[QGIS-Developer] QGIS 3.46 builds to use Qt6?

[Nyall Dawson]

On Tue, 17 Dec 2024 at 23:56, Even Rouault <even.rouault at spatialys.com>
wrote:
>
> I also believe verisonning Qt6 QGIS builds as 4.0 is the right thing to
> do for clarity and from a semantic versioning point of view. This is a
> major API break from plugin perspective.  We don't necessarily need to
> deal with removal of deprecated APIs at that time . That can be
> postponed for a QGIS 5.0. Version numbers are cheap ;-)

While I agree that it's probably correct in theory, my problem with this is
that if we bump to 4.0 now then we've burnt our version break chance for
the next ~5 years or so. This is from a social (not technical) point of
view -- we can't be seen breaking API frequently or it erodes trust in
QGIS' long term stability.

And I'm looking at the recent release of Qt 6.9 beta, and honestly
wondering how long it'll be before Qt 7! Qt 4 only reached 4.8, Qt 5
reached 5.15... it's not unreasonable to guess that we may already be past
the half-way mark before Qt 7 is a thing. 🤣

So that's why I keep leaning toward NOT bumping the version now, and
holding that card back till we REALLY need it (say in 2-3 years, when the
hassle of dragging around deprecated QGIS API is infeasible and we need to
move to Qt 7 or we want to move to GIL-less python or ???).

Nyall


>
> Related to that debate, years ago I wanted to call GDAL 2.5 what has
> finally been released as GDAL 3.0. From a GDAL perspective, the only
> difference between 2.4 and 2.5/3.0 was the PROJ 6 dependency and the
> mundane change that CRS were now authority axis aware by default. I
> don't regret the 2->3 bump because it turned to be a major change for
> users, and the bump was a clear sign they had to do something.
>
> Le 17/12/2024 à 09:17, Nyall Dawson via QGIS-Developer a écrit :
> > On Tue, 17 Dec 2024 at 18:00, Julien Cabieces
> > <julien.cabieces at oslandia.com> wrote:
> >>
> >> Hi,
> >>
> >> I'm also in favor of switching to Qt 6 builds and the timeline you
> >> propose Martin seems reasonnable to me.
> >>
> >> However, I think we should increment the QGIS major version to 4.0 in
order
> >> to stongly advertise our users and plugin developers that there is an
> >> API break. If we are not doing now, when do we plan to increment the
> >> major and clean the code from all deprecated content?
> > I'm SO strongly against breaking QGIS API unless we ABSOLUTELY have
> > to. The pain of the 3.0 transition is still relatively fresh in my
> > memory, and I think we would be doing our users and end user
> > organisations a great disservice by moving to QGIS 4.0 now.
> >
> > I think a 4.0 API bump should be an absolute last resort, when we have
> > no other option.
> >
> >> Regarding the proposal of having 2 versions (Qt 5 and Qt 6), I'm not
> >> completely sure this is a good idea. I'm afraid that plugin developer
> >> would delay the plugin migration if this is still possible to run a
QGIS
> >> with Qt 5.
> > Well, that's entirely their choice. But I'd imagine end user pressure
> > would result in the majority of frequently used plugins being upgraded
> > quite quickly. And if a particular plugin ISN'T compatible with the QT
> > 6 builds, then the user who requires that plugin could still run the
> > Qt 5 build for the foreseeable future.
> >
> >>   It would also be
> >> interesting to display on the plugins platform which are compatible
with
> >> Qt 6 and which are not.
> > Definitely -- that would help users make informed decisions as to
> > which is the best choice for them.
> >
> > Nyall
> >
> >> Regards,
> >> Julien
> >>
> >>
> >>
> >>> On Tue, 17 Dec 2024 at 10:48, Greg Troxel via QGIS-Developer <
qgis-developer at lists.osgeo.org> wrote:
> >>>
> >>>> I wonder about "less stable" qt5 and "more stable" qt6.  Do we really
> >>>> believe that qgis built on qt6, with no plugins will have fewer
crashes
> >>>> and quirks, than the qt5 build?
> >>> Yes, I do. Because Qt 5 is not improving any more, but Qt 6 is. An
example would be when running under Wayland environments on
> >>> linux -- it's a very broken mess on Qt 5 and will never be fixed. On
Qt 6 it's only a slightly-broken mess, and will likely be non-broken within
> >>> the next 12 months. There's a similar situation for apple processors,
which never had full official support on Qt 5 but ARE fully supported
> >>> on Qt 6. This gap is only getting wider as newer operating system
updates and corresponding changes break things underneath Qt 5.
> >>>
> >>> There's also a limited stream of bug fixes getting ported back to Qt
5.15, vs those flowing into the supported Qt 6 releases.
> >>>
> >>>>   That is surprising to me at this point.
> >>>> Do we still believe that if one assumes "qgis with N random plugins
that
> >>>> claim to support qt6"?
> >>> (Well, QGIS + **ANY** plugin = a less stable QGIS. 🤣 But that's a
completely different point)
> >>>
> >>>> I expect a qt6 build is kind of like a .0 release, and we would want
to
> >>>> have qt6 builds widel avaialable and time for feedback before saying
> >>>> it's stable.
> >>> I'd say we're well past the ".0" stage of Qt 6 support. Almost all
the core functionality is quite well tested at this stage, and third party
> >>> clients (like Mergin and QField) switched to Qt 6 based QGIS builds
earlier this year. I'm confident that by the time we hit a (potential)
> >>> October 2025 milestone that we'll have a very stable Qt 6 build
available.
> >>>
> >>> Nyall
> >>>
> >>>> _______________________________________________
> >>>> QGIS-Developer mailing list
> >>>> QGIS-Developer at lists.osgeo.org
> >>>> List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
> >>>> Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer
> >>> _______________________________________________
> >>> QGIS-Developer mailing list
> >>> QGIS-Developer at lists.osgeo.org
> >>> List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
> >>> Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer
> >> --
> >>
> >> Julien Cabieces
> >> Senior Developer at Oslandia
> >> julien.cabieces at oslandia.com
> > _______________________________________________
> > QGIS-Developer mailing list
> > QGIS-Developer at lists.osgeo.org
> > List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
> > Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer
>
> --
> http://www.spatialys.com
> My software is free, but my time generally not.
> Butcher of all kinds of standards, open or closed formats. At the end,
this is just about bytes.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/qgis-developer/attachments/20241218/d7f90dc9/attachment.htm>