Fix ten nightly failures the everyday gate never ran

[leaf-agent]

Problem

ci run 33516407327 came back 10 failed, 1614 passed, 6 skipped. Eight belong to #189, whose ci run was cancelled by #190's push before the nightly leg ever ran on it. #188's test_mcp_app.py case arrived on the same collision as the MCP test below and is carried here rather than left to red the next nightly.

The everyday suite that gates a pull request does not run test_render_* or test_site.py, so a PR can land its nightly half unread — which is how these arrived, and how five more have arrived since, from #193, #194 and #200. main has moved a long way under this branch; the ones it now owns are taken on merge, and three of the new ones are cleared here. #203 reached tests/conftest.py's headless_shell first, so the four test_render_commands.py setup errors are its.

Solution

Grouped by cause rather than by test, since several tests share one.

The example the tests still name. #189 rewrote examples/pr-walkthrough.html into the Worktrunk review packet, and the shipped patch is now a collapsed manifest that builds no lines until a reader opens a file, which the scroller gate read as getComputedStyle(null). The gate reads the diffs that have drawn lines, with its own non-vacuity floor kept.

The new package's chrome and surfaces. .lf-pr-description stacked a second translucent accent onto a card that is already a tint of one, and the layer's own --muted reads 4.32:1 there; the description now sits on the raised-surface token. CallDiff's disclosure was a bare <button> rather than an offer, so an exported copy kept a hand over a press nothing can take. Its location anchor carried an href on the header row, which names no location at all — and reachScrollers reads a candidate for a focusable descendant before granting it a stop, so that hidden anchor answered "there is already a way in here" with a link nobody can reach, for a box whose words run off the side.

A reading that counted clipped text as painted. coveredWords measured each run's whole rect, so an ellipsised name in a narrow column read as covering its neighbour. It now intersects each run with the boxes that clip it, stopping at an out-of-flow ancestor where a hidden overflow further out need not reach. The one test that plants this fault plants it properly now: the runtime's note is parked in a one-pixel box with hidden overflow, so opacity alone never put its characters on the screen.

Three focus rings the corpus stopped painting. code-pre-light and code-pre-shadow are the tab stops on a scrolling code block and on a diff's lines, and with the old example gone no example held either. The page gets one rendered hunk and one code line long enough to scroll. Then .lf-diff-review: it took the outward ring it shared with .lf-diff-next, which stands on its own in the toolbar, while this one rides a file's summary row inside a box that clips at the row's bottom edge — its lower run fell 3px past what the sweep can see. It takes the inset the summary beside it already takes, and splitting the rule made each half declare the ring name neither had.

--diffs-fg-number on changed rows. Surfaced by that rendered hunk: a changed row's number is drawn in the base its own fill is mixed from, green on green at 4.03:1. Pierre's number overrides now take the layer's tinted-surface inks.

A sidebar the page got shorter under. #190's zero-height ToC anchor shortened the release page 58px, putting a named scrollTo(0, 900) past the stretch where the box stands on its own offset. Both edges of that stretch are the page's, so they are read off it, with a floor asserting the stretch exists before a point halfway along it says anything.

Playwright's loop against asyncio.run and anyio.run. sync_playwright() keeps an asyncio loop running in its thread for its whole lifetime, and the browser fixture is session-scoped per xdist worker — so an MCP test's own loop start raises in any worker that has opened a browser and passes in one that has not, leaving the schedule to decide. Verified directly: asyncio.events._get_running_loop() is None before sync_playwright(), the loop inside it, None after. interact_support.run_async runs the entry point on a thread with no loop on it, and both MCP modules go through it — #204's three new exchanges included, since they landed on the direct call while this was in flight.

A declared verb no page replayed. #194 added lf-diff's review to the registry without an event for it on the standing-state page — the one fixture built so that exactly this fails rather than going unexercised. It gets a diff and a standing review.

A focus contract two site tests never heard about. #193 kept the response field passive so a drag leaves the browser's own selection alone and a native copy still has something to take, and aligned the render tests with it. test_site.py is nightly-only. Its label test turned on document.activeElement === field, which is now false for the page's own words and the site's label alike — the discriminator was gone, not inverted — so it reads whether the field was offered at all, which is the difference it was always after.

Testing

uv run pytest tests --run-nightly and pre-commit run --all-files on this head. The everyday suite and lint are green on this branch in CI. Every claim about what main does was taken from a control run of the same tests in a clean checkout of main, not from reading.

Where each failure landed

test	fix
test_render_gate.py::test_the_runtime_holds_a_scroller_the_page_wrote	reads the diffs that drew lines
test_render_export.py::test_an_exported_example_stands_on_its_own[pr-walkthrough], [corpus]	widget, probe, theme
test_render_controls.py::test_every_ring_the_layer_draws_is_shown_whole_somewhere_in_the_corpus	example, then .lf-diff-review's ring and both ring names
test_render_pages.py::test_a_left_sidebar_uses_the_margin_until_the_page_needs_it_back	scroll position read off the page
test_interact_mcp.py::test_stdio_protocol_carries_the_app_resource_and_private_tool_result	own thread
test_mcp_app.py::test_registered_server_prefers_full_page_and_keeps_snapshot_as_fallback	own thread
test_render_projection.py::test_the_render_gate_applies_every_standing_action_a_second_time	standing fixture (#194's)
test_site.py::test_the_label_is_chrome_rather_than_words_to_quote, ::test_a_comment_lands_in_the_thread_with_its_quote	the field is offered, not entered (#193's)
test_render_anchors.py::test_a_data_bound_diff_aims_and_selects_one_source_line	#194's and #193's, taken on merge
test_site.py, test_render_export.py stale <h1>	#202's, taken on merge

The two exported-copy cases each carried four faults, three of them masked: offering asserts before covered, which asserts before the copy's axe run, so each fix uncovered the next. The ring sweep did the same — the geometry fault stood in front of a naming check that had been silent about a rule missing its name since it was written.

Three this leaves, all of them main's

test_render_controls.py::test_examples_have_no_serious_wcag_a_or_aa_violations[pr-walkthrough] stays red. The .lf-pr-description contrast above is fixed and confirmed gone. What holds the test red is 46 nested-interactive findings from #194: each file's "Mark reviewed" <button> sits inside that file's <summary>, and a disclosure with a focusable descendant is a serious WCAG failure. Clearing it means taking the press out of the disclosure and re-laying the row — entry.node is the <details> in ten places, and the theme, print and export rules key on it — which is a change to that widget's shape with visual verification of its own, not a line this branch should slip in.

test_render_anchors.py::test_a_failed_fragment_hydration_waits_for_a_reader_retry fails whenever anything runs before it. Confirmed on unmodified main: three runs of the file at -n4, three failures, and the same failure at -n0 over the whole file. It passes alone every time. assert len(requests) == 2 sees one — the reopened disclosure's second fragment fetch never reaches the route. Arrived with #189.

test_render_anchors.py::test_a_drag_released_mid_word_hugs_words_and_sentences fails over the whole file and passes alone. assert 'inside' == 'inside it' — the selection did not grow to the word boundary. It looks like the next instance of what #191 diagnosed, Chromium 151 collapsing a synthetic range on mouse release.

Also still behind: the gallery's docs/example-pr-walkthrough.jpg. Regenerating here rewrites all nine stills in substituted faces, because theme.css asks for Charter and this runner's fc-match has none of the serif stack — the same reading #175 made. Left for the authoring machine.

---

Automated fix for failed run

This was written by Claude Code on behalf of @max-sixty

[leaf-agent]

Read the merged tree, regenerated examples/corpus.html with scripts/corpus.py (no drift), and traced each of the ten fixes back to the cause the description names. The Playwright-loop diagnosis, the --diffs-fg-number-*-override names against the Pierre bundle, the .lf-call-toggle.lf-ui cascade race against chrome-style.js's .lf-ui { font-size: var(--t-5) }, and the new hunk's @@ -112,4 +112,3 @@ line counts all check out. Two things, both in the widget and probe halves.

The first is a removed-intent gap rather than a regression: toolbar.dataset.lfUi = "" was inert — data-lf-ui is written twice in this tree and read nowhere — so dropping it is right, but nothing re-establishes what it was reaching for, and .lf-call-group-count (outside this diff, count.dataset.lfUi = "" in the same file) is the same inert write left standing.

Not raised, for the record so a later pass doesn't re-derive it: a changed root with no calls beneath it now has no route into the patch at all, since the link moved to the children. That is the bounded edge case the repo's threshold says to leave.

[leaf-agent]

One finding, and it is in the half this PR's own earlier review sent it to — that review is where the mistake came from, so this corrects it rather than restating it.

data-lf-gen does not keep the counts out of a passage. passages.js asks two questions with two lists: quotable() — the reading a drag captures — is uiInside, which tests .lf-ui, [data-lf-said] and never looks at data-lf-gen; GENERATED = ".lf-ui, [data-lf-gen]" is read only by authored(), the version-diff reading. leaf.js's header states the pair outright — "data-lf-gen alone keeps it out of the diff and in reach of the anchor pass" — and passages.js says it again a few lines above GENERATED: "generated text the page authored … is diff-invisible and quotable, which is the pair a user expects". So a reader dragging across 1 changed root · 1 added · 0 removed · 2 items still gets a quotable passage. What the attribute did change is dynamicWords: the run is now its own passage cell, so the quote can no longer run out of it into the page's prose.

The marker is still the right one — it is what lf-diff puts on its +N −M stat, and these counts are the same kind of word. lf-diff also shows what the other reading costs: the line numbers a comment must not quote carry .lf-ui and data-lf-gen, and .lf-call-toggle.lf-ui in this same diff is already the answer to the font-size objection the comment raises against .lf-ui. So keep the attribute; the two places that say what it does are what need correcting, and the second is a test's stated purpose — as written the assertion reads as guarding something it does not guard.

Nothing else. The Playwright-loop diagnosis and the shared run_async thread, the --diffs-fg-number-*-override names against the Pierre bundle, the @@ -112,4 +112,3 @@ counts on the new hunk, the painted climb and its stated over-report direction, and the sidebar test's read-off-the-page scroll all check out; scripts/corpus.py regenerates examples/corpus.html with no drift.

[leaf-agent]

ci run 33537688138 on 2326a8d is the first nightly leg to run with #188 and #189 both on main — #188's own ci was cancelled by #191's push before its nightly leg started, so d55e574 never had a nightly result. It came back 11 failed, and this branch already covers ten of them, including test_mcp_app.py::test_registered_server_prefers_full_page_and_keeps_snapshot_as_fallback, which reaches anyio.run through the same interact_support.run_async helper.

The eleventh, test_render_aim.py::test_a_declared_box_takes_its_comment_on_every_type_that_carries_an_id, is not this branch's: it aims at the next diagram box while the comment it just posted is still arriving, and the arrival closes the composer under the gesture. That one is #196, which touches no file this branch does.

So main needs both, in either order.

[leaf-agent]

Merged main (8aa71b7) into this branch to clear the conflict. Two files collided, and one of them needed a call rather than a mechanical resolution.

tests/test_render_anchors.py — both sides had independently made the same fix to the FAB flow; I kept main's form (expect(page.locator(".lf-fab-input")).to_be_focused()), since reactions.js already focuses that input when the bar appears, so this branch's extra click was redundant.

lf-call-diff.js — main's #194 rewrote travelToLine onto navigateToDatum and added lfPrepareExport, while this branch had switched the toggle to offer(), moved the injected counts to data-lf-gen, and dropped the href from two kinds of row. I kept all of that except the <summary> exclusion: #194 landed a test that clicks the group root's own location link and asserts it opens gateway/limits.py:38 in the patch, including after a data refresh, so removing the link from summary rows now contradicts a tested behaviour on main. The header-row guard survives as record.location &&, which is the dead-anchor case this branch was actually fixing, and standalone.js's painted() fix — the clipped-row misreading — is untouched.

Verification

• scripts/corpus.py regenerates byte-identically after the merge, so no derived corpus output went stale.
• tests/test_render_projection.py tests/test_render_export.py tests/test_render_gate.py plus test_a_data_bound_diff_aims_and_selects_one_source_line: 172 passed, 1 failed.
• The one failure is test_the_render_gate_applies_every_standing_action_a_second_time, and it is not from this merge — it reproduces on a clean 8aa71b7 worktree. Scale PR review data and navigation #194 declared lf-diff/review in the registry, but no example seeds a standing review action, so the gate's declared-verb set no longer matches what the corpus re-applies. tests/test_render_projection.py carries pytestmark = pytest.mark.nightly, so this PR's green test check never ran it — ci only runs it on pushes to main, where the run for f42eb45 should surface it. Tracked separately from this PR.

[leaf-agent]

One finding, and it is about which PR owns a hunk rather than about the code inside it.

tests/conftest.py's headless_shell fix is also #203, and #203 got there first — its commit is 47edb74 at 20:44Z, this branch's 77205a3 at 21:38Z — with the same edit to the same line: chromium.parents[2] / parents[1].name replaced by a search up chromium.parents for the chromium-<build> directory. Whichever lands first conflicts the other, and meanwhile the same change is being reviewed twice. #203 is the narrower home for it: it exists only for #200's follow-ups, it states the macOS layout (chromium-<build>/chrome-mac/Chromium.app/Contents/MacOS/Chromium) as its cause, and it is not carrying thirteen other fixes a maintainer may want to land on a different schedule. Dropping the hunk here leaves this branch's four test_render_commands.py setup errors to #203 — which is where the PR body already attributes them.

If this branch keeps it instead, the half to keep from #203 is rsplit("-", 1). startswith("chromium-") also matches chromium-tip-of-tree-<build>, where split("-", 1)[1] yields tip-of-tree-<build>, chromium_headless_shell-tip-of-tree-<build> never exists, and the candidate loop falls back to the full Chromium binary with nothing said. The explicit AssertionError on this side is the better half of the pair — #203's bare next() raises StopIteration with nothing to read — so neither version is the one to take whole.

Nothing else on the six commits since the last review. Checked: dropping .lf-call-toggle's print rule is covered, since offer() writes data-lf-offer and chrome-style.js's @media print { [data-lf-offer]:not([data-lf-said]) } is the shared rule, while data-lf-said is written only by relabel, which this widget never calls. lf-diff's x-state.review detail schema takes exactly {file, reviewed}, the new ab-patch hunk's @@ -1,2 +1,2 @@ counts match its lines, and the standing floor is set equality, so the verb is now covered rather than merely present. .lf-diff-review's new inset is the same calc(-1 * var(--here-ring-w)) the code-summary rule beside it takes, and both halves of the split rule now name a ring. In test_site.py, a drag over the page's own words reaches openComment(anchor, "", { focus: false }), which opens the composer and so sets fabInput.style.display = "block" — so checkVisibility() is a live discriminator where document.activeElement stopped being one. And the sidebar guard reads both edges off the page before the halfway point is taken.