This is the mail archive of the
mailing list for the GCC project.
Re: Patch to support #pragma pack(push,<n>) and #pragma pack(pop)
- To: law at cygnus dot com
- Subject: Re: Patch to support #pragma pack(push,<n>) and #pragma pack(pop)
- From: Nick Clifton <nickc at cygnus dot com>
- Date: Mon, 21 Sep 1998 10:52:57 -0700
- CC: egcs-patches at cygnus dot com, mark at markmitchell dot com, scox at cygnus dot com, burley at gnu dot org, brolley at cygnus dot com
: Date: Sun, 20 Sep 1998 00:31:29 -0600
: From: Jeffrey A Law <firstname.lastname@example.org>
: In message <199809152125.OAA30243@elmo.cygnus.com>you write:
: > One possibly controversial feature of this patch is that it moves the
: > code for handle_pragma_weak() from gcc/varasm.c into gcc/c-pragma.c.
: > I did this because I feel that it is better to keep as much of the
: > pragma parsing/handling code as possible in one file. The downside of
: > this change is that I had to modify the Makefile.in files for Fortran
: > and Chill in order to have them include ../c-pragma.o in their builds.
: Does it really make sense to allow pragmas in Chill or Fortran? If
: it doesn't, then I'd just assume not have them link in c-pragma.
Probably not - although both parsers already had some primitive code
to handle them.
The main reason that I have to link in c-pragma.o is that it now
defines the global variable 'weak_decls' which is referenced in
varasm.o, *even if* pragmas are not actually being used by the front
I could undo this change, and move 'weak_decls' back into varasm.c
(which is where it was originally defined), which would mean not
having to modify the Chill/Fortran makefiles.