Skip to content

CI: test-webrtc becomes a required core step (#21 closed) - #38

Merged
lann merged 1 commit into
mainfrom
migrate-polymorph-webrtc-datachannel
Aug 10, 2026
Merged

CI: test-webrtc becomes a required core step (#21 closed)#38
lann merged 1 commit into
mainfrom
migrate-polymorph-webrtc-datachannel

Conversation

@lann

@lann lann commented Aug 10, 2026

Copy link
Copy Markdown
Collaborator

With the #21 root causes fixed and runner-validated (#36 latch, #37 test race + teardown settle; evidence in polymorph-webrtc-datachannels#149's probe runs), the webrtc unit suite no longer needs tolerance on runners: gha::core runs test-webrtc as a required step, and the stale runner-networking rationale comments go away. The webkit lane keeps _step-tolerated (its only remaining user).

Closes #21.

The runner-networking theory is disproven and the actual defects are
fixed (#36, #37, root-caused via polymorph-webrtc-datachannels#149's
probe matrix): the unit-test timeouts were a candidate-buffering race in
the tests' own connectPair helper, and the exit-path SIGSEGV was the
addon's thread joins racing isolate teardown after cleanup(). Both fixes
are runner-validated (deno task test exits 0 on ubuntu-latest in the
consumer repo, and their Conformance matrix job runs the same suite
class green). Drop the tolerated wrapper and the stale rationale
comments.

Gate: just test-webrtc.
@lann
lann enabled auto-merge August 10, 2026 02:44
@lann
lann merged commit b396d6a into main Aug 10, 2026
3 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

ports/webrtc loopback tests: no ICE connectivity on GH Actions runners

1 participant