Re: Getting MDR 4.0 working with TCK

Interestingly, the results are different depending on whether I am using 
the memory or b-tree storage.  The results that I had posted in my last 
email are from the memory-based storage.  B-tree storage throws an 
exception with a bad file descriptor and one of the tests fail.  Again, 
I'm not too concerned about it unless someone wants me to put some 
effort into reporting on it...

One of my interests up to this point has been knowing what code is 
executed by the TCK.  I just finished that work, it's available at 
http://codehaus.org/~topping/mdr-clover/. Overall, only a bit over 41% 
of the code is actually covered by the tests.  I realize that the TCK is 
not a unit testing framework, and that many uncovered statements are 
code like error and bounds checking.  I presume that a significant 
amount of code is there for when the MDR is integrated into NetBeans 
itself.  There are roughly 38K non-comment lines of code (NCLOC), so the 
if all the unused code was removed, we would have about 16K NCLOC  in 
the standalone tree.

As an exercise, I had been interested in aiming Guantanamo 
(http://docs.codehaus.org/display/ASH/Guantanamo) at the source and 
seeing what happened, but the tool is not mature yet and the output 
wasn't compilable. 

I guess that's about it from here, thanks a bunch for your help, Martin!

-b