Danek Duvall wrote:
On Wed, Jun 13, 2007 at 04:32:48PM -0400, Richard Lowe wrote:
Where it's being relied upon, we need to be backporting fixes to what we
deliver, not making major bumps to unreleased upstream code.
Then it's highly likely that someone else will need to do the work. I can
roll up a new rev of a stable (or pseudo-stable) tree pretty quickly, but
pulling in cherry-picked revisions and dealing with all the potential merge
issues is not something I'm signing up for.
I was suggesting we need to be way more cautious about such things, post-switch.
If you're implying that you'd be happy to do the work, be my guest. :)
Partly.
If you're implying that you'd prefer I update to hg-stable instead of
mercurial, I can do that pretty simply. It's *supposed* to be bug-fix
only, which implies that internal interfaces are unlikely to change, but I
bet that even that'll happen from time to time.
I thought you *were* only going to hg-stable.
Note that cadmium is depending on interfaces that have no real stability
guarantee, so until Mercurial has a well-defined, stable API, we could be
running into issues like this on a regular basis.
Yes I know, and I don't like it either. I just don't think we have much in
the way of alternative, other than minimizing our exposure.
-- Rich
_______________________________________________
tools-discuss mailing list
[email protected]
