This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [Darwin] PATCH 15544
On Jun 15, 2004, at 2:46 PM, Stan Shebs wrote:
Geoff Keating wrote:
Andrew Pinski <pinskia@physics.uc.edu> writes:
Matt Austern <austern@apple.com> writes:
This patch is a workaround for a Darwin assembler bug. It swaps
the
order of two assembly lines in the case of empty EH symbols. OK to
commit to mainline?
Is there some reason we don't just fix the assembler?
What about old assembler like say on Darwin 6.0 boxes?
Upgrade the assembler at the same time as you upgrade the compiler.
That assumes that upgraded assemblers are available, which isn't
actually the case. There's not even an assurance that the latest
Darwin assembler will even build and run on a Darwin system several
releases back (it might, but it's not a requirement).
I don't think this is a terribly big deal, but I've been trying to avoid
version lock between the compiler and as/ld. After all, they're
separate projects, and it's not always obvious to someone who
grabs a compiler whether it might depend on an updated version
of some other tool.
That's the argument in favor of adopting this patch. The argument
against it is that it's pretty hackish: changing the order of two lines,
whose order shouldn't matter, in a special case where the old
assembler gets something wrong. But as I said, I also don't think
this is a terribly big deal either way. The bug won't affect people
who get a compiler that Apple built, because Apple can distributed
a fixed assembler. It make affect people who get the compiler
directly from the GNU repository and build it themselves.
--Matt