Unity 8 Desktop Still Planned In Time For Ubuntu 16.04 LTS

[zoomblab]

1. Applications decoupled from the OS updates. Traditionally a given release of Ubuntu has shipped with the versions of the applications available at the time of release. Important updates and security fixes are back-ported to older releases where required, but generally you had to wait for the next release to get the latest and greatest set of applications. The new desktop packaging system means that application developers can push updates out when they are ready and the user can benefit right away.

2. Application isolation. Traditionally applications can access anything the user can access; photos, documents, hardware devices, etc. On other platforms this has led to data being stolen or rendered otherwise unusable. Isolation means that without explicit permission any Click packaged application is prevented from accessing data you don’t want it to access.

Even though everyone are interested in the Mir / Wayland, I believe that the greatest progress in relation to the linux distributions we know so far will be exactly these.

[J?rnS]

1. Applications decoupled from the OS updates. Traditionally a given release of Ubuntu has shipped with the versions of the applications available at the time of release. Important updates and security fixes are back-ported to older releases where required, but generally you had to wait for the next release to get the latest and greatest set of applications. The new desktop packaging system means that application developers can push updates out when they are ready and the user can benefit right away.

To solve this issue, it's not needed to create a new packaging system, as .debs are fine. There are 2 possible solutions, that I think would be better:

• a PPA that has the most common software in a recent version, so you can opt-in and have a stable base system with backported applications
• more use of the backports system, which means more backports and an updater that takes care about backports and suggests "there is a new version of Libreoffice in the backports archive 'trusty-backports', do you want to install it?". Lubuntu will have a feature like this in the future, just as a small hint ;-)

As you can see at http://packages.ubuntu.com/de/trusty...ts/allpackages / http://packages.ubuntu.com/de/precis...ts/allpackages, there could be done more. Debian Wheezy has quite a lot backports, including stuff like Libreoffice.

[Malizor]

To solve this issue, it's not needed to create a new packaging system, as .debs are fine.

No they are not. A .deb has to be installed has root and launch scripts written by the packager.

Here we are talking about a random developer pushing its new app version directly to users. You surely don't want to give every developer root access on your system.
What is currently done with traditional backports is a manual review of each change between the old and the new version. This does not scale and the whole bureaucracy does not help attracting new developers.

You can see "click" packages are just dumb deb packages: no dependency, no script, no root access.

[r_a_trip]

By which time other distros will have been shipping Wayland as standard for at least six months, maybe a year. And confronted by that reality, is Canonical really going to stick with Mir all the way to release? Or are they going to decide there's no point in continuing down that path, and spend another year or two porting Unity to Wayland?

It isn't quite as bleak as you think it is. Mir is a compositor which targets EGL. This is what Wayland does as well. Since Unity is written to run a top of Mir and Mir already outputs to EGL, the only thing Canonical/Ubuntu needs to do is write a Wayland backend for Mir to make it a Wayland compositor. How long that takes depends on how well Canonical has written Mir.

My guess is that this is what eventually will happen. Despite Mark shuttleworth declaring a leaders position for Ubuntu a few years back, they are falling behind in advances and they certainly aren't dictating the technology landscape. They already caved on systemd vs Upstart. They have their hands full on realising Unity as a fully converged desktop. Plus they are trying to get an App ecosystem off the ground for their converged efforts.

Mir is an unfortunate distraction in its current form. I still wonder why Canonical wrote Mir as something else than a Wayland compositor. They vastly underestimated the work involved. They didn't get it to market in a timely fashion. They now have a one distro compositor on their hands, which only gets lukewarm support from the Wayland targeting projects. They could plod along, but they might have to keep up with projects adding more and more Wayland-isms to their software. Whatever advantages they imagined, I can't see them at this point in time.

[johnc]

...they need to learn how to work with others and stop these "my way or the highway" projects.

It's interesting how this criticism only has traction against Canonical around here. GNOME, RedHat, systemd, and other open source projects always seem to be exempt from this standard.

[r_a_trip]

It's interesting how this criticism only has traction against Canonical around here. GNOME, RedHat, systemd, and other open source projects always seem to be exempt from this standard.

Other open source projects are (more or less) open for discussion and inclusion of patches/ideas. Canonical from the outside seems rather closed on this. It is be part of Ubuntu or ... (I'd say get lost, but that is a bit harsh).

[Gusar]

It's interesting how this criticism only has traction against Canonical around here. GNOME, RedHat, systemd, and other open source projects always seem to be exempt from this standard.

Gnome has been receiving tons of criticism since the 3.0 release, and because of the Gnome dev's attitude Mate and Cinnamon exist. Also, there's constant flamewars regarding systemd. So what exemptions?

There's one thing interesting though - flamewars on forums aside, the other projects do see adoption - many distros package Gnome, many distros have moved or are moving to systemd. But Canonical's stuff doesn't see adoption outside of Ubuntu. So there must be something those other projects are doing differently than Canonical.

[mvaar]

mature ?

Yep, read the article. Yep, Ubuntu 16.04 will be Mir. No, it won't be nearly as mature as Wayland.

you must be joking. Nobody is using wayland and it has been in development a lot longer and supposedly by the "larger community".

Wayland will NEVER be mature, just as Xorg never has been. Perhaps Mir will meet with the same fate, who knows but wayland will be just another xorg, with small improvements but then, xorg today is X11 with small improvements.

Mir is a paradigm change (API driven, not protocol implementation); you better HOPE it is a success, if linux has to gain any traction on the desktop. Desktop applications on linux have to deal with a major hurdle- the application is strongly linked to the desktop and has to make assumptions about the runtime. You need system level API and an implementation that will at least do window management, irregardless of the runtime. Then you need an openGL implementation ( or directX or whatever) that again, the applications shouldn't care about w.r.t. the implementation.

[mvaar]

they also have to deal with

Perhaps, but if so, I can't say I'm impressed by their organisational abilities... they had well-publicised plans to make Mir the default in 13.10, before backing off after realising how far away they still were. So rather than fix the issues and get it shipped, they've distracted themselves with a desktop rewrite, and pushed it back a few more years.

By which time other distros will have been shipping Wayland as standard for at least six months, maybe a year. And confronted by that reality, is Canonical really going to stick with Mir all the way to release? Or are they going to decide there's no point in continuing down that path, and spend another year or two porting Unity to Wayland?

hecklers from redhat, gnome (usually redhat), Xorg/wayland in addition, which is quite a distraction.

BTW, we will see when any distribution actually ships a fully wayland based system ( no, Xwayland doesn't count - that is cheating and even that hasn't happened). I have been hearing a BTRFS default for 5 years now- how many distros have switched to BTRFS default ?

I don't think that canonical believes in wayland and judging by real world implementations, neither does anyone (talk is cheap, though and so is FUD). Mir is technically preferable to wayland and that is why canonical went with it in the first place, even against the grain so to speak.

[mvaar]

you mean

Canonical is running out of time to pull the plug on Mir and save face.

If they ever want to be a credible upstream again, they need to learn how to work with others and stop these "my way or the highway" projects.

just kow-tow to everything redhat says, right ?