> we should be honest Gtk falls behind very badly these days in almost all category
While I agree you'd have to be absolutely nuts to develop an application using GTK+ directly in C (Yes, I think the whole GNOME project is nuts... but it's telling that they adopted Javascript and Vala), the C++ bindings for GTK[0] weren't all that bad when I last looked at them.
Of course, without a MOC equivalent, you can't really beat the slickness of Qt/QML.
My biggest complaint with GTKmm right now is the lack of composite widget support. It's actually nice to develop a GTK+ application in Vala because composite widgets kick ass and remove a huge pain point for just wanting to throw a window up without all the nasty GTK signal connection maddness. Once GTKmm gets that support I'll probably start using it more.
Well I haven't looked at it since Gtk Builder was called Glade, but QML has convinced me that declarative UI markup with sprinklings of code, and custom widgets implemented on the side, is the way to go.
Agreed, hence why composite widget support in GTKmm would be really nice. In Vala you just make a normal class inheriting from whatever your base widget is, decorate it with a [GtkTemplate] attribute and whenever you construct that widget it will use the gtkbuilder XML you referenced in the attribute to construct the widget, binding all the callbacks referenced in the XML with the respective methods on the class.
Until recently C++ hasn't even had compile time reflection that would allow this to work very well, now that C++14/C++17 are adding that I think this would be very doable.
While I agree you'd have to be absolutely nuts to develop an application using GTK+ directly in C (Yes, I think the whole GNOME project is nuts... but it's telling that they adopted Javascript and Vala), the C++ bindings for GTK[0] weren't all that bad when I last looked at them.
Of course, without a MOC equivalent, you can't really beat the slickness of Qt/QML.
[0] http://www.gtkmm.org/en/