Re: [3.2 PATCH] Handle denormal constants in hexadecimal notation

[Gabriel Dos Reis <gdr at integrable-solutions dot net>]

Jakub Jelinek <jakub@redhat.com> writes:

| On Mon, Feb 17, 2003 at 04:15:33PM +0100, Gabriel Dos Reis wrote:
| > Jakub Jelinek <jakub@redhat.com> writes:
| > 
| > | Hi!
| > | 
| > | This got fixed in 3.3/trunk in Richard's real.c rewrite, but backporting
| > | that to 3.2 seems to be too big change.
| > 
| > Is this fixing a regression?  Do we have a PR for it?  Please include
| > it in the patch for future references.
| 
| It is not a regression, as support for hexadecimal notation was added after
| 2.95.x.
| I don't have a PR for it, would have to file one for that.
| If you think the patch is inappropriate for gcc-3_2-branch, I'll put it just
| to gcc-3_2-rhl8-branch.

In principle, I'm flexible about patches that get committed on
gcc-3_2-branch; it happened that some patches which were taught and
believed safe turned out, unfortunately, to be causing problems on
unrelated hosts and people (including you IIRC ;-) have been bitten and
complaining about applied patches that were -not- fixing regressions...

Past experiences with gcc-2.96 indicate that it is good for nobody
having different-but-nearly-the-same GCC releases floating around.  
The patch is OK for gcc-3_2-branch provided you regtest it on at least
two-different targets.

Thanks,

-- Gaby