Background: at project load time for all layers a copy of the style
is loaded and stored as a DOM document. This was implemented for
broken layers to make it possible to store the style back to the
project if a broken layer was kept in the project.
Of course this is not needed nor useful in a server context.
By introducing a new project load flag and using it in the server
config (projects) cache we allow to skip the initial storage of
styles for the server and cut the startup times of our test "monster"
project (~1000 PG layers) from 26 seconds to 15.
Before you ask: I chose to always store the original style
with the idea that we might have been able to use the original
style copy also to "restore" styles to their initial state.
This was never implemented, so I guess we might be good to
make the initial style copy selectively available for broken
layers only in order to speed up project loading in the general
case (desktop).
Allows for map renders to be filtered by a z or elevation range,
such that only parts of the layer which are considered within
this range will be rendered.
Implements https://github.com/qgis/QGIS-Enhancement-Proposals/issues/201
*Requires support for respecting the QgsRenderContext zRange to be added
to the relevant map layer renderers
value of a parameter for the GUI only
This allows us freedom to change the default settings for an algorithm
shown when opening the algorithm in the toolbox/batch/model without
changing the underlying default value used for the parameter in the
raw API (which we can't do easily without potentially breaking 3rd
party scripts/plugins)
If set, this extent will be used when zooming to full extent (or for the
full extent for a map overview frame) instead of the extent calculated
from all map layers.
The intention is to eventually allow users a way to manually set their
desired "area of interest" for a project, so that zooming to full extent
won't zoom all the way out when the project contains global or national
datasets...
Try to remember/set last timeframe (upon timestep- or timeframe-changes).
Setting the stepsize to a different size, did reset the slider to start
(aka timeframe 0).
Same when you changed the range (data time) extent.
This commit tries to set the slider to the same position as before the
step change, if possible). Else it will take the position of the timeframe
in which the start of the old timeframe fits.
fixes#39994
While it was possible to set the opacity for individual symbol layer
colors via data defined expressions, it's so far been impossible to
set a data defined expression to control the overall symbol opacity.
This commit fixes that omission...
The SIP_FACTORY annotation didn't do the job for a Python implementation
of createChildren(). We must write a custom VirtualCatcherCode to make
sure that if the returned QgsDataItem elements are Python objects they
aren't killed too early.
Fixes the new added test, which uses to crash before the fix,
and corresponds to the use case of a plugin defining its own data items.
Has been checked not to cause memory leaks.
The change of SIP_FACTORY to SIP_TRANSFERBACK is to avoid a memory
leak when Python calls a C++ createChildren() implementation. Was found
by making test_provider_ogr.py::testDataItems() loop on its call to
createChildren()
All the logic for rendering non-opaque layers is handled in a layer
type agnostic way already, so this just allows the existing handling
to be used for mesh/point cloud/vector tile/etc layers also
