From mboxrd@z Thu Jan 1 00:00:00 1970 From: Joe Buck To: dje@watson.ibm.com (David Edelsohn) Cc: amylaar@redhat.com (Joern Rennecke), geoffk@redhat.com (Geoff Keating), jbuck@synopsys.com (Joe Buck), aj@suse.de (Andreas Jaeger), hjl@lucon.org (H . J . Lu), gcc@gcc.gnu.org Subject: Re: Compiler for Red Hat Linux 8 Date: Wed, 18 Jul 2001 15:24:00 -0000 Message-id: <200107182223.PAA08726@atrus.synopsys.com> References: <200107182212.SAA20610@makai.watson.ibm.com> X-SW-Source: 2001-07/msg01311.html David Edelsohn writes: > Even if you do not try to find a way to use the FSF GCC sources as > your base, I hope that you will consider the request from numerous corners > that the version of GCC included in Red Hat Linux 8 be compatible with FSF > GCC releases. It was clear to me from Geoff's message that folks at Red Hat are thinking about the compatibility issue. I wouldn't seek to overconstrain Red Hat; any of several mechanisms with respect to branches may be just fine and choosing what's best is the task of the developers that do the work. It would be easiest for all of us in the long run, however (including Red Hat folks), if the delta between what Red Hat ships and what others have is never too large. Forking off at some point from the 3.0 branch to give a controlled, supportable set of code that won't be too far from FSF 3.0.x might be a good way to do that, but perhaps a decent ia64 backend needs newer code that is in RH's internal tree but not the FSF tree. The thing to avoid is for the two code bases to diverge so far that bug fixes don't port or subtle incompatibilities arise.