Re: Working compiler?
> I've been mulling this whole situation over and it's becoming an ugly
> situation.
>
> What I'm really trying to do is get a stable "newer" version of egcs,
> whether it be from the 1_1 CVS branch or from the snapshots. No matter
> what, though, I need them to compile jade and fix some of the problems
> that were present even back in 1.0.3 (jade wouldn't compile then
> either).
Where there any problems with 1.0.3a to compile C programs, (as opposed
to compile C++)?
>
> I had an idea, though...
>
> I'm tempted to rebuild the hamm egcs package as a secondary compiler and
> making an eg++ as well (just for conformity). I would upload this, have
> the 'egcc' link in the current egcs be diverted, and be able to use both
> while not needing the older libstdc++ or libg++ versions (now available in
> a separate package anyway).
That would be wonderful.
Altough if it is as simple to do, instead of a diversion, I would
suggest we arranged so that the gcc compiler from 1.0.3a, be made the
default, and that any snapshot or the still young 1.1b be
optional for the C part. That would ensure nobody compiles by mistake
kernels, important libraries, or basic binaries with a compiler which
is reported by John to fail with -O2 for one package out of 2.
(altough 1.0.3a mey not be always perfect either).
Anyway in allcases, we would have something very similar to the hamm
situation for i386, if you replace:
gcc -> egcs1.0.3a
egcs -> newest egcs
(expect they are more things to ignore in egcs than in the original
gcc package).
Loic
Reply to: