On Sat, Jul 04, 2009 at 07:56:02AM -0600, Matthew Burgess wrote: > On Sat, 4 Jul 2009 06:52:49 -0700, Bryan Kadzban > <[email protected]> wrote: > > On Fri, Jul 03, 2009 at 08:55:50PM +0100, Guy Dalziel wrote: > >> I don't know how important the difference > >> between _most_ compiles and _all_ compiles is, but T_CFLAGS seems to > > work > >> just as well. > > > > The point of adding that flag was to create byte-for-byte identical > > compiler binaries. Have you verified that hacking on T_CFLAGS (instead > > of XCFLAGS) actually does this? > > To be honest, Bryan, I only care about the fact that the bootstrapped and > non-bootstrapped compilers should be feature/speed compatible, and without > that flag they won't be (as -fomit-frame-pointer speeds the compiler up, > apparently).
Well, only by removing a tiny bit of code from the functions that "don't need it". But yeah, OK. > If it also makes things byte-for-byte compatible, then all the better. I suspect T_CFLAGS won't -- but then again, I was just going by the description given when it was added. If that's not entirely relevant, then whatever. Objection withdrawn. :-P
pgpF0CGULLJ7I.pgp
Description: PGP signature
-- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
