bug#40576: call-process-region does not accept nil as first argument

[Philipp Stephani]

Am So., 12. Apr. 2020 um 18:44 Uhr schrieb Philipp Stephani
<p.stephani2@gmail.com>:
>
> Am So., 12. Apr. 2020 um 18:21 Uhr schrieb Eli Zaretskii <eliz@gnu.org>:
> >
> > > From: Philipp Stephani <p.stephani2@gmail.com>
> > > Date: Sun, 12 Apr 2020 18:01:40 +0200
> > > Cc: 40576@debbugs.gnu.org
> > >
> > > > According to the documentation of call-process-region,
> > > >   If START is nil, that means to use the entire buffer contents; END is
> > > > ignored.
> > > >
> > > > But when I run:
> > > > (call-process-region nil nil "/bin/cat" t (current-buffer))
> > > > I get:
> > > > Debugger entered--Lisp error: (wrong-type-argument integer-or-marker-p 
> > > > nil)
> > > >   call-process-region(nil nil "/bin/cat" t #<buffer *scratch*>)
> > > >   eval((call-process-region nil nil "/bin/cat" t (current-buffer)) nil)
> > > >   elisp--eval-last-sexp(t)
> > > >   eval-last-sexp(t)
> > > >   eval-print-last-sexp(nil)
> > > >   funcall-interactively(eval-print-last-sexp nil)
> > > >   call-interactively(eval-print-last-sexp nil nil)
> > > >   command-execute(eval-print-last-sexp)
> > > >
> > > > If, instead, I run:
> > > > (call-process-region (point-min) (point-max) "/bin/cat" t 
> > > > (current-buffer))
> > > >
> > > > It works.
> > > >
> > > > Am I interpreting the documentation wrong?
> > >
> > > Nope, looks like a genuine bug (that happens only if DELETE is non-nil).
> >
> > Right.  But there's more here than meets the eye, because the change
> > after which we started advertising the special meaning of nil for
> > START exposed a problem: write_region, called from create_temp_file,
> > has special meaning for START = nil: it widens the buffer and writes
> > the entire buffer contents to the temp file.  Which isn't right when
> > write_region is called from call-process-region, as it allows access
> > to inaccessible portion of the buffer, something we shouldn't do.
>
> I think that's pretty much intentional. The documentation says
>
> "If START is nil, that means to use the entire buffer contents"
>
> It specifically doesn't say to only use the accessible portion of the
> buffer. Given that this behavior probably has been in place since
> commit 561cb8e159e7eff7a6487a45a1cfab47ba456030 from 1994, it would be
> rather unwise to introduce such a breaking change.

While the behavior of call-process-region and write-region is
unfortunate in this respect, I think it's way too late to change it
now. We should rather explicitly call out in the docstrings that
buffer restrictions are ignored.