Some filters are fast enough to return results that it's overkill
to run them in a background thread - add a flag to these filters
to allow them to run (blocking) in the main thread instead.
- add a clone() method to filters, and always search using the
clone instead of the original filter
- add a prepare() method to filters, which is always run in the
main thread and can be used to prepare the filter for safe
background execution (e.g. creating feature iterators in advance)
- don't use QtConcurrent to perform searches in background threads,
since it is not safe to use with QObjects
- instead manually create threads and ensure that cloned objects
are always moved to the thread that they will run in, to ensure
that they correctly have thread affinity with the thread in which
they are executed
This is used a lot, yet the current constructor calls the
normalize() method which does a bunch of operations for no
result.
So instead provide a simple optimised null QgsRectangle
constructor and save a lot of cycles.
Refs #17809
This was a missing capability in the processing API - while algorithms
could declare multiple layer input parameters, there was no corresponding
multi-layer output. This meant that algorithms (such as Package Layers,
Vector Split) which create a set of layers which cannot be determined
in advance had no way to pass these generated layers on for further model
processing steps.
It's also useful for algorithms which operate on a specified folder,
processing all layers found there, and allowing these generated
outputs to be utilised in other model steps (e.g. packaging
all of them, merging them, etc)
Since libspatialindex is not thread safe on all platforms, and
have expressed desire to remove the thread safety that they DO
have on remaining platforms, it's safer and easier for us
to manually add locks to QgsSpatialIndex and be gauranteed that
this class is thread safe on all platforms and libspatialindex
versions.
Also improve docs for the class.
Since the underlying libspatialindex is not thread safe
on some platforms (e.g. Windows), manual calls to detach()
must be made if a QgsSpatialIndex is to be accessed across
multiple threads.
Note that for platforms on which libspatialindex is thread
safe, calling detach() has no effect and does not force the
deep copy.
their unique id()
This is used when generating the QgsHelp url for algorithms
attached to the providers.
Implement helpId overrides for the native and 3d providers so
that they return 'qgis' helpIds, meaning that all QGIS processing
algorithm documentation can be kept within the same url path
regardless of which QGIS provider library it sits within.
This also allows us to freely move algorithms from the Python
'qgis' provider to c++ 'native' provider in future releases
without breaking the help URLs.
Fixes#17231
From the clazy docs:
"Warns when a signal, slot or invokable declaration is not using fully-qualified
type names, which will break old-style connects and interacting with QML."
But always prefer case-exact matches for parameter names. Turns
out the grass provider requires use of parameters with the same
name but different case, so we need to be able to handle this.
Since the underlying issues with the Python bindings are now fixed,
in most cases we can safely default to allowing an algorithm to
run in a background thread!!
So now we make this the default, and require individual algorithms
which are NOT thread safe to declare this. This includes algorithms
which directly manipulate the current project or layers (such as
setting layer styles), alter the selections in layers, or which
rely on 3rd party libraries (for now, SAGA and GRASS algorithms
are marked as not thread safe... TODO - someone more familiar with
these libraries can investigate and remove the flag if appropriate).
Also models are marked as non-thread safe. TODO: only flag an
individual model as thread-unsafe if any of its child algorithms
report this flag.
And make this support opt-out, since the vast majority of providers
are based on QGIS API and don't have external dependencies which would
restrict use of memory layers/etc.
Plus, I'd rather see non-compliant providers expose this support when
they can't use non-file-based-outputs (and make this the bug which
needs fixing) then have to rely on plugin providers to discover and
explicitly expose this support.
options should be available for certain model outputs and script
algorithm outputs
We do this by swapping the test for non-file based output support
from checking only the algorithm's provider to instead checking
on a parameter-by-parameter basis.
This is done in order to support models. For models, depending
on what child algorithm a model output is based off, an individual
model may or may not have support for non-file based outputs. E.g
a model may generate outputs from a native qgis alg (supporting
these outputs) AND an output from a GDAL alg (with no support
for these outputs). In this case we need to enable or disable
the ui controls for non-file based outputs on an individual
output basis.
For scripts (for now) we blindly just say all outputs support
non-file based formats. This is going to be the case most of
the time, since scripts will usually be written using PyQGIS
API. For the exceptions (e.g. scripts which call other algs
like GDAL algs) we probably should add some way for the script
to indicate whether an individual output supports this, but
for now we just say they all do.
Fixes#17949
