Ubuntu Touch Q&A 30

All focused on OTA-4
19 June 2018

 
 

Looking for the Audio-only version of the Q&A?

You're just one big orange button away -->

Show Notes

 

News first.

Florian announces his newborn son Lukas. He will be a UT user!

Jan has been pushing Marius to fix things! There are still some issues before RC and 16.04 release.

Marius says progress has been good. Some extremely tricky issues had to be solved in very obscure packages, such as Content Hub. Only one critical issue left before RC. That is the camera issue on BQ devices and Pro 5. RC probably within days.

Jan: Some really advanced stuff needed doing in obscure areas, so this was the most difficult phase so far. Project Sudoku as we called it started 10 months ago. Over 72 issues had to be fixed.

Marius. Only started on the latest phase seriously after Christmas. Every single package has been modified! Canonical did a lot of work on Xenial but it was still in a terrible state.

Jan. We also did experiments, such as Anbox, to sketch out some ideas for the future of Xenial, before we actually got there. Soon we will activate a more normal testing cycle. A week ago we broke all the devel channel builds. That will not happen again.

Marius. We can now see the end of the tunnel. All have had personal issues and that has complicated things even more. OTA-4 now looks like end of June but there are unpredictable items. They can take 2 days or 2 months. Depends. We have the only fully open phone OS. We are the first to do it.

Florian. The OpenStore needed some huge changes. Many thanks to Brian. Testing under RC will be a lot easier than testing on Devel. As an example, one build exceeded the partition size on OnePlus One and 'broke' it for all those testers. Be careful if the phone is your daily driver. Devel is for fun and testing!

Jan: Of course we are not talking about bricking. You can get everything back with a reinstall, including data etc.

Florian: Microsoft bought Github. There has been a lot of discussion

Jan. This is very controversial. Some say Microsoft is the enemy of open source, some say they have changed. An option would be Gitlab. There is a very active discussion on the forum. UBports is not going to make any immediate decisions. There will be a continuing debate in the community and a proper risk assessment.

Florian. It is about what the community wants.

Jan. Gitlab has some great features that Github doesn't have, especially in project management. The biggest downside is that the user base is much smaller. If the developer ecology does not come with us, that would be very difficult for us.

Marius. Debian and Gnome have moved to Gitlab but they are way bigger than us and they moved from ancient developer architectures, so for them the situation was very different.

Jan. Our build system works but we have not polished the setup as much as we could on Github. If we have to tidy anyway, a move might not be that much more disruptive. Anyway, nothing before OTA-4 is settled. It would be a distraction.

Marius. Please tell us, if you are a current Github user - would you leave with us or stay there if we left ?

Florian. Some people have asked about self-hosting. That would take money, time, getting up in the middle of the night, backups etc etc. So we would probably not want to risk that. The most freedom but also the most risk.

Marius. Need to add to search engines. Also you need a login system etc etc. The benefits are not obvious, especially if we use a free service anyway.

Florian. There are some great new community initiatives. The first is Raspberry pi project. Maybe a test build, maybe for Internet of Things but if you add a touch screen, perhaps also some kind of communication device. There is a forum for this now, under Alternative Projects. Also a Github page The great thing is that this grew independently out of the community, showing how vibrant the community has become.

A second project came out of a UBports meeting in Munich. Some guys there got interested in porting the Mycroft voice assistant to Ubuntu Touch. They too have a Mycroft page on the forum. It exists as standalone software. There is a demo click package already. You can test it out. It will go to the OpenStaore sometime. We are always open to more community projects. For the core developer team who don't have time to do these things, it raises their morale to see something happening outside of their routine.

Jan. He was planning to go the Munich meet but then could not, so he felt rather bad about it. He is pleased that the meeting went so well.

Florian and Marius are skeptical about voice assistants day to day. It will be interesting to see if it finds some really valuable purpose with users. Our version will not be 'listening all the time' of course. Even if we removed containment somehow, the interaction would only be local, no 'phoning home' like the Google equivalent.

Questions now

Dohnix asks how we can keep up with so much stuff on Telegram. Jan says he gave up a year ago! He swings by occasionally. In the forum or in Github are good for permanent and non-synchronous discussions

Florian says that Supergroup is general and non persistent. You can't follow everything. In the forum you can select more. Scrolling in Telegram is chaotic. Big thanks to Rodney who answers everything 24/7 !

Mailing lists never quite took off.

Florian can sometimes keep up but sometimes jumps hundreds of messages. What Telegram is great for is building relations with users. It is the marketplace where you can ask for directions.

Hummelbach asks if they are bored with regressions? Marius says they are not fun but they have to be done.

He also asks how general users can help out. There is a list of good issues on the Github tracker for new developers. These are quite easy, not urgent. Those are not regressions, they are features, refinements, non-critical issues. Jan is constantly adding suggestions. For example, a simplification of the UI in the dialer. Good for a beginner with QML.

Marius answers a question about removal of Mir support in LightDM 1.28. This is not an issue for us. The newest version of Mir supports a thing called LoginD which is a daemon which authorizes hardware such as keyboard. With that support, we have the freedom to use any display manager we want, if it supports Wayland. We will be able to do that natively. Actually it was a weird container within a container arrangement before, so the new way simplifies.

Dohnix wants to know if there will be apps that will run equally on UBports and Purism's phone OS? Marius says that the app frameworks are not finalized, so we don't know exactly. Of course, any of our apps will work because they are simply Qt with UTtoolkit running on top. The obstacle for us is really GTK, which doesn't play nice with Mir. With improved Mir integration with Wayland, we would expect some of the GTK issues to be solved. It won't matter until some time next year though, so it isn't urgent.

Dohnix also wants to know if we should move away from UT SDK and use QtQuick instead. Florian and Marius acknowledge that this would be a more universal approach, since the SDK is relevant only for UBports whereas QTQuick can be used by other groups. Over time, we want to bring our stuff to the desktop world and to other phone systems, so that could provide a bridge. UTtoolkit works very nicely on Arch by the way, built by Ivan.Some progress with Mir and Unity8 on Arch too, which is extremely cool! For the future, we want to be closer to the upstream and to what Qt on desktop users are doing. We don't want to scrap UTtoolkit though. Lots of app developers use it and actually it works well and is useful. On the contrary, we would like to strip out some of the dependencies so that it will run nicely on other platforms.By taking out the Content Hub and the phone stack, we can make it something universal. Jan points out that we are already bringing it to Debian.

Dohnix asks how 16.04 is coming along. Jan says "Nicely!"

On channels, Jan corrects the impression that Vivid RC will become Xenial RC somehow. Vivid RC will stay as Vivid. If you are on that,the changeover will not happen until OTA4. To clarify, for a time, there will be six channels. Florian says it will all be explained exactly later.

Jan says some milestones will be unresolved even after OTA4 but those are non-critical. Core team are now impatient and want to move on from the hard work of Xenial stable, to start providing some fun stuff to the community again.

Things preventing Tomas Oqvist switching to Xenial devel now - camera app regression, calendar sometimes crashing. Also will all devices go to 16.04 at the same time. Florian says some users already go onto Xenial for the weekend, then back to 15.04 for Monday! At least that avoids disruptive crashes. Marius says an easier way to do that is not to check for upgrades at all during the week but then get the updates at the weekend. If they crash, it doesn't matter so much. Jan says that even if you have to reinstall, you will not lose data unless you click wipe.

Jan just remembered something important from the news section. Having different release numbers for different devices gets really confusing, so for devel the new name system will be the day, for RC releases it will be the week and for stable it will be just as now, 1,2,3,4 etc. In future, all devices will use the same name for each release. So, '12 July 2018', 'week24 2018', 'OTA5'. QA will be so much easier!

Now a question about New Browser. Will there be more improvement before OTA-5 and will it become possible to create webapps that use it ?

Marius says yes there will be improvements. UI issues such as scaling and keyboard popup need to be soon, maybe at OTA4. We want to get rid of crashes by OTA4.

Jan says it has no core developer. The new one uses QML so it is very open to community developers to do a lot.

Will it support progressive webapps, [service work *?] processes or javascript, native system notifications and push? Wow says Florian "A lot!" Jan says at the moment, no. On Tor support, Marius says it could be done. It is for a developer to come forward and do that but it is not seen as a core aim of UBports. Jan says the same is true of Orbot, for system wide use. Just needs someone to volunteer. It can be done.

A suggestion is made by a user to break with the OTA approach and move to debs instead. Also can the ARM apps for PC be added to what is available?

Marius responds that there is something special about the fact that we are developing a phone. It is essential that it works. It even has an emergency use. You have to rely on it completely, so it must be very solid. The OTA approach does that because basically everyone gets the same system. If you rely on debs, everyone has a different setup and then the issues with conflicts etc multiply. We still use the Android base, with differences for every device and with multiple different partitions. Holding all of that together requires a standard approach, which we get with OTA. It is impossible to update the kernel for debs. No exceptions. Rebooting to Recovery is not an optional. It is the only way it works. Debs on PC have been tested and tested for 30 years. Not on phones. Experimental users can play with deb of course but we build for casual users. How we do it can be tweaked but it has to be read only and image based. What we have talked about is simplifying deb use with a purpose made button, so you can choose.

Jan says that once 16.04 is stable, we will have a big discussion in the community about our next priorities and this will be one of the issues to be discussed. Also Anbox, systemd and a load of other stuff.

Marius points out that you can use debs now, the thing you don't have is the user interface. But those asking about debs don't need that anyway!

Florian says of ARM apps that scaling and UI are the big issues and again, do not expect core developers to put effort into that. If the community wants it, it can do that work. So LibreOffice for example will not be a mainstream project but could be done. Do it as a community project, as with Mycroft.

The team builds a big sandcastle in the middle but you must find your own sand and build the extensions to it.

Will a Print to pdf mechanism be provided. Florian says yes. At first it will just convert to print but later it will convert toi save. It will also not be able to capture screenshots and web pages as pdf straight away but that will come. OTA5 is a possible for some pdf functionality to be incorporated in the system. First will come printing a pdf that you downloaded.

Marius says that there will be a fix soon for an error where xenial installs vivid app versions as an 'upgrade'. By the time of RC, there will be a fix to stop that happening and messing up e.g. Telegram. Just re-issuing Telegram unchanged with a higher version number will fix that for now.

Finally, there was some criticism of lighting, angles etc. in the Q&A. The response was that the team are most interested in doing basic work on UT. Perfecting video skills are not something interesting or worth spending time and money on. You can always listen to sound only if you want!

Raspberry Pi Project https://forums.ubports.com/category/43/ut-for-raspberry-pi

Mycroft on UT https://forums.ubports.com/topic/1329/mycroft-voice-assistant-demo-app

Discussion about moving out or not of GitHub https://forums.ubports.com/topic/1317/move-from-github-to-gitlab