kitemmmodels ought to be built with python bindings to silence cmake warnings
Description
Status | Assigned | Task | ||
---|---|---|---|---|
Open | None | T4483 kitemmodels with python | ||
Resolved | sitter | T4482 pyqt5 needs upgrade | ||
Resolved | jriddell | T5639 pyqt5 literally not rebuilt |
This looks a bit fiddly.
Primarily we'll want to depend on dh-python, python-sip-dev, python3-sip-dev and some supporting plunder. This should get both python2 and 3 modules get built. From what I understand they'll go into python-pykf5.kitemmodels and python3-pykf5.kitemmodels respectively (package pykf5, module kitemmodels). Additionally this also generates sip files which seems a bit novel, outside of Py*/ dirs anyway, so naming is a bit less obvious there. Supposedly pykf5.kitemmodels-dev would be the most obvious as it mixes pyqt5's own pyqt5-dev with the binary naming policy (FTR: there is no overall pykf5-dev)
dh_python2 and dh_python3 manpages are relevant
Best run results by Scott Kitterman for review
The buggers install -./usr/lib/python2.7/dist-packages/PyKF5/__init__.py which is gonna be super conflicting if more than one framework build python bindings :O
wip https://packaging.neon.kde.org/frameworks/kitemmodels.git/log/?h=Neon/python-bindings
also talking about this on debian-kde list
Also see T5016#74679 enabling us to build pykf5 lower in the stack, I haven't managed to try that yest as my build systems keeps catching on fire for various reasons.
I haven't gotten more input from Debian, but I think the changes they suggested put the packaging in a good spot overall.
Unassign. Nobody's asking for it, not worth losing sleep over IMO. Also debian didn't enable it so I guess they too have no requests for it.