Keith Packard's video from 'LinuxConfAu 2018' talking about Virtual Reality and Linux is up

But even now, there's some cool stuff coming out because people are working to get VR working properly on Linux. Basically that was something that simplifies sharing displays on multi-seat setup much easier.

If I'm able to explain it correctly, it's way to lease the display to a program that wants exclusive (but still revocable) access for a display, while making sure that client doesn't have access to the clients running on the real X server. X server or kernel can revoke the access at any time and control comes back to the X server without any disastrous issues.

Sounds interesting, multiple separate X sessions at the same time are a really something that is missing at the moment. Not only for VR, but in general. A few weeks ago, I was in a situation where I would've really needed that...

I don't see where a VR device offers that much more functionality than a monitor + controller/mouse.

Right. It's you who doesn't see this.

For gaming, or generally anything where being "inside the first person" has an advantage, sure. But that's just a small fraction of things that classical monitors cover.

Yeah, and that's why people like to use it for some kind of games. No matter what you're saying about its weight or anything. You seem to be trying to refute that VR devices will replace everything - but I'm asking again, who even said that before you were trying to refute it?

I don't see where a VR device offers that much more functionality than a monitor + controller/mouse.

Right. It's you who doesn't see this.

You pulled that sentence out of context, which is amazing, considering you quoted the next sentence right after that... You can read them as "I don't see... except for...".

You seem to be trying to refute that VR devices will replace everything - but I'm asking again, who even said that before you were trying to refute it?

I already answered that question. What do you want from me? :S:

Also... it was you who can imagine that VR devices will eventually replace everything.


Last edited by TheSHEEEP on 30 January 2018 at 3:07 pm UTC

But even now, there's some cool stuff coming out because people are working to get VR working properly on Linux. Basically that was something that simplifies sharing displays on multi-seat setup much easier.

If I'm able to explain it correctly, it's way to lease the display to a program that wants exclusive (but still revocable) access for a display, while making sure that client doesn't have access to the clients running on the real X server. X server or kernel can revoke the access at any time and control comes back to the X server without any disastrous issues.

Sounds interesting, multiple separate X sessions at the same time are a really something that is missing at the moment. Not only for VR, but in general. A few weeks ago, I was in a situation where I would've really needed that...

In some form it has been possible for long time. Oldest one around is to make all monitors separate X displays. Makes things pretty cumbersome though as clients programs can't be dragged between the displays and which display programs end up is determined with an environment variable. And setup can't be really changed dynamically. Getting rid of the setup requires editing configuration file and the restarting the Xorg server.

Xephyr was somewhat more practical, it's basically a window that contains another X server, ie. nested X server.I don't think I have ever tried to play games with it though. But in theory it could be possible at it should have GLX support now. I don't think mouse is locked inside the window though, which for some games might not be practical.

But even now, there's some cool stuff coming out because people are working to get VR working properly on Linux. Basically that was something that simplifies sharing displays on multi-seat setup much easier.

If I'm able to explain it correctly, it's way to lease the display to a program that wants exclusive (but still revocable) access for a display, while making sure that client doesn't have access to the clients running on the real X server. X server or kernel can revoke the access at any time and control comes back to the X server without any disastrous issues.

Sounds interesting, multiple separate X sessions at the same time are a really something that is missing at the moment. Not only for VR, but in general. A few weeks ago, I was in a situation where I would've really needed that...

In some form it has been possible for long time. Oldest one around is to make all monitors separate X displays. Makes things pretty cumbersome though as clients programs can't be dragged between the displays and which display programs end up is determined with an environment variable. And setup can't be really changed dynamically. Getting rid of the setup requires editing configuration file and the restarting the Xorg server.

Xephyr was somewhat more practical, it's basically a window that contains another X server, ie. nested X server.I don't think I have ever tried to play games with it though. But in theory it could be possible at it should have GLX support now. I don't think mouse is locked inside the window though, which for some games might not be practical.

Ah, what I needed was to have two X sessions with only one monitor (or rather, on a laptop with no spare monitor around). By switching the contexts using the good old Ctrl+Alt+F1-8, I tried to start another X session on one of them.
The funny thing is that I remember me looking around in Lynx (!!!) about how it could be achieved. What I don't remember is why I needed that in the first place...

I don't see where a VR device offers that much more functionality than a monitor + controller/mouse.

Right. It's you who doesn't see this.

You pulled that sentence out of context, which is amazing, considering you quoted the next sentence right after that...

Yes, on purpose, because I think this decsribes the problem better than all the kilobytes you wrote about it.
You don't see it, (many) other people do.
I don't think there's much more about it.
It's nothing objective and all your writing won't make it.


Last edited by Eike on 31 January 2018 at 6:49 am UTC

So they are working on both Vive and Rift compatibility?

For as long as Oculus is owned by Facebook, if the CV1 or anything after it gets Linux support that's not a third-party hack that Facebook tries to shut down in some way, I will somehow eat my nonexistent hat. It's just not happening.

But maybe WINE will work well enough

Damnit I'm working on it! It's just Really, Really, Really Hard because I need either some really obscure DXGI features to be implemented (which may not even be possible via OpenGL), or a purpose-made API inside wined3d/VK9/DXVK/vkd3d. The second is actually likely to happen, due to DXVK hitting walls with mixed D3D9/D3D11 applications, which probably will result in an interconnecting API for VK9/DXVK/vkd3d to cooperate with each other, which would be suitable for WineOpenVR as well.

OpenGL and Vulkan applications (should) work with what I have now, but good luck finding such apps.

The other way around: Games made specifically for VR could always be made working for monitors as well. There is no movement my head can make that cannot be imitated with mouse & keyboard.

I wanted to keep out of the argument but this really hit me.

Hell no.

You clearly haven't played Gorn. Its mechanics are literally impossible without tracked motion controllers, which are uselessly awkward without an HMD.

I have a drone, would love to tie the controller into VR with the HMD and fly one. There are very accurate flight sims out there especially for drones (https://www.realflight.com/ for example). X-Plane does Linux but RF doesn't. Flight time is flight time though.

Have you heard of FPV drones? Literally drones with a small camera mounted in the front that you use with a display or purpose-made goggles.


Last edited by roothorick on 31 January 2018 at 12:25 am UTC