Re: lib/timevar.c patch to avoid warnings on Solaris 8 + GCC 3.2

[Paul Eggert]

> From: Akim Demaille <address@hidden>
> Date: 21 Oct 2002 10:00:24 +0200
> 
> What's our policy wrt GCC and Bison patches?  In particular, what's
> our plan for the bitset patches?

I would like to use the same bitset sources for both GCC and Bison, but
I haven't heard back from any of the GCC guys about this since the
brief bug fix from m.hayes on October 10.

The bitset code is not in GCC yet (at least, it's not in the
2002-10-14 snapshot that I just checked), so perhaps at this point we
could say that the Bison copy is the primary one and any GCC copy
could be derived from the Bison one.  If so, the derivation should be
doable via ansi2knr right now, so should be easily automatable.

There would be one little trick, though: you'd need to ship both K&R
and ANSI versions of the source code with GCC, since the code passes
enum parameters and ansi2knr generates K&R code that does not conform
to the C standard in that case.  If this is a real showstopper, we can
have a single source by backing out the patch
<http://mail.gnu.org/pipermail/bison-patches/2002-October/001225.html>
at the cost of adding confusion to the code.

As far as the timevar patches go, that had already diverged from GCC's
copy before I got to it.  I'll try to start the ball rolling by
submitting patches to the GCC folks to merge the two versions.