Hi there, QWindowSystemInterface is treated as a QPA header since this discussion:
http://thread.gmane.org/gmane.comp.lib.qt.devel/4178/focus=4210 http://thread.gmane.org/gmane.comp.lib.qt.devel/4178/focus=4206 QWindowSystemInterface is different to QPA API because it is just a bunch of static methods on a class, rather that a virtual interface designed to be subclassed. http://thread.gmane.org/gmane.comp.lib.qt.devel/3198/focus=3210 It can be extended in its current form (adding more static methods) in the future, and it can remain an 'internal' implementation detail, but I think it should be made public. If it is not made public, then all CMake based code using QtTest will have to use private QtGui headers. https://codereview.qt-project.org/#change,30483 It of course already breaks KDE tests. It will continue to break other tests where qmake isn't used (which has a workaround for this problem - if qtestlib is used then the -private modules of all Qt modules are used). I don't think it's enough or 'good' to have a workaround for qmake and leave the others out in the cold. I think the best solution is to move QWindowSystemInterface headers out of qpa, making it 'public but internal' API. I haven't seen any reason not to do that - I have only seen 'that shouldn't be public' without a reason why its header must be in a private directory. I have a reason to do it - otherwise all non-qmake code using testlib gets very ugly and has to use 'private' stuff. https://codereview.qt-project.org/30541 Thanks, -- Stephen Kelly <[email protected]> | Software Engineer KDAB (Deutschland) GmbH & Co.KG, a KDAB Group Company www.kdab.com || Germany +49-30-521325470 || Sweden (HQ) +46-563-540090 KDAB - Qt Experts - Platform-Independent Software Solutions
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________ Development mailing list [email protected] http://lists.qt-project.org/mailman/listinfo/development
