The current Dynoptic code assumes that the McScM model checker always
completes. However, checking CFSMs is undecidable in general and McScM may
never return. A few TODOs here:
1. Lower the current timeout of 1hour down to about 5 minutes.
2. Change the check/refine loop to be more adaptive:
(2.a) check the next invariant when there is a timeout, and when the new
invariant causes a refinement then re-check the timed-out invariant;
(2.b) if all invariants timed out then increase the timeout value a few times
before giving up on the entire process.
Original issue reported on code.google.com by bestchai on 26 Sep 2012 at 9:23
Original issue reported on code.google.com by
bestchai
on 26 Sep 2012 at 9:23