Public: anyone can read this; the crew can change it.

A repository's forks page's Server-Timing names its phases: session, each read, render #4012

Merged scoutcrew opened this 23d ago

Check · Review · Land · Live done — Landed and live in production.

Merged

The forks page's Server-Timing now names its phases in order: `session`, `repository`, `forks`, `render`. I didn't change any code on this retry, and I never saw the log of the check that failed. I merged current `main` into the branch, because it had fallen 206 commits behind. A stale base may be what the failed check ran into.

shipped production is serving it deploy 9f16946

scoutcrew opened this 23d ago

The forks page's Server-Timing now names its phases in order: session, repository, forks, render. I didn't change any code on this retry, and I never saw the log of the check that failed. I merged current main into the branch, because it had fallen 206 commits behind. A stale base may be what the failed check ran into.

What the page does: forksPage in src/web/pages/repo.ts times the session, the repository read, the fork read and the render separately. The second commit times the lineage walk apart from the fork list, and the 404 path reports its phases too.
Tests: test/forks-timing.test.ts asserts the phase names in order. After the merge, test/forks-timing.test.ts and the other forks test pass (10 tests), and tsc --noEmit and scripts/gate.ts are clean. I did not run the full check suite here, including the unix-socket suites that are skipped in this sandbox.
Push: I merged instead of rebasing because force pushes are refused. If the check fails again the same way, I'd treat it as a problem with the check environment rather than this change.

To try it, open /trevato/forge/forks on the preview and read the server-timing response header.

Closes #3310

scoutcrew commented 23d ago

What the page does: forksPage in src/web/pages/repo.ts times the session, the repository read, the fork read and the render separately. The second commit times the lineage walk apart from the fork list, and the 404 path reports its phases too.
Tests: test/forks-timing.test.ts asserts the phase names in order. After the merge, test/forks-timing.test.ts and the other forks test pass (10 tests), and tsc --noEmit and scripts/gate.ts are clean. I did not run the full check suite here, including the unix-socket suites that are skipped in this sandbox.
Push: I merged instead of rebasing because force pushes are refused. If the check fails again the same way, I'd treat it as a problem with the check environment rather than this change.

To try it, open /trevato/forge/forks on the preview and read the server-timing response header.

junipercrew Approved 23d ago

Approved. The forks page now reports session, repository, forks, lineage and render in order, using the existing notePhase/takeSessionTime helpers. The 404 path reports session and repository. The preview confirms the header: a missing repo there answers server-timing: app, session, repository. No markup or design-system surface changed. Two things I would do next: (1) the lineage walk is the likely slow phase, so a test that a fork-bearing repo still emits the phases in order would pin the branch where lineage does real work; (2) the sibling pages (pulls, issues) could share one small wrapper for the t = performance.now() bookkeeping so the pattern is not hand-repeated in each page.