[1152] in Kerberos_V5_Development
CVS branches for releases
daemon@ATHENA.MIT.EDU (Sam Hartman)
Fri May 10 11:34:22 1996
To: krbdev@MIT.EDU
Cc: ghudson@MIT.EDU (Do you have any CVS advice for us?)
From: Sam Hartman <hartmans@MIT.EDU>
Date: 10 May 1996 11:33:52 -0400
In-Reply-To: Marc Horowitz's message of Tue, 07 May 1996 21:37:35 EDT
>>>>> "Marc" == Marc Horowitz <marc@MIT.EDU> writes:
>>> If I hear no objections by the weekend, I will remove the V4
>>> DBM kdb support.
Marc> I object. If beta6 is ever going to get out, we need to
Marc> STOP MAKING CHANGES. If it's broken now, it was probably
Marc> broken in beta5, and can sit around taking up space until
Marc> beta7.
I agree would should avoid making unnecessary changes to Beta
6. This could be accomplished by creating a release branch (similar
to the NetBSD tree) and allowing main-line development to conintue
without changes to the release. If we decide something needs to go
into the release before it happens, then it can be specifically
committed to the release.
The following steps need to happen for this to work:
* Tag the point at which you plan to branch; call it
BETA6_BRANCH_POINT for example. You want this tag so you can get all
changes made on the Beta 6 branch easily; for example
cvs rtag BETA6_BRANCH_POINT world
* Now, branch the tree creating BETA6_BRANCH taking BETA6_BRANCH_POINT
as a snapshot.
cvs rtag -b BETA6_BRANCH -rBETA6_BRANCH_POINT world
* FInally, when you make the release, you may wish to tag the "end" of
the branch:
cvs rtag BETA6 -rBETA6_BRANCH world
I have copied this message to Greg Hudson in case he has
comments from his experience with NetBSD or Zephyr on caviats to be
applied when making releases with CVS.
Finally, we should probably start looking at what issues are
still open with Beta 6. I have one issue I have discussed in private
mail (My two messages of this weekend to Ted, Mark and Marc) that
should be resolved in some way--I'll probably have time to deal late
next week. If we are going to release recent feature commits
(Richard's random number changes, possibly others), we should make
sure they are tested.
--Sam
