Canonical Shows Legacy X11 Apps Running On Mir With Unity 8

[phoronix]

Phoronix: Canonical Shows Legacy X11 Apps Running On Mir With Unity 8

Canonical posted a video today showing the state of running a Unity 8 session with Mir while supporting legacy X11 applications that lack a Mir back-end...

http://www.phoronix.com/scan.php?page=news_item&px=Ubuntu-Mir-Legacy-Apps

[pracedru]

I Really thought they where further along in the development of XMir.
Please remind me again... Why didn't they just use Wayland?

[computerquip]

Because they're stubborn and don't want to admit their obvious mistakes.

[boxie]

I Really thought they where further along in the development of XMir.
Please remind me again... Why didn't they just use Wayland?

I thought so too... but it does seem that there is not much to go.

as for why not wayland... (I believe) the decision was based on not wanting a protocol (i.e. passing XML back and forward), but rather an API to implement against.

as to which is better - Mir *might* have an upper hand in terms of performance since it does not have to parse XML. Wayland has the benefit of being more flexible in terms of protocol implementations (i.e. each desktop environment can implement its own custom protocols etc).

At least, that is as far as I understand it.

[sandy.martel]

... (i.e. passing XML back and forward)...

The protocol is *defined* as XML but is *not* XML. That would effectively not be very performant for a local display server. Think of the XML as a schema, not the messages themselves.

http://wayland.freedesktop.org/docs/...ol-Wire-Format

[Scias]

I thought so too... but it does seem that there is not much to go.

as for why not wayland... (I believe) the decision was based on not wanting a protocol (i.e. passing XML back and forward), but rather an API to implement against.

as to which is better - Mir *might* have an upper hand in terms of performance since it does not have to parse XML. Wayland has the benefit of being more flexible in terms of protocol implementations (i.e. each desktop environment can implement its own custom protocols etc).

At least, that is as far as I understand it.

Mir does use a protocol as well (it uses google's protbuf format), it's just not being exposed to the regular clients.

And as for the potential performance gains there's absolutely no evidence of them, quite the contrary actually : https://bugs.launchpad.net/mir/+bug/1352210

I Really thought they where further along in the development of XMir.
Please remind me again... Why didn't they just use Wayland?

Well I can think of one valid reason, is that since they can control both Mir and Unity development they can supposedly work faster and taylor them to their exact needs without having to bother with upstream protocol compatibility and such...
But seeing the current results, numerous issues, delays, how much they had to borrow to libhybris/xwayland and then finally resort to use libinput in the end (which is ironical considering their supposed main concern against Wayland was the security of its input stack) I guess we can safely say that it was all NIH and for being able to lock the whole stack under their shady GPLv3 + harmony CLA combo.

[Scias]

I thought so too... but it does seem that there is not much to go.

as for why not wayland... (I believe) the decision was based on not wanting a protocol (i.e. passing XML back and forward), but rather an API to implement against.

as to which is better - Mir *might* have an upper hand in terms of performance since it does not have to parse XML. Wayland has the benefit of being more flexible in terms of protocol implementations (i.e. each desktop environment can implement its own custom protocols etc).

At least, that is as far as I understand it.

Mir does use a protocol (it uses google protbufs), it's just not being exposed to regular clients.

And as for the supposed performance gains from it, we've yet to see them in action because for now their protocol performance seems disastrous.

I Really thought they where further along in the development of XMir.
Please remind me again... Why didn't they just use Wayland?

NIH, and lock the whole stack under their shady GPLv3 + harmony CLA combo.

Because let's be serious. If their reason really was because they could work faster and not to worry about upstream protocol compatibility and such they wouldn't have suffered so many delays, issues, or had to borrow so much from xwayland/libhybris, or finally had to resort to use libinput for input management, while their main reason for making Mir was because they didn't like Wayland's input model...

[jacob]

I Really thought they where further along in the development of XMir.
Please remind me again... Why didn't they just use Wayland?

Yes, I wonder. They want a display server that is separate from the compositor, which is fair enough, but why does Mir have to have its own protocol instead of supporting Wayland... On the other hand, if the GTK and Qt backends allow apps to actually work seamlessly (which remains to be seen, there is a huge difference between supporting a calculator app and supporting a nontrivial one with its own custom widgets, rendering etc.), then I suppose it's largely immaterial for users and app developers alike.

[Delgarde]

The protocol is *defined* as XML but is *not* XML. That would effectively not be very performant for a local display server. Think of the XML as a schema, not the messages themselves.

Yeah, using XML over the wire would be insane for something like a display server. And since the Wayland developers aren't completely out of their minds, XML is simply the format the protocol is *documented* with.

Edit: you can see the XML in question here... basically, a file that specifies what all of the messages and interfaces in the protocol look like, and which can be used to generate code bindings that conform to that specification.

[Veto]

But... does it run Steam?