There's Been A Bit Of Progress On Vulkan Support In Ubuntu's Mir

[phoronix]

Phoronix: There's Been A Bit Of Progress On Vulkan Support In Ubuntu's Mir

When the Vulkan 1.0 API specification was unveiled last February, we were originally told by Canonical that Mir in Ubuntu 16.04 would have Vulkan support but now one year later, Mir in Ubuntu 17.04 doesn't even look like it will have Vulkan support. But at least progress is being made...

http://www.phoronix.com/scan.php?page=news_item&px=Mir-Vulkan-Bits-March-2017

[lucrus]

The whole story sounds like there is just one dev at work with Mir, he is overbooked and he serves feature request coming directly from Mike Shuttleworth in a LIFO fashion, because they pile up faster than he can manage to find the oldest one down the bunch of post-its.

Have a nice life Mir, and, more important, a fast death: I won't be missing you.

[ElderSnake]

As someone who's not so technically in the know, I still don't understand their problem with Wayland. If Wayland is relatively minimal and doesn't "do much on it's own" couldn't Canonical have made it work for them instead of messing around with Mir for so long?

[lucrus]

couldn't Canonical have made it work for them instead of messing around with Mir for so long?

They could back then, when they thought Wayland wasn't up to the job. Admitting they have been wrong for so many years would cost them too much now, in terms of brand reputation. It is probably better for them to keep their position pretending they have always been right unless someone proves them wrong. And none can prove them wrong until they don't complete Mir, which they will never do.

[FireBurn]

Does anyone actually care?

[theghost]

Serios question: Why would one need Vulkan support on the display server level? And how is the Vulkan state on the Wayland front?

[ua=42]

As someone who's not so technically in the know, I still don't understand their problem with Wayland. If Wayland is relatively minimal and doesn't "do much on it's own" couldn't Canonical have made it work for them instead of messing around with Mir for so long?

Yes. They could of used wayland.

[indepe]

Serios question: Why would one need Vulkan support on the display server level? And how is the Vulkan state on the Wayland front?

I'm not familiar with the specific needs of a display server, but I think Mir, as well as Wayland implementations, are much more than a pure display server.

[smitty3268]

As someone who's not so technically in the know, I still don't understand their problem with Wayland. If Wayland is relatively minimal and doesn't "do much on it's own" couldn't Canonical have made it work for them instead of messing around with Mir for so long?

There's the official Canonical answer, which is frankly nonsense, and the actual truth.

Canonical felt Wayland was a Red Hat project and didn't think changes they wanted to make would be allowed upstream, so they split it off into their own project that they were in charge of so they could do whatever they wanted to it. They had grand visions of finishing it all up in a year while Red Hat was still puttering around and slowing down Wayland adoption.

It turned out that creating a new graphical stack to replace X was a lot more difficult than they realized, and they've ended up spending the last few years basically copying Wayland code and waiting for Red Hat to fix up the rest of the userspace stack.

What they should do going forward is simply announce that Mir (the compositor) will work with the Wayland protocol - this would be very simple for them to do, and it would end the whole nonsense. It doesn't appear that's likely to happen though, largely because they've put a lot of money into it and more importantly a lot of ego. I think they'd see that as an admission of error and don't see it happening.

[Delgarde]

There's the official Canonical answer, which is frankly nonsense, and the actual truth.

Canonical felt Wayland was a Red Hat project and didn't think changes they wanted to make would be allowed upstream, so they split it off into their own project that they were in charge of so they could do whatever they wanted to it. They had grand visions of finishing it all up in a year while Red Hat was still puttering around and slowing down Wayland adoption.

That, and a bunch of bullshit and outright lies (sorry, "alternate facts") about security flaws in the Wayland design. Those claims were later retracted as being a misunderstanding... but even if that's true, it gives you a fair idea just little research they did on Wayland before rejecting it...