fix(swift): preserve declaration identity and compatible call edges

[DavidHLP]

What does this PR do?

Fixes Swift overloads sharing a graph node and inheriting each other's calls in #2061. This builds on the merged #2342 plumbing and enables Swift only.

Declaration identities retain parameter labels and types, generics and constraints, where requirements, async, qualified type paths and explicit return types. Node names stay bare. Call resolution uses labels, arity, defaults and trailing closures, and keeps every compatible candidate when it cannot choose one. It does not infer types or resolve aliases.

Qualified types are written as A/Item; explicit returns as pick()=>Int or pick()async=>String. Format4 rebuilds older indexes once to replace collided nodes and stale calls. I'd appreciate a review of that format boundary before merging.

Head: 6ca78d06e66a8d60c12c72618cbd9fd6cf9f8e21, merging main e71f23ecf40e473a3e0a49cff1ffebcccead384d without rewriting published history. Six conflicts were resolved by retaining Swift signature/default metadata alongside upstream definition test metadata, language-aware registration and SPAWNS handling. Incremental registry seeding restores both language and Swift metadata; unresolved parallel route calls retain upstream self-loop suppression. GitHub now reports this head mergeable.

At this merge head on local Linux/GCC 16.2.1, conflict-file syntax checks, whole-tree formatting, NOLINT and memory-core checks passed. The five-suite ASan/UBSan build was stopped after a prolonged first build in generated LSP standard-library data; no suite result is claimed. The full test entry stopped at a sandbox-denied fixture-server bind. CI lint cannot run locally without cppcheck, and full clang-tidy reports errors. Current-head DCO passed on GitHub; remaining upstream checks and maintainer approval are pending.

The runtime evidence below is historical: tested code 10f4769779c8ec63d509b7884219187a0e7a1939, documented at 35533b2674a09e2e6e9a066faf52e3fb8ccdaada. It has not been rerun at the merge head.

The new tests fail on the old implementation with one node instead of two for both qualified-parameter and return-only overloads. They pass after the fix and check source locations, outgoing calls, candidate counts, serial/parallel indexing, persistence and migration. Existing default-argument, closure, ownership and boundary regressions are retained. The standalone graph/MCP drivers are opt-in.

On remote-dev (Linux x86_64, GCC16.2.1), 909 focused tests and 328 pipeline TSan tests passed. Strict graph10/10 and real MCP5/5, the clean production build and CI lint also passed. CI lint needed a retry with a longer time limit; the timeout log is retained. All49 commits pass DCO.

There are still two ObjectScript UBSan reports, so the focused run is not sanitizer-clean; their cause is unconfirmed. The full C entry stops at the Scoop metadata precheck, full clang-tidy is not passing, and native CLI version/help exit139 with the same startup stack as the earlier main comparison. Current macOS/Windows/Ubuntu CLI semantics have not been checked. Upstream checks and maintainer approval are pending.

Review summary · Commands and raw logs · Validation notes.

Written with Codex on behalf of DavidHLP.

Checklist

• Every commit is signed off (git commit -s) — all49 checked.
• Tests pass locally (make -f Makefile.cbm test) — focused remote runs pass; the full run and sanitizer limitations are noted above.
• Lint passes (make -f Makefile.cbm lint-ci) — historical CI lint passed; current-head upstream lint pending and local cppcheck unavailable.
• New behavior is covered by a test (reproduce-first for bug fixes).

[github-actions]

Thanks for opening this — it has been seen, and it is queued.

This note is automated, but it is not a brush-off: it exists so you know where your PR stands instead of having to guess from silence.

Current review status: working through a backlog. 0.9.1-rc.1 is out, so the release freeze that held reviews is over — but it left a large queue of open pull requests behind it, and we are reading through them oldest-first. The background is in discussion #1144.

What that means for this PR, concretely:

• It will not be closed for inactivity. No stale bot touches pull requests here.
• It may still sit a while before a human reads it. That is on us, not on you.
• Older PRs are read first, so a recent one is not being skipped — it is behind a queue.

Things that will genuinely speed it up whenever review does happen:

• Keep it rebased on main — the tree is moving quickly right now, and a conflicting branch cannot be reviewed as the diff you intended.
• Get CI green, or say which failures you believe are pre-existing.
• Keep the change to one claim. Bundled features and refactors get split before they get merged, which costs you a round trip.
• Every commit needs a sign-off (git commit -s) — CI enforces DCO.

If this fixes a bug, a reproduction we can run is worth more than a description of the symptom.

Thanks for contributing, and sorry in advance for the wait.

[DeusData]

Thank you for the detailed status update and for clearly distinguishing the actual PR head from unpublished work and unverified results. The changed-file scope includes callable identity, resolution and persisted metadata, so we need more time for a complete review together with the parent change. This remains queued for deeper review; this comment is not an approval or a request for another routine rebase. The queue is full, and we appreciate your patience and the care taken to document the remaining work.

[DavidHLP]

I've finished the Swift fixes on top of #2342. The current head is 35533b2674a09e2e6e9a066faf52e3fb8ccdaada; the code was tested at 10f47697, and the last commit only updates the validation notes.

The two remaining collisions now have separate identities: A.Item versus B.Item, and functions that differ only in their declared return type. The tests reproduce each collision before the fix, then check separate nodes, source locations and outgoing calls, including parallel indexing and reopening the index. Call resolution still keeps all compatible candidates rather than guessing from types.

The main thing I'd like feedback on is the identity spelling: qualified types use /, explicit returns use =>Type, and format4 rebuilds older indexes once. The implementation and migration tests are in callable_sig.c, registry.c and test_index_format.c.

On remote-dev, 909 focused tests and 328 pipeline TSan tests passed, along with the separate graph and MCP checks, production build and CI lint. There are still two ObjectScript UBSan reports, so the focused run isn't sanitizer-clean. The full C run stops at the Scoop metadata precheck, full clang-tidy still reports errors, and the native CLI crashes at startup with the same stack seen in the earlier main comparison. Current macOS, Windows and Ubuntu CLI checks haven't been run.

Test commands and full logs are available here. I removed the duplicate progress comments; their original text is archived. This is ready for review, with the remaining checks and approval still pending.