Re: GCC 3.0 Release Criteria

["Bryan W. Headley" <bheadley at prismalink dot com>]

Jeffrey A Law wrote:

>   In message <20000426135643B.mitchell@codesourcery.com>you write:
>   > >>>>> "Geoff" == Geoff Keating <geoffk@cygnus.com> writes:
>   >
>   >     Geoff> I would be interested in knowing what benefits are hoped to
>   >     Geoff> flow from this change.
>   >
>   > Me, too.  I'm acting here on input from Jeff and RTH, who have
>   > indicated that this is the Right Thing.
>   >
>   > Richard, would you mind summarizing the problems with the current
>   > scheme and how you making libgcc.so a shared library will help?  Also,
>   > what negative consequences are likely?
> Actually Alexandre  & Jason can probably handle this better I can.  Basically
> it avoids most of the insane problems we have with libgcc symbols ending up
> in user applications/libraries and conflicting with other copies that are
> in other shared libraries.

If that's what you are solving, then I'm all for it. I had to implement a "fake"
eh_handler because I had a gcc library built with 2.7x and trying to link with
2.95 on an Ultra Sparc. Given the source to the library, I could've fixed it
easier...

Nice thing was, all code in question was C. Who knows why eh primitives were in
there?

Next thing is: use Linux 2-2-14 as the criteria? Let's get Linus' input. Since
he's about to release 2.4, I'm sure he'd want that as the benchmark, or one of the
later 2.3.99 pre-releases. You remember the pain and argument during the early
2.3x releases when egcs started giving him issues? A dollup for avoiding that.


--
Bryan W. Headley                     Phone: 312/913-9158
Chief Technology Officer             Fax:   312/913-9155
Prisma & Company                     Email: bheadley@prismalink.com