I suggest we do not have such "giant" branches as bigbang. I think it may prevent some developers working on the project, at least it may make people think twice with what they are doing (in a stressful manner). Instead of trying to keep trunk perfectly building, we can always work on the trunk and avoid merging problems; and we can sometimes (svn) tag the trunk when it perfectly builds and makes sense as a whole. So people can be referred to those tagged versions if they want to go with the edge version of the server.
Just an idea,
To prepare for a new ApacheDS release cycle and a bit of a different mode of working we've swapped out the trunks with bigbang branches for the following projects:
What is now in trunk is a stable build that works. Unlike before we intend to work in trunks now to add what we need for the 1.5.3 release. Before we were just keeping the trunk stable for users but the changes we need there for remaining features will not be large and can be safely applied in trunk without breakage. We will merge these changes back into the bigbang after making them. The following features remain for 1.5.3 on our roadmap:
Index Rebuilding - Fix CLI based index command for adding new indices.
UserPassword - make sure userPassword cannot be searched
New Installers - installers for Solaris and the tarball remain + making the *.bin installer work on all supported *NIX platforms
We will also most likely attack the most critical bugs.
Shortly we will change the bigbang to use the 1.5.4-SNAPSHOT with respective versions for dependent projects even though we intend to release 1.5.3 from trunk. We're doing this because there are issues with jar collisions due to the fact that we now have a build server that makes sure our snapshot repository is fresh. So for this reason we decided to bump the revision for bigbang to 1.5.4-SNAPSHOT while working on 1.5.3-SNAPSHOT in trunk.
Also we're going to get seriously medievil withing the bigbang: almost to the point where it may not compile for extended periods of time. For these reasons we will switch continuum to build from the trunk now. I'm going to go ahead and try to make the changes for continuum now. BTW the following features will be incorporated into the 1.5.4 release:
DoS Safeguards - add safeguards to prevent size based DoS attacks
DSML w/ Jetty - add Jetty container for self service apps and DSML support
Nested Partitions - DIRSERVER-465
Encrypted Attributes - add support for encrypted attributes
Snapshotting & Restoration - add a way to restore a crashed database (restore command as a CLI tool)
JNDI - Complete JNDI removal
New Default Partition - decouple the leafEvaluator from the Partition / Using Cursors
MINA 2.0 - Move to MINA 2.0
This should be the last of the changes we will do as part of the bigbang effort. The we can just keep working out of the trunk and get to 2.0 as soon as possible.
 Roadmap - http://cwiki.apache.org/confluence/display/DIRxPMGT/2.0+Roadmap
 DIRSERVER-465 - https://issues.apache.org/jira/browse/DIRSERVER-465