This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: PATCH: better cpplex.c patch
- To: law at cygnus dot com
- Subject: Re: PATCH: better cpplex.c patch
- From: Greg McGary <greg at mcgary dot org>
- Date: 14 Sep 2000 16:54:46 -0700
- Cc: gcc-patches at gcc dot gnu dot org
- References: <7551.968973366@upchuck>
Jeffrey A Law <law@cygnus.com> writes:
> Is there some reason why we need two bounded-ptr.h files? Why can't we
> have one which always does the right thing?
My first response didn't directly address the second question, which
I'll do now: I want a simple wrapper header that can be distributed
with any random package that will define stubs for all non
BP-compiles. I don't want the wrapper header to contain any details
of the BP implementation that might be subject to change. The real
non-stub definitions should always come from a header supplied with
gcc itself, that matches gcc's implementation.
I plan to add some other things to the real gcc/ginclude version in
the near future: (1) Macros that give access to the name-mangling done
on functions that have BPs in their sigs. (2) Machine-specific macros
for checking bounds from assembler code. I already have these for
ix86 and PPC in the glibc tree, but wish to migrate them to
gcc/ginclude so that all packages, not just glibc, can conveniently
wire assembler code for BPs.
Perhaps we could rename the real header if you think that the
name-duplication causes confusion. I'd say keep bounded-ptr.h for the
user-visible wrapper header, and use another name, say
`bp-internal.h', or whatever, for the real one in gcc/ginclude.
Greg