Closed anitagraser closed 4 years ago
The last status overview I could find was by @pcav https://lists.osgeo.org/pipermail/qgis-psc/2018-September/006556.html
- increased stability for Processing: unknown
- OSGeo4W updates: unknown
- Processing R provider: unknown
- OpenCL for Processing: completed
- QGIS server OGC: partly done, resources terminated
- Charts in forms: completed
- Training manual: 40% completed
Response by Nyall: https://lists.osgeo.org/pipermail/qgis-psc/2018-September/006601.html
They are all completed, except for the last one and the OGC WFS compliance (which ran out of resources).
@nyalldawson Is there any page I could link to for a summary of the "Increased stability for Processing GUI and External Providers" proposal?
@anitagraser
Here's the text from the report sent to PSC:
"Processing GUI and maintenance work:
Many bugs in the QGIS 3rd party providers have been fixed, with lots of new unit tests added. I've cleared out most of the outstanding issues against the SAGA provider (in most cases adding one or more test cases for these). I've also refactored the GDAL algorithm tests to test the GDAL command generated instead of the resultant output. This allows for much easier test creation (and there's now dozens of tests covering most of the GDAL algorithms, with tests for all the individual parameter options for these), and is much easier to detect when things go wrong vs the old approach of testing the created files. This also means we're no longer testing GDAL's operation itself, just making sure that what QGIS sends to GDAL is correct (so the tests are more robust against different GDAL versions). Additionally, lots of GRASS unit tests which were disabled in the lead-up to 3.0 have been resurrected, and we now test have some GRASS vector algorithms unit tests too. I'll continue to track the different channels for bug reports against individual algorithms and fix these.
The GUI refactoring has proceeded nicely, with the new c++ classes and framework in place for QGIS 3.4. This new framework is soaked in unit tests. So far the string, CRS and boolean parameters have been ported to the new API, with numeric and distance parameters ported in a (currently unmerged) PR. The new API brings many improvements besides the added stability, and should make it easier to utilise expression-based parameter values within models and make support for these automatic for all ported parameter types (vs lots of duplicate/fragile code in the older Python implementations). The porting work will continue during freeze and 3.6 (next up are the file, range and expression parameter widgets)."
There's no blog post write up of this, just the email report.
@nyalldawson Thank you, Nyall! I was under the impression that those paragraphs were merely a status update for work in progress but if they represent the final report, I'll use them.
@anitagraser
It's a little murky I guess -- that was the report initially sent to PSC, but there has been many follow ups since then.
E.g. for the gui refactoring, the point parameter, file parameter, range parameter, expression parameter, enum parameter, field parameter and fixed table (matrix) parameters have all been ported to date. Not all of these were part of the original proposal however!
I've published a summary blog post now: http://blog.qgis.org/2020/01/06/reports-from-the-winning-grant-proposals-2018/
Please let me know if anything should be added or needs correction.
Thanks Anita!
Deadline
2020-01-31
Notes
There were seven successful proposals, checkmarks indicate finished projects: