Skip to content

[docker] bump z3 and alt-ergo patches - #1231

Open
fdupress wants to merge 1 commit into
mainfrom
docker-z3-5.3.4
Open

fdupress wants to merge 1 commit into
mainfrom
docker-z3-5.3.4

Conversation

@fdupress

Copy link
Copy Markdown
Member

No description provided.

@fdupress fdupress self-assigned this Nov 10, 2026
@fdupress fdupress added the yolo-pr Don't bother reviewing, I will merge label Nov 10, 2026
@fdupress fdupress changed the title [docker] bump z3 5.3 to 5.3.4 [docker] bump z3 and alt-ergo patches Nov 10, 2026
@fdupress

Copy link
Copy Markdown
Member Author

The new alt-ergo patch fixes a soundness problem with 2.6.4 that looks like it would've pervasive and affect pretty much all of our SMT calls.

I would like to blacklist to force awareness in downstream users, but also admit we can only do this because the AE team was so transparent in disclosing it. @strub can you remind me where to find the blacklist?

@strub

strub commented Nov 11, 2026

Copy link
Copy Markdown
Member

Search for eviction in ecProvers.ml

@fdupress

Copy link
Copy Markdown
Member Author

Looks like Alt-Ergo 2.6.5 changed the way it reports its version in a way that's incompatible with why3's detection (like cvc5) and I will have to also merge in a process to patch up the why3.conf.

@fdupress

Copy link
Copy Markdown
Member Author

And I can confirm that the proof goes through with 2.6.5 and a patched why3.conf. Now I just need to patch why3.conf in the docker...

@fdupress
fdupress force-pushed the docker-z3-5.3.4 branch 3 times, most recently from 69be95b to 8aec2f6 Compare November 20, 2026 14:20
z3 5.4.1
alt-ergo 2.6.4 -> 2.6.5

evict alt-ergo 2.6.4

include patch for why3.conf for configuration of recent cvc5, alt-ergo
@fdupress

Copy link
Copy Markdown
Member Author

@strub this is now working, although I don't enjoy the fix I had to fold in there to accommodate the broken prover detection. I know it's marked yolo, but if you could give it some eyeballs and confirm there's not much else we can do short of a PR to Why3 (or all the provers that broke its detection, I guess ..), I'd be appreciative.

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

Labels

yolo-pr Don't bother reviewing, I will merge

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants