Re: master 4803fba487 1/2: 'C-x v v' on a diff buffer commits it as a patch (bug#52349)

[Dmitry Gutov]

On 31.08.2022 09:53, Juri Linkov wrote:
> > > > I can look into adapting this approach for other VCS if nobody beats
> > > > me to it. As long as we fundamentally agree on the behavior.
> > > Thank you.
> > This seems to work.
> >
> > Might use some extra polish, though, if people also encounter the one weird
> > error about "buffer already killed" that I saw a few times but then stopped
> > after a restart.
> Thanks, looks universal enough to handle all backends.
> I've tried it out, but it gets stuck asking the same question:
>
>    Hunk has already been applied; undo it? (y or n)
>
> because of the infinite loop in:
>
>              (while (not (eobp))
>                (diff-apply-hunk))
>
> After removing the loop, everything works fine.

Do you perhaps have customized diff-advance-after-apply-hunk to nil? I 
guess a let-binding is in order.

But a loop is necessary because the diff can have more than one hunk, 
isn't it?

> > +(defun vc-default-checkin-patch (backend patch-string comment)
> > +  (with-current-buffer (get-buffer-create "*vc-checkin-patch*")
> > +    (erase-buffer)
> > +    (insert patch-string)
> > +    (diff-mode)
> > +    (pcase-let ((`(,backend ,files) (diff-vc-deduce-fileset))
> > +                (root (vc-call-backend backend 'root default-directory))
> > +                (tmpdir (make-temp-file "vc-checkin-patch" t)))
> I'm worried that after a crash, reboot might wipe out the /tmp dir.

Creating the tmp directory inside root is also an option.