- make it independent on the current view's Z range
- simplify scaling code
- reverse Z handling (higher Z value == closer to "camera")
- remove unused API functions and member variables
* PointCloudLayerExporter class
* set points limit
* saveas dialog
* qickndirty task
* wip
* add pdal files
* transform context
* add getters
* try to use some subclasses
* use nested classes
* don't default layername
* limit supported formats
* error reporting
* limit formats in gui too
* format filters and source/target crs
* use default infinite extent
* set pdal crs
* add pdrf but keep 3 for now
* properly handle LAZ attributes
* catch coordinate transform exceptions
* use default destructor, fix leak, docs
* trim includes, fix typos
* use qgsspinboxes, add points limit to the gui
* fix limit exporting logic
* avoid auto
* fix builds
* don't reuse lastVectorFormat setting
* Rework layer name handling
* Rage against the elevation
* invert attribute activation logic
* use wkt for laz srs
* use original offset/scale if not reprojecting
* move created memory layer to main thread when done
* allways add memory layer to map
* oops
* properly handle existing vector files
* don't add layer if cancelled
* create the memory layer on the main thread
* remove colon
* remove duplicate code and avoid crashing on mesh and vectortiles
* fix leak
* handle case blunder
* address review
QgsProviderSqlQueryBuilder provides an interface for creation of SQL queries, which
can be overridden for backend provider specific SQL syntax.
Initially add a virtual method for creation of row limited select *
queries.
This method restores the Qt 5 logic for testing for null variants,
where for core Qt types the isNull check is forwarded to the actual
data type.
E.g. on Qt 5: QVariant( QDateTime()).isNull() is true, but
on Qt 6 it's false
This breaks a LOT of assumptions made throughout QGIS. The new
helper method avoids this breakage by ensuring we follow the Qt 5
logic also on Qt 6 builds, and also gives us the option to extend
this logic for user types (e.g QVariant( QgsGeometry() ).isNull() could
return true). (That's not included here)
on newer qt/sip versions
This was incorrectly returning a malformed/semi-null QVariant
...something?!?
We probably need to add more cases here, but this one was directly
causing a test failure and so was easy to identify
of resolved relationships, so that it can handle resolving
a many-to-many weak relationship to two QgsRelations (which is
required to represent a many-to-many relationship using QgsRelation)
