The timeline drew a tick for every registered mark, and the weather clock registered every radar, satellite and lightning frame, which come every few minutes, so the track was a wall of dashes and the arrow keys crept from frame to frame. Weather products now register only the span of their frames (start and end, as before) with no marks. Holds, the gate and the next-frame wait while the world plays are unchanged. A weather row's step param goes to the previous or next weather frame (every shown product's frames, and the row's own while it is suspended) rather than to the timeline's marks. The Earthquakes layer registers with the world clock once, at init, and unregisters on destroy. While on, its quakes of magnitude 6 and up (EARTHQUAKE_MARK_MAGNITUDE) are marks, and its coverage is their own span, or none without any, so it never stretches the timeline beyond the feed. Off, it has none. It tells the clock when the set of marks changes. The map still draws every M2.5+ quake. Launches stay marks. With no mark ahead or behind, the arrow keys step along the track as shown.
Changelog fragments
One file per entry. A pull request adds a file here instead of editing
CHANGELOG.md, and a release concatenates them into it.
changelog.d/<pull request number>-<slug>.md
723-atc-decision-layer.md, for example. The slug is lowercase, words separated
by -. Two pull requests never pick the same filename, because the pull request
number is already unique — which is the point: new CHANGELOG.md entries go at
the top of the file, so two pull requests in flight write into the same few lines
and git has nothing to merge on.
Write the file exactly as the entry should read in CHANGELOG.md: a - bullet,
wrapped the way the surrounding entries are, crediting yourself and the pull
request.
- Report which upstream declined a Street Traffic road load. The layer row now
reads `Overpass rate-limited` instead of a general "Road data temporarily
unavailable" (yourname, #665).
The text is spliced in unchanged apart from trailing whitespace, so what you
write here is what ships. Fragments are deliberately outside
scripts/format-scope.json, so npm run format never reflows one.
At release time:
npm run changelog:check # list what would move, change nothing
npm run changelog # move every fragment into CHANGELOG.md
The assembler refuses a batch rather than guessing: a filename that is not
<pull request>-<slug>.md, or two fragments claiming one pull request number,
stops the run before anything is written and names the files.
If npm run changelog fails part-way (a full disk, a killed process), run it
again. The batch is staged in changelog.d/.assembling/ and only lands when the
assembled file replaces CHANGELOG.md in a single rename, so the rerun either
finishes that batch or puts its fragments back first. It never adds an entry
twice or drops one. npm run changelog:check names a staged batch, if one is
left over, instead of reporting past it.
A documentation claim that needs a regression test — the way
src/annotations/drawTool.test.mjs asserts against CHANGELOG.md — should
assert against docs/CURRENT-STATE.md or README.md instead. A pull request
still edits those directly, so the text is in the file while the change is under
review; a fragment does not reach CHANGELOG.md until release.