test(interop): go-libp2p WebRTC-Direct interop tests

[yashksaini-coder]

The formal go-libp2p interop tests for #1437.

A small pinned go-libp2p v0.49 harness (tests/interop/go_libp2p/webrtc_direct/, listen / dial -version N) is built on demand under a file lock and skipped when the Go toolchain is missing, then driven against our WebRTC-Direct transport in both directions and both versions.

• py → go: passes live for v1 and v2 (dials a go listener, completes Noise, checks the peer ID).
• go → py: xfail (strict) — the inbound path tears the peer connection down right after ICE completes, tracked in WebRTC-Direct: go-libp2p dialer can't connect to the py listener (ICE succeeds, then the connection is torn down) #1470. The strict xfail means the suite will flag it the moment that's fixed.

Best on top of #1459 (needs the v2 dialer + framing fixes), but the harness/tests stand alone. go → py is a real listener-side bug, not something this PR introduces.

Refs #1437

🤖 Generated with Claude Code

[acul71]

🤖 Generated with Claude Code

Hi @yashksaini-coder , you used claude code with which LLM ?

[acul71]

Review readiness

Thanks for the clean interop harness — scope matches the description (pinned go-libp2p v0.49 listen/dial, py→go live, go→py strict xfail for #1470), and the #1459 dependency is already on main.

Not ready to dig into a full review pass yet; a couple of blockers first:

Blockers

1. Lint is red. Recent CI fails typecheck on test_webrtc_direct_interop.py:

   Object of class `NoneType` has no attribute `receive_some`
   

   (proc.stdout is optional after open_process). Please assert/narrow stdout (or otherwise satisfy the checker) and get lint green.

2. Maintainer question still open — which Claude model did you use with Claude Code? (comment above)

Follow-ups (can land with or right after the lint fix)

• Go is not installed in CI. .github/workflows/tox.yml sets up Nim/Node for interop, but not Go. Without actions/setup-go (matching go.mod / go-libp2p v0.49), these tests will skip on Actions and CI won’t actually exercise the harness.
• Newsfragment is 1470.internal.rst (issue id); consider 1471 if that’s the project convention for the PR that adds the tests.
• Commit history is noisy (already-merged framing/v2 work + merges); net diff vs main is fine — squash-on-merge is enough.

Happy to do a deeper review once lint is green (and ideally once Go is wired into the interop job so CI runs the new tests for real).

[yashksaini-coder]

Thanks! Addressed the blockers and follow-ups:

• Lint green — narrowed proc.stdout (it's Optional after open_process); pyrefly is now clean on the interop module.
• Go in CI — added an actions/setup-go@v5 step (go 1.25, matching go.mod) to the interop tox job, right after the Node.js setup, so the harness actually builds and the tests run on Actions instead of skipping.
• Newsfragment renamed 1470.internal.rst → 1471.internal.rst (PR id).
• History — rebased onto current main (v2/framing already landed), so the branch is now a single clean interop commit; net diff is just tests/interop/go_libp2p/ + the CI/newsfragment bits.

Local: 2 passed (py→go v1+v2) / 2 xfailed (go→py v1+v2 → #1470). Ready for a deeper pass whenever you are.

[yashksaini-coder]

🤖 Generated with Claude Code

Hi @yashksaini-coder , you used claude code with which LLM ?

I recently got hands on claude code, mainly for implementing and testing the main feature changes affect on other module features. It also helps me explain for other cases and scenraios, which helped me not to raise any more follow up PRs, after a big feature or PR lands, as I used to do before

[acul71]

Re-review

Previous blockers are addressed:

• Lint: proc.stdout narrowed with assert stdout is not None — lint green
• Go in CI: actions/setup-go@v5 (go-version: '1.25') on the interop tox job; harness builds and tests run (py→go PASS v1/v2, go→py strict XFAIL → #1470)
• Newsfragment renamed to 1471.internal.rst
• CI currently all green; merges cleanly with current main (1 commit behind, no conflicts)

Harness design looks good: pinned go-libp2p v0.49, session fixture + file lock for xdist, skip when Go missing, both directions/versions covered. Strict xfail on go→py is the right call until #1470 lands.

Non-blocking follow-ups

1. _read_lines drops leftover LISTEN lines. In test_py_dials_go, the first _read_lines(..., count=1) can consume a multi-line burst and discard extra LISTEN lines still in the local buffer; the second call (count=3) then sits on the 45s move_on_after timeout. That matches the ~45s py→go call times in CI and can defeat _lan_addr’s non-loopback preference. Prefer one read that collects all LISTEN lines (short settle after the first), or preserve leftover buffer across calls.
2. Branch is 1 commit behind main — optional merge/rebase before squash; not required.
3. Claude model question still unanswered by name — optional.

LGTM — OK to squash-merge.

[acul71]

Follow-up to the review note on _read_lines / LISTEN collection:

Pushed a fix that waits for the first LISTEN line, then briefly settles (~250ms) to drain sibling addresses from the same burst in one buffer — instead of a second count=3 wait that discarded already-read lines and burned ~45s per py→go case.

Locally: test_py_dials_go v1/v2 both PASS in ~0.3–0.4s each (was ~45s call time). Also merged latest main.