On 9 Jun 2015 1:07 pm, "Larry Shaffer" <lar...@dakotacarto.com> wrote: > > Hi, > > This can be fixed for Qt 4.8.x, but requires direct access to the Cocoa API when needed to introduce the non-Qt-source workarounds noted in the QTBUG-40449 [0]. > > Btw, this is an example of my old PR [1] to introduce abstracted calls to the OS (Mac for example, with Objective-C++) to handle changes in OS API libraries and keep such workarounds in one place instead of peppered throughout the source tree. Especially for those instances where Qt fails us, but a direct call to the OS could be abstracted for a workaround where needed. > > Unfortunately, it is a significant change that needs really tested well, and certainly not reasonable to introduce into 2.10 as a fix, as it is more of a feature. > > I never upgraded to 10.10 for several reasons, but am about to, or I would have noticed this nasty bug sooner. > > [0] https://bugreports.qt.io/browse/QTBUG-40449 > [1] https://github.com/qgis/QGIS/pull/1084
Thanks for looking into this Larry, I'm wondering - is there anything stopping the OSX builds from moving to the Qt5/pyqt4 combination? I gather qt4 on 10.10 is unsupported and probably has a lot of little quirks like this we can't easily fix. I'm tipping the next OSX release will have even more. But since we're not relying on a distro and are more or less free to choose dependencies on OSX, couldn't we switch the OSX build across now? Nyall
_______________________________________________ Qgis-developer mailing list Qgis-developer@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/qgis-developer