Repository navigation
Freeze consolidation: durable gunbc work (#5402/#5395/#5407/#5400/#5396) on one branch - #5413
Merged
Merged
Conversation
`.length()` is a §3 nickname for the cardinality of a sequence; the canonical op is `count` (the only one in the compiler method table — 04_method.dag — and the only one that gets descent/termination inference via classify_size_expr). `.length()` falls through the compiler's unknown-method path and emits a call to a nonexistent `v1_rt::length`, which breaks the v1 seed (E0425). Dissolve it to `.count()`, which emits the valid host `.len() as i64` for List receivers. Scope: the two files in the v1 seed closure (the only seed .rs that emit v1_rt::length): extdeps/version/semver.dag (8) + std/realization_schedule.dag (13) = 21 sites, ALL List receivers (semantics-preserving count-nickname; no String receivers in scope). The remaining corpus .length() sites (pep440, yaml, cache_interface, compute_fabric, workflow_yaml) are outside the seed closure and land with the separate unknown-method fail-closed hardening work-item (which also needs the count-on-String host realization). Verified by execution (authoritative, not proxy — source change, count is an existing builtin so one regen takes effect): control vs treatment on the same main base — E0425 36 -> 17 (exactly -19, the v1_rt::length cluster), every other error class identical, zero regressions. v1_rt::length count in the seed: 0. emitter-completeness slice, NOT seed-green; seed-green lands in #5325. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…rkers Enum variants used as type-level kinds (the Quantity/Scale tags Memory/One/ Power/Gibi/... in Measure<Memory, One, Nat>) are emitted as bare unit structs (`pub struct Memory;`) with no derives. Any carrier that mentions them then fails its own derive closed: Measure<Q,S,M> derives Debug/Clone/PartialEq + serde, which requires Q,S: those traits, so the seed gets E0277 (`Memory` doesn't implement `Debug`) and E0369 (`==` cannot be applied to a Measure). Give the markers the matching derive set (Debug, Clone, Copy, PartialEq, Eq, serde::Serialize, serde::Deserialize) via a single-authority helper `rust_phantom_zst_marker_def`, used by both the per-enum and per-module marker emitters. Verified by execution (2-stage proxy on the integration seed): E0277 3 -> 0 and E0369 2 -> 0 (-5), no new error classes. Helper confirmed emitted into the regenerated v1_compiler_emit_rust.rs. emitter-completeness slice, NOT seed-green; seed-green lands in #5325. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The auto-commit captured a local regen of src/v1/stage0/. Per the regen-lockstep repair hygiene, the committed seed must stay main's until the green co-land (PR 5325); this slice ships the emitter .dag change only. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
17 labelled field sites (10 positive anemia violations, 7 §5-negatives) as the ground-truth set for the anemia-lens CONFIRM judge. Consumed cross-repo (ctrl coherence gate) from this fixed submodule path — not vendored. Mechanical split only (decided_by: deterministic vs haiku_confirm); no severity tiering — one violation class, everything fires blocks. haiku CONFIRM rows are a hard blocking dependency run in observe mode as a rollout phase. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Primary-precedence pool index (opt-in; strict default). Transport + witness four teeth; dependency_pool_index_compile_test Rust receipts. Gate-only on current main — no cli_run.rs, reducible, or stale self_host files. Co-authored-by: Cursor <cursoragent@cursor.com>
…eader dup. §6 scaffold: name extdeps.shell de-fork (S2) as dissolution lane for the transitional primary-precedence policy. Reframe dsl-shell receipt as §5 fail-closed guard, not dual-authority celebration. Drop duplicated phrase in transport header comment. Co-authored-by: Cursor <cursoragent@cursor.com>
…g doc pointer Addresses cursor/composer-2.5 REQUEST_CHANGES on #5407: - every `located` now resolves to a real current in-tree site (SqlMigrationStep, ServedStaticRoute, CloudAuthScheme.SigV4, OciDescriptor@container.oci, GitRemote, GitHubErrorShape, CargoDepSource.RegistryDep, Cpu/DramModuleCatalogRow, tokenize param, source_authority_contract source). - new `grounded_now` column: the audited positive sites have SINCE been grounded by the extdeps cleanup program, so the live field carries the grounded type, not the audit-time `declared_type`. README now states the harness reconstructs judge input from the row fields (declared_type/signal/coincides_with) and does NOT re-bind `located` live; the merged grounding is the ground-truth evidence for the REAL label. - README design-doc pointer no longer dangles (cites PR #5302, not a path absent on main). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…ource_root Execution proof (merry-owl-830): the within_root.contains_key guard before index.contains_key/continue never fires when the path is already in the primary root; insert_module_path remains the real within-root detector. Restore a206519 index_source_root pool_fill_only branch; remove the Rust test that asserted the dead path. Co-authored-by: Cursor <cursoragent@cursor.com>
… struct (cursor 31502) Construction peel kept naming/boxing/phantom on the AUTHORED alias (tn, e.g. Gibibyte) while feeding the expanded resolved_type — so for a struct alias the phantom field was skipped (E0063) and display could become Measure::Gibibyte. Fix: pass the peeled concrete struct name (from expanded_rt) as emit_typed_record_lit's type_name so the literal name, Box<M> boxing, and PhantomData population all agree on Measure; and guard the effective_parent fallback with variant_belongs_to_enum so a struct is never prefixed as Enum::Variant. .dag-only; seed stays main's. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…sources The discriminating witness ran the real emit pipeline via the interpreter, but `compile_sources` hits 'not yet implemented: method chars' in the interpreter, so it cannot execute (and would RED the CI floor if enrolled). Removing it; execution-verification of the construction peel rides the integration-branch full build. Keeps the PR .dag-only. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
rust_monolith_gate failed on cargo fmt --check after dependency_pool_index_compile_test mod registration. Co-authored-by: Cursor <cursoragent@cursor.com>
This was referenced Jun 20, 2026
6 tasks
Pre-existing defect from #5406, not introduced by this consolidation. `main`'s rust_monolith gate is skipped on main pushes (should_run_gates fires only when the diff touches a .rs), so #5406's lint went uncaught; the prior emit PRs predate #5406. #5413 is the first post-#5406 PR to touch a .rs, so it is the first to run `cargo clippy --all-targets -D warnings` against this file. - ok % 25 == 0 -> ok.is_multiple_of(25) (clippy::manual_is_multiple_of) - cargo fmt reflow (discover_source_root_reads chain; progress eprintln) Local receipt: cargo fmt --all --check clean; cargo clippy --all-targets -- -D warnings exit 0; cargo test -p v1-compiler-tests interp_recorded_ fixture/wet_hermetic/resolve_expr_types_retraversal = 28 passed, 0 failed. Co-Authored-By: Claude Opus 4.8 (1M context) <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
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.
Freeze consolidation — durable gunbc work on one branch
Single branch consolidating the durable, conflict-free gunbc lanes from the 2026-06-20 all-freeze digest, per operator request ("integrate all this into a single branch, finish the low-hanging fruit"). Each constituent merged conflict-free.
Constituents (each closed in favor of this branch)
dsl_compile_cleanto dual-root (the keystone gate-widen).length()nickname → canonical.count()Debug/PartialEq/serde on phantom type-tag markersPlus one carried fix (pre-existing main defect)
src/v1/stage0/src/bin/v2_whole_tree_parse_scan.rs:clippy::manual_is_multiple_of(ok % 25 == 0→ok.is_multiple_of(25)) +cargo fmtreflow. This defect was introduced by v2 self-parse closure (post-#5369 successor): whole-v2-tree parse regression on the merged gap-4 arm-body engine change (zero new rejects), then locate+fix the next self-ingest gaps through parse->normalize->infer->emit #5406 and is onmain, butmainnever caught it: therust_monolith_gateis skipped on main pushes (it fires only when the diff touches a.rs). This PR is the first post-v2 self-parse closure (post-#5369 successor): whole-v2-tree parse regression on the merged gap-4 arm-body engine change (zero new rejects), then locate+fix the next self-ingest gaps through parse->normalize->infer->emit #5406 PR to touch a.rs(via Widen dsl_compile_clean to dual-root (Strategy A gate) #5402), so it is the first to actually runcargo clippy --all-targets -- -D warningsagainst that file. Fixing it here unblocks this PR and carries the fix to main on merge.Base + local verification (ground-truth, not CI rollup)
Built off
origin/main@f08c263c1d(CIsuccess,PASS source_root_ingest_gate_passes— the digest's "main RED" was stale; #5406 closed that gap). Local receipt on the merged tree:cargo fmt --all --check→ cleancargo clippy --all-targets -- -D warnings→ exit 0cargo test -p v1-compiler-tests -- interp_recorded_fixture wet_hermetic resolve_expr_types_retraversal→ 28 passed, 0 failedcargo check -p v1-compiler→ cleanNotes
--target rustcompile-clean gate is SCAFFOLD (theCompilation failurescount is the known, tolerated M1 rust-emit residual); intermittentsccache: failed to spawnis a documented infra flake.Excluded
WIP: tmp_census.pycommit; the actualdsl/std/reducible.dagdelete + consumer migration is not committed (and dsl→v2 import feasibility is unresolved). Left open for its owner.🤖 Generated with Claude Code