You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We have a test suite-produced script timeout-4.13.4.smt2.txt that times out with z3 v4.13.4 in the ite statement evaluation.
In the issue we opened in the z3 repository, sat.smt=true was recommended as a way to avoid this issue. Adding the option in haskell backend/booster by @jberthold caused the prover to loop and the = sign to be missing in the output.
The legacy backend does not use smtlib-backends-process (directly uses System.Process) and does not seem to have this problem.
Investigate why the solver cannot be used by booster with the sat.smt=true option
(We already have a CLI option to pass options to the SMT solver). Test what happens when passing the option via CLI.
Once the solver can be started and used with this option: Evaluate performance impact of the option. @PetarMax mentioned that this sat.smt=true option used to slow down Z3 operations.
The text was updated successfully, but these errors were encountered:
Related: Z3Prover/z3#7507
We have a test suite-produced script timeout-4.13.4.smt2.txt that times out with z3
v4.13.4
in theite
statement evaluation.In the issue we opened in the z3 repository,
sat.smt=true
was recommended as a way to avoid this issue. Adding the option in haskell backend/booster by @jberthold caused the prover to loop and the=
sign to be missing in the output.The legacy backend does not use
smtlib-backends-process
(directly usesSystem.Process
) and does not seem to have this problem.sat.smt=true
optionsat.smt=true
option used to slow down Z3 operations.The text was updated successfully, but these errors were encountered: