[Qgis-psc] Plans to move to Qt6

[Nyall Dawson]

On Sun, 4 Dec 2022 at 20:00, Richard Duivenvoorde <rdmailings at duif.net> wrote:
>
> Hi psc/devs,
>
> (Let me start to say: no pressure! I know we are all busy....)
>
> As an avid Linux user, trying to support the Gnome/Wayland windowing system, I see more and more glitches appearing in Qt5 applications recently (Gnome 43, Debian Testing).

Just to restate my position so that it's clear for anyone coming
across this thread: !!!!!! Don't use QGIS under Wayland -- there's SO
many bugs and limitations (which aren't our fault) :) !!!!!!

> Do we have some (version) planning about moving to Qt6?nOr are we maybe waiting for distro packagers to package certain (for us) essential Qt6 packages?

See https://github.com/qgis/QGIS-Enhancement-Proposals/issues/198 for
the current "grand plan" of QGIS + Qt6.

It's currently blocked by the lack of Python support. It's a work in
progress - see https://github.com/qgis/QGIS/pull/51060#issuecomment-1332232828
for the current state/sticking point. We were blocked by missing PyQt6
binaries until Sandro Mani put in a great effort and created these, so
now we're just(?) waiting on someone with time/sponsorship available
to do some complex work adapting the python bindings to make them
QT5/6 compatible. That's probably a good 3-4 days work if we're lucky.

> I know we are already building core parts for Qt6, and I'm not aware of all Qt6 work (failed to get stuff compiled some months ago...).

Everything should work on current master versions, with the exceptions of:

- Python/all plugins
- Oracle provider
- oauth support
- everything related to webkit (ie html widgets, layout html items)
- GPS serial port devices

Almost all c++ tests are passing, but obviously Python tests can't be
run yet, so the overall stability is a bit of an unknown.

>
> I do understand that it will not be easy, and will probably cost a lot of energy/time/money...
> But as 3.28 will be LTR, I was wondering if we maybe could take that moment to (some idea's):
>
> - create a funding round for Qt6 (I'd happy to chip in for such 'labeled' funding)? It's the 'gift season' isn't it?  🎵"All I want for Christmas is 🎵 .... you and Qt6/QGIS6'🎵 ?
>
> - maybe gather as much as 'Qt6-able' people for the next Hackfest.. to give it a boost? Or maybe theme the hackfest a 'moving to Qt6' ? Even wilder idea: invite Yorik van Havre (Freecad) to the hackfest as the also have to move up....?
>
> - to do a minimal gui branch, in which we piece by piece pull ui functionality in (if I am correct one of the long standing wishes is to have an even better ui/logic separation)?

You actually can already run the full QGIS application (gui, 3d and
all) under Qt 6.

>
> I know: "idea's are cheap" Off course this will all come with a cost, I'm sort of asking if there is a plan to chop the work up in pieces so we can maybe (depending on money/time) grow to a QGIS6 ;-)

Well, the TLDR is that *someone* needs to spend the better part of a
week getting Qt6 Python support in place, whoever that is/however we
make it possible. :)

Thanks for keeping the discussion flowing!

Nyall


>
> And as I am just partially aware of the deep technical details, this email can be ignored if I'm too eager...
>
> Regards (and again: not meant to give pressure!!),
>
> Richard Duivenvoorde
>
> _______________________________________________
> Qgis-psc mailing list
> Qgis-psc at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/qgis-psc