[PATCH] Fix PR target/16304: AIX 4.x forward reference proble

[John David Anglin]

> > 	GCC releases often require newer versions of Binutils, GDB, or
> > Glibc.  Without it, GCC may not build or be fully operational.  A
> > GNU/Linux distributor may not support a system without the default system
> > toolchain, requiring a complete system upgrade and new license for
> > support.  This does not seem different than the AIX situation to me.
> 
> I agree.

The one difference is that AIX is not free software.

> There's nothing wrong, in principle, with GCC declaring AIX 4.3 unsupported.
> 
> (In fact, for some time I've suggested that we should actively remove 
> support for relatively obscure systems, rather than the traditional 
> policy of "allow bitrot, but if someone steps forward to maintain it, 
> let them."  I'm not necessarily saying that AIX 4.3 is a relatively 
> obscure system; I don't know much about that.)

I agree with this approach but it will obsolete my position as maintainer
for various obscure systems ;-(

Albert Chin asked me to look at the AIX assembler problem because
he supports various AIX clients using AIX 4.3.  He specifically
needs GCC C++ to build a number of the packages that he distributes.
He also supports clients using HP-UX 10.20.  Thus, he still has a
support needs for various old systems even if his clients don't use
GCC themselves.

As far as I can tell, the principle metric is the presence of a
maintainer and reasonably active maintenance.  I don't have the time
or the systems to adopt AIX 4.x but I do have a few more patches in
the pipeline.  I believe Albert would like to see GCC support on AIX
4.3 for another couple of years.

For some reason, I spent the past couple of days trying to work
through various bugs on the vax.  I believe that I now have a set
of patches that will allow a full bootstrap except for the Ada
runtime on the 3.3 and 3.4 branches.  Thus, it's possible to keep
old systems going if there's a will.

I still get emails from various vax adicts trying to get netbsd and
linux going on the vax.  These activities certainly aren't mainstream
but they do provide a training ground.  This is how I came to GCC.
GCC isn't getting any simpler and the demands for a quality product
probably will increase.  Thus, I believe there is a role for a set
of minor ports.  On the other hand, rot has to be controlled, so
housekeeping is needed from time to time.

Happy New Year,
Dave
-- 
J. David Anglin                                  dave.anglin@nrc-cnrc.gc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6602)