Repository navigation
Conversation
…drift step reads clean (Closes #13004) The step "claw drift" of gate-topology fails on master head 21e20c4 with three problems: SPEC-ONLY t27os, SPEC-ONLY t27os-release, DIR-ONLY stale-arm-repair. The directory holds 70 workflow files, the inventory claimed 71. - t27os.yml and t27os-release.yml were deleted by 453fe31 (#12167, a spec PR whose description names no workflow change), after the inventory was regenerated at 71 rows in 06458ed (#12821). - stale-arm-repair.yml was added by 1c94936 (#12846) without a row. Regenerated by the recipe in docs/claw/README.md: - specs/claw/gates.t27: the two t27os rows are dropped, the row stale-arm-repair is added in alphabetical place in all seven parallel arrays (blocking by the class rule: no continue-on-error; trigger set schedule,workflow_dispatch; a NOT_MERGE_CRITICAL reason of 137 bytes). COUNT, the array sizes and every loop bound say 70. The census tests now pin 63 blocking + 3 soft + 4 reporting, the coverage lines say n=63, the hole test points at row 69. - specs/claw/gates_pins.t27: both length pins now cover all 70 rows (they stopped at 69, two short of the old table). - docs/claw/README.md: [70] arrays, census 63/3/4. The other 69 rows were re-measured with the README awk recipe against the live on: blocks: names, trigger sets and lengths are unchanged. This does not restore the two deleted workflows; whether #12167 deleted them on purpose is the owner's call, and restoring them re-adds two rows. The seals claw_ClawGates and claw_ClawGatesPins are stale after this edit and are not re-saved here (a seal saved from a worktree under /Users/... can embed the path); that is a separate step with the post-#12824 t27c through a T27 OS shell-cell pane. Measured without a build: typecheck of both specs with the kept t27c (t27c-bc19b54e6) exits 0; the spec NAMES equal ls .github/workflows (70 lines, zero diff); NAME_LENS, EVENT_LENS and REASON_LENS equal the byte lengths of their strings; the seven non-zero class rows are the same seven workflows as before. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Contributor
… the drift step compiles the new table (Refs #13004) The first push of this PR changed specs/claw/gates.t27 only, and the step "claw drift" of gate-topology still printed the same three problems (SPEC-ONLY t27os, SPEC-ONLY t27os-release, DIR-ONLY stale-arm-repair): the step does not generate C, it compiles the committed gen/c/claw/gates.c (tracked, last written by 06458ed), so the stale table was still the one it audited. My issue text said no gen/ file may be edited; that was wrong, a generated file follows its spec and is rewritten by the generator, never by hand. gen/c/claw/gates.c is now the stdout of `t27c gen-c specs/claw/gates.t27`. The same binary (the kept post-#12824 t27c-bc19b54e6) first reproduced the committed file byte for byte from the previous spec (cmp identical, 732 lines), so the 53 changed lines are only the new table: NAMES, NAME_LENS, CLASSES, EVENT_LENS, EVENTS, REASONS and REASON_LENS at 70 rows, `size_t n[70]`. The 70 NAMES equal `ls .github/workflows` with zero diff. No hand edit, no non-ASCII line, no path of the machine in the file. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Contributor
PR DashboardGenerated at: 2026-10-31 11:21:15 UTC
Summary
Seal Status
|
Collaborator
Author
|
Closing as superseded: PR 12967 merged 2026-10-31T12:10:51Z (8296ba4) with the same gates.t27, gates_pins.t27, gen/c/claw/gates.c and claw seals. Owner word: close. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #13004
What
Regenerates the gate inventory
specs/claw/gates.t27from 71 to 70 workflows so the step "claw drift" ofgate-topologyreadsclaw: matrix is currentinstead of three problems:SPEC-ONLY t27osandSPEC-ONLY t27os-release: both workflow files are gone (deleted by 453fe31, spec(fpga): wasm board P6 -- the 64-bit datapath design floor (STACK_2x32, OP_RUN_USER64) #12167, "wasm board P6"; the inventory had been regenerated at 71 rows in 06458ed, ci: unblock master -- the classification moves into gates.t27, 26 workflows classified #12821).DIR-ONLY stale-arm-repair: the workflow file landed with 1c94936 (crons: the stale-arm repair workflow live -- paid by the check-conflicts.sh deletion (Refs #12606, #12813) #12846) without a row.How
By the recipe in
docs/claw/README.md, no build:specs/claw/gates.t27: drop the two t27os rows, addstale-arm-repairin alphabetical place (index 59) in all seven parallel arrays. COUNT, array sizes and loop bounds say 70. Census tests pin 63 blocking + 3 soft + 4 reporting; the coverage lines say n=63; the hole test points at row 69.gen/c/claw/gates.c: regenerated witht27c gen-c specs/claw/gates.t27. The drift step compiles this committed file, not the spec, so the first push (spec only) still failed with the same three problems. The same binary first reproduced the old file byte for byte from the old spec, so the 53 changed lines are only the new table.specs/claw/gates_pins.t27: both length pins now cover all 70 rows (they stopped at 69).docs/claw/README.md:[70]arrays, census 63/3/4.The new row: class blocking by the class rule (no
continue-on-error), eventsschedule,workflow_dispatch, a 137-byte NOT_MERGE_CRITICAL reason. If the owner prefers it reporting (it arms auto-merge and comments, like pr-queue-drain), the census becomes 62/3/5; that is a one-digit change.The other 69 rows were re-measured against the live
on:blocks: names, trigger sets and lengths are unchanged.Measured (no build, no test run on the Mac)
t27c typecheckof both specs with the kept t27c (t27c-bc19b54e6): exit 0, 0 errors, 0 warnings.ls .github/workflows(70 lines, zero diff)..github/workflowsis touched; no hand edit undergen/; no added non-ASCII line. The 70 names ingen/c/claw/gates.cequalls .github/workflows(zero diff).t27c test,gen-cand the C compile of the drift step are left to CI.Not done here (owner's word needed)
claw_ClawGatesandclaw_ClawGatesPinsare stale after this edit and are not re-saved: a seal saved from a worktree under/Users/...can embed the path, and the re-save needs the post-fix(#11182): gen-rust lowers a slice-field capture to a copy + report-only seal resync #12824 t27c through a T27 OS shell-cell pane. Until then the seal and coverage checks of this PR read red; they are the only expected reds.t27os.ymlandt27os-release.yml. spec(fpga): wasm board P6 -- the 64-bit datapath design floor (STACK_2x32, OP_RUN_USER64) #12167 deleting them looks accidental (its description names no workflow change); if the owner wants them back, the same regeneration re-adds two rows (72 in all).Landing
Draft. Auto-merge is dead (GH_AGENT_TOKEN expired), so the merge is by hand on the owner's word. PR 13002 (documented-commands gate) is a separate draft and is untouched.
🤖 Generated with Claude Code