Re: test of results of building graphs

[Navindra Umanee <navindra@cs.mcgill.ca>]

Lin LI <lli31@cs.mcgill.ca> wrote:
> If you have some idea about how to test the results or experimental
> framework, please let me know.

I have no idea how easy this is to try and test (and I don't remember
all the details about PEG) but if you can come up with two simple Java
programs:

(1) Has a locked critical region, and two threads, but the two threads
    never attempt to access the the critical region at the same time.

(2) Has a locked critical region, and two threads, and the two threads
    may attempt to access the critical region at the same time.

If your graph can detect (or eventually be used to show) that in (1)
the locks can be removed but not in (2), I think that would be pretty
cool.

Btw, I guess you don't have time to dwelve into this, but I've found
soot.util.CFGViewer to be very useful for visualising graphs... having
something similar for PEG would be neat.

Cheers,
Navin.