On 05/04/2016 11:00 AM, Matthias Kuhn wrote: > Hi, > > I have just published a post with the state of Python 3 and Qt 5 migration. > http://www.opengis.ch/2016/05/04/qgis-qt5-and-python3-migration-current-state/ > > In general I am very happy with what we have achieved so far. > > There is one problem that I think needs broader discussion which is an > issue about the handling of NULL values in PyQt5. With the equivalent to > the SIP API V2 we don't have PyQtNullVariant up to PyQt 5.6. This > basically means, no NULL for python plugins. And I consider this a blocker. > > We could > > * Switch to another API (sip.enableautoconversion(false), similar to > SIP API V1). Resulting in countless lines of code in plugins requiring > to be updated. without any chance of automation? > * Bump the minimum Qt version to 5.7 (and hope that PyQtNullVariant > will be fixed there, chances are good, see > https://www.riverbankcomputing.com/pipermail/pyqt/2016-April/037336.html). > > I tend towards Qt >= 5.7 - on Windows and Mac we control the > dependencies, so that should be ok anyway, on Linux we can say "if you > want a recent QGIS, get a recent distro". And by the time we actually > ship QGIS with a Qt5 dependency there should be a number of distros with > Qt 5.7. > > Any opinion on this? > Matthias > _______________________________________________ > Qgis-developer mailing list > [email protected] > List info: http://lists.osgeo.org/mailman/listinfo/qgis-developer > Unsubscribe: http://lists.osgeo.org/mailman/listinfo/qgis-developer
_______________________________________________ Qgis-developer mailing list [email protected] List info: http://lists.osgeo.org/mailman/listinfo/qgis-developer Unsubscribe: http://lists.osgeo.org/mailman/listinfo/qgis-developer
