Jolla Brings Wayland Atop Android GPU Drivers

[Teho]

Reduce pressure on creating free drivers or on open the hardware seems a bad idea.

I think effect might actually be positive in that area. First of all this there's multiple problems with the closed source drivers on ARM space, one of the more important ones being that they usually work for only couple of kernel releases. That doesn't really fit the fast moving Linux developement model so open drivers is needed for that. But also I think it might be more interesting to develop these drivers when you already have a proper GNU/Linux stack running on top of it. Also the closed source drivers for AMD and NVIDIA haven't really stopped the open source driver developement for their cards. These Ubuntu Touch / Sailfish / Firefox OS projects are so insignificant currently that they hardly create any "pressure" anyway. They might actually do if they first got out of the door but that then again is very difficult without having good support for closed source drivers.

Your argument really holds no water here.

There might be some configurations of Tizen (or previously MeeGo) that allows/allowed L/GPLv3; GPLv3 code was removed from MeeGo TV and it was a no-go for IVI systems as well. Yocto has "non-GPLv3" option. One can only wonder why Apple wanted to avoid GPLv3 to the point of writing alternative for Samba (and GCC?). Projects like FreeBSD are also "fine" with L/GPLv2 but GPLv3 just goes too far. Tizen coreutils seems to be based on the pre-GPLv3 release of GNU coreutils too.

[argeep]

Using Wayland with Android drivers is not new. Collabora ported Wayland to Android months ago: http://ppaalanen.blogspot.de/2012/09...o-404-and.html
This here is obviously a step further in that Wayland runs not only on an Android system but now on a glibc system with libhybris but Android drivers are used in both cases (and this Jolla works probably builds on the Collabora work).

Indeed. Carsten gives credit in his blog for this...

...

This is the entry point used to implement Wayland on top of Android GPU drivers on glibc based systems. Some fantastic work in this area has already been done by Pekka Paalanen (pq) as part of his work for Collabora Ltd. (Telepathy, GStreamer, WebKit, X11 experts) which proved that this is possible. Parts of the solution I will publish is based on their work - their work was groundbreaking in this field and made all this possible.

...

[maksml]

Jolla looks good

Its good to see companies embracing Wayland.
I've been fond of Jolla since the beginning.
This just makes me more fond of them.

[renox]

This means: the only legitimate reason to support Mir has disappeared overnight.

That's not fully true: the Mir developers claim that server side allocation of buffer allow them to better manage the (scarces in a phone) resources of the GPU (video memory).
Wayland does normally client side allocation of buffer.
Thought at least one Wayland developer has turned the client side allocation into a "request a buffer to the server" so I'm not fully sure that this was indeed a legitimated reason..

[przemoli]

My thoughts exactly. Why did they even create Mir, if it was that easy to make wayland run on android drivers also?

What part of "Decision made THEN, circumstances changing NOW" you do not understand?

[przemoli]

On several occasions one or two Mir main developers stated they didn't participate in the decision to create Mir. And Mir is developed under CLA and GPLv3(fact: a lot of companies avoid GPLv3 in their products or simply ban it).

<tinfoil> I bet the reasons aren't technical at all. They want control and a restrictive license, so they can sell proprietary licenses to manufacturers. Google sells services and Canonical sells proprietary license to GLPv3 code. </tinfoil>

FACT: Weston is on MIT. You can fork Weston and sell it as proprietary.

[przemoli]

Getting unified Android/Linux/ChromeOS drivers would be ideal, whoever manages to bring us that. It should've been Google's project from day one of Android, but I guess it wasn't their priority, which is too bad because that has led to some of the biggest fragmentation issues of Android.

Not going to happen.

Not all features from mobile are needed on desktop (OpenVG?)
Not all features from desktop are wonted on mobile. In fact some are most unwelcome (power hungry ones).

Some well defined and limited feature set that would be supported across devices and gpus is probable though.

[przemoli]

Canonical has done nothing to get rid of closed drivers.

Actually Canonical is only major backer of any Linux distro who is fully supporting Intel FLOSS driver efforts.

Eg. Fedora castrate Intel drivers.


There are many angles you need to consider to give verdicts.

[przemoli]

Carsten did say they are being quite cooperative now, participating in discussions and submitting patches upstream. This is the way it should have been from the beginning.

Its called product anouncment. If they worked on libhybris from the beginning in the open. Somebody would ask question WHY?
And most probably Canonical would receive salvos of bad PR from people who would assume that Canonical is working on mobile and abandon (or neglect) desktop.

FLOSS wont resolve such problems. Its politics and marketing.

[Teho]

Actually Canonical is only major backer of any Linux distro who is fully supporting Intel FLOSS driver efforts.

Eg. Fedora castrate Intel drivers.

What exactly do you mean? Sounds like complete bullshit. Red Hat also has many Mesa developers on their pay role. The are actually developing the driver infrastucture that Intel also uses.

FACT: Weston is on MIT. You can fork Weston and sell it as proprietary.

Yes, everyone can but only Canonical can sell Mir. It's unfair advantage and one of the many reasons why it will never be accepted by the community. Also the code in Weston/Wayland is resuable because MIT is very liberal license. For example projects like KWin and Mutter are licensed under GPLv2 hence they can't use code from Mir.

What part of "Decision made THEN, circumstances changing NOW" you do not understand?

Because it was already shown that Wayland could be used with Android drivers almost a year ago? Because Canonical didn't even contact the Wayland developers about the subject?

Its called product anouncment. If they worked on libhybris from the beginning in the open. Somebody would ask question WHY?

Canonical had announced that they have plans for mobile a long time ago.

And most probably Canonical would receive salvos of bad PR from people who would assume that Canonical is working on mobile and abandon (or neglect) desktop.

How is that any different from what has happened? If anything they have once again shown how they are incapable in working with the community. That if something is bad "PR".