Open jad-hamza opened 3 years ago
I checked again by running just with --solvers=smt-cvc4
, and CVC4
is in fact able to solve the VC instantly (so last sentence of previous post was wrong). But for some reason the portfolio solver keeps on running until the timeout.
Running the example from https://github.com/epfl-lara/stainless/issues/979 in portfolio mode with
--vc-cache=false --solvers=smt-z3,smt-cvc4 --timeout=N
takes aroundN
seconds (+loading) and succeeds with the (body assertion) VC being verified in exactlyN
seconds. I tried withN = 10
,N = 18
,N = 24
, orN = 100
, and the VC consistently takesN
seconds to be verified. I wonder if perhaps the VC is displayed as valid while it's actually timing out.