The most important scientific libraries have pledged to drop support before 2020, and are all python3-ready
I used to work on a GUI app in Python. I ported it to Python 3, then switched OSes for various reasons. 5 years on, on Ubuntu Xenial (so new I can't even use it in Travis, but that's a separate whine), I install pykdeuic4 and it's using Python 2. So I've basically abandoned that project for 5 years now, because every time I looked at it I thought "surely Python 3 will be here in a few months, I don't want to port backwards to Python 2".
(Serious question: is there a PPA or anything I can use to get these things for Python 3? I need PyQwt as well as PyKDE)
Seriously: if you can, try building your thing with tkinter. It's the standard lib, and will "work"
Qt's binding support has been it's biggest issue since forever.
It has quite a few bindings already, here are the python ones: https://github.com/joaoventura/pylibui .
I love catching errors at compile time, not runtime :-)
Plus, it runs faster
If there were an ML-family language with good Qt support I'd use that for my GUIs - I use ML-based languages for almost all my non-GUI work. There's plenty I dislike about Python but it's still the best desktop GUI experience I've found.
% python --version
Python 3.5.2
% pip install numpy
Collecting numpy
Using cached numpy-1.12.0-cp35-cp35m-macosx_10_6_intel.macosx_10_9_intel.macosx_10_9_x86_64.macosx_10_10_intel.macosx_10_10_x86_64.whl
Installing collected packages: numpy
Successfully installed numpy-1.12.0Not numpy, scipy, tensorflow...