Re: Criteria for GCC 4.0

[Scott Robert Ladd <coyote at coyotegulch dot com>]

Joe Buck wrote:
> I won't deny that there are some things that would be nice, that haven't
> been done because no developer is interested.  But it would help if you
> were more specific about what isn't getting done.  So far you have
> identified documentation.

In a previous message, I identified three items that I would like to see 
in a GCC 4.0:

1) Improved Documentation (as you noted)

Quite frankly, I'm not sure how this would get funded. It's hard enough 
finding a budget for docs at for-profit companies; I know, it's one of 
the things I "do." Everyone knows that documentation is important, but 
it's difficult to make a business case.

"Documentation" also encompasses a lot of different topics, ranging from 
user stuff to internals algorithm descriptions.

2) OpenMP

Actually, I've twice thought I'd found folk who would fund some work on 
this, and in both cases, they decided to go with commercial compilers. I 
can't blame them (and I got paid anyway) -- but it's cheaper to buy the 
commercial compiler than it is to fund a couple of developers for many 
months.

3) Faster compiles

Some work is going on in this area; the frustration level is quite high 
among many types of users, ranging from kernel developers to people with 
massive applications.

And I'll add one more:

4) Automatic Vectorization

I know this is being worked on.

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