Files
TREK/client/tests
Maurice bd347f553b fix(maps): the satellite switch answers on the phone, and the compass stops hiding under it
Two faults behind one round button on the phone map.

The base layer switcher that both engines draw in the bottom left corner
sat on top of the GL compass, which started 8px to its left and ran on
under the frosted shell, so it read as a second button showing through.
The compass now stands beside the switcher, placed off the switcher's own
exported inset and size so the two cannot drift apart again.

Satellite never drew on GL because applySatellite gave up whenever
isStyleLoaded() was false and only retried on styledata. That flag is false
while any source has tiles or a setData in flight, which on a phone is the
usual state, and styledata does not fire when those finish. On the road
trip stage every tap met a busy style, because the settings write
re-rendered the shell and handed the map fresh stage arrays. The guard is
gone (the pass only runs after load, and addSource/addLayer only refuse a
style document that is not in yet), an existing layer just flips its
visibility, and a source left without its layer still gets the layer.

The stage data in MMapArea is memoised, so a store write no longer refits
the camera or sets the route, access and place sources again.
2026-09-16 22:22:46 +00:00
..
2026-07-18 22:13:52 +02:00