This content has been marked as final. Show 1 reply
lets establish together what we need for the testing of the VFS bug.
A simple answer would be:
we need something that would access nested resources every
* (cache limit time) + 1 period - if timed cache
* (cache size limit) + 1 times - if LRU cache
* weak/soft cache release - if we have some way of doing this (JBoss Profiler - Clebert's work)
Nested resources should be:
* packages (from package-info.java)
* metadata files
* some configuration files
* via classloader
* direct URL usage
* via VFS api
Each '?' means this should be made flexible,
easy to add new type (resource, access, ...).
The test should prove:
* the temp resources are not growing beyond resonable size/number
* cache entry instances are expected
I am running long tests for the clustering test, maybe we can add it to the test scenario,
As long as we can properly monitor it.
but i guess it will be easier to create one from the scrach.
But depends what we need indeed.
It's up to you to decide that,
as only you (and probably Brian) know what you're doing there.
Ales, maybe you can send your "some quick/crappy test on my 2nd machine" and I can have a look at that.
Not something I'm proud to share. :-)
I'll keep it as a last resort if you really have no other idea. ;-)