Re: GCC 3.0 Status Report

[Stan Shebs <shebs at apple dot com>]

Mark Mitchell wrote:
> 
> Of course, if 3.0 is unusably slower than 2.95 (and some have claimed
> that), then what we do in 3.1 doesn't really matter much.

As usual with FSF releases, there is a wide spectrum of reactions.
x.0 releases typically get a lot of slack; everybody realizes that
they're a step forward, and are not going to insist on perfection.
Back in the 1 -> 2 transition, 1.x went through several more releases
after 2.0 came out, and a 2.x compiler didn't get really wide usage
until 2.3 or so.  The key thing is to hear from the clients (users
and redistributors) whether they're satisfied with the branch as it
stands.

In Apple's own case, we're unlikely to use 3.0 for production, but a
3.1 in the fall would be pretty interesting, especially with a good
PCH implementation.  So it doesn't really matter to us if 3.0 has
problems, because we'll be working on the latest code, but a long
delayed release will push back the dates for followon, and that would
be more of a concern.

> Also, it's fair to point out that there's no guarantee that your PCH
> work will ever go into the FSF GCC.  The design will have to be vetted
> on the FSF lists before that happens.  I don't expect there to be
> problems, but, as with any contribution, all the issues have to be
> weighed before anything on this scale is accepted.

I'd certainly like to take a look!  In OS X we have a working example
of precompiled headers to compare for functionality and ease of use,
and I'd like to know if the Red Hat version will be adequate.

Stan