Patches to 2.95.2 for CenterLine on Sparc

[Bruce Korb]

Kaveh writes:
>I have some comments:
>
>1.  I think the patches against language subdirs aren't necessary
>since we require gcc for those, right?


That makes sense, except that once GCC shifted into using "gccx",
the problems stopped.  I probably could have avoided some of the
language specific stuff if I had started trying to build only gcc
instead of loading up the whole suite.  Is there a faq that I did not
read that says to not try building everything unless you use gcc for
the bootstrap?

>2.  Its not clear whether you intended these for inclusion or not
>since there are some "shortcuts".

I don't really expect them to be blindly applied.  My focus was on
getting something working on my way out the door.  Monday is
my last day at DDS and, for various reasons, the system where
my legacy tools will need to be built did not have a compiler.
(Other than the CenterLine one that was licensed to only one person.)
While at DDS, I used Linux, but they will be dismantling that box.

>I'm curious though, why doesn't it see the [sbrk] decl in unistd.h?


Beats me.  Oh, on thinking about it, I bet the reason is is that
CLCC pre-defines __STDC__ to be 1 and I did not think to configure
with CC="/import/CenterLine/bin/clcc -ansi -D__EXTENSIONS__".
I already had it configured when I worked out that I had to compile
that way.  I bet that is the reason.

>3.  Many compilers have a switch to relax some of their diagnostics.
>[[...]] You may wish to
>investigate whether such a flag exists on your compiler too.

Humm.  It might.  There won't be time to mess with it Monday tho.