Repository navigation
perf(graph): share applicability inputs during prerequisite verification - #116
Merged
Merged
Conversation
This was referenced Nov 10, 2026
This was referenced Nov 11, 2026
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
A single prerequisite verification recompiles several historical/current canonical frames. Each compilation previously rediscovered and reparsed the same applicability mappings, provenance and active Memory evidence. This change captures those ambient inputs once per verified context while freshly compiling every graph frame and retaining every ordered reduction/reinheritance check.
The snapshot binds exact file bytes, directory membership and missing optional inputs. Changed/new/deleted inputs, expired/forged handles, or changed loaded proof/auditor/compiler code fail before a comparison capability is returned. No proof result, graph result or parent-history pass is cached; the multi-context regression ladder remains mandatory. The five new counterexamples run first in the existing Curriculum CI
test:transitive-minimalitycommand.Local validation passed: five snapshot regressions, ten source/applicability regressions, twelve actual CLI cases, the existing exhaustive graph/origin counterexamples, and ordinary eslint. Three complete outputs across all 21 catalogs / 6,524 goals match exactly, including evidence and findings; the normalized serialization also matches the archived Main report. A local measurement was 11.55 s for independent compilation, 12.00 s for initial snapshot loading, and 8.55 s for fresh compilation reusing inputs. This is one bounded measurement, not a whole-CI speedup claim.
Astra reviewed the exact implementation and generated-child fixture correction. The actual current V7 semantic-kind bridge passed in 254.86 seconds on this head: its complete declared 21-catalog context verified 881 affected sources and emitted all 766 Physics goal bindings, with all bound inputs unchanged. This is one successful real verification context, not the full repeated historical regression ladder. Full proof-regression and fresh remote CI acceptance remain pending. Canonical content, proof grammar, milestone floors and standard G17 scope are unchanged. This independent Main-based performance delivery is compatible with the separate optional assessment-basis work in #114.