Re: Beyond GCC 3.0

[Mark Mitchell <mark at codesourcery dot com>]

> What count as distinct platforms here?  Different target triples, or
> different CPUs?  Do one native and two different simulator crosses
> suffice?  Is it OK if some of the platforms do not support libgcj (say)?
> When Ada is in the tree, is that required to be included in the tests on
> some or all platforms?

Yes, all these are details that need answers, but not now.  Solaris 2.5,
2.6, 2.7 would not be a good choice.  The important thing is not the
rigidity of the rules; it is

>
>>   - Patches that cause regressions, even on the mainline, must
>>     be in the process of being fixed within 48 hours, or else can be
>>     reverted by anyone with global write privileges, if they think
>
> Do regressions here include performance regressions?  Do they include
> documentation regressions?

I don't know.  Almost any back-end change can cause performance
regressions.  Some performance regressions are more compelling than
others.  Documentation regressions concern me less *if* the people
who checked in the patch can be trusted to fix the docs relatively
shortly.  The problem we are trying to address is that people cannot
get their work done because the mainline is broken as often as not.

Statistically, the mainline typically does not build on multiple
platforms that did in the previous release.  The set of platforms
oscillates, but there is always something that doesn't work.

-- 
Mark Mitchell                mark@codesourcery.com
CodeSourcery, LLC            http://www.codesourcery.com