Hi Wendel,
> Yes. Can you comment more on the JMX issues in 1.7.3? Any thoughts on when 1.7.4 will be released?
Some caching code was added that damaged weaving when jmx was turned on, it is discussed here:
I'd be building 1.7.4 if I wasn't answering this post :)
> I referenced before is an article about using the built in timers to assess whether your pointcuts are taking a long time to match.
Basically you need to try and ensure there is a fast matching component to your point cut so a quick decision is possible on things that definitely can't match. If you want to email me your pointcuts privately I am happy to take a look over them.
Did you post on the spring forum about your closure being treated as a bean issue?
There are different degrees of caching. I accepted a contribution a while back from someone about a cache that worked well if all your class loaders are using the same aspects. Some use cases utilize different aspects at different levels in the class loader hierarchy, those can utilize caching but it doesn't work as efficiently as the simple case. If all the class loaders in a system are using the same aspects you know that whichever person loads a type, it will be woven exactly the same - this means if class loader loads a type T and weaves it, any other class loader that subsequently loads T can used the cached version. This shared fast cache is discussed here:
cheers,
Andy