Skip to content
Permalink
Browse files
Add difference image
  • Loading branch information
m-kuhn committed Apr 23, 2016
1 parent adf42da commit da0682d730c381fdd7224bd67f0da576e8a4cb53
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.

3 comments on commit da0682d

@nyalldawson

This comment has been minimized.

Copy link
Collaborator

@nyalldawson nyalldawson replied Apr 24, 2016

@m-kuhn what's your thoughts regarding the opacity/blend mode images? The color differences from qt4 seem extreme to me. But I can't find anything in the qt5 release notes which would explain this difference as an intentional change.

@m-kuhn

This comment has been minimized.

Copy link
Member Author

@m-kuhn m-kuhn replied Apr 24, 2016

@m-kuhn

This comment has been minimized.

Copy link
Member Author

@m-kuhn m-kuhn replied Apr 24, 2016

In general, the biggest remaining issue is probably that NULL information is not preserved on attributes but 0 or '' returned instead. It's likely that this will be reintroduced with PyQt5.7 after a check on the pyqt mailing list.

Another strange thing is, that in the Processing tests, vector layers seem not to be converted to QgsVectorLayer but are a QgsMapLayer. Not sure where the problem here is, IIRC, QObject subclasses should be converted to the proper subclass automatically.

Then we have some floating point issues ( 6.999999999998 != 7.0 ) which I'm not sure if they should be treated as issues or not and where this difference comes from.

Please sign in to comment.