Closed ivanazuzic closed 3 years ago
The current octApron (to be renamed to apron in #335) even when switching from Oct.t
to Polka.loose Polka.t
(to use polyhedra for things like r = x + y
) cannot prove the asserts, but at least reaches a fixpoint. I guess whatever the original issue was, got fixed at some point. If these asserts should go through with polyhedra, I guess one has to investigate again, why they currently wouldn't.
The current implementation of Poly analysis (uses polyhedra from Apron) doesn't reach the fixpoint for the following example: