This functionality is duplicated by the PointsAlongGeometry algorithm
Keep the algorithm version, since it has unit tests and supports things
like progress reports.
(marked as feature for inclusion in release notes)
The Fill_holes script is duplicated by the DeleteHoles algorithm,
and the DeleteHoles algorithm has many advantages:
- it's a full algorithm rather than a script
- uses qgis geometry classes and maintains curves/z/m
- has unit tests
- doesn't rely on non-standard external library (shapely)
(marked as feature for inclusion in release notes)
Copy min area parameter from 'Fill holes' algorithm to 'delete
holes' algorithm.
Also:
- make algorithm maintain z/m values
- make algorithm work with curved geometries
- add unit tests
Since both input and intersect layers have only one field (fid), the result is the same for both (existing and new) tests. The result now comes with the intersect field, renamed to fid_1 (previously not kept).
Rename algorithm SplitLinesWithLines to SplitWithLines
Accept polygon as input, too
Use only selected lines to split with (if processing is set to use selection only)
Issue log message if trying to split multi geometries
Update help
Instead of copyign a pre-existing .db shipped with QGIS, create
a new .db and import an .xml file containing our default symbols.
On the UX front, the main benefit is being able to ship with
pre-defined tags and favorite flags for the default symbols.
On the developer side, it means we get rid of the requirement
to maintain both an .xml and a binary .db in the source tree.
We also say aurevoir to the symbol_xml2db.py script and the
need to update the (obsolete) script every time we add a new
feature to QgsStyle.
These variables take a lot of time to calculate and cause lots of
lengthy hangs in processing. (Eg add some moderately large rasters
to a project, then try to run any processing algorithm and QGIS
will freeze).
The layer extent can already be used in expressions via the
layer_property function, which only evalutes the extent if required
and only for layers it is used for.
The band stats for raster layers should be moved to a
band_statistic function in core which behaves the same way.
