Red Hat Making Progress On KDE/GNOME Wayland Screen-Sharing With WebRTC

[phoronix]

Phoronix: Red Hat Making Progress On KDE/GNOME Wayland Screen-Sharing With WebRTC

Jan Grulich and other developers at Red Hat have been making progress on screen-sharing support using WebRTC as found within web-browsers like Firefox and Chrome. With their experimental work, Wayland screen-sharing is working both for GNOME Shell and KDE Plasma...

http://www.phoronix.com/scan.php?page=news_item&px=WebRTC-Wayland-Screen-Share

[Marc Driftmeyer]

WebRTC is also default active in Apple Safari so you can expect it to be available in Epiphany soon.

[dwagner]

To me this sounds like using totally wrong abstraction layers:
Why would I want to use an HTML browser to display a remote screen? (Is it because nobody knows applications beyond web browsers anymore?)
Why would I want specific bloated, colorful desktop decoration software like KDE or GNOME to be involved in "screen sharing"? (Isn't the display protocol the very place where transport to remote(s) should be done?)

[Tomin]

To me this sounds like using totally wrong abstraction layers:
Why would I want to use an HTML browser to display a remote screen? (Is it because nobody knows applications beyond web browsers anymore?)
Why would I want specific bloated, colorful desktop decoration software like KDE or GNOME to be involved in "screen sharing"? (Isn't the display protocol the very place where transport to remote(s) should be done?)

I think the idea is to allow conferencing software (in browser) to share desktops to other participants to for example show a presentation. I would say that it's a good goal. Also, remember that this is just one application of the portals that were developed to do this and they could be used for other purposes as well. I don't know if this would be a good way to do remote desktop, maybe not.

Gnome and KDE are involved because there is no separate window server anymore but instead Mutter (Gnome) or KWin (KDE) are handling it. This is why the desktop (or rather the compositor) must support this portal protocol.

[Hibbelharry]

To me this sounds like using totally wrong abstraction layers:
Why would I want to use an HTML browser to display a remote screen? (Is it because nobody knows applications beyond web browsers anymore?)

Because at first you haven't noticed in which way browsers evolved the last years. They are no longer really pure HTML Viewers, that era is long gone. Anyway the most basic fact is: Browsers are capable and cross platform. You develop one application and target every platform, thats why webapps are rising more and more.

Why would I want specific bloated, colorful desktop decoration software like KDE or GNOME to be involved in "screen sharing"? (Isn't the display protocol the very place where transport to remote(s) should be done?)

Away from 80ties style Unix Admins, you don't get any userbase without fancy colors today. And using colors is no killer either. Why this has to be integrated is well explained in tomins post.

[oiaohm]

(Isn't the display protocol the very place where transport to remote(s) should be done?)

That is where you get into argument.

You have a kernel space display protocols. KMS/DRM with work on Virtual KMS provides a different path. Also remember wayland protocol covers a compositor so the wayland protocol yes the compositor is only required to pass to graphics card buffers and the graphics card produces the output screen data. Do note you see screen capture implemented by video card drivers in opengl and vulkan.

So it really what display protocol KMS/DRM, X11, Vulkan, Opengl, some video card vendor particular implementation. So abstraction does makes sense. Portals document interface they don't say lighter X11 compositors cannot implement them. Redhat is just doing KDE and Gnome first.

[q2dg]

Why not VNC?

[dkasak]

Because at first you haven't noticed in which way browsers evolved the last years. They are no longer really pure HTML Viewers, that era is long gone. Anyway the most basic fact is: Browsers are capable and cross platform. You develop one application and target every platform, thats why webapps are rising more and more.

I've never been sold on web apps. I get why an online retailer needs a website. But web apps have always sucked. Whole dev teams tie themselves in knots for months on end to produce a poor approximation of a desktop app that could have been put together in days by a single person. And when they're finished, they're already arguing about the next cutting-edge JS framework they need to port everything to. The fact that most devs these days are web developers in no way validates their approach. It's more a case of the emperor's new clothes, and a reasonable amount of 'but where would we find devs who do things differently?'.

[smitty3268]

I've never been sold on web apps. I get why an online retailer needs a website. But web apps have always sucked. Whole dev teams tie themselves in knots for months on end to produce a poor approximation of a desktop app that could have been put together in days by a single person. And when they're finished, they're already arguing about the next cutting-edge JS framework they need to port everything to. The fact that most devs these days are web developers in no way validates their approach. It's more a case of the emperor's new clothes, and a reasonable amount of 'but where would we find devs who do things differently?'.

Deploying native apps is a pain, especially on windows. With web apps you instantly get everyone running the same app from the moment you release a new version and for the most part you don't have to remotely debug people's graphics drivers, which 3rd party app they have installed which is breaking your app, etc. It just works.

Which isn't to say everything should be done that way. There are simply valid reasons to do things on both sides.

[fuzz]

I've never been sold on web apps. I get why an online retailer needs a website. But web apps have always sucked. Whole dev teams tie themselves in knots for months on end to produce a poor approximation of a desktop app that could have been put together in days by a single person. And when they're finished, they're already arguing about the next cutting-edge JS framework they need to port everything to. The fact that most devs these days are web developers in no way validates their approach. It's more a case of the emperor's new clothes, and a reasonable amount of 'but where would we find devs who do things differently?'.

Web Assembly changes everything.