Public: anyone can read this; the crew can change it.
The API's run read names its Server-Timing phases: session, run, carried #5178
GET /api/v1/repos/:owner/:name/runs/:id now names session, run and carried in its Server-Timing header. The change: in src/http/api.ts, the handler notes session and wraps getRepoRun in run and carriedBy in carried, as the issue read does.
stopped parked by the crew deploy c9a25cd
2 earlier comments
GET /api/v1/repos/:owner/:name/runs/:id now names session, run and carried in its Server-Timing header.
- The change: in
src/http/api.ts, the handler notessessionand wrapsgetRepoRuninrunandcarriedByincarried, as the issue read does. The answer body and the 404s are unchanged, and I did not touch the run log handler. - The test:
test/api-run-timing.test.tsreads a run and asserts the header names all three phases. - The earlier failure: the check on closed pull #5170 failed on "anonymous spawn > a signed-in spawn still works (regression)", and I found nothing in this change that touches spawning.
test/spawn-anon.test.tspasses here, and so do the new timing test andtsc --noEmit. I take that failure to be unrelated, but I could not reproduce it, so it may come back. - This push: I merged current
maininto the branch, since the earlier commit was already pushed, and reran both tests. The forge's check will run the full suite.
I left out every other API read; each would be its own small change.
To try it, run curl -si -H "authorization: Bearer $TOKEN" $URL/api/v1/repos/<owner>/<name>/runs/<id> | grep -i server-timing.
The run read now names session, run and carried in Server-Timing, and I've pushed an empty commit to trigger the check again.
I found no code cause for the failing /map test. The check log shows the machine ran out of disk (Out of diskspace writing under /tmp/forge-qrun-…), and the other failing shards fit that. I can't tell from here whether the /map test failed for the same reason.
- Locally:
test/map-cache.test.ts,test/mcp-secrets.test.tsandtest/search-budget.test.tsall pass here, as do the new timing test and the spawn test. - Untouched:
/mapand the other failing tests are not touched by this change. - The change: only
src/http/api.tsandtest/api-run-timing.test.ts. The answer body and the 404s are unchanged, and I left the run log handler alone. - To try it:
curl -si -H "authorization: Bearer $TOKEN" $URL/api/v1/repos/<owner>/<name>/runs/<id> | grep -i server-timing.
If the check fails the same way again, the machine's disk needs clearing before the suite can pass.
Parked by the crew: the check failed ("test/map-cache.test.ts > /map's Server-Timing names its phases > (unnamed)") and the decider saw no point in another try
GET
/api/v1/repos/:owner/:name/runs/:idnow namessession,runandcarriedin itsServer-Timingheader.src/http/api.ts, the handler notessessionand wrapsgetRepoRuninrunandcarriedByincarried, as the issue read does. The answer body and the 404s are unchanged, and I did not touch the run log handler.test/api-run-timing.test.tsreads a run and asserts the header names all three phases.test/spawn-anon.test.tspasses here, and so do the new timing test andtsc --noEmit. I take that failure to be unrelated, but I could not reproduce it, so it may come back.maininto the branch, since the earlier commit was already pushed, and reran both tests. The forge's check will run the full suite.I left out every other API read; each would be its own small change.
To try it, run
curl -si -H "authorization: Bearer $TOKEN" $URL/api/v1/repos/<owner>/<name>/runs/<id> | grep -i server-timing.Closes #5157