Re: problems building 1.0.2 and 1.0.3a

[Craig Burley <burley at gnu dot org>]

>also, you asked if the compiler is broken.  I'm not sure how to tell, but
>it compiles Mesa, and the bintutils 2.9...  is is version ...
>
>pomerol-2% gcc -v
>Reading specs from
>/usr/local/gtl2/lib/gcc-lib/sparc-sun-solaris2.5.1/2.7.2.3.f.1/specs
>gcc version 2.7.2.3.f.1
>pomerol-3% 

Oh, right off the bat, any "....f.N" version of gcc might well be
buggy, since that's a version that's been modified by g77 patches,
which included a whole bunch of patches from GNAT.

Either re-install gcc 2.7.2.3, then build egcs (any version) with that,
or install gcc 2.8.1 first.

Either way, you'll end up with a working (almost certainly better-
working) gcc *and* g77.

The first way, though, you don't get a new g77 until you build and
install egcs, because there's no g77 for gcc 2.7.2.3 that *doesn't*
patch the gcc back end.

If you start with gcc 2.8.1, though, you can pop g77 0.5.23 into
it before you build and install without risking breaking anything
in that version of gcc itself, because g77 0.5.23 is the first
non-egcs-based release of g77 in a long time that doesn't patch
the gcc back end at all.  That way, you'd have g77 available even
before building egcs with gcc 2.8.1.

Of course, you might not want or need g77, but somebody thought
you might, which is why they built and installed it *plus* the
modified version of gcc it produces, labeled 2.7.2.3.f.1.

(Yay for modified version numbers.  I still dream about somebody
having the GNU project design and build a simple system for
tracking product names, version numbers, and patches even in
heirarchical situations like gcc/g77/GNAT, etc.  :)

        tq vm, (burley)