@DJac

Sur UT 24.04-2.x?

@DJac said:

En pratique, l appli est bonnes, mais je n ai pas accès aux messages :
La boite de reception ne correspond pas au contenu de celle du site. je n y trouve que les messages 'officiels' de blablacar et aucun des messages de mes passagers.

Je serai intéressé de savoir ce que donne l'application sur UT 24.04-2.x si quelqu'un test à un moment ou à un autre! Est-ce que la messagerie fonctionne? ( Ca indiquerai que l'app bénéficie du nouveau moteur Qtwebengine 6 )

@domubpkm said:

@pparent You asked to @naz.r why he use XWayland on Mererhyvo and no problem with the virtual keyboard with 24.04-2.x.

Cos he implemented his own keyboard inside the app, and never has used the Ubuntu Touch Keyboard in the first place!

@klh said:

Launching apps outside lomiri-app-launch is not going to help debug things, you are introducing a whole lot of additional uncertainties. I asked about immodules to know where to look, not assuming it's the reason.

Then we can always create a .click that will launch and XWayland Qml app. I suspect it will be the same result, but it could be tested to be sure.

@klh said:

Someone will look into it.

Great!

@klh said:

@pparent maybe inform the people you are asking to run random commands that this will run said app unconfined.

I've added a warning notice on my previous posts. Thank you for reporting!

@brenno.almeida said:

@pparent did you use the restart button in system settings? If so its a bug that needs to be fixed. That button should remove everything from your system. If you remember, give me more details, I'll try to reproduce it.

Anyway the only file outside the app data/cache/config directory is a unit file in your user systemd directory, you can see it by running sysremctl --user status greenline.service

Thank you for your reply, and sorry for my late reply.

What I recall having done is disconnecting my account from the greeline menu, I dont recall the exact button name. But when opening the app it proposed me to make the association again with a whatsapp account

then I have found a solution to the problem:

rm  ~/.config/systemd/user/greenline.service

The file was left on the system.

I hope to have more time later to do more testing, but right now it's a bit complicated!

Thank you for your work!

@Ida_ said:

@pparent I tried your little experiment with Maps Exporter. The app launched in an x11 window, and the OSK was gone! This was a fresh install for me as I don't use Google Maps, but I verified that the OSK worked first when I launched it as a click the usual way by pressing the icon.

Ok so when you launch "Maps exporter" it in XWayland mode with the commands above you have no keyboard. Can you confirm it still is the case after refocusing on the XWayland window created by the adb/terminal commands ( by switching to another windows and going back, or by locking and unlocking the phone )?

@ida_ Thank you very much!

@klh said:
immodules.cache hack

It seems to show that the bug is not specific to the "immodules.cache" configuration or even to maliit-inputcontext-gtk, but concerns any XWayland window using maliit.

@Ida_

Thank you very much!

Unfortunately, at the moment, this does not help me understand what happens on 24.04-2.x

@klh said:

immodules.cache hack

Here is an interesting test that can be done to see if the bug is specific to the use of immodules.cache or not. It is possible to start a QML app in XWayland mode.

Here is an example on how to do it with "Maps exporter", but it can be adapted with any QML app: You can enter those commands in abd or in the terminal

WARNING: this will run the app unconfined, only do that with an app that you fully trust, only for testing purposes, and run it preferably only on a testing device with no personal data rather than on your daily driver

export QT_QPA_PLATFORM=xcb
export DISPLAY=:0
export QT_SCALE_FACTOR=2 
export QT_IM_MODULE=maliit
cd /opt/click.ubuntu.com/mapsexporter.pparent/current/
qmlscene app/Main.qml

What I observe on 24.04-1.x is that the application starts in XWayland mode and the On-Screen-Keyboard do work, but sometimes it works only after refocusing on the window: by switching to another windows and going back, or by locking and unlocking the phone.(This points once again towards something working wierdly with the focus detection of XWayland windows)

It would be extremely interesting to know how this behaves on 24.04-2.x, if someone can test. Can we get the OSK with a QML app sarted in XWayland mode (with the above commands)?

Also I'm wondering if on existing XWayland apps ( like uWolf, or Min browser or Signal UT ) refocusing on the window ( by switching to another windows and going back, or by locking and unlocking the phone ) could possibly allow to have the OSK appear again in 24.04-2.x , in any way shape or form?

@messayisto @gpatel-fr @equareo @ida_ ?

@Ida_

Thank you! What you describe is kind of expected. Currently it seems the keyboard does not work with any XWayland app on 24.04-2.x. Normally the messages are sent by typing Enter/Return on the virtual keyboard.

I could help if you give the output when you run these adb commands, and try to open the keyboard (click on the edit area)

WARNING: this will run Signal UT in an unconfined way, only do that if you fully trust Signal UT, only for testing purposes, and run it preferably only on a testing device with no personal data rather than on your daily driver

cd /opt/click.ubuntu.com/signalut.pparent/current/
export DISPLAY=:0
export LD_LIBRARY_PATH=lib/aarch64-linux-gnu/
export MALIIT_DEBUG=1 
export G_MESSAGES_DEBUG=all 
./launcher.sh

@gpatel-fr said:

min starts on 26.02 build 104. Log: minlog.txt

the repeated lines at the end were happening when I was tapping in the URL bar.

It starts but not OSK right?

I see on your logs and on the logs of @equareo that line:

qml: Application requested te be focused but no window for it. What should we do?

So I think your hypothesis might be right! Given that a patch was added in Maliit to "Verify an application has focus before granting it input", it seems possible that the XWayland windows are not detected correctly in Lomiri as giving the focus to the app and therefore block the virtual keyboard. This is only an hypothesis but seems very possible.

https://gitlab.com/ubports/development/core/packaging/maliit-framework/-/commit/5a767263a7d82b6194ad4c7f8548e8895bfb5d47

https://gitlab.com/ubports/development/core/packaging/maliit-framework/-/commit/6bd655621fea66af7b9bb3a477b9de072b4939db

It makes me think as well that on the left panel we still have the X icon instead of the app icon, which seems a bit strange to me as the window does now have the right icon under it, on the spread view. But it might be another sign that the XWayland apps might never really get the focus, from Lomiri standpoint.

@messayisto said:

Min Broeser is crashing after it started,

@messayisto
Side note: the latest version of Min browser should not crash anymore on 24.04-2.x. (But most probably the OSK will not work anymore, like on other XWayland apps).

@gpatel-fr

For fun I tried to replace in the launcher.sh for the piano app of @pparent the qmlscene6 (provided by the app, qt6 is packaged into the app) by the system one and it failed miserably so at the moment I think that QT6 is not available for apps, at least not in a straightforward way)

Possibly it is because you are using the qmlscene6 from the system but still the Qt6 library bundled into the app. It might not be compatible. You may try to remove the libraries form the app repository (or temporarily move them elsewhere) to see if it works better.

@Equareo

Do you have more detailed logs? Could you post your

journalctl -r

after trying to open the keyboard in Signal UT?

Thank you!

EDIT: Here is a way to get more log with Signal UT, if someone can post the logs coming from those adb commands, it could be useful. ( It will start Signal and you can try to open the keyboard )

WARNING: this will run Signal UT in an unconfined way, only do that if you fully trust Signal UT, only for testing purposes, and run it preferably only on a testing device with no personal data rather than on your daily driver

cd /opt/click.ubuntu.com/signalut.pparent/current/
export DISPLAY=:0
export LD_LIBRARY_PATH=lib/aarch64-linux-gnu/
export MALIIT_DEBUG=1 
export G_MESSAGES_DEBUG=all 
./launcher.sh

@messayisto said:

Min Broeser is crashing after it st

This is normal I will realease very soon (possibly tomorow) a new version that will not crash with 24.04-2.x, this is another bug that I can and will fix easily, that should currently only impact Min and Rocket.Chat .

The real problem is the OSK, I guess if it does not work with uWolf, it is the same with Signal UT or Whatsnew

Thank's

@pparent have you reported the issue anywhere? Not sure if anyone worked on that. Did this testing result in >any logs or anything that could point us to a cause? Are these apps still using my immodules.cache hack from >back in xenial?

I did not open an issue, because I did not have the time to really dig. But some users did report me the issue about my apps.

But I remember we discussed with @gpatel-fr about this issue on this forum, and I think he had an hypothesis about a commit that might have provoked it, but I did not have the time to dig. I will try to find again this topic!

@klh said:

Are these apps still using my immodules.cache hack from back in xenial?

Possibly yes. My apps are indeed using a immodules.cache file in GTK_IM_MODULE_FILE to point to the maliit-inputcontext-gtk library path form the app. Its probable that the other apps from other authors do the same.

Edit:

@gpatel-fr said:

this could be linked to this change:

https://gitlab.com/ubports/development/core/packaging/maliit-framework/-/commit/6bd655621fea66af7b9bb3a477b9de072b4939db

if it is, you may find warnings such as 'Application attempted to activate itself without focus' in the journal.

https://forums.ubports.com/post/93101

If anyone can report the current status of the app on 24.04-2.x, it would help, as I currently don't have a testing device compatible with it!

https://forums.ubports.com/topic/12379/ubuntu-touch-24.04-2.0-call-for-testing/10

Thank's!

@lsitongia

Unfortunately I cannot really test now because my testing device is Oneplus Nord N10 5G, and I can't risk my daily driver at the moment.

My concern is about XWayland apps, is there still the problem of OSK not showing up on these apps ? (uWolf, uFirefox, Chromium for UT, Signal UT, Whatsnew, Min Browser, Rocket.chat)? I say that because I could test previous revisions of 24.04-2.x on my device and those app could not access the OSK anymore. Other people reported the same.

Can any one report the current status of these apps? Has it improved or is it still the same?

( I feel it would be a shame to loose usage of these apps with upgrade to 24.04-2.x, they are currently indispensable to my personal daily usage on my volaphone 22 ).

Thank you!

@gpatel-fr said:

@pparent

Sorry for the aside, but what is the status for using Signal from a desktop app (without having a phone number) ?

Well I guess it will probably be a reality one day, but for now we have to be patient, as no official announcement has been made about if and when a standalone desktop version (Without needing Android or iOS) will be available.

Also from my understanding it would not be without needing a phone number. I think you will still need to receive a code by SMS. Just that you won't need to install Signal on an Android/iOS smartphone, and you will be able to directly make the installation process (that requires SMS authentication) on your desktop (or in this case on you Ubuntu Touch).

It seems Signal-Desktop will support linux-arm64 officially in upcoming releases!

https://github.com/signalapp/Signal-Desktop/commit/8e1728ff634734a5e30fd314ab4c246434f0aa2e

Which is a nice to have! (That does not exclude the need to patch the code for use on UT )

@domubpkm said:

Note that in 24.04-1.2 dev, the problem is exactly the same on volla 22. And my perception is that the problem is magnified

Do you mean in 24-04-2.x?

Now I'm that I'm running the script it's much better for me. But I still observe the following:

-Sometimes the connexion is abnormally slow, I'm not sure if it comes from the operator, the coverage or the phone. Now the priority is maybe given to 5G...

-LTE does not reconnect properly most of the time after a connexion loss.