runtimeverification / runtimeverification/haskell-backend
Make the `"simplify`" endpoint in Booster evaluate inconsisten predicates and patterns to `#Bottom`
Nobody has claimed this yet.
- Dominant language
- Haskell
- Stars
- 224
- Forks
- 43
- PR merge metrics
- No merged PRs in 30d
Description
The "simplify" endpoint in Booster, is different from what Kore does. When asked to simplify the state state-vacuous-but-rewritten.execute:
Term:
<generatedTop>(
<k>(kseq("b", dotk())),
<int>(N:SortInt{}),
<generatedCounter>("0")
)
Conditions:
_=/=Int_(N:SortInt{}, "0")
_==Int_(N:SortInt{}, "0")
which is a pattern with contradictory constraints, booster-dev returns the pattern almost unchanged:
Term:
<generatedTop>(
<k>(kseq("b", dotk())),
<int>(N:SortInt{}),
<generatedCounter>("0")
)
Conditions:
_==Int_(N:SortInt{}, "0")
notBool_(_==Int_(N:SortInt{}, "0"))
while kore-rpc-booster will return #Bottom (via Kore):
[request 1][proxy][timing] Performed SimplifyM in 0.45s (0.41s kore time)
[request 1][proxy][abort][detail] Kore simplification: Diff (< before - > after)
*** /tmp/extra-dir-51565986628734/diff_file1.txt 2024-08-01 08:37:36.610508904 +0200
--- /tmp/extra-dir-51565986628734/diff_file2.txt 2024-08-01 08:37:36.610508904 +0200
***************
*** 1,9 ****
- Term:
- <generatedTop>(
- <k>(kseq("b", dotk())),
- <int>(N:SortInt{}),
- <generatedCounter>("0")
- )
Conditions:
! _==Int_(N:SortInt{}, "0")
! notBool_(_==Int_(N:SortInt{}, "0"))
\ No newline at end of file
--- 1,4 ----
Conditions:
! Ceil conditions:
! Substitutions:Unsupported parts:
! {"sort":{"args":[],"name":"SortGeneratedTopCell","tag":"SortApp"},"tag":"Bottom"}
that means that ApplyEquations.evaluatePattern --- the function behind the "simplify" endpoint in Booster, is unable to detect a contradiction in constraints. And this is indeed the case, since it evaluates every constraint assuming all the others, but never calls Z3 on them together as we do when rewriting.
Same happens if we isolate the constraints and ask the servers to simplify that:
request 1][proxy][timing] Performed SimplifyM in 0.49s (0.47s kore time)
[request 1][proxy][abort][detail] Kore simplification: Diff (< before - > after)
*** /tmp/extra-dir-51565986628735/diff_file1.txt 2024-08-01 08:40:47.352321295 +0200
--- /tmp/extra-dir-51565986628735/diff_file2.txt 2024-08-01 08:40:47.352321295 +0200
***************
*** 1,5 ****
Conditions:
- _==Int_(N:SortInt{}, "0")
- notBool_(_==Int_(N:SortInt{}, "0"))
Ceil conditions:
! Substitutions:
\ No newline at end of file
--- 1,4 ----
Conditions:
Ceil conditions:
! Substitutions:Unsupported parts:
! {"sort":{"args":[],"name":"SortGeneratedTopCell","tag":"SortApp"},"tag":"Bottom"}
and indeed, the "simplify" endpoint, when it receives predicates only, simplifies them in isolation and never checks their combined consistency.
I understand that this behaviour of the "simplify" endpoint in Booster is intentional, since, for most of its lifetime, Booster did not have an SMT solver and thus we've geared to towards simplifying constraints in isolation. It's time to change that.
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start at ApplyEquations.evaluatePattern, the function behind Booster's "simplify" endpoint, and reproduce the state-vacuous-but-rewritten.execute example with its contradictory predicates. Compare Booster's result with Kore's #Bottom result; done means combined constraints are checked so inconsistent predicates simplify to #Bottom.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- haskell
- Domain
- backend
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100