splitting this list

[Joseph S. Myers]

On Tue, 3 Jun 2003, Ben Elliston wrote:

> What do folks think about trying to alleviate the increasingly heavy
> volume on this list into something more manageable?  I was thinking
> that, for example, the most common target architectures (i386, alpha,
> sparc, ppc) could get their own "gcc-<foo>" list for target specific
> discussions.  Any architecture without a dedicated list could default
> to the gcc@ list and, if an architecture got sufficiently noisy, a
> target-specific list could be created for it at that stage.

So, you haven't heard of the gcc-ppc list :-)?  (I'll add gcc-cvs-testrun
and gcc-testlist to complete the list of undocumented GCC lists.)

I don't see sufficiently large and active development communities for
individual architectures to be worth the fragmentation (and
correspondingly larger number of lists to follow to follow GCC
development, notice people forgetting coding standards anywhere in the
compiler, etc.).  I think the only subcommunity large enough is tree-ssa,
and that is only temporary before it becomes part of mainline.

> It might also help to have separated lists for documentation/WWW
> changes and for testing.  Comments?

Documentation/WWW changes are not something that should be separate, it is
fundamental that every change to the compiler comes with consideration of
what documentation changes are needed.  Likewise testing is fundamental to
development.  These are central issues, not ones for separate lists.  
(Whereas *if* there are large and separate enough communities - which I
don't think is the case for any targets - language front ends and target
architectures can reasonably be separated.)

(gcc-testresults and gcc-regression are already separate; I assume you're
referring to human disucssion of testing though.)

-- 
Joseph S. Myers
jsm28@cam.ac.uk