developer schedule/misc.

[R Clint Whaley <rwhaley@cs.utk.edu>]

Guys,

I have not yet got a developer release out, but I think we are ready for one.
First, however, I have to get the next stable out, 3.2.1.  With the memory
leak bug, it becomes critical to get a fixed tarfile out, and the number
of fixes was onerous anyway.

As soon as that is done, I will be looking at a developer release.  It is then
that those of you who have submitted new stuff will hear from me.  Right now,
I want to get with the program on the new SSE stuff submitted by Camm and
Peter, as well as the parallel make stuff Carl is doing.  I also have speedups
for small-case LU.  That and LLt needs some more work, so I'm not sure I'll
hold up the first developer release to get it all in . . . 
My guess is it will be early next week at best before I begin looking at the
submissions in order to put together the new developer release . . .

With me being more and more a chokepoint in the process, it seems possible that
we will one day want to have some CVS access, or something.  I'm not sure of
what form, or anything else, though even allowing read only would be a boon,
as submitted patches could be against the newest stuff, not the last release.
However, if we do something like that, you'll have to know that ATLAS is not
maintained in normal source code format.  It is kept in basefiles, using a
tool called extract that I developed a long time ago.  Anyone interested
can scope out the extract homepage at:
   http://www.cs.utk.edu/~rwhaley/extract/
ATLAS's basefiles are not currently kept using CVS, so if someone scopes
this out and wants to see how it is used in ATLAS, I can probably post a
couple of the basefiles somewhere . . .

Cheers,
Clint