Patch to projects.html: glibc's string function macros

[Mark Mitchell]

>>>>> "Joseph" == Joseph S Myers <jsm28@cam.ac.uk> writes:

    Joseph> This patch to projects.html documents the
    Joseph> architecture-independent optimizations in glibc's
    Joseph> <bits/string2.h> macros as possible projects for
    Joseph> implementing in GCC.  In particular, it would be desirable
    Joseph> if the memset, strcmp and strcpy optimizations could be
    Joseph> implemented in GCC, as those macros in glibc have huge
    Joseph> expansions, and could be disabled if GCC did all that
    Joseph> glibc tries to do.  OK to commit?

Yes, this is fine.

I'm a huge proponent of doing this stuff in the compiler.  That will
yield better code, simpler C libraries, faster compile-times, and
benefits for people not using libc.  In general, library folks should
not spend effort doing this kind of thing in the library; it would be
better to ask the compiler folks to make these improvements.

Also, would you enter your report on the code-quality regression in
GNATS?  Make it a "high" priority bug, meaning that we need to fix it
before the next release.  (Generally, the authority to make things
`high' is reserved to maintainers, but in this case you've got my
blessing!)  We definitely need to do something about this before the
next release, and I hope it will be to fix the builtins.

Thanks,

--
Mark Mitchell                   mark@codesourcery.com
CodeSourcery, LLC               http://www.codesourcery.com