Skip to content

BS-50: the tables and predicates of the lexicon entry schema as source (#269) - #500

Draft
humuhumu33 wants to merge 7 commits into
release/v0.4from
claude/issue-269-lexicon-schema
Draft

humuhumu33 wants to merge 7 commits into
release/v0.4from
claude/issue-269-lexicon-schema

Conversation

@humuhumu33

@humuhumu33 humuhumu33 commented Nov 12, 2026 •

Copy link
Copy Markdown
Contributor

First slice of the lexicon closure of #269. bootstrap/src/LexiconSchema.lex.tex holds six production roots to lexicon::entry and lexicon::package:

  • tokenIndex: the category, frame, channel, associativity and feature tokens
  • shapeFacts: frame_permitted, Frame::fixed_arity and the category facts
  • surfaceSafety: surface_safety over the atoms of BS-03, with is_math_safe_scalar
  • isLeanName, tomlCommentAt, entryPath

None admits recursion; all declare allocation and overflow.

Case (conformance_bs_50, run on WSL over 26e5e0ace). 838 named vectors, the committed lexicon entries (1,064 forms, 725 Lean names, 1,007 entry IDs) and 2,738 TOML files, plus seeded inputs. The evaluator agrees on 8,317 runs and rustc on 18,500; all 24 planted defects are detected by their own vectors.

Regenerated. 25a32dba5a applies regenerate.yml run 43335516555 (base 20cc174194): the export with LexiconSchema's six packages, the oracles, the verify records, and the counts of BS-05 to BS-08 and BS-11 to BS-15 (manifests 330 -> 336). Two commits before it: max_child_output_bytes = 268435456 (the maintainer's decision), and segAt, because the kernel rejected certificate B of an if whose condition is (second st) of a pair (LLV7015 in run 43316802288; the message names the root CalculusCheck.checkProgram, but the function is LexiconSchema.leanStep). Ready for the parent's batch landing; it stays a draft.

🤖 Generated with Claude Code

humuhumu33 and others added 4 commits November 12, 2026 21:05
#269)

LexiconSchema holds six production roots to lexicon::entry and
lexicon::package: tokenIndex (the category, frame, channel, associativity
and feature tokens), shapeFacts (frame_permitted, fixed_arity and the
category facts), surfaceSafety (surface_safety over the atoms of BS-03, with
is_math_safe_scalar), isLeanName, tomlCommentAt and entryPath. None admits
recursion. The case holds them to the oracle on 838 named vectors, the
committed lexicon entries and TOML files, and seeded inputs; 24 planted
defects are each detected by their own vector.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…iner's decision)

The Lean extraction of the production export with BS-50 writes more than the
33,554,432 bytes max_child_output_bytes allowed (LLS8002 in the regeneration
run 43308029226); Ilya chose 268435456, as for BS-65 and RB-24. A semantic
input, so the lock moves.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…ificate checks

Certificate B of leanStep was rejected by the kernel (LLV7015 in regeneration run
43316802288): the preservation proof of an `if` whose condition is `(second st)` of a
pair does not type-check. The condition now reads the flag through `segAt`, as
every certified condition that is a call does; the two plant sites follow it.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@afflom afflom mentioned this pull request Nov 13, 2026
Applied from regenerate.yml run 43335516555 (base 20cc174): the bootstrap
export with the six packages of LexiconSchema, the build oracles, the verify records,
and the counts of BS-05 to BS-08 and BS-11 to BS-15 the count writer moved
(manifests 330 -> 336).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

@afflom afflom left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Bounded product-adoption review credits #50025a32dba and #5028911e442. The previous blanket missing-products finding is superseded for these owner branches: current module build/map/manifest inputs, sampled packages/provenance and recorded audit exit0 bind respectively to semantic d2c3e484 and a42e49b8 with compiler635a25ac. #502's committed stack rows include2031616/2228224. Complete PR-vs-base filename inventories match1596/1532 paths; the commit/compare300-file cap is explicit, so those inventories are not asserted to be exact newly generated deltas or complete byte validation.

Reuse these products, alongside #501's previously credited adoption, in the existing canonical parent batch where their input bindings remain valid. Shared manifests/root inventories/counts/helper visibility still need coherent composition and required resulting-head qualification. Sample bindings and recorded audits do not independently establish all certificates/21-definition axiom policies or runtime correspondence. No producer, proof check or gate was repeated, and no new standalone qualification wave is requested. Both drafts retain parent-batch ownership; #269 remains open.

Generated files take release's side for the regeneration to rewrite; release
carries max_child_output_bytes = 268435456, so this slice's own sentence and
entry for it give way to RB-24's. The register is 495 IDs; the lock follows the
new language-1.2 compiler-semantics ID.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

afflom commented Nov 13, 2026 •

Copy link
Copy Markdown
Member

Updated bounded owner checkpoint, 2026-11-13 13:06 UTC: #500 is2dbd89e8, #501 is77e108d9, #502 is7b924ebb. Each new commit resets earlier slice-generated products/counts to the release b8b6 baseline pending regeneration. This is baseline preparation, not adoption of the three completed output patches.

The inspected bootstrap source/config/lock, compiler implementation and regeneration workflow remain unchanged. The commit rationale is baseline/count consistency. Please explicitly disposition the prior successful source-bound artifacts against that rationale and the new outputs: resetting committed records alone does not demonstrate that every completed product is invalid. Consolidate the applicable products and necessary changes in your existing canonical batch; no additional standalone wave is requested. Current runs43401865485/43402123118/43402370319 are now all terminal SUCCESS, each5/5 jobs. Final nonexpired source-bound patch artifacts are14031778340 (2dbd),14033532029 (77e1),14036802681 (7b924). The third patch digest is44c7f7583c7b3987bdeb1081b32133302ed5af26e5b97e69ea8d31a333397224. Heads remain unchanged: successful artifact production has not published/adopted these products. No additional regeneration is needed merely to complete that publication step.

Retain the successful prior runs43373252951/43373638120/43374086737 (each5/5 jobs54/54 steps, final source-bound patch available) as scoped evidence. Their read-only workflows return artifacts rather than publishing commits. Review base-commit/file/hash records before adoption. Fast count sampling and artifact presence are not whole-case/final-head acceptance.

The prior fold-laws/compiler reuse finding remains limited to exact project/compiler/input bindings; distinct bootstrap inputs prohibit blindly overlaying one branch's manifests/maps/certificates into another or the union. Current tooling has no only:bootstrap selector. No gate exemption or #508 cross-run policy is authorized.

The concrete next owner checkpoint is which retained artifacts are adopted, which are inapplicable and why, and the canonical publication step or actual blocker. No completed check, producer, log or native measurement was repeated by this review.

The merge of release/v0.4 kept this slice's earlier regenerated records,
exports and counts where release had moved under RB-24's compiler-semantics
ID. They are release's again, and the slice is registered on top (register
495 IDs), so the regeneration writes everything from one tree.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants