[QGIS-Developer] Processing model improvement suggestions

[johnrobot]

Hi
I have been using processing models in QGIS 3.10.0 a bit lately and I have
some additional improvement suggestions (on top of my previous collection,
http://osgeo-org.1560.x6.nabble.com/QGIS-Developer-Thoughts-on-using-processing-models-for-ETL-tasks-td5365767.html#a5368491).
In no particual order:

- Support for drag and drop of model files (.model3) in a QGIS window, that
should open the model.

- Improved documentation for algorithms: All algorithms should include
examples and images that show the data before and after the algorithm. That
would help understanding what it does to the data.

- The help/documentation is always visible to the right when I open an
algorithm separately (not in a model), it doesn´t remember its state from
the last time I used it. That could be added, or even move the help to a
separate window. The "Help" button always seems to give me an 404 error.
Also, I don´t see how to access the algorithm documentation from a model.
The pane to the right is missing and the "Help" button gives me the same 404
error.

- Improved error messages when an algorithm fails. Perhaps even add
suggestions on how to fix the issue?

- I would like a model to be able to read multiple files stored in
compressed files, such as .zip and .rar. I have a use case where multiple
GeoTIFF files with elevation data are stored in rar archives and I would
like to read all of them, without having to unpack the rar files manually
before running a model.

- The number of features read and written (for each data set and the total
amount) should be specified in the log file. That would help me detect
potential data loss.

- Expose incoming attributes in the data used in the model, much like FME
does. At current, it is unnecessarily hard to view how attributes are
processed/changed throughout the model.

- I would be great with an option for dynamically separating output into
separate files/tables/layers (depending on the type of output) based on an
attribute value, for example one file for the features containing "Canada"
and a separate one with "Mexico".

- Support for reading and processing non-geo JSON data, as described here,
http://osgeo-org.1560.x6.nabble.com/Read-non-geo-JSON-data-in-QGIS-td5420262.html.

- Support for writing model result as TopoJSON, the format used by Power BI.

- In the algorithm for calculating the shortest path (point to point), it
seems that I have to define start and end points manually. In my case, I
have about 2 million sets of start and end points. Perhaps add support for
reading these coordinates from attributes instead of picking manually in the
map? Also, support for routing using more than one processor core, and even
a GPU (using OpenCL?) would probably speed up these calculations
considerably.

Any thoughts on these suggestions?

Regards,

Magnus



--
Sent from: http://osgeo-org.1560.x6.nabble.com/QGIS-Developer-f4099106.html