[openib-general] Re: Testing of 1.0 release candidates

[Bryan O'Sullivan]

On Thu, 2006-03-16 at 18:06 +0200, Tziporet Koren wrote:

> This will help but it should include also the the core and not mthca and 
> libibverbs only.

Perhaps.  I'm open to that.  But I don't really care what kernel stuff
makes it into the 1.0 branch, since people should be testing with the
current 2.6.16 -rc or -git kernel in any case, not bits out of SVN.

> I think maybe the branch was opened too early and we need to take 
> another branch these days.

No, this is absolutely not the case.

Development on the trunk proceeds at such a variety of different rates
that if we tried to choose a single point to branch at, it would
*always* be too early for some feature or constituency that hadn't quite
made it in, and *always* too late because some other feature got
snapshotted at a point where it was in the middle of a destabilising
change.

As long as there are multiple projects using a single Subversion tree to
do different things, each with its own level of urgency and at its own
speed, the *only* rational choice is to branch at a fixed point in time
and then bring stability or feature fixes into the stable branch as we
discover a need to do so.

	<b