fix(commonlisp): derive node ids from the full path stem

[guitelesc]

extract_commonlisp builds its id prefix from the bare path.stem, so a/sample.lisp and b/sample.lisp both mint sample / sample_init. Every other extractor uses _file_stem(path), which preserves the full repo-relative path for exactly this reason (#1504) — commonlisp.py never imported it. It looks like the line predates the port out of extract.py and came along unchanged.

Impact

Not cosmetic — nodes are lost. The in-batch collision rescue hides it when both files land in one extract() call, but incremental graphify update extracts in separate batches:

=== SEPARATE batches (what incremental update does) ===
  [a] 'sample'  'sample_init'   <- a/sample.lisp
  [b] 'sample'  'sample_init'   <- b/sample.lisp
  COLLIDING IDS ACROSS BATCHES: ['sample', 'sample_init']

=== merged graph: 2 nodes (expected 4) ===

build() merges last-writer-wins and b/sample.lisp disappears. It does warn loudly, which is how I found it.

The id-remap post-pass doesn't cover this either: it rewrites absolute-path-derived ids (#502), and a bare stem matches none of its keys.

This also means any graph containing a .lisp file trips graph_has_legacy_ids, so query/serve print the "pre-#1504 node-ID scheme" nudge on a freshly built graph. On this repo, tests/fixtures/sample.lisp alone was enough.

The fix

stem = _make_id(_file_stem(path))

Pre-collapsed through _make_id rather than passed to _cl_id raw, because _CL_CHAR_MAP maps / to _slash — the naive stem = _file_stem(path) that matches the other extractors yields tests_slashfixtures_slashsample_init. Pre-collapsing is idempotent under _cl_id and leaves the CL operator-char handling intact (upi= -> upi_eq, verified via header_eq / header_lt in the existing fixture).

Testing

Added test_cl_ids_are_path_qualified_across_directories, modelled on the existing test_decldef_merge_does_not_merge_across_directories. Red/green verified by reverting the one-line change: fails with overlap ['sample', 'sample_init'], passes with it.

uv run pytest tests/ -q — 5086 passed, 11 skipped, 0 failed.

One note for reviewers: the AST cache is keyed by content hash, so after this lands a graph built with an older graphify keeps the old ids until graphify-out/cache/ast is cleared. The cache dir is version-stamped (v0.9.50-s2), so a release bump handles it.

🤖 Generated with Claude Code

[graphify-labs]

Graphify reviewed this change.

Looks safe to merge — no coupling regressions and no blocking issues, checked against the code graph (not a self-assessment).

Formal verification. No changes could be formally verified in this run.

---

Graphify review — findings

Fixes Common Lisp node id collisions between same-named .lisp files in different directories by deriving the id prefix from the path-qualified _file_stem (run through _make_id) instead of the bare path.stem, so separate extract batches no longer merge and drop one file's nodes (#1504). Adds a regression test asserting extract_commonlisp mints disjoint ids for a/sample.lisp and b/sample.lisp.

No blocking issues surfaced. 2 lower-confidence candidates did not survive cross-model review.

Analysis details — impact, health, verification

Impact & health

Graphify review

Impact — 547 functions depend on the 547 functions this change touches.

Health — this change adds coupling hotspots:

• new: extract_commonlisp() — 28 callers, 6 callees
• new: _process_form() — 1 callers, 10 callees
• new: _handle_defpackage() — 1 callers, 7 callees
• new: _handle_def_form() — 1 callers, 6 callees

Verification — 547 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: 547 function(s) in the blast radius were not formally verified this run

Formal verification

Could not verify: Could not verify extract\_commonlisp.

The verifier did not have enough to check extract\_commonlisp, so it is saying so rather than guessing. No false assurance is the whole point.

Guarantee: No guarantee either way, this is an honest abstention, not a pass.

Note: Reason: parameter `path` is annotated `Path` — outside the synthesizable primitive/collection set

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