Repository navigation
kernel: re-pin resource-algebra to f0818f2 and occupancy-types to ead4478 - #290
Conversation
…4478 scripts/fetch-siblings.sh pins resource-algebra at f0818f2 (was 0b338c9) and occupancy-types at ead4478 (was 8f09c0b). The script asks for one commit per pin; these two must move together. At f0818f2 the kernel's _≤_ is a record over a relation defined by recursion on ℕ (so that it checks under --cubical-compatible); z≤n and s≤s are functions there, and occupancy-types builds against it only from ead4478 (occupancy-types#25). Either pin alone fails the gate. Upstream restated proofs, added lemmas and adopted the PROOF-POLICY §2.1 flags. The definitions and law statements the grade kernel transcribes are unchanged, as are occupancy-types' Timeline, Reclaim, Witness and both expected rejections; the transcription table regenerates identically. The module headers that cite a pin name the one the gate checks, and say what is unchanged since the commit each was ported from. Nat.idr no longer claims _≤_ and LTE agree clause for clause: they are logically equivalent. LEVEL-STATUS names the pins and the §2.1 flags; its v1.2 history row keeps af95728. tools/check-kernel-proofs.sh: OK — build, 16 rejections, table, and 4 negative control rows (Idris2 0.8.0, Agda 2.6.3 locally; CI runs 2.6.4.3). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Y6FD63UrriRBHqmhhEca9t
|
Warning Review limit reachedYou've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository. Next included review available in 49 minutes. View limit details
No actionable comments were generated in the recent review. Summary
Merge Risk:
|
|
No fix exists yet. The log redacts the matched value, so the owner needs to look at that file. Generated by Claude Code |
CHANGELOG: keep both entries. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Y6FD63UrriRBHqmhhEca9t
CHANGELOG: keep both entries. LEVEL-STATUS merged cleanly. Kernel proof gate and grade-kernel gate pass on the merge. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Y6FD63UrriRBHqmhhEca9t
|



Summary
This PR moves two sibling pins in
scripts/fetch-siblings.sh:0b338c9tof0818f2;8f09c0btoead4478, which is occupancy-types#25.The script asks for one commit per pin, but these two have to move together. At
f0818f2, the kernel's_≤_is a record over a relation defined by recursion on ℕ, so that it checks under--cubical-compatible. There,z≤nands≤sare functions rather than constructors, and occupancy-types builds against that only fromead4478. Moving either pin alone fails the kernel gate.Changes
scripts/fetch-siblings.sh: the two new pins.ResourceAlgebra*.idr,typed-wasm.ipkg,tests/reject/MinMaxIsLoss.idr): they namef0818f2. They also say what is unchanged sincecf9235c, the commit first transcribed: the definitions and law statements. Upstream has since restated proofs, added lemmas and adopted the §2.1 flags.ResourceAlgebra/Nat.idr: the header no longer claims that_≤_and Idris2'sLTEagree clause for clause. They are logically equivalent, and every lemma statement reads the same.Occupancy.idr,HWM.idr,typed-wasm.ipkg,tests/reject/{CostIsOccupancy,DriftIsPeak}.idr): each namesead4478as the checked pin andaf95728as the commit the code was ported from. Timeline, Reclaim, Witness and both rejection files are byte-identical between those two commits.tests/transcription/ops.tsv: only the header pin changes. The regenerated rows are identical.LEVEL-STATUS.adoc: names the pins and the §2.1 flags. The v1.2 history row keepsaf95728.CHANGELOG.adoc: new entry.RSR Quality Checklist
Required
As Applicable
CHANGELOGupdatedTesting
I ran these locally with Idris2 0.8.0 and Agda 2.6.3, using
ALLOW_ANY_AGDA=1. CI runs Agda 2.6.4.3.tools/check-kernel-proofs.sh --regenprintedOK: kernel proofs — build, 16 rejection(s), table, and 4 negative control row(s). That run covers:refland by Idris2Refl.tools/check-grade-kernel.shpasses.f0818f2and occupancy-types still at8f09c0bfailed withCould not parse the left-hand side ≤-step (s≤s p). That failure is why the pins move together.This PR edits a
LEVEL-STATUS.adocline next to one that typed-wasm#289 changes. Whichever merges second may need a small merge.Generated by Claude Code