This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: rearranging REG_ALLOC_ORDER in gcc/config/rs6000/rs6000.h
- From: Mike Stump <mrs at apple dot com>
- To: Jim Wilson <wilson at tuliptree dot org>
- Cc: gp at qnx dot com, gcc at gcc dot gnu dot org
- Date: Wed, 11 Jun 2003 14:08:43 -0700
- Subject: Re: rearranging REG_ALLOC_ORDER in gcc/config/rs6000/rs6000.h
On Wednesday, June 11, 2003, at 12:44 PM, Jim Wilson wrote:
Whether the kernel can be FP reg free, in which case the kernel
doesn't have to save/restore FP regs for its own use. In both cases,
you have to get this exactly right or you can't do it at all.
Hum... Isn't that like saying, I'm sorry, we don't accept ports with
any bugs in there. Surely they all do, so surely we should remove them
all, or optimization passes with bugs.
If this were mandated for complex features, the compiler would have no
complex features. EH would not exist, for sure.
gcc has a legacy of half ideas in it. g++ is such a half idea. C99
support is such a half idea. Templates certainly are. Anyway, my
point is that gcc is stronger for having these things, despite the fact
that they all are all there yet, and never may be.
Ideas that cannot be made to work, or are at odds with the design of
the compiler, or bad ideas... should be rejected, but on those
grounds...