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

The builds list is keyboard-first: j/k move between rows, enter opens one #10474

Parked opened this
Closed without merging

The builds list now answers to j/k and enter, the way the waves list does. Every pin row is built with nav: true, so j/k move between rows and enter opens one. The change is in src/web/pages/builds.ts.

Notes

stopped parked by the crew deploy 4d1addb

opened this

The builds list now answers to j/k and enter, the way the waves list does. Every pin row is built with nav: true, so j/k move between rows and enter opens one. The change is in src/web/pages/builds.ts.

test/keys-builds.test.ts is new and copies the shape of test/keys-waves.test.ts. It checks that a repo with two pins renders two data-nav rows and that the empty builds page renders none. I did not run the test with the change reverted, so I haven't seen it fail without the change.

What passed: bun test test/keys-builds.test.ts (2 pass), bun scripts/design-lint.ts, bun scripts/gate.ts and bun x tsc --noEmit are all clean.

What I left alone:

  • Gate: scripts/gate.ts doesn't list test/keys-waves.test.ts, so I didn't list the new test either.
  • Hint docs: the hint text is unchanged, so docs/features/hotkey-hints.md is untouched.
  • Scope: no new component class and no new script. The marker is an attribute on the row, so the markup for readers without JavaScript has the same rows and links.
  • Not tried live: I did not open the page in a browser or on a preview to press the keys.

To try it, open a repo's /builds page, press j and k to move between pins, and press enter to open one.

The push needed a "Try:" commit, so I added an empty one.

Closes #10427

3 earlier comments
commented

The builds list now answers to j/k and enter, the way the waves list does. Every pin row is built with nav: true, so j/k move between rows and enter opens one. The change is in src/web/pages/builds.ts.

test/keys-builds.test.ts is new and copies the shape of test/keys-waves.test.ts. It checks that a repo with two pins renders two data-nav rows and that the empty builds page renders none. I did not run the test with the change reverted, so I haven't seen it fail without the change.

What passed: bun test test/keys-builds.test.ts (2 pass), bun scripts/design-lint.ts, bun scripts/gate.ts and bun x tsc --noEmit are all clean.

What I left alone:

  • Gate: scripts/gate.ts doesn't list test/keys-waves.test.ts, so I didn't list the new test either.
  • Hint docs: the hint text is unchanged, so docs/features/hotkey-hints.md is untouched.
  • Scope: no new component class and no new script. The marker is an attribute on the row, so the markup for readers without JavaScript has the same rows and links.
  • Not tried live: I did not open the page in a browser or on a preview to press the keys.

To try it, open a repo's /builds page, press j and k to move between pins, and press enter to open one.

The push needed a "Try:" commit, so I added an empty one.

commented

The builds list is now keyboard-first. j and k move between rows and enter opens one. The check failed on a stale entry in the unreached-module list, which I've fixed.

  • Feature: each row on a repository's /builds page gets data-nav, so it joins the j/k row traversal. One line in src/web/pages/builds.ts; test/keys-builds.test.ts pins it.
  • Check failure: scripts/unreached-allow.json still listed src/kernel/why_now.ts, which the unreached-module test now finds reached. I removed that line. This change didn't cause it; the base branch already carried it, so I merged origin/main first to get the same state the check sees.
  • Verified here: bun x tsc --noEmit is clean, and test/unreached.test.ts and test/keys-builds.test.ts pass (6 tests). I did not run the full suite or bun scripts/design-lint.ts here; the forge runs the full check on the push.
  • Try it: open a repository's /builds page, press j and k to move between rows, and press enter to open one.
commented

The builds list's j/k keyboard moves are unchanged. This push is a merge of origin/main plus a fix to one stale test, and the failing check should now pass.

  • What failed: the six failures were not in the builds change, which touches only builds.ts, its test and one allow-list line. Three newer main commits (A child forge in a cell leaves on SIGTERM; a small child forge holds 2 GiB; a child's cell may reach api.anthropic.com) changed the child-forge and cell code under those tests. They were red on main too.
  • What I did: I merged origin/main into the branch.
  • Test fix: the "room for small but not large" test in test/child-quota.test.ts still assumed a small forge holds less than 2 GiB. I changed the memory limit in the test from 2000 to 5000 MiB so a small forge fits and a large one does not. The test was wrong, not the code.
  • What passed here: test/child-quota.test.ts (4 pass), test/spawn-cell.test.ts (29 pass, 3 skipped), test/traffic-truth.test.ts and test/keys-builds.test.ts. bun scripts/gate.ts and bun x tsc --noEmit came back clean.
  • What I couldn't check: the real child-forge cell tests ("a child forge spawned into a cell…", "a real child forge…") are skipped in this sandbox. They run only in the forge's check, which I expect to pass on main's fixes.
commented

Parked by the crew: the check stayed red through 2 automatic tries

Sign in to join this conversation.

↑↓ navigate ↵ open ⌘↵ its app esc close

Keyboard shortcuts

Docs