On Tuesday 23 May 2006 20:36, Henrique de Moraes Holschuh wrote: > On Tue, 23 May 2006, Anish Mistry wrote: > > On a better note I've added support for FreeBSD 4.x that uses > > gcc 2.95. The patches are in the linked tarball. These changes > > should be merged into the FreeBSD ports tree in the next week or > > two. > > gcc 2.95 is thruly ancient, and the gcc 2.95 fixes in your patch > are just working around gcc 2.95's inability to deal with perfectly > valid C99 code (the hunks just shuffle variable initiation around). Right.
> > The current code (which gcc 2.95 doesn't like) has better variable > locality, which might even help the optimizer, and certainly makes > for more readable code... > > Is there a *real* reason to apply those specific 2.95-compatiblity > hunks upstream? No new Linux *or* FreeBSD 5.x install should need > them, and neither does MacOS. To me it looks like something that > should remain in the FreeBSD 4 ports patch. > > Of course, I am *not* talking about the FreeBSD *support* hunks in > your patches -- these clearly should get merged upstream. I just > wonder about the hunks specific to gcc 2.95 support. Actually the FreeBSD stdint.h #ifdefs don't need to be merged either if there is no intent to do the gcc 2.95 changes, since 5.x and above contain stdint.h. I'll keep those changes locally in the port skeleton. 4.x is EOL'd in 7 months anyway. -- Anish Mistry [EMAIL PROTECTED] AM Productions http://am-productions.biz/
pgpFJOHhzIwcD.pgp
Description: PGP signature
