occur within autosync-magit-push-debounce, the commit message is the buffer-local value of the first file saved.

Fetch then merge (if needed) from REPO_DIR. This interactive function is not throttled, it is executed as soon as it called. Merges are synchronous, to minimize possible conflicts with files modified by Emacs in the repository.

Stage any change, create commit with MESSAGE and push to REPO_DIR. This interactive function is not debounced, it is executed asynchronously, as soon as it called.