On Wed, 2006-08-23 at 13:29 +0200, Bo Ørsted Andresen wrote: > On Wednesday 23 August 2006 09:43, Alan Mckinnon wrote: > > I believe the gcc ebuild does do a bootstrap compile, which does imply > > that emerge -e system is redundant. > > > > But, we had a huge long rambling thread on this point recently which I'd > > rather not rehash again :-) and IIRC the general consensus was that > > 'emerge -e system ; emerge -e world' was the way to go, for reasons > > broader than just gcc. I forget the details, maybe we should both check > > the archives. > > Back then we were speaking of a gcc upgrade in which case Richard doesn't > disagree with the emerge -e system step which is also mentioned in [1]. We > were definitely NOT talking about a minor modification of CFLAGS between -Os > and -O3! That's where you're wrong.
Can you be absolutely certain I'm wrong? Can you absolutely guarantee that a recompile with a different config MUST be treated differently to an upgrade? As I see it the only difference I can count on is that the first just happens to not involve a change in version number. In both cases a new binary is made which replaces the old one, and said new binary may or may not include features/configurations that are incompatible with other already existing software. The only way you could guarantee that is if the gcc sources ignore CFLAGS and other settings when compiling gcc in an effort to build a compiler that is always in a known state with known output. That would be an awfully good idea for a compiler but I haven't seen a clear statement anywhere that this is in fact what is done, so I don't assume that it is that way. Unless you have assumed that gcc is not to be recompiled, but only given different options to use in the future when doing it's thing, in which case we are both right, but talking about different things alan > > [1] http://www.gentoo.org/doc/en/gcc-upgrading.xml > -- [email protected] mailing list
