Re: emacs' turn: remove useless if-before-free tests

[Jim Meyering <jim@meyering.net>]

Richard M Stallman <rms@gnu.org> wrote:
>     Why *not* remove them?
>
> It is an unnecessary change and a risk of breaking on some platform.

In my initial message, I included what I thought was enough evidence
to demonstrate that such concerns are anachronistic.

> By contrsat, making xfree accept a null argument is unproblematic,
> since the function and its callers are 100% under our control.

Stefan,
I didn't think there could be a serious objection,
so went ahead and committed the change a few hours ago.

Let me know if you'd like me to revert that part.

If desired, I can propose a patch to do what gnulib's "free" module does,
i.e., detect (at configure time) the losing free function and work around
it via a replacement function that makes "free(NULL)" as a no-op.

That way, we can keep the clean-up without fear that it will cause
trouble even on crufty old systems like SunOS 4.  Though, even emacs
admits that SunOS4 is not supported -- it was removed from emacs' own
list of supported system back in January.  So I'd like to know what
system I'm trying to accommodate if I take the time to write such a patch.