On 15 August 2016 at 23:45, Denis Rouzaud <[email protected]> wrote: > Hi all, > > While looking at failing tests on Qt5, it seems that QgsJSONUtils test is > failing due to a regression on Qt 5.5. > Conversion from QVariant to QString is broken for double and float values > [0]. > It seems to have been fixed in Qt 5.7 > > Especially QgsJSONUtils.encodeValue(5.9) returns '5.9000000000000004'. > > What would you suggest? > > * Don't care and discard failing test with Qt 5.5 - 5.6? (is this really an > issue?) > * Use a custom toString method to fix this? > * Bump min Qt version to 5.7? > * ....?
IMHO Qt really dropped the ball on this one. Such an annoying blocking bug to have present in their LTS release! I'd go with: - minimum recommended version >= 5.7 - allow builds with 5.5/5.6, but with a prominent warning in the release note that strings such as map grids, etc will be ugly! I'd love to just say we put a hard minimum version of 5.7 (and avoid all the numerous bug reports which are bound to come in!), but I don't think that's realistic given the glacial pace some of our supported distributions move at... Nyall > > Cheers, > Denis > > > [0] https://bugreports.qt.io/browse/QTBUG-47192 > > _______________________________________________ > 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
