feat(cobol): migrate COBOL to scope-based resolution (regex provider) (#941)#1835
Conversation
|
@prajapatisparsh is attempting to deploy a commit to the NexusCore Team on Vercel. A member of the Team first needs to authorize it. |
✨ PR AutofixFound fixable formatting / unused-import issues across 257 changed lines. Comment |
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
prajapatisparsh
force-pushed
the
feat/cobol-scope-resolution
branch
from
0c56b01 to
ead3972
Compare
Migrate COBOL to scope-based registry resolution, validating the parse-source-agnostic contract — COBOL uses regex, not tree-sitter, but implements the same LanguageProvider interface via emitScopeCaptures. Phase 1-5 complete per abhigyanpatwari#941 DoD. New files: languages/cobol/captures.ts — emitScopeCaptures wrapping regex tagger languages/cobol/interpret.ts — import/type-binding/receiver hooks languages/cobol/index.ts — barrel export languages/cobol/scope-resolver.ts — ScopeResolver wiring (9 fields, 3 toggles) Modified files: languages/cobol.ts — wire 4 scope-resolution hooks registry.ts — register cobolScopeResolver registry-primary-flag.ts — document REGISTRY_PRIMARY_COBOL Fixtures: 17 fixture files, 30 test cases across 11 required classes test/integration/resolvers/cobol-scope.test.ts Tests: 24/24 pass (default + REGISTRY_PRIMARY_COBOL=0) tsc: zero cobol-specific errors Shadow mode (GITNEXUS_SHADOW_MODE=1): zero crashes Regex perf: 10K-line file in 408ms (threshold: 2000ms) NOT added to MIGRATED_LANGUAGES — REGISTRY_PRIMARY_COBOL env var only.
prajapatisparsh
force-pushed
the
feat/cobol-scope-resolution
branch
from
ead3972 to
fb00e19
Compare
|
Could you please achieve 100% parity? We need to add cobol into the primary registry as part of this PR. |
yes ill work on it |
|
Added COBOL to MIGRATED_LANGUAGES per your request. 24/24 tests pass under both REGISTRY_PRIMARY_COBOL=0 and REGISTRY_PRIMARY_COBOL=1 — capture-level parity is clean. Note: because COBOL uses |
|
claude is surely taking some time 😶🌫️ |
It seems GitHub is experiencing an outage. - https://www.githubstatus.com/ |
|
Reviewer persona You are a senior GitNexus production-readiness reviewer and COBOL expert. Review this PR as someone fluent in fixed-format COBOL, Area A/B layout, copybooks, Be adversarial but fair. Do not assume the PR description is true. Verify directly from files, tests, CI, and repository history. PR identity Review GitNexus PR #1835: The PR initially stated that Problem the PR is trying to solve This PR claims to migrate COBOL to GitNexus’s scope-based registry-resolution architecture under RFC #909 and issue #941, using a regex provider rather than tree-sitter. Issue #941 describes COBOL as a Ring 3 production language migration and calls for a parse-source-agnostic The production question is not merely “do captures exist?” The question is whether real COBOL ingestion, registry-primary routing, graph emission, shadow/parity validation, and same-graph behavior are correct when COBOL is enrolled in Repository history gathered RFC issue #909 says the project is moving from legacy call-resolution DAG behavior toward pure scope-based resolution, with rollout gates and parity expectations before language flips. It also identifies related infrastructure issues for shadow harnessing, registry-primary routing, and ReferenceIndex emission. ([GitHub][3]) Issue #941 is the direct COBOL migration issue. Verify that this PR satisfies its deliverables, especially Repo docs are part of the review bar. Current known state Visible changed-file domains: Runtime/provider code:
Tests and fixtures:
Visible CI/check state includes CI, CodeQL, Gitleaks, Dependency Review, Docker Build & Push, PR Autofix, scope-parity, e2e, platform-sensitive tests, and CI Gate. CI Gate, scope-parity, and PR Autofix were visible as succeeded in opened check pages, but the checks UI also showed jobs being re-run. Docker metadata warnings were visible about no Docker image/tag version being generated. ([GitHub][6]) Visible review/bot state includes a Vercel authorization notice, earlier autofix/formatting concerns that later appeared clean, Advanced Security AI issues marked fixed, maintainer feedback requesting 100% parity and adding COBOL to the primary registry, force-pushes, and a GitHub hidden/bidirectional Unicode warning in the PR UI. ([GitHub][1]) Merge status and mergeability Merge-state classification: checks pending. Reason: visible checks included successful jobs, but the GitHub checks UI still showed jobs being re-run. The final merge box, conflict status, exact required-review status, and branch-protection conclusion were not visible. Before final approval, verify the latest head SHA, every required check, branch protection, conflict-free mergeability, and whether review is blocked by unresolved conversations. Branch hygiene verdict Branch hygiene verdict: clean feature/fix PR. Reason: visible commits and files appear concentrated on the COBOL scope-resolution migration, registry flagging, resolver registration, fixtures, and integration tests. Do not treat this as final until you verify there are no hidden merge-from-main commits, unrelated generated churn, broad workflow changes, or non-COBOL behavior changes. Required repo docs Read these first and cite any missing or conflicting guidance in your review:
If any are missing locally, say so and use the closest available project guidance. Required repo context Inspect the current changed files plus the surrounding ingestion and scope-resolution contracts. At minimum inspect:
Files to inspect Inspect every changed file exactly, not only the summary:
Related files to inspect Also inspect likely integration points:
Search terms Search the repository for:
Related PRs/issues/commits to verify Verify these references directly:
Review objective Determine whether this PR is production-ready for COBOL scope-based resolution and registry-primary enrollment. The highest-risk failure mode is that COBOL is added to Do not approve based only on fixture-level capture counts. Approve only if real ingestion behavior, same-graph semantics, registry-primary routing, and parity coverage are verified. PR-specific Definition of Done This PR meets the bar only if all of the following are true:
Review method Start by reading the required docs. Then inspect the latest PR head locally, not stale rendered snippets. Suggested verification commands, adjusting only if repo scripts differ:
If any command does not exist, use the closest documented project command and state exactly what was substituted. Deep-dive assessment sections Assess these lanes separately:
Specific likely issues to verify
Hidden Unicode/hygiene checks Run these or equivalent checks:
Also run a Python scan if shell Unicode matching is unreliable: Do not block ordinary visible punctuation or non-ASCII in fixture data if the repo style allows it and the test purpose is clear. Do block hidden/bidi controls in executable code, tests, YAML, Dockerfiles, regexes, query strings, security comments, or misleading text. Output format Return your review with exactly these sections: Review bar for this PR Problem being solved Current PR state Merge status and mergeability Repository history considered Branch hygiene assessment Understanding of the change Findings PR-specific assessment sections Back-and-forth avoided by verifying Open questions that remain only if unavoidable Final verdict For every finding, use this format:
Distinguish confirmed findings from unverified suspicions. Cite files, line ranges, check names, issue/PR references, commands, and command output. Do not present a suspicion as a confirmed bug. If no issues are found, say exactly: No production-readiness issues found against the current DoD bar. Final verdict rules Choose exactly one final verdict:
Justify the verdict in 3–6 sentences. Do not choose
Review behavior guidance Focus on production readiness, not style preferences. Do not ask the author questions that you can answer by inspecting files, tests, CI, docs, or history. Convert incomplete visibility into verification steps. Block the PR only on concrete production, correctness, parity, CI, security, or branch-hygiene risks. |
|
Claude finished @magyargergo's task in 10m 36s —— View job COBOL Scope-Based Resolution — Production Readiness ReviewReview bar for this PRDoD.md §2.1 (end-to-end correct, no partial wiring), §2.7 (tests fail loudly on regressions), TESTING.md § Scope-resolution parity (migrated language must pass both Problem being solvedMigrate COBOL to GitNexus's scope-based registry-resolution architecture (RFC #909 Ring 3), implementing the parse-source-agnostic Current PR stateOpen. Latest visible commits: Merge status and mergeabilityNot mergeable as-is. CI parity status could not be retrieved (API returned 0 runs for the latest SHA). The findings below constitute production-readiness blockers regardless of CI green/red. Repository history considered
Branch hygiene assessmentDiff is scoped to COBOL files. No unrelated churn, no workflow changes, no generated garbage. Unicode scan was gated by permissions; however, Understanding of the changeThe PR adds:
FindingsFinding 1 — CRITICAL: Double graph emission when
|
|
i am looking into it |
This reverts commit f234330.
|
Don't listen to it we must the the full migration here. |
prajapatisparsh
force-pushed
the
feat/cobol-scope-resolution
branch
from
656073a to
dcf16ca
Compare
|
Reverted f234330 — COBOL removed from MIGRATED_LANGUAGES. Fixed the duplicated JSDoc and inverted arityCompatibility. The provider contract (captures, interpret, scope-resolver, registry.ts registration) is ready to merge. The MIGRATED_LANGUAGES flip will come in a follow-up PR after cobolPhase is gated with isRegistryPrimary(). |
whatt ? i am confused now |
prajapatisparsh
force-pushed
the
feat/cobol-scope-resolution
branch
from
6049b44 to
dcf16ca
Compare
|
I attempted the full migration as requested. Here's what I found:
Root cause: The scope-resolution pipeline (parse-worker) has a hard gate that skips parseStrategy: 'standalone' files entirely. COBOL is standalone (regex, not tree-sitter), so no ParsedFile is ever created → no scopes → no edges. This isn't a COBOL issue — it's a missing pipeline path for ALL standalone languages. Fixing this requires changes in parse-worker.ts / scope-extractor-bridge.ts to route standalone providers through emitScopeCaptures. That's a cross-cutting change affecting all standalone languages, not in scope for this COBOL provider PR. The provider contract (captures, interpret, scope-resolver, registry.ts registration) is complete and mergeable. The pipeline wiring is a follow-up. |
CI Report✅ All checks passed Pipeline Status
Test Results
✅ All 9958 tests passed 4 test(s) skipped — expand for details
Code CoverageTests
📋 View full run · Generated by CI |
|
@prajapatisparsh So the plan is to sort out the primary registry in a separate PR? |
yes 👉👈 |
|
@prajapatisparsh please carry on with the primary registry work 🙏 |
|
Update: opened follow-up PR #1842 ( Standalone provider wiring in parse-worker — replaced bare Test parity results:
The scope-resolution pipeline still doesn't graph CALLS/CONTAINS/ACCESSES edges for standalone languages — those produce 0 under =1. Full parity (59/59) is a separate follow-up. This PR provides the wiring infrastructure so other standalone languages can use the same pattern. |
3 participants
Summary
Migrate COBOL to scope-based registry resolution (RFC #909 Ring 3), validating the parse-source-agnostic contract — COBOL uses regex, not tree-sitter, but implements the same LanguageProvider interface via
emitScopeCaptures(sourceText) => Capture[].Changes
New modules (
languages/cobol/)captures.tsemitCobolScopeCaptures— wrapsextractCobolSymbolsWithRegexto produce RFC §5.1CaptureMatch[]interpret.tsscope-resolver.tsScopeResolverimplementation — 9 required fields, 3 togglesindex.tsWiring
languages/cobol.tsemitScopeCaptures,interpretImport,importOwningScope,receiverBindingregistry.ts[SupportedLanguages.Cobol, cobolScopeResolver]registry-primary-flag.tsREGISTRY_PRIMARY_COBOLenv varCapture vocabulary
@scope.module+@declaration.program+@declaration.name@scope.function+@declaration.function@scope.function+@declaration.function@import.statement+@import.name@reference.call+@reference.name+@reference.arity@reference.call+@reference.nameTesting
30 test cases across 17 fixture files, covering all 11 required classes (basic, COPY, CALL USING, dynamic CALL, nested programs, SECTION/PARAGRAPH, single-quoted, fixed-format, malformed, empty file, legacy parity). All pass in both default and
REGISTRY_PRIMARY_COBOL=0modes.Validation matrix:
REGISTRY_PRIMARY_COBOL=0modeGITNEXUS_SHADOW_MODE=1npx tsc --noEmit