Convergent UI for Linux with Qt Quick or Kirigami

[bam80]

There was a discussion on Flathub repo:
flathub/app.organicmaps.desktop#26 (comment)

It's better keep it here.

[RedAuburn]

if we're doing a convergent UI, wouldn't gtk (+libadwaita?) be a good choice? especially with all the work making gtk apps scale nicely for mobile

[biodranik]

I’m familiar with QML and its pros and cons. JavaScript’s performance and battery consumption are significant cons. It contradicts the current OM idea of being blazing fast and battery-friendly.

[baarkerlounger]

GTK + Libadwaita would be particularly nice for mobile Linux since Phosh (GTK) is by far the most popular shell for mobile linux currently.

[biodranik]

It would be great to focus on the universal, multiplatform experience, not only on Linux.

[bam80]

GTK + Libadwaita would be particularly nice for mobile Linux since Phosh (GTK) is by far the most popular shell for mobile linux currently.

That would be a separate project I guess, as I personally don't have willingness to port to GTK things, and don't have experience either.
Also, I'm not sure if GTK + Libadwaita solution is doable for Android.

So let's dedicate this discussion to QML UI. If someone has ability to implement GTK port, or discuss which is better - please crate a separate discussion.

[pravi]

Just wanted to mention there is draft pr to support building gtk apps on android https://gitlab.gnome.org/GNOME/gtk/-/merge_requests/7555 but as you suggested we can discuss that on a separate issue if anyone is interested in exploring that.

[bam80]

I found an interested developer for this task.
Could it be accepted as GSoC 2023 project?
I could be a mentor.

@biodranik

[biodranik]

Raising a bounty only for a specific feature is ok. As rewriting UI in a user-friendly way while keeping the desktop version convenient for development is a complex/large/mid-long term task that will compete with other prioritized tasks, it may be more efficient to cover also testing/reviewing resources in addition to development time.

[pravi]

Thanks. Can we start with a rough estimate for time requirement and then based on that an initial budget estimation? We can include testing/reviewing resources in the budget. If current core team is not able to dedicate resources, we can try to find developers / freelancers from outside. We can also explore possibility of getting funds from organizations that generally fund Free Software.

[bam80]

Hi @pravi, thank you for your interest.
I'm the one who originated this initiative. As you can see, it was done mostly publicly, but we also had few private conversations with the team to coordinate our efforts. At the time, we estimated the project roughly 6 months.
I'm still interested in this, and can act as a developer, too.
Currently, I'm searching for funds, so your interest is highly welcomed!

Looking forward for the cooperation,
thank you.

[pravi]

Since we have an interested developer already, next step would be to setup a targeted fund raising project at opencollective.com/organicmaps We can explain about the task, who is going to work on it, benefits of this project etc. Once we have this setup, we can promote this in our communities and ask people to donate.

[pravi]

@bam80 @biodranik we can try to get some funds from https://floss.fund/blog/announcing-floss-fund/

@biodranik can we create a project on organic maps opencollective page for this ? (we can do that in parallel to floss.fund application).

[bam80]

We could borrow Pure Maps approach which supports either QtControls or Kirigami at build time:
https://github.com/rinigus/pure-maps#building

Also it splits QML code into platform-specific and platform-independent parts. We could use it to unify QML support on both Android and Linux (again, optionally):
https://github.com/rinigus/pure-maps#platforms

[bam80]

Does OMap do it's map rendering with OpenGL?
Seems it's not easy to overlay QML UI on top of non-OpenGL-based scene:
https://stackoverflow.com/questions/27440157/qt-widgets-mixed-with-qml-on-top

UPDATE:
OK, according to #2947, it's OpenGL(ES) already, so shouldn't be a problem.

[biodranik]

What are the pros and cons? In the ideal world, the UI is written once, works on all desktop and related mobile platforms without any additional exceptions (excluding iOS and Android, of course), is blazing fast, user-friendly, and easy to support.

[bam80]

Pros and cons of supporting both of Qt Widget and QML UI?
I think once QML UI gets mature, there would be no good reason to support others.

Kirigami apps are good citizens on Android, too. There is a showcase and other apps from KDE Community:
https://play.google.com/store/apps/details?id=org.kde.kirigamigallery&hl=en_US&pli=1

[bam80]

I wrote a text to be included in the GSoC 2023 ideas Wiki.
Feel free to make a suggestion/corrections, if needed:
https://github.com/orgs/organicmaps/discussions/4342#discussioncomment-5453522