Repository navigation
scroll: coast-frame-pacing - #889
Conversation
The coast stepped after the children were laid out, so each frame drew where the clock was one gap ago and uneven gaps showed as uneven steps. It now steps in init, before layout. Agent-Team: linux Worker: claude Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
ce95595 to
79a595e
Compare
PM note: how this PR will be handledWhat it is for: scroll: coast-frame-pacing Rules that apply: none special. Things to watch (checked by rule):
What happens next:
A living status comment below is kept current. Nothing here is a merge decision. |
Where this PR standsStage: Done: approved What the PM needs before approving (ticks as it is met; the full CI runs only at the end, after the reviews are clean):
Reviews at the current commit (
2 earlier reviews are for older commits and will be redone. Latest fixes comment: https://github.com/fizzyedit/work/issues/419#issuecomment-6438916902 (2 findings answered) Updated 2026-10-27 11:45 UTC by the PM. This comment is edited in place. |
Agent-Team: linux Worker: codex
Agent-Team: linux Worker: codex
Part of fizzyedit/work#435
Goal
Keep touch coast drawing at the current frame's timestamp, and answer windows/codex F1 at 3314dc7 (Part of fizzyedit/work#435).
What changed and why
Removed unconditional stderr measurement output from the coast displacement regression. Its motion assertions remain intact. This is test-output cleanup, not a demonstrated repair of the reported Windows unit failure.
The reported 152/181 Windows unit result cannot be diagnosed from a step count alone. Linux unit tests pass 181/181 on both the supplied head and this edit. The new ScrollArea regression is imported by core/sizing_tests.zig, whose run artifact belongs to test-integration in build/app.zig, not the unit gate. FIXES.md declines attributing that unidentified failure to this test.
How it was verified
Linux headless coast displacement regression passes 94/94 build steps, 14/14 tests. Formatting, lint, native build and unit tests pass; full results and sandbox socket failures are in GATES.md. No device presentation or Windows execution was available.
Temporary measurement instrumentation captured the following headless per-frame gaps and child displacement (viewport points at scale 1). Before means coast restored to its pre-fix position in deinit after release handling; after means coast advanced before child layout. The instrumentation was removed after capture.
The regression fails with the pre-fix placement (92/94 steps, 13/14 tests) and passes with the fix (94/94 steps, 14/14 tests). This retains the per-frame evidence requested by the earlier F1 at 79a595e.
Not done / follow-ups
The Windows unit failure remains unverified and needs the first failing command and full diagnostic from a Windows rerun. The harness commits, pushes, and posts the single FIXES comment; this worker has no network or git write access. Maintainer device/web presentation validation remains outstanding.
Changes (63 files, +557/-1681)
Gates: 6 of 6 passed
scripts/check-fmt.sh(exit 0)scripts/zb build --summary all(exit 0)scripts/zb build test --summary all(exit 0)scripts/zb build test-integration --summary all(exit 0)scripts/zb build check-web --summary all(exit 0)scripts/zb build test-sdk-version --summary all(exit 0)