> > > > I would like to roll out a v6.4.3 the day we make v6.5 beta,
> > > > myself ... since only bug fixes were supposed to go into the
> > > > v6.4.3 source tree, and everyone would obviously have followed
> > > > that simple rule to the letter, there shouldn't be any more bugs
> > > > in v6.4.3 then v6.4.2...right? :)
Nope. We had a *firm* declaration that the v6.4.x tree was dead. That
usually means done, finished, kaput. I stopped committing changes to
that tree, but instead posted patches in /pub/patches/ as I would think
the unwritten procedure would call for.
A long time later we then declared that there *would* be a v6.4.3, and I
was forced to try retrofitting the patches onto the dead source tree. I
don't know what other problems other people introduced, but it is a bit
of work to raise the dead, and I'll agree with Bruce that it is a
distraction.
However, we also have the feeling that v6.5 is a bit riskier than
previous recent versions, so want to continue the v6.4.x thread. I would
suggest that those who feel that way adopt the v6.4.x tree and ensure
that it is a solid release. Asking everyone to devote the same attention
to both v6.4.3 and to v6.5 will lead to gaps in test coverage which we
should avoid.
Or, perhaps we should declare a v6.4.3 beta period in which we do some
focused testing, then pick up v6.5 afterwards.
> > > Of course, the fact that the REL6_4 tree has been broken more than
> > > once says that we've been sloppy :-(.
Sloppy? See above...
- Tom