Ok for all
Regarding overloaded method of course you are right and I perfectly know that. Megawatts model are a very first implementation and not yet completed. I've been in doubt if commit it yesterday or still keep it on my local git. I've preferred to commit an incomplete version to avoid eventual conflict if someone was doing something else n the source (in fact I was hoping that)
The same for builders interface, we will take a look toghetr.
Cu online
S.
- regarding the metadata model, afaics we're not supporting overloaded methods (the simple method name might not be unique give a class)
- the current initial implementation for the instrumentation based metadata collection is based on a transformer (Agent) that modifies classes adding a call to an InvocationTracker to every method. The InvocationTracker keeps track of the called classes/methods during a given test execution. When the tests run is over, I plan to build the metadata model from the tracker's content. The InvocationTracker is currently implemented leveraging a ThreadLocal, but I still need to test/verify the whole process to see if that's the most suitable solution here. Btw, I assume the tests' runner will set the current running test name in the tracker... but we'll see once I implement that in details
- I still need to understand how to fit my future metadata builder to the current MetaDataBuilder interface, perhaps we can review that together later, once I have clearer idea.
I am currently working on a project that could use some "testedby"-magic
and I intend to start testing it for that purpose as soon as we got some
running code in.
CU
Jochen
--
Jochen Mader - Technical Lead
Pramari Europe GmbH
Bruderwöhrdstrasse 15b
93055 Regensburg
p: +49 941 604889 - 701
f: +49 941 604889 - 700
joc...@pramari.com
http://www.pramari.com
Amtsgericht Regensburg
HRB 11343
Geschäftsführer Andreas Hübner