fix: persist resolution failure metadata

[hopstreax]

Summary

Fixes #3667

Cross-file resolution failures were previously caught and logged, but the extraction artifacts gave no indication that a resolver had failed. This could leave graphs silently missing call/reference edges while appearing healthy.

This PR persists resolver failure metadata as degraded_passes and exposes it through the graph, reports, and diagnostics.

Changes

• Record failures from language resolver registry and hand-wired cross-file resolution passes.
• Persist degraded_passes through extraction, build, watch, and CLI flows.
• Serialize degraded resolution metadata in graph.json.
• Display failed resolution passes in GRAPH_REPORT.md.
• Surface degraded passes through extraction diagnostics.
• Preserve degraded state across partial incremental re-extractions.
• Clear degraded state when the affected language corpus is successfully re-extracted.
• Clear stale failures on full rebuilds.
• Maintain backward compatibility with existing graphs.
• Keep clean graph.json and GRAPH_REPORT.md output unchanged.
• Normalize degraded metadata for deterministic watch/topology comparisons.

Verification

• 16 degraded-resolution regression tests passed.
• tests/test_build.py: 94 passed.
• tests/test_watch.py: 234 passed, 13 skipped, 2 deselected.
• Export/report/diagnostics/resolver suites passed.
• git diff --check passes.
• python -m graphify update . completed successfully.

No unrelated changes are included.

[graphify-labs]

Graphify reviewed this change.

Worth a look — the grounded gate found no coupling regressions or blocking issues, but 2 advisory finding(s) below merit a look before merge.

---

Graphify review — findings

Adds first-class tracking of degraded resolution passes end-to-end: extract records each pass that fails via _record_degraded_pass (capturing the pass name, error type/message, and affected file suffixes) into a degraded_passes list that flows through extract_parallel and safe_extract_with_xaml_root. Propagates degraded_passes through build, build_from_json, and the CLI extract merge, and reconciles them on incremental merges in merge_raw_extraction and build_merge — carrying forward an existing pass only when its suffixes still have surviving AST nodes and a fresh pass hasn't superseded it, then re-sorting by pass name. On export, to_json lifts degraded_passes to the top level (dropping it from the nested graph dict to avoid dual persistence) only when non-empty, and diagnose_extraction/format_diagnostic_report surface the count and per-pass detail lines.

Worth a look

• build_from_json rejects previously ignored non-iterable degraded_passes metadata — graphify/build.py:1481 · Escalate · medium
  • agreed by 2 of 2 members but NOT verified (no proof, no reproducing execution) — consensus is not a verdict; needs human review
• build() drops degraded metadata for later suffixes with the same pass — graphify/build.py:1527 · Escalate · medium
  • agreed by 2 of 2 members but NOT verified (no proof, no reproducing execution) — consensus is not a verdict; needs human review

Analysis details — impact, health, verification

Impact & health

Graphify review

Impact — 3926 functions depend on the 1702 functions this change touches.

Health — this change adds coupling hotspots:

• new: extract() — 657 callers, 46 callees
• new: _rebuild_code() — 142 callers, 54 callees
• new: build_from_json() — 219 callers, 20 callees
• new: build_merge() — 77 callers, 14 callees
• new: to_obsidian() — 38 callers, 14 callees
• new: to_json() — 62 callers, 7 callees
• new: extract_js() — 87 callers, 4 callees
• new: extract_files_direct() — 17 callers, 20 callees
• …and 77 more — each is listed as a finding

Verification — 3926 functions in the blast radius were not formally verified this run (proofs are advisory here).

Gate & verification

graphify gate

PASS — objectively clean (no health regressions, tests not run — proofs not run this pass (advisory)). Grounded, not self-assessed.

Advisory (not blocking):

• verification_scope: 3725 function(s) in the blast radius were not formally verified this run

Test selection

Test selection

211 of 290 test file(s) selected (73%) via static blast radius.

• tests/test_affected_cli.py — impact
• tests/test_agents_platform.py — impact
• tests/test_analyze.py — impact
• tests/test_astro_extraction.py — impact
• tests/test_astro_import_ids.py — impact
• tests/test_atomic_canvas_export.py — impact
• tests/test_atomic_writes.py — impact
• tests/test_benchmark.py — impact
• tests/test_benchmark_raw_graph.py — impact
• tests/test_build.py — impact, changed-test
• tests/test_build_merge_dedup_scope.py — impact
• tests/test_build_merge_hyperedges_and_prune.py — impact
• tests/test_build_merge_shrink_guard.py — impact
• tests/test_builtin_global_type_refs.py — impact
• tests/test_carried_hyperedge_remap.py — impact
• tests/test_case_sensitive_resolution.py — impact
• tests/test_charmap_encoding.py — impact
• tests/test_chunking.py — impact
• tests/test_cjs_module_extension.py — impact
• tests/test_claude_cli_backend.py — impact
• tests/test_cli_export.py — impact
• tests/test_cluster.py — impact
• tests/test_codebuddy.py — impact
• tests/test_community_labels_skill.py — impact
• tests/test_confidence.py — impact
• tests/test_corrupt_graph_json.py — impact
• tests/test_cpp_nested_and_cli.py — impact
• tests/test_cpp_objc_cross_file_calls.py — impact
• tests/test_cross_extension_reexport_self_cycle.py — impact
• tests/test_cross_language_call_resolution.py — impact
• tests/test_cross_repo_external_call_guards.py — impact
• tests/test_cross_repo_member_calls.py — impact
• tests/test_csharp_call_site_generic_args.py — impact
• tests/test_csharp_enum_members.py — impact
• tests/test_csharp_field_generic_args.py — impact
• tests/test_csharp_generic_callsites.py — impact
• tests/test_csharp_interface_dispatch.py — impact
• tests/test_csharp_member_calls.py — impact
• tests/test_csharp_member_nodes.py — impact
• tests/test_csharp_object_creation.py — impact
• tests/test_csharp_partial_classes.py — impact
• tests/test_csharp_type_resolution.py — impact
• tests/test_dedup.py — impact
• tests/test_dedup_remaps_hyperedges.py — impact
• tests/test_definition_file_portability.py — impact
• tests/test_detect.py — impact
• tests/test_devin.py — impact
• tests/test_dotnet.py — impact
• tests/test_duplicate_annotation_edges.py — impact
• tests/test_elixir_import_resolution.py — impact
• … and 161 more

Selection is safe under the controlled-regression assumption; always-run tests + a periodic full run are the backstops. Advisory — it never changes the check verdict.

· 85 more finding(s) on lines outside this diff (see the check run).