• Home
  • Features
  • Pricing
  • Docs
  • Announcements
  • Sign In

floatdrop / moq-go / 31674203330
89%

Build:
DEFAULT BRANCH: draft-19
Ran 13 Aug 2026 06:33AM UTC
Jobs 1
Files 100
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst

13 Aug 2026 06:32AM UTC coverage: 87.07% (-0.1%) from 87.187%
31674203330

push

github

web-flow
test: fix two flaky tests at their preconditions (#72)

Both failed CI on draft-19 while passing locally, in different jobs and for
different reasons.

* TestFetch_UnknownMarkerWhenUpstreamCapsEndLocation. unknownGapTopology's
own comment says it populates the relay's cache with the live tail, but it
returned as soon as Subscribe did, while the upstream writes those groups
afterwards from its own goroutine. A fetch landing early is not merely
incomplete — the unknown-range floor sits wherever the cache happened to
reach, so the relay gives a legitimately different answer, which the retry
loop papered over. The helper now waits until every group has been forwarded
to the live subscriber, which cannot happen until the relay has ingested it.

I could not reproduce this one: 90 runs under -race with forced contention,
40 at -cpu=1, and six whole-package runs all passed on a 14-core machine
against a 4-vCPU runner. I also disproved the obvious alternative — that
out-of-order arrival sets the floor wrong — by writing group 6 before group
5, which the test survives. A real defect in the helper, fixed at the root,
but not a proven fix for that CI failure.

* TestReproRelayMutualConference had a bug independent of timing.
awaitPeerAndFetchCatalog reports failures with t.Fatalf and the mutual test
runs it inside two goroutines, where t.Fatalf calls runtime.Goexit on the
wrong goroutine — so done never receives and the real error is replaced by
the outer "mutual catalog fetch did not complete within 8s". Every failure of
this test has been reported as a timeout regardless of cause, which is why
the CI log said so little. The helper now returns its error and the
goroutines carry it out; pointing one side at a peer that never publishes
fails in 1s naming the announce, where before it took 8s and named the wrong
thing. Budgets also move from 3s per step to a shared 15s, since loopback
QUIC on a busy runner does real handshakes across competing sessions.

Co-authore... (continued)

8902 of 10224 relevant lines covered (87.07%)

1.99 hits per line

Coverage Regressions

Lines Coverage ∆ File
3
86.15
-0.61% pkg/relay/handler_subscribe.go
2
81.48
-3.7% pkg/moqt/session/control.go
2
86.31
-0.76% pkg/moqt/session/request.go
2
57.14
-28.57% pkg/relay/discovery/discovery.go
2
60.53
-5.26% pkg/relay/relay_namespace_watch.go
1
92.91
-0.71% pkg/moqt/session/datastream_out.go
Jobs
ID Job ID Ran Files Coverage
1 31674203330.1 13 Aug 2026 06:33AM UTC 100
87.07
GitHub Action Run
Source Files on build 31674203330
  • Tree
  • List 100
  • Changed 6
  • Source Changed 0
  • Coverage Changed 6
Coverage ∆ File Lines Relevant Covered Missed Hits/Line
  • Back to Repo
  • Github Actions Build #31674203330
  • ba99a599 on github
  • Prev Build on draft-19 (#31673100289)
  • Next Build on draft-19 (#31676741265)
  • Delete
STATUS · Troubleshooting · Open an Issue · Sales · Support · CAREERS · ENTERPRISE · START FREE TRIAL · SCHEDULE DEMO
ANNOUNCEMENTS · TWITTER · TOS & SLA · Supported CI Services · What's a CI service? · Automated Testing

© 2026 Coveralls, Inc