[lubuntu-devel] [Lubuntu 17.04 Discussion] LXQt

[Julien Lavergne]

Hi,

2016-10-29 1:02 GMT+02:00 Simon Quigley <tsimonq2 at lubuntu.me>:
> Hi everyone,
>
> So I'd like to talk about LXQt. Initially, if I remember correctly, our
> plan was to at the very minimum have an LXQt image for 16.10, but that
> didn't happen due to trouble with the default settings being applied
> incorrectly, a bug which has been fixed in Qt.

Yes the plan was to have an image, so at least we can compare LXQt
image with the LXDE image. I think it's the very minim before even
thinking of switching.


> So here's a progress report of my recent work with that and different
> pieces of the puzzle which have come together.
>
> I worked with the Debian KDE/Qt team to get that patch in their
> packaging for Qt 5.7.1 (which Ubuntu pulls from) and will be included in
> the next Qt upload.

Finally, we will be able to apply some custom settings ... That's also
a need item. Thanks for this.

>
> I also talked to Timo Jyrinki, the usual Qt uploader, and Qt 5.7.1 will
> begin the staging process as soon as it's released. So the faster we
> help him get that into the archive, the better. Until then, once
> uploaded to the usual Qt staging PPA, we could rebuild all of LXQt
> against it in a PPA (which would have a dependency set on the Qt PPA)
> and do any needed modifications in order to get those default settings
> and everything else done and ready to go, so as soon as Qt 5.7.1 lands
> in zesty-release, we can get that in the archive and images building.

I'm sure Debian LXQt team will also have a look at the build test with
Qt 5.7.1. That's should be harmless.

>
> Lastly, last I heard, LXQt 0.11 is currently in Debian Sid, and Julien
> and myself have started discussions for getting that in Ubuntu. There's
> a bug report open against liblxqt[1] (which is the first step in getting
> a new version of LXQt in the archive) which we were discussing. Today I
> plan to look over what's left for that and see what to do from there.
> The Lubuntu Packages Team is subscribed to that if you wish to take a
> look at our progress there.

As I noted on the bug report, it needs libqtxdg sync before everything
else. Do you build test it on zesty ?

>
> After liblxqt is synced, the rest of the components (with an exception)
> can be synced from Debian. The exception to that is lxqt-session, which
> needs to be manually synced due to some changes Julien made in Ubuntu
> that upstream won't accept.
>
> Either way, automatic autosyncing from Debian hasn't been turned on yet
> due to a Perl transition in Ubuntu. The Ubuntu Release Team hasn't
> automatically turned that on because they don't want MoM to blow up. So
> Julien, if you could look at the LXQt packages that could be synced,
> and, well, sync them, that would be great.

Like libqtxdg, they need to be build test before (like in a zesty
pbuilder). If you want to speed up the process, you can do this test.

> So it seems we have the following items left before we can think about
> moving Lubuntu to LXQt:
>  1. Get Qt 5.7.1 in the archive.
>  2. Apply new default settings and separate that into a separate binary
> package.
>  3. Do some minor tweaks to the seed that will become apparent later
> (I'm not remembering at the moment).
>  4. Get a daily image building.
>  5. Get the Lubuntu QA team to thoroughly test it and fix any errors
> that arise.
>  6. Ask for testing and public opinion on the packages that we have
> installed by default.
>  7. While #6 is going on, we should consider looking at the Ubiquity
> slideshow and removing any KDE-specific components.
>  8. Decide on a final set of packages.
>  9. Get Package QA tests written and approved by the Lubuntu QA Team.
>  10. One last round of thorough testing by the QA team.
>
> The main question I have here is, do you think that we can get
> everything on this list done before 17.04 is released (keeping in mind
> archive freezes)? At the very minimum, does that TODO list sound
> reasonable enough to get done before 17.04 and at least provide an image?
>
> If not, what timeline do you propose?

I'm more for the option 2 : Get the most TODO items on the list, so we
get an image to compare. Remember that it needs to be as good as the
LXDE version to be candidate for the switch. That mean we have to look
at the performance, features, usability ... versus the cost of the
switch (all the documentation, tutorial on the web, the habit of the
users will have to change ...).

> There's no doubt in my mind that LXQt is stable enough for daily use.
> I've been using it for months now with no major problems, and I've heard
> LXQt 0.11 will be even better than that. So in my honest opinion after
> using it for months, it is ready for Lubuntu to switch to. The only
> concern I might have is the lack of a development team in LXQt, and once
> we get more public interest, given collaboration with the LXQt team on
> making contributions easy, I don't see a problem there.

I'm not worry about the stability of LXQt. The 0.11 version should be
OK. However, I'm more worried about the others packages that we have
to add to make a full distribution. That's why it's important to get
an image, to see the gap between the LXDE version, and the LXQt
version.

To be honest, I'm not very confident with this switch anymore. Years
ago, I was a great fan of the LXQt initiative. But with some time, we
see the similar problems (release cycle, getting all the desktop
features, translation system ...), and the Qt-5-only ecosystem is not
very big. Also, one of the reason to replace LXDE was that it's
unmaintained, but that's not entirely true (there are still recent
commits on upstream git). So for now, I'm very undecided. I really
wait for the image testing result to see the among of work we have to
put to make it a real distribution.

Finally, there is also an option possible : the split. As we may
consider that the LXQt version is not good enough to replace the LXDE
version, or that the LXDE version is still good to go), that doesn't
mean this version doesn't deserve to exist on its own.

So, to conclude, I think we have to do the TODO list Simon described,
as far as we can, and let discuss it again with 2 images ready to run.

Regards,
Julien Lavergne