Would distributions have an issue if KF 5.77+ would require Qt >= 5.14 (instead of >= 5.13 as of now)?

[Nicolas Fella]

Hi,

with my KDE Android hat on this would be fine, we use Qt 5.15. For KDE's
own Windows/Mac builds I'd expect it to be similar.

What most/all not-traditional-Linux-distro users have in common is that
they are not bound to the specific Qt version decided and shipped by a
vendor and instead either build themselves or download official binaries.

The two major "distributions" that are potential frameworks users but
can't because of Qt versions (SailfishOS and Ubuntu Touch) don't ship
5.13 either, so it doesn't make the problem worse.

Cheers

Nico


On 12/1/20 1:29 PM, Jonathan Riddell wrote:
> Not from KDE neon of course, we're on 5.15.  And not from the KDE
> snaps build either.  But I suppose there's more than just Linux
> distros to consider as we ship apps using KDE frameworks on Flatpak,
> Android, Windows, even Mac to ponder too.
>
> Jonathan
>
>
> On Tue, 1 Dec 2020 at 12:14, Friedrich W. H. Kossebau
> <kossebau at kde.org <mailto:kossebau at kde.org>> wrote:
>
>     Hi,
>
>     last week KDE Frameworks master saw a bump in the
>     required/expected minimal Qt
>     version to Qt 5.13, following rules once agreed and noted here:
>     https://community.kde.org/Frameworks/Policies#Frameworks_Qt_requirements
>     <https://community.kde.org/Frameworks/Policies#Frameworks_Qt_requirements>
>
>     I would like to challenge that former decision though and propose
>     to instead
>     go straight to Qt 5.14 as minimum requirement now.
>
>
>     QUESTION:
>     Would any of the distributions have an issue with requiring Qt
>     5.14 instead of
>     Qt 5.13?
>
>
>     From some quick checks using https://repology.org/
>     <https://repology.org/> it seems that any
>     distribution versions which currently use Qt 5.13 have also
>     settled on some
>     older KF version, so will not update to just KF 5.77 and thus be
>     screwed.
>
>     Motivation:
>     * KDE CI not setup ATM to cover builds with Qt 5.13 (no build, no
>     unit tests)
>     * Qt 5.14 added some new API, chance to miss out when using that
>     in new code
>     * C++: no need to write #if QT_VERSION < QT_VERSION_CHECK(5, 14,
>     0) variants
>     * QML: no need to do hard-to-read generation tricks to support <
>     Qt 5.14
>     * Qt 5.13 went out-of-support in June
>     * App bundle builders would rather use some recent Qt 5.14/5.15
>
>     So by restraining to Qt 5.13 as minimum version IMHO we would
>     make/keep life
>     complicated for KF contributors without adding any value for anyone.
>
>     With most of KDE Frameworks in my local checkout:
>         grep "QT_VERSION_CHECK(5, 14, 0)"  frameworks/*/src -r
>     2>/dev/null | \
>             grep "QT_VERSION " | wc -l
>     gives me "92", so there are quite some code variants which need
>     support in
>     current code.
>
>     From the emails at least in
>     https://mail.kde.org/pipermail/kde-frameworks-devel/2020-July/112712.html
>     <https://mail.kde.org/pipermail/kde-frameworks-devel/2020-July/112712.html>
>     I could not see a discussion whether Qt 5.13 makes
>     sense at all now, seems mainly the algorithm was applied. I
>     propose to match
>     the result to known real world needs now. Or teach me what I have
>     missed here
>     :)
>
>     Cheers
>     Friedrich
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kde-frameworks-devel/attachments/20201201/8b6ae7a2/attachment.htm>