> Does the OKTech profiler make use of a native JPDA interface for the
> stack sampling?
At the moment OKTECH Profiler uses only the JMX infrastructure to get
samples from the running application, and currently there is no JPDA
interface used at all. It might be used later, but the actual profile
scenarios we are facing are just fine with the (local or remote) JMX
methods. Could you please describe why you think JPDA should be used
instead of JMX?
> The reason I'm asking this is that I want to add profiling to my
> software transactional memory implementation called Multiverse
> http://multiverse.googlecode.com and I'm thinking about different ways
> to realize this. If I could extend somehow from oktech-profiler, it
> would save me a lot of work.
I think you can profile it right now. The easiest way to start is to
check the following tutorial:
http://code.google.com/p/oktech-profiler/wiki/2MinuteGuide
When you start you application, just add the following JVM argument:
java -javaagent:hu.oktech.profiler.jar [other parameters]
And you can check the results this way:
java -jar hu.oktech.profiler.jar input=inputfile
This is the very basic setup to start with, but it will give you an
idea how it works. Please give us a feedback how you like it, what you
are missing...
Thanks and regards,
Istvan