News and Update
This session was hosted by Marius and Alfred.
Marius first explained that he has not been active in the community for several months and that has resulted from some tragic events in his family.
A mini news announcement followed from Alfred. Responsibility for the Ubports shop has moved and is now run by community member Moem. Items will be sent out soon, our apologies for the wait.
Coming events and development budget
Ubports will be represented at two upcoming events – the Festival of Software Libre and Ubucon. There are details in the show notes.
Alfred reminded us that as Ubports doesn’t have a proper development budget and relies on volunteers, we have to accommodate the private work that those in our community do.
Join Alfred personnel project
Since it relies on Qt6 it cannot be ported to Ubuntu Touch at the moment but in future it may be possible.
For more details, please visit GitHub page: https://github.com/fredldotme/Tide and Apps Apple Store.
Halium QG has been brewing for about a year and it now brings immense performance improvements. Qt provides basic functionality to put things on the screen. Controls, buttons, text. Since Halium QG works as a Qt plugin, there is nothing that developers have to do because their apps already sit on top of Qt.
Porters do need to enable it, if it would assist their build. There will be some documentation on how to do that soon. The Fairphone 4 and Pixel 3a already have it enabled. It is built for multicore use. With all the tweaks in place, the performance benchmark for UT on the FP4 has increased 10x! A video was shown which showcases the impressive results but note that the video was made before binary blobs and all the tweaks were in place. Performance is now even better.
Part of the changes to Halium made by Alfred harness the power of the Android binary blobs. Morph browser shows those changes strongly. For porters the top priority should always be stability and they should work on that first but if they have done that, there are performance tweaks like this they could work on too.
The benchmark rating on the FP4 is very stable at 200 milliseconds. If there are limiting factors they are somewhere else. The particular approach used features built into Android phones so they are not applicable to Mainline but there may be other approaches that Mainline developers could take.
Qtwebengine for UT has been improved a lot by Alfred. He has used vendor elements to provide hardware video acceleration. This works alongside the general Qt improvements to give an even bigger boost and that really ups the framerate. Before, 4K video wasn’t even possible but now it is pretty good at 60 fps. This brings us up to level of Android and iOS in terms of capabilities.
During development of those tweaks, Alfred discovered that Lomiri could sometimes crash as a result of memory leaks, so he has made some modifications to stop that happening.
Marius has been working on a stability issue where the cell provider is not the same as the cell tower operator. Data can drop when moving between the two. Unfortunately to fix it he has to go for a drive! Work on that is not finished but the changes made so far should help a lot of people..
Pixel 3a and Fairphone 4
The Pixel 3a and FP4 are the standout devices for UT but they are very different. The user repair and sustainability aspects of the FP4 make it quite a big, heavy slab, while the Pixel is light and compact.
Aethercast on UT
Rachanan and LionelD have been doing a lot of work.Utilisation of the APN configuration files from cell providers is one development. We use an upstream which is constantly updated. Marius has been working on assisted GPS, which is much faster because it uses local prompts in the form of distance from cell towers. We would prefer to work with more open systems like the Mozilla location service but for now we have to do it with what is available. At the moment that service doesn’t seem to be functional. It is sad that we have to rely on a Google service but for now we do, although it will be an opt-in. We may be able to use some sort of proxy, to access Google indirectly.
Ubuntu Touch development "roadmap"
The News section of our Forum is the best place to post questions for the Q&A. YouTube live chat,Telegram and Matrix are other places to post a question.
If you didn't know, the Forum questions get priority.
The first question was about agps and whether it needs data to get a fix. Yes, agps needs either cellular data or wifi.
Hugehead asked how the manpower issue in the dev team is going? Volunteers make up almost all of that. Only Rachanan and Marius get funding from the foundation so that they can work on core projects. There were some temporary paid contributions during the massive Focal upgrade work.
Transition to 20.04
Hugehead asked how did the transition to 20.04 go and what was learned during the process? The task was so immense that some things have broken along the way, so we have a new crop of bugs to address. That has underlined that we still don’t have enough people involved in formal testing. That slows us down. New volunteers at the Telegram QA group would make a big difference. We need systematic tests and proper reporting through the right communication channels. A lot has been learned and that has shown us where changes can be made and where improvements are possible. We survived and that demonstrates our underlying strength as a project. To make it happen at all we had to up our game on teamwork too.
Next upgrade 24.04?
Will you be going for 24.04? Yes. The changes we have already made simplify future upgrades this will hopefully avoid the ‘indigestion’ caused by the recent fundamental changes. We are now part of the Debian ecosystem so we are constantly updating packages to the latest versions. A majority of the content of 24.04 will already be in UT distributions. The compiler errors for those will already have been fixed.
Raspberry Pi 4 Aury88 announced in the forum that the transition to Focal also enables UT to be run on Raspberry Pi 4. Rather than a functional system, it can act as a testbed for UT functions and potential changes, feeding into the project. There are Forum and Telegram group for UT on Pi.
Mr10001 asked about Googles anti-rollback features. Will their action limit UT in the future? Alfred thought not. Those issues arise at bootloader level so before any device is chosen for porting it is essential to check that the bootloader can be disabled. There are some security benefits to the anti-rollback and as with other things we can probably integrate those features inside UT in a way that suits us but still gets the security protection. Concern about it seems to be overblown.
Arudy asked whether Gallery app will gain some edit features such as simple crop? What do you think about using third party apps for that? This is one of those things that would be great and would be warmly welcomed if some volunteers want to put it together. It is a ‘nice to have’ though, not a core priority.
Alfred has gained some insights into iOS with his project and he has found that Apple have created a an ecosystem a bit like our ContentHub, allowing apps to work together in a sandboxed way, with shared protocols. Some people like having a very basic gallery, so feature creep may not be welcomed by them. More is not absolutely better. Crop or resize would be basic features belonging in a simple Gallery app. Advanced processing and filters ought to go in a free-standing app.
Click app format
Are we going to stay with Click as our app package format? Alfred likes Snaps and has created a Snap enabler which is available from OpenStore. It doesn’t have much value on a phone but on a tablet it can start to make sense. Having said all that, Click is not a problem and not something we are desperate to get away from. Ideally we would just add formats and allow users to decide. The basic point though is that all formats should ‘just work’. Users should not have to struggle with them. The philosophy with UT is to aim it at non-technical users, not geeks.
Any thoughts on Widevine? Marius has done something (for ARM64 only). You can find it in OpenStore. It is still very experimental and certainly not something for your daily driver device. It patches a lot on the system including libC, so very low level. Netflix or Disney+ does work.
Snap on Ubuntu Touch
Going back to Snap, Alfred appealed to any Canonical devs watching to maybe take a look at his PRs.
On the subject of Snap, are we going to continue following LTS or might we switch to core? Just on the basis of the resources we have, the work involved in dropping LTS is something we could not take on. Increasing patching is not a direction we really want to go in.
Any news on FP5 for UT? This is a new phone. If anyone from Fairphone is listening in, we would love to support the new phone. It hasn’t shipped yet so we haven’t had an opportunity to look at it. Marius certainly wants one :)
See you next time :-)