This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Support Sun as visibility syntax
Eric Botcazou <ebotcazou@adacore.com> writes:
>> And why would this matter? You're using make for the build anyway, and
>> everything happens automagically. If you can convince the build
>> maintainers that tm_defines is a bad idea, see that it is removed
>> wholesale, otherwise I'd like to use it in cases like this.
>
> Well, I'm just asking you to follow the conventional wisdom; AFAICS no other
> configurations define USE_GAS in config.gcc, they either include usegas.h or
> xxx-gas.h files.
I wouldn't call this conventional wisdom, but just another case of
inconsistencies between ports. E.g. i?86-*-solaris2* uses tm_defines
for TARGET_GNU_LD=1 and TARGET_BI_ARCH=1, m68k*-*-openbsd* for
OBSD_OLD_GAS, mips-sgi-irix[56]* for IRIX_USING_GNU_LD, and many more.
I'm just trying to get away from the huge number of single-line headers
which can just as well be handled by tm_defines.
As soon as there is more to define (like multi-line macros, any stuff
needing a comment), *-gas.h/*-gld.h headers might be fine, although they
have problems themselves, like overriding some ELF default in
sparc/sol2.h and returning to the default again in sparc/sol2-gas.h. It
would be much clearer to have both variants, side by side, in
sparc/sol2.h (or even sol2.h, if it's common to both sparc and i386),
protected by USE_GAS/USE_GLD.
I plan to look into those inconsistencies in the 4.6 timeframe.
Rainer
--
-----------------------------------------------------------------------------
Rainer Orth, Center for Biotechnology, Bielefeld University