GitLab

This avoids the problem of signals being emitted before we could connect
to them, in particular in a QML context where we have less control over
the execution order.

Useful for testing against a test server prior to deploying server updates.

This seems to happen for polygons referenced by multiple multi-polygons,
and was massively confusing our geometry merging code.

Fixes the inner geometry in Berlin central station for example.

The old approach would not allow a seamless transition, but always produce
a one level gap.

Makes the structure in the info box a bit more clear.

This allows custom display of those values, which are out of scope for
this library.

This will replace previous "rowCount() > 1" checks with something that
automatically updates when data is loaded delayed.

Fixes a few unit tests.

Namely, this now yields the expected results in the face of OSM data with
incomplete or otherwise imperfect layering information, resulting in a
messed-up z order. In some cases this doesn't even have significant visual
impact (e.g. when transparent or large area boundary elements are drawn on
top), but it resulted in unexpected picking results. Taking into account
the alpha value and the item size addresses that.

That's often the case for tram or subway platforms that don't follow the
a simple numbering scheme. This gives us a more complete platform list
on all tested stations, without creating additional noise.

Needed for line icon downloads to work.

Used in the map example to show line logos for lines departing from a
platform.

Fixes the FRA center runway being misrendered in some sections.

This should allow us to show proper line icons in the map next to their
respective stops/platforms.

This is necessary as Marble OSM tiles do not guarantee its synthetic
ids are globally unique. And while unlikely to happen in close spatial
proximity under normal conditions, it it practically possible to run
into this, showing as various tears in the map display.

Right now both are subject to the same handling still, but we want to only
apply remapping of synthetic node ids to the current merge buffer.

Besides being a tiny bit more efficient, this prepares for decoupling new
and deferred merges, which we need for remapping colliding synthetic node
ids. That's not a new problem, but one that has become more likely to
happen with the new Marble tile generator.

This is part of allowing KDE Itinerary to move the map to the correct
arrival/departure platforms.

This will allow deeper integration into KDE Itinerary.

Will be needed for domain-specific content models.