A year ago, the Avalon project went through major problems when two of its main developers went into an irrational fight. One was fired out of the ASF (yes, his commiter rights were revoked) and the other one stayed.

So the "winner" became the project leader. IMO, he should have been fired also as more than a leader, he became a dictator. No future out of Merlin, and Fortress should be considered as an interim solution on the way that leads from the deprecated ECM to Merlin.

Now consider Cocoon: we've been very happy with ECM up to now and, due to its deprecation, consider moving to Fortress in Cocoon 2.2. Fortress adds some performance improvements and better object pool management. But it also comes with the need to use metadata to describe components instead of marker interfaces and explicit class naming in configuration files that are so easy to understand.

This meta layer adds some magic to the whole system that makes it less understandable. As I like to say, "too much magic kills the confidence". And it seems that Merlin adds again more abstraction in the metadata compared to Fortress. Questions are: do we need it? are "average" people able to understand it (on this subject, read the last paragraph here)?

Community wise, Avalon is on a dead end. Some of its major open-minded contributors have left the project: Berin Loritsch, and now Leo Simons. And they were also the ones bringing counter-power to the dictatorship of Merlin's almost single developer.

Consider also the discussions going on on avalon-dev: more time is spent arguing about people divergences and relational problems than actually doing something concrete. Definitely not appealing for people interested in being involved.

So does Avalon have a future, both technically-wise and community-wise? I'm doubtful about this...

Update: according to some Avalon folks, my fears seem to be unjustified (see comments on this post and Leo's clarification). I sincerely hope this is true.