> > William Harrington wrote: > > > Sometimes you can't build mpfr mpc and gmp within the gcc source tree > > for some targets. We found that out in CLFS. That's we we don't build > > gmp mpc and mpfr within the tree. Works okay for x86 and x86_64, > > however, when you start building for other targets, it becomes hairy. > > You may want to try to build gmp mpfr and mpc separately. > > I've been mostly quietly keeping up with this question on several lists: > is it better to build those programs in the source tree or separately? > There's a discussion somewhere (on a gcc list?) that comes down strongly on > the side of building them in the source tree. It even asks why anyone would > want to do it differently. Yet the programs contain instructions for > tuning, which requires them to be built separately. > > During the 1 1/2 years I've been playing around with compiling gcc and all > of the LFS programs, I've experimented with building separately many times. > No problem building them, but gcc usually fails to find at least one of > them. > > Now, I'm a real newbie when it comes to all this, but if anyone knows why > gcc can't seem to find the programs, I'd sure like to know. I'm trying to > understand all the ins and outs of everything covered by LFS and a lot more > besides. > > Alan > > Finally -- some recognition that there is a potential problem with this build. I disagree that it "works okay for x86 and x86_64" because I reported here that the MPC configure error appeared randomly on my x86_64 build, as, indeed, others have. I've already suggested building outside the tree in chapter 5, as we do in chapter 6. I still believe that there is some sort of race condition happening by building the three packages in the tree at the same time, but I've been too busy of late to test for that.
Richard
-- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page