bug#9031: Installed rest of patch

[Paul Eggert <eggert@cs.ucla.edu>]

On 07/11/11 05:49, Stefan Monnier wrote:
> it's better to keep code cleanups for later

This particular fix is both a code cleanup and a bug fix.
The latter, because often compilers complain about unused
local variables, and we'd rather have Emacs builds that
don't have useless chatter and generate bug reports that
waste everybody's time later.

Compiler warnings are a judgment call.  Pacifying
every compiler would be a never-ending task, and we shouldn't
try to do that.  But this particular case is a trivial and
obviously-safe change, and it fixes GCC warnings about
code that was introduced after the feature freeze.
The (admittedly minor) benefit of such a change surely
exceeds its cost.

I'd prefer it if, when one builds Emacs, one gets no more
compile-time diagnostics than one did when the feature
freeze was introduced.  This shouldn't be an absolute goal
and it shouldn't stand in the way of more-important issues,
but surely it's not an unreasonable aspiration.

> you could
> try to install them in the `pending' branch, but such code cleanups are
> likely to generate merge conflicts

Does this advice mean that code cleanups should not be installed
anywhere now, not even in the 'pending' branch?  I ask because I
have some integer-overflow-and-signedness-related cleanups and
bug-fixes in my private copy that should really get incorporated
at some point.

(Sorry, I don't know the current practice with regards to the
'pending' branch; maybe 'pending' should be documented somewhere
under admin/?)