Criteria for GCC 4.0

[Scott Robert Ladd]

Steven Bosscher wrote:
> FWIW, many vendors, like SUSE, allow their people to work on what
> they think is necessary.  Take for example the call graph code, which
> wasn't a vendor idea, but just an inspired moment by Honza.  I think
> that in many cases the vendors don't drive the development of GCC at all.

But the kind of people who work for distro vendors have a specific range 
of experience and knowledge; that's no knock against them, just a fact. 
SuSE isn't going to hire a scientist who write research code; it isn't 
in SuSE's interest. I doubt SuSE would fund a developer who, say, went 
off trying to add an esoteric frontend to GCC, since that doesn't 
accomplish anything for SuSE.

That doesn't say SuSE is wrong -- I'm merely recognizing that the 
environment focuses development and information into certain quarters. 
Kind of like the heat and humidity keep people indoors in Florida -- 
it's the environment they live in. That doesn't make a moral statement 
about the universe; it's merely a recognition of the facts.

Hence the perception among some GCC developers that GCC "tree-ssa" 
really isn't going to impact people -- their companies don't have a 
great interest in Fortran or Java, and changes in those languages don't 
affect C programmers.

What is wrong with trying to communicate better? Doing so would benefit 
everyone.

..Scott

-- 
Scott Robert Ladd
Coyote Gulch Productions (http://www.coyotegulch.com)
Software Invention for High-Performance Computing