Versioning Teleconference
Monday, December 13, 1999

Attending: Jim Amsden, Geoff Clemm, Eric Sedlar, Neil Weber, Jim
Whitehead, Tim Ellison, Chris Kaler

DAV:predecessor/DAV:merge-predecessors vs. DAV:predecessors issue:

Issue: Why you want this: need to know where to merge to. Might also
want to distinguish between a checkout for an edit, and a checkout for
a merge. Rebuttal: this is an activity to activity relationship, not a
version tree relationship. It's undesirable to use the predecessor
relationships to determine where to merge, since there are some cases
where this will lead to merging the wrong version.

Participants agreed to only use DAV:predecessors, and not have a
special DAV:merge-predecessors relationship. However, Chris joined
late, and disagrees.  Chris and Geoff will discuss this offline. All
agreed to use a MERGE method to create the multiple DAV:predecessors
relationships, and to create a working resource where the client will
deposit the merged results. It will also be possible to delete some
DAV:predecessors relationships (but only before the merge working
resource is checked in).

Versioning locks issue. Some concern over the computational cost of
recomputing revision selection rules after they are modified by a
lock. Some discussion on whether a lock should prevent merging within
a versioned resource, or just on a particular working resource, or
just on a particular branch, or just for a particular revision.  No
agreement reached, will discuss this topic during the next call.

*** Meeting adjourned ***