3.4 PATCH: Support Solaris 10

[Bruce Korb]

Rainer Orth wrote:
> 
> The next version of Solaris will be identified as *-*-solaris2.10.  Of
> course it needs to be treated very much like Solaris 8/9, not (by matching
> wildcards) like Solaris 2.1 :-)

Alternatively, we could treat Solaris 2.1 systems as if they were 2.10. :-D

> The following straightforward patch does this by handling it like Solaris
> 8/9 where appropriate and removing Solaris version wildcards except where
> micro releases existed.  Up to date, this holds only for Solaris 2.5/2.5.1,
> which is treated appropriately.  By being careful about wildcard use, we
> avoid problems with Solaris 20 etc. in the future :-)

I think Solaris 2.2 will be deemed obsolete before we get to 2.20.
We'll have to solve the UNIX Y2K problem before we see that issue come up.

> Bootstrapped with only a single regression compared to a
> sparc-sun-solaris2.8 on sparc-sun-solaris2.10.
> 
> +FAIL: g++.old-deja/g++.eh/badalloc1.C execution test
> 
> I'm not yet sure what's going on here, but this should nevertheless be good
> enough for now.

Particularly since you're working with a beta release.

> Ok for mainline?

Looks fine to me.  I think GCC 3.3 may still be alive by the time
Sol 2.10 is released, so you might consider that branch, too.