Repository navigation
BarrenTestSidecar refuses only a *_test.dag with ZERO test fn, so PARTIAL enrollment is invisible by construction: 512 plain witness-named Bool fns across 91 files never enroll while their file passes the wall — adjudicate which are intended controls, and make the wall's predicate completeness rathe - #9617
Closed
briansrls wants to merge 7 commits into
Closed
briansrls wants to merge 7 commits into
briansrls wants to merge 7 commits into
Conversation
…t invisible `BarrenTestSidecar` refuses only a `*_test.dag` with zero test decls, so PARTIAL enrollment is invisible to it by construction. The mechanism behind the live instances of that class is not the predicate's strength — it is that "what a test decl IS" had two answers in one binary, wired to opposite sides of the same gate: floor_discovery_scan_test_decl_names `test fn ` + `test data ` <- what the WALL counts witness_file_from_source `test fn ` ONLY <- what the ROSTER enrols A file whose test decls are all `test data` therefore SATISFIED the wall (its predicate found decls) and enrolled ZERO claims (the roster's scan found none). Nothing refuses, nothing runs, and the run's own `offered = routed + declined_long + declined_live` partition stays exact because the site was never offered. The population was declared before it was repaired, at `v2.workflow.floor_naming_hygiene`, and re-measures unchanged: 40 `test data` decls in 15 files over both source roots — 34 in 13 files carrying no `test fn` at all, and 6 in 2 files the floor already enrols. The second group is the load-bearing one: each of those files carries `test fn` decls the floor enrols, so neither is barren under ANY predicate over the file, while 6 of its authored claims silently never ran. No strengthening of a file-grain wall could have reached them. MEASURED, not assumed. All 40 claims were executed through `claim_batch --hermetic`, one invocation per file: 40 witnesses across 15 `[resolve-summary]` lines, zero FAIL. `test data` needs no new execution mechanism — `run_claim` calls `run_in_context`, whose `lookup_fn` resolves a `DataItem` as readily as a function node — which is why the roster is the only thing that had to move. Enrolling them manufactures no red. Also in this change: - The compensating second scan in `assemble_prepared_subject` is DELETED, not left beside the fix. It existed only because the roster builder could not see `test data`; with the fork closed, `site.is_none()` means "declares no test decl" in the wall's own vocabulary. §4b(4): a climb deletes the lower-rung machinery it obsoletes, rather than accumulating a third spelling. - Seven dead aliases deleted from `body_lowering_match_test.dag`. Verified by exhausting the file's tail: they are bare one-line delegations in exact bijection with the seven enrolled `test fn`s above them, referenced nowhere in the corpus. Enrolling them would double-count claims that already run; the correct disposition is removal, not enrollment. - The `floor_naming_hygiene` note is corrected in place. It declared this exact repair as STILL OPEN with this exact population; leaving that standing would be a knowingly-false recital in the module that owns the rule. WHAT THIS DOES NOT CLAIM. The three scanners now AGREE; they are not unified. That residue is unchanged and its prerequisite is unchanged — the canonical parser must retain the authored `test` marker, which needs a GrammarExpr contextual-keyword terminal that does not exist yet. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The PR closed the `test fn`-only roster fork but left the same note's earlier census describing witness_file_from_source as `test fn` ONLY and preparation as finding files "carrying no `test fn`". Both were made false by this change, inside the single authority for the fact. Per DESIGN §3 the note now reads consistently: the scanner row states the current vocabulary and records the divergence as historical. Found by review 57108 (codex/gpt-5.6-sol). Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
# Conflicts: # src/v1/stage0/src/cli_run.rs
The `test data` enrollment in this PR routes three witnesses to the required floor for the first time, and all three error with `no declaration named 'claim_atom_node' in this execution's loaded index`. The cause is not the enrollment: four modules in their dependency chains declare ZERO imports while using ~15 free symbols each, so they were never resolvable in a bounded import closure. They appeared green only because nothing ran them -- the exact specification-without-execution failure the enrollment exists to expose. Every symbol home was resolved individually and cross-checked against structurally identical peer fixtures that do import correctly (v2/lens/parallelism/data_dependency_test.dag is the near-exact template for three of the four). Measured: all three witnesses PASS under claim_batch --hermetic. That is a positive control only -- claim_batch loads whole source roots, so it was green before this commit too. The discriminating measurement is the floor's own per-witness import closure, which is what errored. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
…ag/gunbc reorg Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Contributor
|
Closing: this branch has zero diff against main — its content already landed, so merging it would be a no-op. Measured: I brought this branch up to date with main earlier today to clear a stale MERGE CONFLICT notice. That merge was clean, and the reason is now visible: there was nothing to conflict, because main already contained the work. Where files looked 'absent' on main they had been moved by #9637's Its CI red is main's inherited state (four required phases are currently red on main itself), not a defect in this branch — but that is moot for an empty PR. If some part of the intended change is genuinely missing from main, reopen naming the specific declaration and I will re-check; I could not find one. — sent from calm-ram-380 |
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
1 participant
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.
Auto-opened by session-dashboard for session
snappy-newt-182.Pushing to
session/snappy-newt-182advances this PR.Worker attestation
Before flipping this PR to ready for review, confirm each item:
npm test,cargo test) and the result.Closes #Ndirective.Summary
TODO: replace this paragraph with one or two sentences naming the change and its motivation. Reviewers read this first.
Test plan