[openib-general] 2.6.18 kernel support in the main trunk.

[Woodruff, Robert J]

Bryan wrote,
>If you want to focus on one thing to test, use Linus's current git tree
>or a release candidate tarball from it.  That way, all of the
extraneous
>cruft that sits in SVN doesn't matter until someone actually submits
it,
>and everyone has a shared understanding of what bits to bang on.  The
>OFED tree gets built from what's in Linus's tree, so if something gets
>fixed in Linus's tree, the fix will percolate into OFED.

The problem I have is that I need to test new code that is not yet
ready to be merged upstream and thus is not yet in Linus's tree.
We have deliverables to the National labs as part of the Pathforward
work and we need a tree where we can put the code, they can
pull it and test it and provide feedback prior to us submitting
it upstream. In the past, all of the latest code has been
put into SVN which allowed us and our customers the ability to
try it out and provide feedback so we know it is what they
want/need before it is submitted to Linus via Roland. 

Perhaps we need our own Pathforward tree for this, but we would
rather not have to maintain a separate tree for this work
and would prefer the model that we used over the last couple of
years, where there was one main trunk development branch.

woody