Repository navigation
BS-33: the canonical project manifest and its digest as source (ProjectText) - #463
humuhumu33 wants to merge 2 commits into
Conversation
…manifest The next slice of the project manager as source (#274), on BS-31: the row (register 479 -> 480, the §31 table, the RP-07 check and its case), the scenario, the §17.23 prose, the CHANGELOG entry, the README's claim, the CI part and its Intel lane, and the case, conformance_bs_33, which holds ProjectText.canonicalProject and configSha256 to ProjectConfig::canonical_toml and config_sha256 on hand-built configurations, on every committed lexlean.toml that parses, and on seeded ones, with 12 planted defects each named by its vector. VERIFICATION.md records the first run: the evaluator agrees on 561 and 553 runs, rustc on 2,714, and every plant is detected. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
bootstrap/src/ProjectText.lex.tex over LockText: canonicalProject writes a configuration as lexlean lock canonicalizes it, and configSha256 is the hex SHA-256 of that text. Lean verifies the module and each root is eligible with exactly allocation and overflow. The entrypoint, the lock, the two export roots, and the project README follow. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
afflom
left a comment
There was a problem hiding this comment.
Bounded source review of exact 70b3377 against stacked #458 base a5bfa93: no new concrete source defect found in this review. ProjectText’s field encoding, canonical header/limit/PDF ordering, stable equal-package ordering and direct SHA-256 composition agree with the public config.rs oracle on inspection. Expected differential values call the public oracle directly. The BS-33 part/vv row and Intel print-lane addition retain the existing entries.
This is source evidence only. The reported counts reconcile arithmetically at 351 parsed manifests (561 text + 553 digest evaluator cases; 2,714 native cases), but I did not reproduce measurements, Lean/policy checks or plants. The corpus walker scans the filesystem and silently ignores read_dir failures; this review therefore does not establish an exact tracked-manifest inventory.
The explicitly missing canonical_project/config_sha256 packages, required generated products, exact export equality and final combined-tree qualification remain outstanding. Current CI42825985627 is draft/stacked development with host/Atlas groups skipped; repro42825985661 passing does not make it full release acceptance. Keep this draft and #274 open pending complete reviewed delivery. Existing contributor ownership is preserved; no competing implementation, regeneration or rerun was started.
|
Landed in #480 (merge |
Part of #274: the canonical text of the project manifest and its digest (
BS-33,ProjectConfig::canonical_tomlandconfig_sha256) as language-1.2 source, with rootscanonicalProjectandconfigSha256ofProjectText. With BS-31 and BS-32, it covers every input of the source ID that is written canonically.Draft, stacked on #458 (
BS-31).ProjectTextwrites throughLockText's TOML string, array and line writers.Design
canonicalProjectwrites a configuration exactly aslexlean lockcanonicalizes it:Order, orlexicon_source = []when there is none;[limits]table, with eleven rows in their fixed order;configSha256is the lower-case hex SHA-256 of that text. Each root admits exactlyallocationandoverflow, and neither admitsrecursion. Reading and validating a manifest stay in the host.The oracle is the existing public code of
config.rs. No Rust outside the case changes.Evidence
Measured on the normative host on #458 over
release/v0.4atcf76090d4:Each of the 351 committed manifests that parse is already its own canonical text. Lean verifies the module, and the eligibility report gives each root exactly
allocationandoverflow.Gates at the head
cargo fmt --check,validate-model(480 IDs),validate-spec-links(480 rows), clippy on the workspace with-D warnings, the Intel lane plan test, and the 17 RP cases pass.CI is expected red until the regeneration
The two packages are not committed yet, so
conformance_bs_33stops at the stack-bound step after the differentials above have agreed. The regeneration will commit them, together with #458's.🤖 Generated with Claude Code