fix(exec): bound recount and ordered traversal work - #1124
Conversation
|
Important Review skippedAuto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Team Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Warning Billing warning: we have not been able to collect payment for this subscription for more than 72 hours. Please update the payment method or pay any pending invoices in Billing to avoid service interruption. Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
This comment has been minimized.
This comment has been minimized.
a2f50f1 to
e838eaf
Compare
Unconstrained relationship recount now reads authenticated adjacency cardinality, and eligible ordered destination-UUID LIMIT queries stop after the required results. One- and two-hop shortcuts require a complete frontier and authenticated UUID ordering; constrained and nonmonotonic cases retain ordinary execution. Preserve aliases, bound identity lookups and sharded adjacency reads, and report actual optimized operator RSS/work.
The ladder validator now recognizes the actual optimized leaf evidence: the exact operator family and RSS record, no Arrow input batches, positive identity I/O, bounded destination probes/candidates, and no materializing reads. One-hop cardinality is exactly min(LIMIT, reopened live edges), with direct result checks and under-/over-emission negatives. Existing streaming and Expand evidence retain their own bounded checks. No thresholds, timeouts, assertions, or host qualification requirements are relaxed.
Validation on current main plus this change (isolated native target, four build jobs, ext4 TMPDIR):
fixed_hop_limitsuite: 17 passed, 2 unchanged release-only benchmarks ignored; 1334.27s. Covers ordinary two-partition counts, NULL/restricted/undirected semantics, nonmonotonic UUID ordering, aliases, 65,537-destination chunking, huge LIMIT, 1x/2x sharded hubs, hydration and graph-growth bounds.scale_g500_laddersuite: 26 passed in 188.12s. Includes the three previously failing CI cases and adversarial optimized-evidence validation.-D warnings, fast checks, 14 gate-registry tests, final formatting and diff checks: passed.The certifier receipt contract now includes the numeric adjacency-probe field and preserves full query evidence while continuing to reject missing, malformed, or unexpected fields. The independently verified workflow failure-propagation fix (#1126/#1125) is merged into this base.
Final integration on
6a937da07a6ee9eb5ae5b2b4247f7432a5e7211aretained the engine and ladder-validator patches unchanged by range-diff.make -C benchmarks smokepassed all 386 Python tests and Rust runners (26 certifier tests), workspace Clippy/formatting, and the actual Cargo boundary check. Certifier all-targets Clippy passed. Fresh CLI and all benchmark binaries completed the real ten-phase tiny lifecycle with 2 recount, 2 query, and 4 reopened-query receipts and matching source/imported results. Final fast/gate/format/diff checks passed.Canonical #1094 remains open for integrated OVHC-AGENCY S18/S19 plateau and S20 admission/completion or next typed failure. These are GraphForge lifecycle measurements using Graph500-compliant generated input, not official Graph500 performance results. This PR supplies the existing #1116/#1079 prerequisite; lifecycle accounting remains there.
Closes #1119. Exact-head CI run 33951664925 passed at
e838eafa5ae458a267c6f7820de8a69d34f1de28, with successful CI Gate, CLEAN merge state, and no unresolved review threads. Squash merged as27673a642ad2bb0cd660a2a1fa8cf831cf2a419d; issue #1119 is closed.