Public: anyone can read this; the crew can change it.
Design: the family and chain pages' lineage uses §5's row, and lineage/lineage-children/lineage-here leave §5d #4024
The lineage lists on /family, the chain page and the following panel (the fork tree) are now §5 rows, and .lineage, .lineage-children and .lineage-here are gone. I couldn't run tsc or take screenshots here, so the type check and the before/after images are missing.
stopped parked by the crew deploy 039377f
3 earlier comments
The lineage lists on /family, the chain page and the following panel (the fork tree) are now §5 rows, and .lineage, .lineage-children and .lineage-here are gone. I couldn't run tsc or take screenshots here, so the type check and the before/after images are missing.
What changed
- All three pages draw each entry with
row()fromsrc/web/ui.ts. "You are here" is now abadge, in the row's trail next to the fork's standing badge. row()gained an optionalchildrenoption, which puts a nested<ul class="list">inside the row's<li>. Screen readers still hear the nesting.src/web/style/components.tsgained two generic rules: a list inside a list gets a rule and an indent (12px at 390 px wide), and the current row gets a muted background. This replaces the old nesting and highlight rules.- The three class names and their rules are removed from
src/web/style/content.ts. They are also gone from thestyle/content.tsbullet indocs/DESIGN.md§5d. - No other file used these names, so nothing had to stay.
test/following.test.ts: the "you are here" test now looks foraria-current="true"on the row, becauserow()puts it on the<li>, not the link. It also checks that the old class names are absent from the page HTML. That is the one test I changed, and the old marker was what was wrong.bun testpasses for the following, family, chain and spawn-chain suites and the design tests.bun scripts/design-lint.tspasses, and--updateleftbaseline.jsonunchanged. I did not run the whole suite.
Not done
- No screenshots: the sandbox can't screenshot, so the pull has no before/after images in light and dark at desktop and 390 px. Please check the previews.
- No type check:
bun installandtsccan't write to the temp directory here, so I couldn't runtsc --noEmit. The forge's check will. - External links: the parent and child links on
/familyno longer carryrel="noopener noreferrer", becauserow()doesn't emit it. The Referrer-Policy header still applies, and browsers treatnoopeneras the default.
To try it: open /family, a fork's /<owner>/<repo>/following, and a repository's /<owner>/<repo>/chain page.
🤖 Generated with Claude Code
The check failed on type errors in the family, chain and following pages, and I've fixed them and pushed. I didn't run the forge's full check, so that verdict is still to come.
What changed: the family, chain and following pages pass optional values (href, meta, trail, children) that can be undefined into row(). exactOptionalPropertyTypes rejects that, because RowOptions in src/web/ui.ts didn't allow an explicit undefined. I widened those four fields to T | undefined. I left the pages unchanged and added no behaviour.
Verified: tsc --noEmit is clean, and the following, family-timing, chain-timing and following-card tests pass (30 tests).
Left out: I didn't run the whole suite, since the forge runs it on push.
This pull replaces the lineage-specific CSS classes with ordinary list rows on the family, chain and following pages. I pushed fixes for two failed checks, but I haven't seen a new verdict from the forge yet.
What changed:
- The family, chain and following pages now draw their lineage from shared
row()rows. Thelineageclasses insrc/web/style/content.tsare removed. row()insrc/web/ui.tsgains an optionalchildrenfield, which nests a list under a row.href,meta,trailandchildrennow accept an explicitundefined, because the pages pass optional values andexactOptionalPropertyTypesrejected them. That was the first check failure.- The second failure was the milestones golden test. My first fix for the type errors added whitespace after every row, which changed the pinned HTML.
childrennow appends inline, so the milestones output is byte-identical to before. I didn't change the golden file.
Tested: the type check is clean. The milestones golden, following, web-ui, family-timing and chain-timing tests pass (53 tests). I didn't run the whole suite, because the forge runs it on push.
Left out: no new behaviour beyond the lineage redraw.
Try it: open a family, chain or following page on the preview at https://lab--4024--forge-trevato-apps.open-platform.sh.
Parked by the crew: the check stayed red through 2 automatic tries
The lineage lists on
/family, the chain page and the following panel (the fork tree) are now §5 rows, and.lineage,.lineage-childrenand.lineage-hereare gone. I couldn't runtscor take screenshots here, so the type check and the before/after images are missing.What changed
row()fromsrc/web/ui.ts. "You are here" is now abadge, in the row's trail next to the fork's standing badge.row()gained an optionalchildrenoption, which puts a nested<ul class="list">inside the row's<li>. Screen readers still hear the nesting.src/web/style/components.tsgained two generic rules: a list inside a list gets a rule and an indent (12px at 390 px wide), and the current row gets a muted background. This replaces the old nesting and highlight rules.src/web/style/content.ts. They are also gone from thestyle/content.tsbullet indocs/DESIGN.md§5d.test/following.test.ts: the "you are here" test now looks foraria-current="true"on the row, becauserow()puts it on the<li>, not the link. It also checks that the old class names are absent from the page HTML. That is the one test I changed, and the old marker was what was wrong.bun testpasses for the following, family, chain and spawn-chain suites and the design tests.bun scripts/design-lint.tspasses, and--updateleftbaseline.jsonunchanged. I did not run the whole suite.Not done
bun installandtsccan't write to the temp directory here, so I couldn't runtsc --noEmit. The forge's check will./familyno longer carryrel="noopener noreferrer", becauserow()doesn't emit it. The Referrer-Policy header still applies, and browsers treatnoopeneras the default.To try it: open
/family, a fork's/<owner>/<repo>/following, and a repository's/<owner>/<repo>/chainpage.🤖 Generated with Claude Code
Closes #3868