Re: GCC 3.0 Status Report

[Craig Rodrigues <rodrigc at mediaone dot net>]

On Mon, Mar 19, 2001 at 12:28:04PM -0800, Mark Mitchell wrote:
> Overall
> -------
> 
>   The Steering Committee has concluded that the release criteria
>   are somewhat over-ambitious given the lack of resources available
>   to complete some of the tasks.
> 
>   Therefore, the GCC 3.0 release will be largely a functionality
>   release, featuring the rewritten x86 back-end, improved Java
>   compiler, new C++ ABI, and new C++ standard library.  Stability
>   will remain a focus, but quality of the generated code will
>   become a secondary priority.  This withstanding, we believe that
>   GCC 3.0 will generate better code than GCC 2.95.x in many 
>   circumstances.

Has the Committee decided to reduce the number of applications
which must compile as an acceptance criteria for a gcc 3.0 release?
(I'm referring to the ones on: http://gcc.gnu.org/gcc-3.0/criteria.html).

The criteria web page lists a set of platforms on which gcc 3.0 must
build and work in order for the 3.0 release to be accepted.
Must the applications listed on this web page compile and run on
all the acceptance platforms?  (this does not make sense for things
like compiling the Linux kernel under Solaris, but it may make sense for
Emacs or POOMA).

Has any thought been given to using something like
Blitz (http://oonumerics.org/blitz/, a C++ library for doing matrix
mathematics which makes very aggressive use of templates) for
testing the template features in gcc 3.0?

Thanks.
-- 
Craig Rodrigues        
http://www.gis.net/~craigr    
rodrigc@mediaone.net