* It's now implicitly shared, meaning that changes to the config
can be supervised and a signal is sent from QgsVectorLayer and
there is no risk of elements suddenly being deleted.
* Remove a bunch of methods that were in QgsVectorLayer for legacy
reasons.
* Save more data to QML
* Virtual fields
* Map tips
* Display expression
* Read only flag
* Streamline expression context generation
Whenever an object is able to generate an expression context it
implements the method createExpressionContext() declared in
QgsExpressionContextGenerator.
This makes a cleaner API and allows using QgsFieldExpressionWidget and
QgsDataDefinedButton from python because standard OO programming
approaches are used instead of callbacks and void pointers.
* Colorize output of doc and sip tests
* Fix build
* Fix sip complaints
* Fix rebase problems
* Workaround failing bindings test
The improvement allows immediate cancellation of raster rendering
in progress. Until now, even when map rendering got cancelled
(e.g. by zooming of panning canvas), the GUI got blocked while waiting
for raster layers to finish their rendering (only vector layers have
had support for cancellation). This should allow for much smoother
user experience while browsing maps including rasters.
The cancellation is supported currently by WMS/WMTS and WCS providers.
GDAL provider may also get support thanks to improvements in GDAL 2.
Funded by Land Information New Zealand.
Fixes#11371 (crash on ungrouping after moving the group) and more
undo/redo related issues.
Enable pending test for the crash (now passing) and add many more
undo/redo related ones (including signals testing).
Includes a new QgsGroupUngroupItemsCommand class
and its SIP bindings.
Move the QgsExpression::Interval class out to its own QgsInterval
class, extend with new methods and add tests
Add a typedef to keep API compatibility for 2.16
The original name was not representative for the scope of the class
and misleading at best.
The class manages actions for a given layer and it's not a single action
like the former name suggested.
There is a typedef introduced to keep it in the API until we can remove
it with QGIS 3.
Using this renderer no symbol will be drawn for features, but labeling,
diagrams and other non-symbol parts will still be shown.
Selections can still be made on the layer in the canvas and selected
features will be rendered with a default symbol. Features being edited
will also be shown.
This is intended as a handy shortcut for layers which you only want
to show labels or diagrams for, and avoids the need to render
symbols with totally transparent fill/border to achieve this.
(fix#12131)
This adds a configuration interface and renderer that makes it easy to
put all the pieces together which are required to get a 2.5D effect.
It allow for configuring some of the styling and is meant to create an
easy-to-use setup.
Since every part of the system is built around QGIS' internal rendering
and symbology engine, there is much to fine tune. To get all the
possibilities, just change the renderer to a graduated, categorized or
single symbol renderer upon creation and you will find full access to
improve the style to your needs.
Funded by
* Regional Council of Picardy
* ADUGA
* Ville de Nyon
* Wetu GIT cc
[FEATURE] Tracing of features (digitizing)
Tracing can be now used in various capturing map tools (add feature, add part, ...) including reshape and split tools.
Tracing is simply a new mode for these tools - when tracing is not enabled, the tools work as usual. When tracing is enabled (by clicking the new magnet icon or pressing T key), tools switch to tracing behavior:
- first click on a vertex/edge (must be snapped!) will start tracing - moving mouse on top of the map continuously updates the trace
- next click will confirm the trace and mark start of a new trace Tracing can be enabled/disabled anytime even while digitizing one feature, so it is possible to digitize some parts of the feature with tracing enabled and other parts with tracing disabled.
Tracing respects snapping configuration for the list of traceable layers.
If there are too many features in map display, tracing is disabled to avoid potentially long tracing structure preparation and large memory overhead. After zooming in or disabling some layers the tracing is enabled again.
Internally, things work like this:
- when tracing is requested, linestrings are extracted from vector layers, then noded (using GEOSNode to resolve all intersections) and finally a simple planar graph is built (vertices + edges)
- when tracing, endpoints are temporarily added to the graph (if not equal to one of existing vertices already) and Dijkstra's algorithm is run to get shortest path
Original specs for the curious ones (the interaction with QGIS is slightly improved from what has been specified): http://www.lutraconsulting.co.uk/crowdfunding/autotrace-phase-2/specification.pdf
Now all classes and members are either exposed to bindings or marked
as "not available in Python bindings" in the docs.
Drop test thresholds to 0. Now it should be much easier to determine
what missing members have been added which are causing test
failures.
