Repository navigation
ci: repair #13283 master reds: build trees, tri gen files, stale seals (Refs #13283) - #13289
Merged
Merged
Conversation
#13283 (75cf1a7) committed two cargo build-output trees, target-lab-retire/ (813 files) and target-m2/ (369 files). Their .d and fingerprint files carry /Users/playra/ paths, so the secret-scan "scan" step (hardcoded home path) fails on master and on every open PR. - git rm both trees (build output, not source) - .gitignore: /target-*/ beside target/, so a renamed cargo target dir cannot be committed again - REPORT-AUTONOMOUS-LOOP-2026-11-07.md line 7: the absolute home path becomes ~/t27 (the one non-artifact hit of the same scan step) Refs #13283 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Contributor
cli/tri/src/main.rs:95-98 includes search_cmd, math_verify, issue_cmd and sync_cmd from gen/rust/tri through #[path]. #13283 never committed them (gen/ is gitignored, the #11324/#11328 trap), so tri does not compile on master. Regenerated with t27c gen-rust from specs/tri/<name>/<name>.t27 in a T27 OS shell cell, not edited by hand, and added with git add -f. Refs #13283 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
#13283 edited specs/pins/parser.t27, specs/tri/net/http.t27, specs/tri/trees/trie.t27, specs/sandbox/https_enforce.t27 and docs/nona-03-manifest/{SOUL,PHI_LOOP_CONTRACT}.md without re-saving their seals. Re-saved with ~/t27/target/release/t27c seal --save in a T27 OS shell cell; the command rewrote each twin (<dir>_<Module>.json) naming the same spec. Tests: TriHttp 26/26, TriTrie 13/13, sandbox.https_enforce 10/10. PinsParser stays BLOCKED by the same comptime assertion its old seal recorded. The two doc seals were saved with --force, as before: a .md generates nothing, so every gen hash is none. Refs #13283 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
gHashTag
changed the title
Contributor
PR DashboardGenerated at: 2026-11-07 18:18:57 UTC
Summary
Seal Status
|
gHashTag
marked this pull request as ready for review
November 7, 2026 18:19
The L2 GENERATION gate maps gen/<backend>/<stem>.<ext> to specs/<stem>.t27. cli/tri/src/main.rs includes the tri command cores as gen/rust/tri/<name>.rs, but #13283 put their specs at specs/tri/<name>/<name>.t27, so the four committed copies failed as HAND-MADE COPY ("no spec at specs/tri/<name>.t27"). - git mv specs/tri/{search_cmd,math_verify,issue_cmd,sync_cmd}/<name>.t27 to specs/tri/<name>.t27, beside the flat fmt_source.t27 - header comment line 2 of each spec names its new path - the gen/rust/tri/*.rs copies are unchanged: t27c gen-rust embeds no spec path and no comment, so the bytes L2 regenerates are the same Moving the specs keeps this change in .t27; the other way (renaming the four #[path] lines in cli/tri/src/main.rs) is a hand edit of Rust. Refs #13283 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Contributor
PR DashboardGenerated at: 2026-11-07 18:25:43 UTC
Summary
Seal Status
|
…13283) duplicate-bodies has been red on master since afb3243 (the #13287 merge of #13283; bb5a029 before it was green): "the corpus improved and the ledger did not move. bytes_equal [10] -> [9]". The ratchet refuses a ceiling left above the real number, so it fails on every open PR. python3 tools/dupe_scan.py --bless, after merging master e599065. The ledger diff is the one count line, bytes_equal 10 -> 9: no group added, none grown. The re-run says "ok: no new duplicate body". Refs #13283 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Contributor
PR DashboardGenerated at: 2026-11-07 18:32:15 UTC
Summary
Seal Status
|
This was referenced Nov 7, 2026
Contributor
PR DashboardGenerated at: 2026-11-07 19:05:35 UTC
Summary
Seal Status
|
…#13283) spec-guards lists both as "STALE generated-code hash" on every PR. #13288, which carried them, was closed unmerged, and #13290 covers only the trios-chat seals, so the owner added these two to this PR's scope. t27c seal <spec> --save in a T27 OS shell cell, with the master-built t27c (cb1c1d6) that saved the other seals here: - specs/analog/linear.t27: tests 26/26 - specs/analog/linear_zoh.t27: tests 22/22 All eight gen hashes equal the "current" values spec-guards printed for 948d8c3; each spec_hash equals shasum -a 256 of its source. No other seal names these specs, so there is no twin, and neither file holds a home path. Refs #13283 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Contributor
PR DashboardGenerated at: 2026-11-07 19:13:55 UTC
Summary
Seal Status
|
This was referenced Nov 7, 2026
github-actions Bot
pushed a commit
that referenced
this pull request
Nov 8, 2026
…13285) (#13306) * seals: re-save the two analog seals #13289 saved with the pre-#13290 compiler (Closes #13285) #13289 re-saved analog_AnalogLinear and analog_AnalogLinearZoh with a t27c built from cb1c1d6, the #12695 parser before the #13290 fix. With #13290 on master the generated code is back to its pre-#12695 form, so Spec Guards reports all four gen hashes of both seals as STALE (run 42406614640). Re-saved with `t27c seal <spec> --save` in T27 OS shell-cell pane 12, using a t27c built from 2cf81c1 (the #13290 head; bootstrap/ is identical to master f2d3228). Tests: AnalogLinear 26/26, AnalogLinearZoh 22/22. The four gen hashes of each seal equal the CI current= values; no spec, closure or toolchain field changed. Neither spec has a twin seal, so sync-twins has nothing to do. No spec edit, no bless, no baseline entry. Closes #13285 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * seals: re-save compiler_ZigArityOneTypedLiteral with the master compiler (Refs #13285) Went stale with #13327. Re-saved with the same t27c used for the analog seals (2cf81c1; bootstrap/ differs from master only in generated header comments). zig/c/verilog now match the CI current hashes; tests 3/3 instead of blocked. No spec text changed. Refs #13285 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
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.
Repairs the master reds that #13283 (75cf1a7) introduced. Every open PR inherits them.
1. Build artifacts committed (8cf3266)
#13283 committed two cargo build-output trees:
target-lab-retire/(813 files) andtarget-m2/(369 files). Their.dand fingerprint files contain absolute home paths, so the secret-scanscanstep ("Hardcoded developer home path") fails.git rm -r. They are build output, not source./target-*/to.gitignorenext totarget/, so a renamed cargo target directory cannot be committed again.REPORT-AUTONOMOUS-LOOP-2026-11-07.md:7: changed the absolute home path to~/t27. It was the only non-artifact hit of the same scan step.2. Generated Rust files never committed (ac8a6d4)
cli/tri/src/main.rs:95-98includes fourgen/rust/tri/*.rsmodules through#[path]:search_cmd,math_verify,issue_cmdandsync_cmd. None was committed, becausegen/is gitignored (the #11324/#11328 trap). Sotridid not compile (error: couldn't find file .../gen/rust/tri/search_cmd.rs), which turned cli-tri, Corpus ratchet, Documented t27c subcommands, Every source file is reachable, and No test binary shares one scratch directory red.I regenerated all four with
t27c gen-rust specs/tri/<name>/<name>.t27in a T27 OS shell cell, with no hand edits, and added them withgit add -f. Every symbolcli/triuses is defined in them:search_cmd::find,math_verify::verify_*,issue_cmd::title_ok,sync_cmd::{DOC,ISSUES,PRS,entity_count}.3. Seals left stale (4965c59)
#13283 edited 4 specs and 2 docs without re-saving their seals. I re-saved them with
t27c seal <spec> --save(built from master after cb1c1d6) in the same shell cell. The command also rewrote each twin that names the same spec. That makes 11 files:--forceas before; every gen hash is none)Each file's
spec_hashequalsshasum -a 256of its source. No seal contains a home path.Not touched here: analog_AnalogLinear and analog_AnalogLinearZoh (#13288), and the four trios-chat CR-CHAT seals, which another session owns.
4. Spec paths the L2 gate can map (a7bda1e)
Once the four copies were committed, L2 GENERATION refused them as
HAND-MADE COPY ... no spec at specs/tri/<name>.t27. The gate mapsgen/<backend>/<stem>.<ext>tospecs/<stem>.t27, andcli/triincludesgen/rust/tri/<name>.rs, but #13283 put the specs atspecs/tri/<name>/<name>.t27.I moved the four specs flat to
specs/tri/<name>.t27(next tofmt_source.t27) withgit mv, and fixed the path in each header comment. The gen copies did not change, becauset27c gen-rustembeds neither the spec path nor its comments. The other way to fix this, renaming the four#[path]lines incli/tri/src/main.rs, would mean hand-editing Rust. Those lines' trailing comments still name the old nested paths.5. Duplicate-body ceiling (5b3f3c7)
duplicate-bodies has been red on master since afb3243, the #13287 merge of #13283. The commit before it, bb5a029, was green. The error reads
the corpus improved and the ledger did not move. bytes_equal [10] -> [9]. After merging master e599065, I ranpython3 tools/dupe_scan.py --bless. The ledger diff is one count line,bytes_equal 10to9: no group was added and none grew, and the re-run reportsok.6. Orphan analog seals (426d730)
spec-guards listed
analog_AnalogLinearandanalog_AnalogLinearZohasSTALE generated-code hash. #13288 carried them but was closed unmerged, so the owner added both to this PR. I re-saved them with the same master-built t27c (cb1c1d6) in the shell cell:specs/analog/linear.t2726/26,specs/analog/linear_zoh.t2722/22. All eight gen hashes equal thecurrentvalues spec-guards printed, eachspec_hashequals the source's sha256, and no other seal names these specs.Reds this PR does not address
spec.zig:767:59: error: encountered @panic at comptime: assertion failed. It was already recorded in the old seal (at line 781).Refs #13283
🤖 Generated with Claude Code