Vulkan Support Still Isn't Ready For Ubuntu's Mir

[starshipeleven]

afaik mir implements more things in server side than wayland, although i could be wrong

for now there is only a mir compositor, (the unity one), but features could be shared IF they were more compositors

What I knew was that both Mir and Wayland compositors there is no "server", the DE's compositor is the "server" and it offloads directly to kernel/mesa.

[rabcor]

It's ok, they're just gonna support DX12 instead, no one will use it on linux either but that's ok, nobody uses Mir either so they don't care.

[bregma]

What I don't understand is why bother with Vulkan? Desktop compositors aren't complex enough to warrant this kind of attention right now.

That's just a failure to understand what Vulkan is.

Unless an application, such as a game or a frame renderer in the movie industry, that needs to use Vulkan can get a rendering context, it can't do anything. What's required for Mir is to provide WSI (window system integration) hooks for a surface. The rest is just the usual chaining buffers and loading command queues.

So, to say there's no need for Vulkan on the desktop is a little odd. If not on the desktop, where do you expect it to run?

[bregma]

When is Canonical going to stop throwing good money after bad and end development of Mir? Is saving face really worth that much money? Is there any technical reason not to switch to Wayland?

They could branch libweston and play catch up within a year.

Is there any technical reason to switch to Weston or is there only an ardent political bias on your part?

Do you have an actual accountancy basis behind your speculation that bad money is being thrown after Mir? Do you know something about Canonical finances the rest of us are not privy to? I, in particular, have an keen interest in knowing.

You seem to have a great deal of knowledge about Mir and the entire shipping Ubuntu Touch stack to be able to estimate it would take a year to rewrite. Have you contributed code?

[starshipeleven]

Is there any technical reason to switch to Weston or is there only an ardent political bias on your part?

1. switch to Wayland. Weston is an example implementation, not used by any DE.

2. Wayland is the protocol most applications will probably use in the future as everyone apart ubuntu is going to use that, not doing so would force such application to fallback to X11 rendering code that goes through xMir.

Do you have an actual accountancy basis behind your speculation that bad money is being thrown after Mir?

He said "good money" after "bad dead end development of Mir".

You seem to have a great deal of knowledge about Mir and the entire shipping Ubuntu Touch stack to be able to estimate it would take a year to rewrite. Have you contributed code?

He is saying that with libweston making a new Wayland compositor from scratch is very easy, as that's the point of that library anyway.

[Cerberus]

https://www.youtube.com/watch?v=oDgNueE36tY

Here is a nice video of Ubuntu 16.10 with Unity 8 and Mir display server with QT, SDL and GTK applications running natively on Mir, looks pretty good except the GTK integration needs more work though I am suspecting a bug because GTK applications I saw running on the phones and tablets are better integrated.

[SaucyJack]

Guys, please have faith. As soon as we have Unity 8 with Mir as the default, we can finally have Vulkan support. And with soon, I mean very soon, it's just a matter of a few years. It will be sooooo great, really.
Meanwhile look at Canonical's great success on the mobile front or IOT. They are simply everywhere...

As much as I see Canonical trying to Microsoft the Linux sphere, I really wish they succeed in the mobile space so I can have a real linux phone.

[starshipeleven]

As much as I see Canonical trying to Microsoft the Linux sphere, I really wish they succeed in the mobile space so I can have a real linux phone.

Ubuntu phones are full of blobs, "real linux phone"... yeah right.

[Khrundel]

When is Canonical going to stop throwing good money after bad and end development of Mir? Is saving face really worth that much money? Is there any technical reason not to switch to Wayland?

They could branch libweston and play catch up within a year.

Sure.
Gnome developers don't use weston, they are implementing their own compositor which will work with wayland and x11 client apps. They are smart guys.
Plasma developers don't use weston, they are implementing their own compositor which will work with wayland and x11 client apps. They are smart guys too.
Unity developers don't use weston, they are implementing their own compositor which will work with wayland and x11 client apps. They are bunch of morons.

[SaucyJack]

Sure.
Gnome developers don't use weston, they are implementing their own compositor which will work with wayland and x11 client apps. They are smart guys.
Plasma developers don't use weston, they are implementing their own compositor which will work with wayland and x11 client apps. They are smart guys too.
Unity developers don't use weston, they are implementing their own compositor which will work with wayland and x11 client apps. They are bunch of morons.

That would make sense if your third sentence was accurate in any way. Mir replaces wayland, not weston. And yes, rewriting wayland makes them a bunch of morons.