Repository navigation
v2.std.collection: primitive-backed map_insert/map_lookup delegates - #11121
Conversation
…closure chain. Keyed lookup is now the same HostRealizedSeam shape as empty_map, bound to the declared insert and lookup contracts, so native maps stay HAMT on both the interpreter and emitted-Rust paths. Co-authored-by: Cursor <cursoragent@cursor.com>
The build lane refused generated surface drift: the emitter orders primitive_lookup before empty_map, which the hand patch did not. Co-authored-by: Cursor <cursoragent@cursor.com>
The seam still refuses wrap-body fallthrough; insert on a non-native carrier returns the carrier unchanged so lookup stays Absent, which is a Bool verdict (a TypeError is RuntimeErrored and cannot stay enrolled). Co-authored-by: Cursor <cursoragent@cursor.com>
Keep the inverted record-shaped control as a Bool predicate over the same Value::Map match; a throw cannot stay enrolled as a floor verdict. Co-authored-by: Cursor <cursoragent@cursor.com>
The predicate now runs map_insert grounding instead of a parallel Value::Map match, so restoring Ok(None) fallthrough turns the control red. Co-authored-by: Cursor <cursoragent@cursor.com>
|
review 64214 (#11121, dashboard artifact On Calling |
|
review 64244 (dashboard artifact
— sent from clever-fox-89 |
A record-shaped insert refuses with InterpError::TypeError; n HAMT inserts and sublinear first-key lookup steps witness the delegate vs wrap. Co-authored-by: Cursor <cursoragent@cursor.com>
|
review 64269 (dashboard artifact That arm is gone on head — sent from clever-fox-89 |
|
review 64373 (dashboard artifact That declaration is deleted on head — sent from clever-fox-89 |
… TypeError control. Co-authored-by: Cursor <cursoragent@cursor.com>
|
review 64382 (dashboard artifact Head |
Co-authored-by: Cursor <cursoragent@cursor.com>
|
review 64382 (dashboard artifact On current head
The wrap-test file remains tracked; clippy — sent from clever-fox-89 |
Keep NAMESPACE_TRANSITION_ADMISSIONS empty after #11137 retired on both sides. Co-authored-by: Cursor <cursoragent@cursor.com>
… replay Three conflicts. namespace_wave_admission.rs is doc-comment only again -- main recorded a thirty-sixth dissolution via #11121 and the roster is empty on both sides -- so main's text is taken. compiler_tests.rs and v1_compiler_compiler_tests_rust.rs are GENERATED. Main changed their .dag generator (#78e8a188aad) and so did this branch (the five-argument parse_with_table_in_occurrence_scope call); the .dag merged cleanly, so the authority is right and only the derived files needed re-deriving. With no build host available, both are booted from main and the emitter's one-row delta replayed onto each -- the same move that CI's regen phase confirmed byte-correct on #10850's parser mirror. The generator mirror is emitted as one left-nested v1_rt::concat chain per function, so the replay there splices the env row into the chain (one more concat( at the head, ", E)" after the allocator row) and asserts paren balance is unchanged. The build lane's regen phase adjudicates; a drift report means the emitter disagreed and its bytes win. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01NQnvBvFGNu9tNL544NJe2j
…re its row MERGE of origin/main dfac90c (#11121, #11142, #11154, #10986). Three conflicts. namespace_wave_admission.rs: main's THIRTY-SIXTH DISSOLUTION record kept, this branch's 181 still-required rows kept with their 181 expected_candidates. accepted_source_emits_uncompilable_target.dag: took MAIN's side -- their rung claim is mechanically preventable backed by executing evidence, this branch's said still-mitigatable and is superseded, and their four added receipts are preserved. v1_compiler_emit_rust.rs: resolved BY REGEN, never text-merged; main's bytes stood as a placeholder and the regenerated mirror overwrote them from the merged .dag authority. THE WALK IS REPAIRED RATHER THAN DEFERRED, because the primitive it was waiting for is now in the tree. #11121 landed as dfac90c (v2.std.collection primitive-backed map_insert/map_lookup delegates), which is the capability the declared drop named. DESIGN section 6: a proven cost-shape defect is always fixed, and a trigger amended after the capability arrives is a deferral with better wording. SHAPE. native_lane_facts_index builds module -> imports ONCE, before the recursion. native_lane_module_reachable looks up only what the frontier names and carries seen_set through the recursion. Neither is rebuilt per round: rebuilding either would reintroduce the cost under a keyed spelling. A MODULE DECLARED TWICE APPENDS, AND THIS IS THE CASE A REVIEW WOULD HAVE PLANTED. The fold this replaces visited every fact whose module the frontier named, so two facts declaring one module contributed BOTH import lists. A naive map_insert keeps the last and SHRINKS the closure -- a behaviour change wearing a performance change's clothes, and the same silent narrowing this PR has already repaired twice. The index appends on a duplicate key. PRESERVED: declared membership; the refusal arms (fuel exhaustion still refuses the whole derivation by identity with budget and frontier); last-round completion. DIVERGED, DELIBERATELY AND STATED ON THE CARRIER: discovery order is FRONTIER order, not FACT order. Preserving fact order needs a per-round pass over all modules to re-derive it -- the repeated scan this repair removes. It is unobservable, checked rather than assumed: native_lane_closure_ingest filters by membership (the ingest supplies its own order), both halves of native_lane_ingest_matches_closure are membership tests, and the emitted receipt carries `closure.len()`, a count, never the sequence. If a consumer that reads the sequence ever appears, THAT change owns this order fact; it cannot be inherited silently from here. NOT CLAIMED: that the route is now O(closure). One walk changed. The instrument is the universe_derivation span -- 192.4 s at 36e6ad9 -- and the successor run on this head is the before/after. No cost witness is added: no corpus home can hold a planted-quadratic control for this walk inside the 500 ms line without a synthetic population that exceeds it, and a witness that cannot discriminate is worse than none. ROW RETIRED BY ITS TRIGGER, with the repair as the discharge, in this same commit -- never in an intermediate state with the original scan still standing. Retiring on the capability alone would have been the 4b(3) inflation: a row marked discharged with the quadratic walk intact. CITATIONS (review 64576). The row's population named native_lane_closure_grow, which exists nowhere -- a section 3 citation defect, and worst in that field, because 4b(3) requires a BOUNDED population and an unreadable member means whoever discharges the row cannot enumerate what to delete. Corrected to the real symbols. Every candidate symbol in both rung-drop rows and the seed-growth row was then swept by git grep: 26 checked, all resolve. AND THE REPAIR RE-STALED A CITATION FIXED MINUTES EARLIER: once the walk stopped calling native_lane_facts_module_named, the `Consumers:` comment naming it was wrong again. Corrected, and it now says the frontier walk no longer reads it. Both projections regenerated mechanically on the final tree: the stage0 mirror (installed from the candidate) and docs/design-rung-drops.md (4 added, 2 removed, carrying the retirement). Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013aZDLk2CxsCDznqn49Xhe8
… new witnesses at the 100ms-equivalent Operator ruling via fierce-lark-661, folded into the same change. A single ceiling has to be either loose enough for the population that grew up under 500ms or tight enough to be a standard for new work, and it cannot be both. So the floor now carries a debt and a standard, not two standards. THE ROSTER IS A MONOTONE DEBT CONTRACT AND ALL THREE OF DESIGN §5'S CONDITIONS ARE MET RATHER THAN ASSERTED. The subject universe is independently discovered and closed — 3,795 identities generated from run 34726300841's own required_floor_disposition artifact, never hand-typed. Membership is at identity grain. Rows may only LEAVE, each removal carrying a typed disposition; an ADDITION REFUSES, which is what makes it a debt rather than a list that grows. THE TIER IS DERIVED FROM MEMBERSHIP AND FROM NOTHING A CALLER SUPPLIES. No flag, no `legacy` column, no label. An identity outside the roster cannot acquire the larger budget by landing, by appearing in the tree, or by having been accepted once. A rename is a deletion plus an addition, so a renamed witness is new — that falls out of identity-grain membership rather than being a rule on top, and it closes the one move that would launder an expensive witness into a fresh identity at the old ceiling. CALIBRATION. 723 steps/ms — 110,011/152ms, the slowest reading from the machine that runs the floor, across five readings on two architectures that all returned an identical eval_steps. Budgets 361,500 and 72,300; neither is a literal, both are the tier's policy milliseconds through one conversion. THE SPECIMEN WAS SHRUNK to 55,011 steps so it sits under the new ceiling it helps derive: grandfathering the measuring stick would exempt it from the tier it defines. TWO THINGS MEASURED RATHER THAN ASSUMED, BOTH OF WHICH CHANGED THE WORK. - My own evidence breached the tier it installs. the_roster_identity_join_... cost 90,086 steps folded together, over 72,300, and these witnesses are NEW. Split into the join and a separate non-truncation check. - Main landed 54 identities after the cut, so they classify NEW. Ran them: all 54 under the ceiling. The 8 witnesses in that set that DO exceed 72,300 are every one of them grandfathered, so the tier does not red the floor on someone else's landings. THE MAP-PRIMITIVE MEASUREMENT. The roster was built three ways: `any` over a flattened list (92,490 steps / 148ms), map-backed via empty_map/map_insert/ map_lookup (39,940 steps — the FEWEST — and 36,000ms CPU), and a short-circuiting descent (8,066/19ms hit, 43,583/76ms miss). map_insert returns a new map, so folding n rows is O(n^2) host work and none of it is an eval step: the prescribed repair for a nested scan reads as cheapest on the gated clock while costing 400x the CPU. Per fierce-lark's ruling that is filed as a REALIZATION DEFECT in the primitive (gunbc.recurring_failure_mode nested_membership_scan_where_a_map_primitive_exists, owner clever-fox-89/#11121), NOT as author guidance — a per-site "check your n" rule prices an exception around a cost-shape defect. The descent stands as a stated interim. REVIEW: the blocking bare-Int millisecond finding is fixed — specimen_cpu is std.measure Millisecond and the conversion takes a Millisecond envelope. 90/90 witnesses green, clippy --all-targets -D warnings clean, --required-regen first_generation_equal=true planned=155, both docs projections regenerated. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BRTesxTRxhRGKX1py2pHVC
…ed standing review 65283 on #11254: the appended receipt said the specimen is reproduced while the older sentences still said NOT REPRODUCED and that no pre-#11121 binary had been run, so one identity answered both ways about one fact. The specimen line now reads reproduced and names where the reproduction lives, keeping the prior wording visible as prior wording; the uncertainties row says which uncertainty is discharged and leaves the emitted-path and denominator ones standing unchanged. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01JKwxo5SvZ7qqsqEpEpEwVm
1 participant
Summary
Map { lookup: ... }closure-chain bodies with HostRealizedSeam delegates matchingempty_map_primitive_delegate, bound tomap_insert_contractandlookup_contract.v1_rt::rc_map_insert/v1_rt::lookup.Closes the capability named in #10940 review 64181 finding 1. The native-lane membership join stays open until it consumes this.
Test plan
cargo test -p v1-compiler --test interpreter_dispatch_authority --releasev2.test.claim.map_carrier_shape_gateholds as a changed witnessMade with Cursor