Use a tree display for processing history entries, where the root
item for each entry shows the full algorithm log when clicked,
and the python/qgis_process commands are instead shown as child
items
This provides more useful information for users browsing the history,
while still making the all the previous information available
This action shows the log of that child step, regardless of whether
or not it failed. This is handy for debugging model errors after
testing, when you've already closed the algorithm window...!
When editing a model through the designer (and after having run
that model), you can now right click any child step in the model
and select "View Output Layers". This will add the output layers
from that step as new layers in the current QGIS project.
This action is available for ALL child algorithms in the model,
even if the model is not configured to use the outputs from those
children as model outputs.
This is designed as a helpful debugging action. If a user's model
fails (or gives unexpected results), they can then trace through
the model and view the outputs for suspected problematic steps.
It avoids the need to add temporary outputs to a model and re-run
to test.
Additionally, this action is always available after running the model,
EVEN if the model itself failed (eg because of a misconfigured
step later in the model).
Sponsored by City of Canning
Adds drop support for layers in the multiple layer selection widgets to more efficiently select the desired elements. Especially useful in big project of with duplicated layer names.
Instead of forcing all steps in the batch processing dialog to execute
in the main thread, we now run each step as a separate task whenever
possible. This keeps the UI nice and responsive, and permits
responsive cancelation and progress reporting.
Individual steps are still run sequentially, not in parallel (yet!)
accepts the prompt to save the changes BUT then cancels the file dialog
asking for the destination file name, don't treat this as though the
user has opted to discard the model
- When saving a model to a file, don't require that the model already
has a name entered. Instead, if no model name has been entered then
set the model name automatically to match the selected save filename.
E.g. if the user saves the model as "Process incoming features.model3",
then the model name will be set to "Process incoming features".
- When first saving a model, IF the user HAS already entered a model
name then make the default file name suggested by the dialog match
this model name
when configuring a new algorithm
While the previous behaviour of defaulting to a static value makes
sense for things like numeric parameters, this is a very rare use
case for map layer parameters. For better new user experience we can
default instead to showing model inputs for these parameter types.
No user visible changes, but we gain the flexibility for better
history storage, including the ability to store the outputs
calculated by the algorithm execution...!
current settings as JSON and paste json settings from clipboard
The utility of this is two fold:
1. It provides a way for users to copy the settings defined in the
dialog in a text format, so they can store these easily and then
restore later by pasting the value
2. It provides an easy way for users to copy the settings into
the json format consumed by qgis_process (following
https://github.com/qgis/QGIS/pull/46497), so that it is easy
for users to see the expected format even for complex parameters
(like tin interpolation parameters)
