Skip to content

Conversation

@mtzguido
Copy link
Member

This proof regressed again. I tried to stabilize it by splittin queries, but ran into this:

(EverParse3d.InputStream.Extern.fst(246,1-271,5))	Query-stats (EverParse3d.InputStream.Extern.peep, 63)	failed {reason-unknown=unknown because Overflow encountered when expanding old_vector} in 42 milliseconds with fuel 0 and ifuel 1 and rlimit 256 

which is a z3 bug. To work around it I also added --z3refresh, and now this works reliably, albeit slowly.

I wanted to add hints but noticed there are none in the repo?

@nikswamy
Copy link
Contributor

We have not been using hints in EverParse. The proofs have generally been stable without them.
But this particular proof is a repeat offender.
@tahina-pro can you take a look at this proof and see if it can be made more explicit/split into smaller parts etc.?

@nikswamy nikswamy merged commit b6680cf into project-everest:master Nov 28, 2023
@mtzguido mtzguido deleted the fix branch November 28, 2023 19:02
mtzguido added a commit to mtzguido/FStar that referenced this pull request Nov 28, 2023
See project-everest/everparse#114 for an
example of the error, visible only via --hint_info or --query_stats.
This adds a separate error if that happens.
@tahina-pro
Copy link
Member

But this particular proof is a repeat offender. @tahina-pro can you take a look at this proof and see if it can be made more explicit/split into smaller parts etc.?

I claim #115 should make this proof much more robust.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants