We're trying to use subversion for real. Our previous mode of operation (with clearcase)
was a separate branch per task, with each developer managing possibly several branches.
Periodically, it might be necessary to mergefrom the trunk into task branches (as other
developers update the trunk), and as tasks were completed they would be mergedto the
trunk.
We tried to do this with subversion but some of the issues have made it difficult.
Because subversion isn't yet managing the merge for you, you have to keep track of where
the fork (svn cp) happened, and also keep track of any mergefrom trunks (and skip these
when merging back to the trunk). We wrote some scripts that tag the commits and copies
with recognizable labels, so that the log can easily be used to find the fork and other
merges. But a variety of things can happen where things get into a rather confused state
and it takes a fair amount of time to sort it all out, sometimes with desired changes
misplaced.
We're probably going to switch to a "all developers off the trunk" mode, with branches
used for "significant" milestones, like releases. I'm just wondering if that's the mode
most people are using. (I understand that svn is going to fix all this at some point.)
marvin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri May 16 21:32:00 2003