Repository navigation
ci(kani): run the proofs on PRs that touch mlog or minvmd; drop the dead sessions path - #3118
Conversation
…ead sessions path scripts/kani.sh expects 9 harnesses in mlog and 5 in minvmd, but ci-kani.yml's paths-filter listed neither, so a PR changing only one of those crates merged with its proofs unrun; they ran again only on the push to main. crates/sessions/** no longer exists. Closes #3117 Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
Reviewed: no findings; approval withheld, 1 thread unresolved at APPROVE withheld: 1 unresolved thread — norrietaylor .github/workflows/ci-kani.yml. Resolving the last thread releases the approval on this head. Resolve the open threads. Resolving the last one releases the approval on this head; no new review is required. Clobber: none. |
norrietaylor
left a comment
There was a problem hiding this comment.
Scope matches the title. The PR changes one hunk in .github/workflows/ci-kani.yml and no Rust. Clobber none. git log on the workflow shows the last related edits are #1537 (bep path) and #2200. This diff keeps both. Correctness the filter now matches the crates that scripts/kani.sh proves: boxes, rcache, mlog, minvmd and bep (the expect lines at scripts/kani.sh 161-165). crates/sessions does not exist in the tree, so the PR removes a dead path. Every harness lives in its own crate (mlog/src/otel/{spool,switches}.rs, minvmd/src/vm.rs), and the harnesses import only super:: and std or third-party items. So the two new globs cover the proof closure, with no other workspace crate needed. AGENTS.md calls .github/workflows/ frozen and CODEOWNER-gated. This PR needs a CODEOWNER approval for that reason. Outside the diff: the workflow header comment (lines 3-6) still names only rcache and PathDecision, and the scripts/kani.sh cost table still says sessions. Both are stale, and a follow-up can fix them. Not checked: an actual lane run. That needs CI.
| # scripts/kani.sh also proves mlog (the OTel spool and | ||
| # switch harnesses) and minvmd (vm.rs); without these | ||
| # two a PR touching only one of them merged with its | ||
| # proofs unrun (#3117). |
There was a problem hiding this comment.
nit · The comment tells the history of the bug ("merged with its proofs unrun (#3117)"). It does not state the rule a future editor must follow. Say what keeps the filter correct: the filter must list every crate that has an expect line in scripts/kani.sh. Then the next crate that gains a harness gets a filter path too.
There was a problem hiding this comment.
Done in cf4d73a: the comment now states the rule (the filter lists every crate with an expect line in scripts/kani.sh, five today, and a crate that gains a harness gets a path in the same change), with the miss as the example rather than the subject.
…the miss Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Closes #3117.
scripts/kani.shexpects 9 harnesses inmlogand 5 inminvmd, butci-kani.yml'schangesfilter listed neither, so a PR changing only one of those crates merged with its proofs unrun; they ran again only on the push tomain. This addscrates/mlog/**andcrates/minvmd/**to the filter and dropscrates/sessions/**, which no longer exists.One hunk in the workflow, no Rust.
actionlintis clean. Found while measuring the lab's prepush runs (min-regress #93): the lab's own Kani gate ran on three PR tips where this filter would not have.🤖 Generated with Claude Code
Summary by CodeRabbit