Re: gnustep on redhat 7.0

[Stephen G. Walizer]

Georg Huettenegger wrote:

> stephen,
>
> > has this been reported to RedHat's bug tracker? I did a quick,
> > non-in depth search and didn't turn it up....
>
> if nobody else will do it i will file a bug report on friday or saturday
> (after taking a second look at the problem).
>

excellent, as I'm a little time-pressed these days (I have a beta release of
the
product for my day job scheduled for 10/15 and there's still a lot to do)
that
would be great...

>
> to your problems building egcs/gcc-2.95.3 (also an unsupported,
> experimental version) => it should be straightforward to build these
> compilers either using the compat-egcs or the kgcc. the problem is to make
> a binary rpm that can be installed additionally to the gcc-2.96 that ships
> with rh 7.0 (that is the main problem in my eyes).
>

hehe I'm not even really concerned with an in-addition-to RPM... I'm willing
to
replace it completely, as everything I've played with compiles fine under
2.95.2
(oh, and the pgcc-2.95.3 is actually gcc-2.95.2 with the Pentium Compiler
Group's
optimization patches... not an unsupported cvs snapshot... though the pgcc
patches
aren't supported either =)) although one possible option is to examone the
compat-egcs
compiler... I'm willing to bet it's a relocated egcs from 6.2... if so it
should (in theory)
be possible to hack the rpm from 6.2 and also include the Obj-C runtime into
the compat
package and then our problems are solved.... No promises, but I may have time
to look at
this over the weekend... as before if I get something working I'll post here
and put it
up on my website...

>
> i fully understand why the redhat folks choose gcc-2.96 and glibc-2.1.92
> and why they risked all the problems that we see now (and when looking in
> the bugzilla database on sees other compiling problems too, but these
> mostly show c++ coding problems: through the transition from gcc 2.7 to
> egcs and now to gcc 2.95/96 the compiler became more and more standard
> compliant and often c++ source is not standard compliant and newer gcc
> releases do not compile them successfully. of course the snapshot
> delivered now with rh 7.0 will need some patches to become a more usable
> compiler but it is probably the best version available now delivering a
> state-of-the-art compiler). other linux distributions have the choice to
> change (or at least they partly do it) of changing their compiler within a
> major release, redhat is committed to use gcc-2.96 (or something fully
> compatible) throughout 7.1/7.2 approximately till next october (if the
> past tells us anything about the future).
>

yes, it makes sense, unfortunately it's lethal from GNUstep's perspective =)
and since Objective-C is not a priority for gcc anymore (grumble) it may
be a challenge to get this fixed... and while I had a compiler design class
in
college, I definitely am not qualified to hack gcc ;-)

>
> to put it in a nutshell if nothing relevant comes up until the weekend i
> will take another shot at the problem and report my findings.
>

excelllent, and if I get time I'll look at getting an RPM for Obj-C for the
egcs-compat system...

Steve