Public: anyone can read this; the crew can change it.
The roadmap list is keyboard-first: j/k move between rows, enter opens one #10480
The roadmap page now joins the shell's j/k row traversal. test/keys-roadmap.test.ts and bun scripts/design-lint.ts pass and bun x tsc --noEmit is clean. I did not run bun scripts/gate.ts for its verdict; I only ran it as part of the commit command and never checked its output.
stopped parked by the crew deploy d0059de
3 earlier comments
The roadmap page now joins the shell's j/k row traversal. test/keys-roadmap.test.ts and bun scripts/design-lint.ts pass and bun x tsc --noEmit is clean. I did not run bun scripts/gate.ts for its verdict; I only ran it as part of the commit command and never checked its output. The forge's check on this push has not come back yet.
What changed
- In
src/web/pages/roadmap.ts, the "Needs you" rows and the per-issue rows inside each pillar's fold now carrynav: true. Both are single-link rows, so enter opens them. - The pillar rows themselves get no
nav. Each one is a<summary>that folds and unfolds the pillar and has no link, so enter would do nothing there. This follows the issue's risk note. test/keys-roadmap.test.tscallsroadmapBodywith hand-built roadmap data rather than a seeded app. That is a difference fromtest/keys-flags-list.test.ts.
- One test checks that exactly four rows carry data-nav (one "Needs you" row and three issue rows), that a row with a pull links to that pull, and that no <summary> has data-nav.
- The other checks that an empty roadmap has no data-nav at all.
- I did not run the test against the old code.
Not done
- Nothing else was changed. The "Draw the graph" and "The pillar's issue" links inside a fold are plain links and stay outside the traversal.
How to try it
- Open
/<owner>/<repo>/map, pressjandkto move between rows, and press enter to open one. - The pre-push hook refused the push until a commit said how to try the change, so I added an empty "Try:" commit.
🤖 Generated with Claude Code
The roadmap keyboard change itself is unchanged: j/k move between rows and enter opens one. Both failures were in tests my change doesn't touch, and I fixed them in this push. I ran only the three affected test files, and they passed after the fix. I did not run the full suite.
The two failing tests were already stale against origin/main, which I merged into the branch before pushing:
child quota > room for small but not large is refused only for large: a recent main commit made a small forge hold 2 GiB, but the test set the host memory limit to 2000 MiB, so even a small forge was refused. I raised the limit intest/child-quota.test.tsto 5000 MiB. That leaves room for a small forge on top of what's already held, and still refuses a large one. The test was wrong, not the code.an allowed module that has become reached leaves the list:scripts/unreached-allow.jsonstill namedsrc/kernel/why_now.ts, a file that no longer exists, because an earlier main commit retired it. I removed the entry.
I did not add anything beyond those two fixes.
To try the roadmap: open /<owner>/<repo>/map on the preview and press j and k to move between rows, enter to open one.
The roadmap list now joins the shell's j/k row traversal, so j and k move between rows and enter opens one. The check on the last push failed on two tests that this change can't have caused. I couldn't fix them, so I pushed an empty commit to run the check again.
- Roadmap change: the one-line addition in
src/web/pages/roadmap.tsis unchanged.test/keys-roadmap.test.tscovers it, and it passed in the check (shard 75 exited 0). - The failures: "a child forge spawned into a cell serves its app at <app>--<child>--<parent>" and "a real child forge > boots in its own forge:<name> cell…" both boot a child forge inside a bwrap cell. The check log doesn't print their failure output, so I don't know why they failed.
- Why I couldn't reproduce: both are skipped in my sandbox because
FORGE_TEST_SOCKET=portis set, so I couldn't run them. Both failed after about 14.5 s. - My guess: the check host's free disk was at 16%, and the cells ask for 2 GiB. Main's recent cell changes (2 GiB small cells, SIGTERM then a 15 s grace on stop) may also be involved. Both are guesses.
- What I changed on the branch: the only edits outside the roadmap are from the earlier merge, two repairs to tests main left red (
test/child-quota.test.tsand an entry inscripts/unreached-allow.json). Nothing here touches the cell code.
What I chose not to do: edit the cell code or its tests without seeing why they failed. If they fail again, the log needs to show the helper's output (test/helpers/real-child-forge.ts and real-spawn-e2e.ts), and the fix belongs in a separate pull on the cell code.
To try it: open /roadmap, press j and k to move between rows, and press enter to open the focused one.
🤖 Generated with Claude Code
Parked by the crew: the check stayed red through 2 automatic tries
The roadmap page now joins the shell's j/k row traversal.
test/keys-roadmap.test.tsandbun scripts/design-lint.tspass andbun x tsc --noEmitis clean. I did not runbun scripts/gate.tsfor its verdict; I only ran it as part of the commit command and never checked its output. The forge's check on this push has not come back yet.What changed
src/web/pages/roadmap.ts, the "Needs you" rows and the per-issue rows inside each pillar's fold now carrynav: true. Both are single-link rows, so enter opens them.nav. Each one is a<summary>that folds and unfolds the pillar and has no link, so enter would do nothing there. This follows the issue's risk note.test/keys-roadmap.test.tscallsroadmapBodywith hand-built roadmap data rather than a seeded app. That is a difference fromtest/keys-flags-list.test.ts.- One test checks that exactly four rows carry
data-nav(one "Needs you" row and three issue rows), that a row with a pull links to that pull, and that no<summary>hasdata-nav. - The other checks that an empty roadmap has nodata-navat all. - I did not run the test against the old code.Not done
How to try it
/<owner>/<repo>/map, pressjandkto move between rows, and press enter to open one.🤖 Generated with Claude Code
Closes #10406