over widgets
Hopefully helps Python processing users match the parameters
in the gui to the parameter names to use when calling via
API
(Blender style!)
While the toolbar buttons were shown in the correct size, the
actual icon image in the buttons was not correctly scaling to
match the button size.
Hardcoding the svg icons in the py file seems to be the only
way to avoid this.
processing tests
Some algorithms will return results in different orders, e.g.
due to the use of dicts or other methods which do not guarantee
a fixed return order.
Using a primary key to do the feature match allows us to flexibly
handle these situations and provide tests for these algorithms.
tests
Some algorithms are non-deterministic and the results may vary from
run to run. In this case we allow specifying multiple possible valid
results, and the test will pass if the result layer matches any of these.
Differences in gdal libraries mean the hash value may differ between
platforms. Allow multiple acceptable hashes to be listed for
expected test results
Improvements:
- Maintain Z/M values
- Keep original data type for group/order fields
- Group field is optional
- Added unit tests
- Don't export text files for features by default
Improvements:
- transparent reprojection to match hub/spoke CRS
- keep all attributes from matched hub/spoke features
- don't break after matching one hub point to spoke - instead
join ALL hub/spoke points with matching id values
keep the algorithm dialog open after execution
Otherwise it's hard to see the error - you have to know to check
the python log. Keeping the dialog open at the log makes the
error immediately visible to the user
Improvements:
- slight optimisation to feature requests - don't request attributes
which are not used
- Remove "method" param. Now the decision to group by field or
not is made only on whether a class field was selected or not
