Editorconfig causes 'Revert buffer from file ...'

[melias122]

Hey found that editroconfig started to cause this when doing magit interactive rebase. xref-find-references is also affected in some modes in elisp.

Found that something was already done in #240. Tried to with miniconfig:

init-mini.el

(setq package-archives
      '(("gnu"   . "http://elpa.gnu.org/packages/")
        ("melpa" . "http://melpa.org/packages/")))

(package-initialize)

;; editorconfig here
(editorconfig-mode 1)
(require 'magit)

Steps to reproduce similar as in #240

• emacs -Q -l init-mini.el
• install editorconfig and `magit
• in git repo M-x magit-status and r i (rebase interactive) on any git hash.
• you should see somethig like Revert buffer from file .../.emacs.d/.git/rebase-merge/git-rebase-todo? (y or n)

[stsquad]

So I've run into this and I think the problem is a combination of having a [*] entry in the project's .editorconfig which causes editorconfig to attempt to reset the mode every time the git-rebase-todo file is loaded. The short term fix is to exclude git-rebase-mode:

(add-to-list 'editorconfig-exclude-modes 'git-rebase-mode)

However I think editorconfig could be a lot smarter here. git-rebase-mode along with other "interactive" style modes are derived from special-mode. I think it would probably make sense for editorconfig-mode-apply to skip settings if any mode is (derived-mode-p 'special-mode).

[melias122]

Hey @stsquad, I tried your fix. It helps with magit, but xref-find-references e.g. in .emacs.d is still affected. I tried to remove [*] from editorconfig but that does not help.

For me until now I will stick to d73333c until we have proper fix.

[stsquad]

@melias122 I can use xref-find-references fine in my .emacs.d but perhaps it's a function of how it's setup. I don't have .editorconfig files for it for example. Can you explain in more detail what goes wrong?

[melias122]

@stsquad, so I tried everything again and you were right with [*] after removing

[*]
- charset = utf-8

xref works fine on f830b86,

but I still believe this is regression and charset = utf-8 should works just fine with "temporary" buffers like xref.

[10sr]

Thanks for reporting your issue, and sorry for inconvenience.
I'm planning to change the way how charset will be applied to buffers, so that issues caused by setting buffer charset should be fixed...

[NicolasPetton]

I haven't looked into exactly which commit introduced the issue, but editorconfig seems to be breaking https://github.com/NicolasPetton/Indium as well.

It seems that when entering indium-debugger-mode, with editorconfig-mode turned on, the buffer gets reverted, losing the debugger mode, which in turn immediately resumes the JavaScript runtime.

I'm not sure if it's the exact same issue as reported by the OP, and I could git bisect the problem if that helps.

[10sr]

I started testing a new implementation of editorconfig-mode, in hope that the issue reported in #242, #244 and #246 would be solved.
This can be enabled by setting (setq editorconfig--enable-20210221-testing t) before enabling editorconfig-mode.

It is not so tested very well yet, but currently I'm using this impl several days without any problem.
If you are ok with it, I'd be very glad if you try this new version and see if it works better.
Thanks!

[melias122]

Thanks @10sr, I will do testing and let you know if something is broken (in my case).

[10sr]

I just merged #263 so this is now enabled by default 🙌 (set editorconfig--legacy-version to disable this and use old version)

[benmezger]

Thank you for this. 10sr ***@***.***> writes:

…

I just merged #263 so this is now enabled by default 🙌 (set editorconfig--legacy-version to disable this and use old version) — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

-- Kind regards, Met een vriendelijke groet, Atenciosamente, Ben Mezger https://seds.nl https://github.com/benmezger

[10sr]

Closing this now.
Please reopen this issue if the problem still exists, thanks!