formatAge and formatAgeLabel repeated the validity guard, the <1h case,
the rounding and the 48-hour switch, which is why the rounding fix had to
land twice. Both now read { unit, count } from a private ageParts, so
they cannot pick different units again. Output is unchanged.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The disk-cache test asserts the cache directory is 0o700 and its files
0o600. Windows has no POSIX mode bits (the directory stats as 0o666), so
the test always failed there. Both checks now run only off Windows, as
localServices.test.mjs and mcpPanelKey.test.mjs already do for their
0o600 checks. The cache code is unchanged.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
formatAge and formatAgeLabel picked the hour unit on the raw value and
rounded afterwards, so a detection between 47.5 and 48 hours old read
48h, while one exactly 48 hours old read 2d. Round to whole hours first,
then choose the unit.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The query tools rounded courses, headings, bearings and azimuths with
Number(value.toFixed(0)) and no wrap, so a direction of 359.5 degrees or
more came back as 360: find_vessel with course 359.7 and heading 359.6
returned course_deg 360 and heading_deg 360.
Round them through the shared roundHeadingDegrees helper the layers and
UI already use (src/data/headingDegrees.js), which wraps into 0 to 359:
course_deg and heading_deg (vessels), course_deg (aircraft), bearing_deg
(transit), heading_deg (cameras), and rise_azimuth_deg and azimuth_deg
(satellites). Other rounding is unchanged.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Starting the upstream timeout after the outbound queue let a request run
up to about 19 s, past the Directions layer's 15 s; the budget again runs
from arrival, queue wait included, and a budget spent before the request
leaves the queue answers 429 "routing busy" rather than an outage.
plan_route threw "Routing HTTP 502" for an outage, and voice passes only
the message on; it now throws "Routing service unavailable (HTTP 502)".
Co-Authored-By: Claude Opus 5.5 <[email protected]>
When the OSRM routing service failed (a 5xx, a 403 block, a 404), answered
with a body that was not JSON, or did not answer at all (timeout, network
error), /api/route returned HTTP 200 with 'no route found' or 'route proxy
error'. plan_route turned that into invalid_arguments ("No walk route from
...: no route found"), blaming the caller for an outage, and the directions
layer reported "No route found between A and B".
The proxy now answers all of these with 502 {ok:false, error:'routing
service unavailable'}: a rejected upstream fetch, any failure status but
400, a non-JSON content type, and a body read that fails or does not parse.
An exception in the proxy's own code after that point is still a 200 'route
proxy error'. createRouteService throws for a 5xx before reading the body's
ok:false, so plan_route reports unavailable, and the directions source
treats any 5xx as "Routing unavailable (HTTP n)", also when the failure
page in front of the proxy is not JSON. OSRM's own refusal (a 400 such as
NoRoute, or a 200 without a route) still reads "no route found".
The 12 s upstream timeout now starts when the request leaves rather than
before it waits for an outbound slot, so a wait in the proxy's own queue
(up to about 8 s at the 1 s spacing) is not reported as an outage.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A one_side line that doubles back within 5 m of itself, such as a
turnaround drawn as one line, matched its own return leg and took a
two-way half share with no partner. Flow index pieces now carry their
source line, and a radiusFor of 0 leaves a piece out of nearest(), so
pairing skips the line's own pieces.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
TomTom draws a two-way street whose directions differ as two one_side
lines on one centreline, and each got a one-way road's full share of
simulated vehicles, so the street drew twice the dots of the same street as
one full-coverage line or a two-way OSM road. A one_side line with an
opposite-direction one_side line within 5 m of at least two of three points
along it now gets a two-way direction's half share. On the decoded Austin
flow fixture every one_side line is paired, and road length weighted by
direction share falls from 553 to 417 km. Pairing a 16-tile area takes
about 41 ms per road pass. Refs #1455.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The ellipsoidal horizon test hid ground below the ellipsoid (low-geoid
coasts, the Dead Sea) from a camera near it, so it is now tested at 0 m.
With the globe hidden the ground is tested as the vertical from 0 to
9,000 m against the frustum planes, in place of a 4.5 km sphere that let
in much unseen ground.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
An unsampled base sits at ellipsoid height 0. Over high terrain seen at an
angle that point projects below the screen while the ground above it is in
view, so the gate skipped ground the camera could see and never sampled it.
The gate now tests loaded globe terrain when there is some, and otherwise
the whole span of heights a sample can accept (0 to 9,000 m).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
scene.sampleHeight reads only surface rendered for the current view. For a
data centre or dam base outside the frustum or over the horizon it read
through the Earth (about -6,344 km over Ashburn on Esri), failed the 9,000 m
check and retried every 2 s indefinitely, one offscreen render per record.
Over Ashburn that held rest frames at 58-589 ms. Unseen bases are now
skipped without being stamped, so each is sampled on the first walk that
sees it, and they no longer arm the retry render.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The set_layer_visibility layerId hint listed aliases such as "dams" for
local-dams and "weather radar" for weather-radar, which the id in the
enum already says. Leaving those out, and a layer with no other alias,
cuts the Realtime prompt from 52,000 to 51,695 characters. Aliases the id
does not spell (aqhi, hurricanes, clouds) stay, and the runner still
accepts every alias. Refs #1363.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The test passed URL.pathname to sharp, which is /D:/... on Windows, so the
faces were not found and npm test failed on every Windows clone.
fileURLToPath gives a path that works on every platform.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A request with any point the proxy cannot answer gets 502, and the page
used to put every point of that chunk on its geoid, including the ones the
proxy had real (cached or stale) heights for. During an upstream 429 pause
that drops cached ground under every chunk that also holds a new point.
The 502 body now carries `partial`: the request's real heights in order,
null for points without one, left out when there are none. The terrain
request service attaches it to the error it throws, and the page caches
those heights and uses its geoid only for the rest. `get_terrain_height`
still reports the failure as before.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Once the TomTom tile budget is spent, or TomTom fails, the proxy serves
cached tiles marked STALE-BUDGET / STALE-ERROR. flowSource works out
`stale`, but the layer fetched flow through fetchFlowForBounds, which
returns only the segments, so getStats() and the sync chip still read
"LIVE · Roads: …". #1239 fixed the get_traffic_flow tool the same way; the
layer now uses fetchFlowDetail too, reports stale and oldestTileAt in its
stats, and the chip reads "STALE · Roads: <source> · Flow: cached TomTom
tiles". A source without fetchFlowDetail keeps the previous behaviour.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
When WFIGS was unreachable the proxy served its last snapshot, marked stale
(X-Data-Stale, body.stale), with no age limit, and the client source kept
only the rows. The layer then read healthy with a fresh update time and
get_fire_perimeters reported the perimeters as current; three days into an
outage it still said "1 mapped wildfire perimeter in the requested box."
The source now offers getSnapshotDetail ({rows, stale, fetchedAt}); the
layer reports stale with the snapshot's fetch time, the tool adds stale,
retrieved_at and "(data may be stale)", and the proxy serves a stale
snapshot for at most 24 hours, as the GNSS proxy caps its own.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Review follow-ups: the upstream error now carries its HTTP status and the
pause checks error.status instead of the message text; concurrent 429s
extend the pause with Math.max rather than a shorter Retry-After cutting a
longer one short; the per-chunk re-check says it is for multi-chunk
requests; the first 429 test runs its backoff on mocked timers.
CURRENT-STATE describes the pause.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A query with a comma counted as "place, region", so a nearby place whose
own name has a comma ("Bed, Bath & Beyond") no longer won the nearby pass
and the unbiased pass could fly to a namesake far away. Its whole-name
match is now kept as the last choice, after the region checks of both
passes. CURRENT-STATE describes the place-region rule.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A name that only started with the place word also matched, so from Lyon
"Paris, France" picked "Paris Gare de Lyon" in the biased pass, where main
fell through to the unbiased answer and reached Paris. Only an exact name
matches now. Adds a test for the UK alias.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A rate-limited upstream got asked again by every page request for new
points (about once a second while the camera moves), each with its own
retries, which kept the limit in force and logged "refresh incomplete" each
time. After a 429 the proxy now leaves the upstream alone for its
Retry-After (at most 10 minutes), or 60 s when it sends none, serving
cached points meanwhile. The warning that starts the pause says
"(HTTP 429; pausing upstream for N s)"; requests refused during the pause
don't log.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
From Denver, 'Paris, Texas' flew to a pub in Budapest at rooftop range. The
camera-biased Photon answer listed the town first, but the pub's name equals
the whole query, and the biased path matched names only. A query with a
region now takes a candidate named for the place whose city, county, state,
country or country code (or USA/UK for US/GB) contains every word of the
region, in both the biased and the unbiased pass; with no such candidate
the existing name rules decide. A query with no region after its comma
keeps the biased name match.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
