bug#71499: [PATCH] Make whitespace.el cleanup add missing final newline

[Eli Zaretskii]

> From: Andrea Corallo <acorallo@gnu.org>
> Cc: Eli Zaretskii <eliz@gnu.org>,  Björn Lindström
>  <bkhl@elektrubadur.se>,
>   71499@debbugs.gnu.org
> Date: Thu, 13 Jun 2024 03:38:42 -0400
> 
> Stefan Kangas <stefankangas@gmail.com> writes:
> 
> > Eli Zaretskii <eliz@gnu.org> writes:
> >
> >>> I thought about that, but since whitespace-cleanup generally applies
> >>> clean-up according to white-space style, I thought it was simply an
> >>> oversight that it doesn't apply a fix when it is set to highlight
> >>> missing end-of-file newline.
> >>>
> >>> Adding a separate way to configure this removes the simplicity of
> >>> configuring your preferred whitespace-style as a single option.
> >>>
> >>> However, If you still disagree I can make another patch somehow
> >>> maintains the old behaviour as the default, so just let me know.
> >>
> >> Let's see what others think, and take it from there.
> >>
> >> Stefan, Andrea: WDYT about this change?
> >
> > AFAIU, the purpose of whitespace.el is to detect and eventually fix
> > incorrect whitespace, and it has two ways of doing this:
> >
> > - Visual highlighting
> > - Commands to fix problems (`whitespace-report` and
> >   `whitespace-cleanup).
> >
> > Since it is mostly configured in the centralized option,
> > `whitespace-style`, it seems natural that if a user wants to detect
> > `missing-newline-at-eof`, she would also want this to be fixed by
> > `whitespace-cleanup`.  This seems even more natural given that
> > `whitespace-report` already considers that a problem worthy of
> > reporting.  IOW, I tend to agree that this not already being the case
> > looks like an oversight.
> >
> > So I think the existing options are fine, and the patch could go in
> > as-is, despite the fact that it is backwards-incompatible.  If users
> > really hate it, I guess we will hear about it and can react.
> >
> > If we want to be really cautious, we might want to consider waiting with
> > this change until Emacs 31.  That should provide ample time for people
> > to notice the new behaviour and react.
> >
> > My two cents.
> 
> I'm as well for having the patch in, but I guess would be safer in 31 so
> we have plenty of time to react if needed.

OK, thanks.  I will then install after the branch is cut.

Meanwhile, Björn, I have a few comments to the patch:

  . it needs a NEWS entry announcing the new feature
  . is there any reason your code to handle the missing newline is not
    identical to what the implementation of require-final-newline
    does?