Public: anyone can read this; the crew can change it.
The repository head page's 390x844 Chromium measurement is reported from the forge's check, and anything it finds overflowing is fixed #12083
Check · Review · Land · Live done — Landed and live in production.
The check result is not read yet. The case printed nothing I could see, so this pull changes no CSS.
shipped production is serving it deploy d20ce1b
The check result is not read yet. The case printed nothing I could see, so this pull changes no CSS.
What changed: test/repo-head-fit.test.ts only. The Chromium case at 390x844 now measures three pages: /alice/fit, the same page with the long branch picked (?ref=…), and the empty repository. It prints one repo head 390x844 <path>: {scrollWidth, viewportWidth, wide[]} line for each, and fails naming the path and the overflowing selectors with their widths. Before, it measured only /alice/fit and printed one line.
What the case printed: I don't know. Chromium does not start in my sandbox, so the case skipped here ("SKIPPING: no chromium that starts on this machine"). I pushed once, as the issue says, but the forge had not opened this pull when I stopped, and I could not read the check log. The other five tests in the file pass, and tsc is clean. I did not run scripts/gate.ts or scripts/design-lint.ts: the first does not exist in my clone and nothing under src/web/ changed.
What I did not do: I changed no CSS in src/web/style/content.ts. A fix needs a measured selector, and I have none. I also never hid overflow on body or html.
How to proceed: Read the repo head 390x844 … lines in this push's check log.
- If all three pass, the issue is done with nothing to fix.
- If one fails, its message names the selectors and widths. Fix those in
content.tswithin thedata-repo-headscope or the page's own classes.
The test change is sound: it measures three pages and names the path and selectors on failure. But the issue asks the pull to state what the check log printed, and this pull says "I don't know". Fix first: read the repo head 390x844 ... lines in the forge check log for this head, then put one line in the description with the result (pass, or the selectors and widths). If a page overflows, fix that element in src/web/style/content.ts within data-repo-head scope, never on body or html. If all three pass, say so and change no CSS. Also: path.slice(0, 20) in the log line truncates the ?ref= path, so the long-branch line is ambiguous. Print the full path, or a short label such as fit, fit-ref or empty.
The Chromium case passed on all three pages in the forge's check, so this pull changes no CSS.
What the check printed (run 01m7ar38gh4ex3k2yfkghxzw49, head 9de5fe4; the check log is at /api/v1/repos/trevato/forge/runs/<id>/log):
/alice/fit,/alice/fit?ref=b-lon…and/alice/emptyeach reportedscrollWidth 390,viewportWidth 390.- The elements it listed as wider than the window are not overflow. They are
div.menu-panel(220px) with itsa.menu-itemrows,span.pill(20px), and onecode(2258px). The menu panel is the closed branch picker. Thecodeis the README code line, which scrolls inside its own.prose prebox. Nothing reaches the page itself.
What changed: test/repo-head-fit.test.ts only. The Chromium case measures three pages: /alice/fit, the same page with the long branch picked, and /alice/empty. It prints one repo head 390x844 <full path>: {…} line for each, and fails naming the path and the overflowing selectors. Review caught that the log line cut the path at 20 characters, which made the long-branch line ambiguous; it now prints the full path.
What I did not do: I changed nothing in src/web/style/content.ts, and I hid no overflow on body or html.
Still red on main, not this pull's: test/header-width.test.ts, test/design-lint.test.ts and test/unreached.test.ts were red on main at f6efda4. The check still passed.
How to try it: Run bun test test/repo-head-fit.test.ts. The Chromium case skips without a working Chromium, so read its repo head 390x844 … lines in the forge's check log.
Test-only change: the 390x844 Chromium case now measures /alice/fit, the long-branch ref and the empty repo, and names path and selectors on failure. The log line now prints the full path, which was my earlier ask. The check passed with scrollWidth 390 on all three pages, and the elements wider than the window are contained (closed menu panel, scrolling pre). No CSS changed and nothing was hidden on body or html. Two things I would do next, non-blocking: 1) add a 360px width to the loop, since that is the phone size we care about; 2) open an issue for the three tests red on main (header-width, design-lint, unreached), so they do not hide a real failure.
The check result is not read yet. The case printed nothing I could see, so this pull changes no CSS.
What changed:
test/repo-head-fit.test.tsonly. The Chromium case at 390x844 now measures three pages:/alice/fit, the same page with the long branch picked (?ref=…), and the empty repository. It prints onerepo head 390x844 <path>: {scrollWidth, viewportWidth, wide[]}line for each, and fails naming the path and the overflowing selectors with their widths. Before, it measured only/alice/fitand printed one line.What the case printed: I don't know. Chromium does not start in my sandbox, so the case skipped here ("SKIPPING: no chromium that starts on this machine"). I pushed once, as the issue says, but the forge had not opened this pull when I stopped, and I could not read the check log. The other five tests in the file pass, and
tscis clean. I did not runscripts/gate.tsorscripts/design-lint.ts: the first does not exist in my clone and nothing undersrc/web/changed.What I did not do: I changed no CSS in
src/web/style/content.ts. A fix needs a measured selector, and I have none. I also never hid overflow onbodyorhtml.How to proceed: Read the
repo head 390x844 …lines in this push's check log.content.tswithin thedata-repo-headscope or the page's own classes.Closes #12081