bug#71504: 30.0.50; FR: Fix suggestions ("quick fix") for Flymake diagnostics

[Eshel Yaron via "Bug reports for GNU Emacs, the Swiss army knife of text editors" <bug-gnu-emacs@gnu.org>]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Eshel Yaron <me@eshelyaron.com>
>> Cc: sbaugh@janestreet.com,  71504@debbugs.gnu.org
>> Date: Sun, 07 Jul 2024 10:53:12 +0200
>> 
>> >> From: Eshel Yaron <me@eshelyaron.com>
>> >> Cc: Eli Zaretskii <eliz@gnu.org>,  71504@debbugs.gnu.org
>> >> Date: Thu, 27 Jun 2024 20:15:37 +0200
>> >> 
>> >> Spencer Baugh <sbaugh@janestreet.com> writes:
>> >>
>> >> > For example, maybe we want to have a command which can accept fixes
>> >> > output by a process running in M-x compile.  Baking the UI into flymake
>> >> > would make that impossible, wouldn't it?
>> >> 
>> >> I don't think adding a command for fixing the diagnostic at point should
>> >> preclude any other developments or explorations.  It's a useful thing to
>> >> have, and many Flymake backends have the needed data readily available.
>> >> 
>> >> > So before any change in flymake I would like to see much more
>> >> > exploration of "fix" UIs which are genuinely flymake-independent.
>> >> 
>> >> Flymake shows diagnostics, and "fixing" is what we do to diagnostics.
>> >> What would be the benefit of a Flymake-independent UI for fixing the
>> >> diagnostics that Flymake already shows?
>> >
>> > The benefit would be that we will be able to use that UI when "fixes"
>> > are shown in, for example, the *compilation* buffer.
>> 
>> It'd be good to enhance compilation buffers as well, but this feature
>> request is about interaction with Flymake diagnostics, that are shown in
>> the diagnosed buffer: I'd like to have a standard way to act on (fix)
>> the diagnostic at point.
>
> I frankly don't understand what you are saying here.  Several people
> opined that we should take a broader view on the fixes and how to
> handle them, but you insist that Flymake should have its own solution?

No.  I only insist that there should be a command for fixing the
Flymake diagnostic at point.  If it's part of a "broader solution",
that's swell.

> IOW, the "fixes" diagnostic shown by Flymake is not just diagnostic,
> it's a suggestion to make some change in the source code.

I think there is a misunderstanding here: it's not about specific
diagnostics which represent fixes, this is about enriching
(potentially) all diagnostics with backend-provided fix suggestions,
and adding a command that applies such fixes.  For example, with my
implementation I use the same command for fixing checkdoc, shellcheck
and LSP diagnostics.

> So supporting that cannot be separated from the more general concept
> of making changes proposed by some external tool.  Or what am I
> missing?

IIUC, I think I agree.  In my implementation, Flymake delegates the
application of the code changes to another library, that includes a
general purpose function for applying code changes.

> Or maybe this is a simple misunderstanding: what do you mean by
> "acting on diagnostic at point"

Applying a suggested code change that resolves the diagnostic.

> , and how could such an act be indifferent to what and how is fixed?

A single command should let you fix diagnostics from different sources
(backends).  It doesn't need to be indifferent, just consistent.

Does that make sense?