I like the idea of making matching behaviour more a framework (by some naming conventions) and using reflection for some bits. The only thing I would be aware of is the amount of reflection being used - in our project we used heavily reflections in different bits of the system, which in fact was "very nice" from the conceptual point of view, but we were trapped inside performance issues. Switching to normal API from reflection-based improved the performance of some modules by factor of >1000 I would rather make ANT fast.. Of course with minimal amount of reflection it is not the case.
> I really like the worklist idea. That would allow me to specify some > patterns at the top of the file and reuse them all the way down, sort of I like the idea of worklist only if we will decide to use nested tags. I do not really like those Makefile-like "everything is global" way... maybe by using nested tags we can improve the encapsulation and ease of use for all those "matching" things. The "worklist-like" model also seperates matching behaviour from tasks.. I like that. I do not like however making it a global-scoped property-like things. Nested tags would be the way to go.. maybe... have to think it over. mariusz
