Open danieldietsch opened 7 years ago
This is definitely a lasso program. The problem is that the new block encoding does not seem to compose sequences that contain calls an returns. A workaround could be to use the procedure inlining as well.
I can add this composition. But I have to check Kojaks support for it first.
The termination analysis in general would benefit from this composition.
With the new ICFG and the new block encoding, LassoRanker seems to have trouble finding lassos in the CFG.
For example, consider the (formerly working) input:
LassoRanker reports "This is not a lasso program".
The NWA constructed from the ICFG looks as follows:
L#16true
contains cycles, is accepting, and is connected as far as I can tell.