Open ehildenb opened 4 years ago
@ehildenb Can you confirm that the evm-semantics
build system still runs this spec without the SMT solver?
@ttuegel yes the build system passes --haskell-backend-command 'kore-exec --smt=none'
, and if you look at kore-exec.sh
, you can see the --smt none
option. Anything else you'd like me to verify about that?
This was probably triggered by #2170 and would be fixed by #2174. I need to investigate that, but that's where I'll start.
I'm can't really explain how this ever worked. The cited rule requires the SMT solver to be applied. My hope for fixing this right now is that, as part of #1986, the backend will apply simplification
rules before and after function definitions, so we can avoid that rule with the appropriate lemma.
kevm-bug.tar.gz
This file comes from a simple proof on https://github.com/kframework/evm-semantics/pull/912/commits/06c9c8b514b544057e27a81e568acd89eda58654.
The proof is from
tests/specs/functional/lemmas-no-smt-spec.k
, and looks like this:The error produced looks like this: