@ajs6f can you please review this?

Has this been finished? I was waiting for @fasseg to confirm that...

yeah this is finished, and works for me, the question would just be if that would not bring us into difficulties later since the now the propagation of the event only depends on the contents of the path.

JCR 2.1 is going to add

Event.getMixinNodeTypes() returns the declared mixin node types of the node that was added or removed.

I wonder if we should reject this pull request, and instead wait for JCR 2.1 to address this performance issue (rather than accepting this PR and then having to essentially revert it later)?

This addition in JCR 2.1 sounds really helpful.
But public review of JSR-333 (http://jcp.org/en/jsr/detail?id=333) just closed, so there is no final draft yet and it will take some time until this is implemented by ModeShape.
So I'll gather some numbers on the actual differences on the SCC cluster which I am currently playing around with, in order to get an impression of the perf impact in a clustered environment.

Any new status on this? If the performance hit is having an impact, then accepting this request now and updating when Mode catches up may make sense.

IIRC, this improves performance <%5 (before e.g. the leveldb change, which will make this patch's performance increase that much less.)

That being the case, I would suggest not merging this and creating a new item to update DefaultFilter when Mode catches up with this aspect of JCR 2.1.