KWinFT: KDE's KWin Forked To Focus On Better Wayland Support, Modern Technologies

[Volta]

https://www.gamingonlinux.com/render_chart.php?id=801&type=stats

Nvidia is 60% of gpu marketshare.
I anticipate the question - who cares? Roman's employer cares.

Too bad nvidia doesn't care about them.

[enrico.tagliavini]

Well that's nothing short but awesome. I just tried Plasma on Wayland a few minutes ago, before reading this post.... it's a wreck. It's not usable really.

KwinFT announcement is very well laid down, Roman Sir, well done. I wish you the best of luck and I thank you for your effort from the bottom of my heart!

[R41N3R]

I've upgraded my Arch Linux desktop with an AMD GPU to kwinft (Wayland) and good news is that it works :-)

Some issues I've noticed so far. If I move the Firefox XWayland window (which is slightly dimmed) will create some flickering. ALT + TAB shows the virtual keyboard partially so it seems to recognize it as a normal window. But not sure if any of these issues are related to kwinft.

[romangg]

I plan to drop the lowlatency codebase if Gilg's codebase is low-latency enough and I am able to merge everything.

I know they won't add full-screen unredirection to KWin so, most likely yes.

Hi tildearrow,

I won't allow full-screen unredirection to go into KWinFT since its focus in the future is on Wayland but you are most welcome to base a fork providing this functionality on the KWinFT code base.

You are of course also welcome to work on KWinFT directly if some Wayland functionality would spark your interest.

In any case thanks for your work on kwin-lowlatency. It was certainly a motivating factor for me to take a deep dive into KWin's languishing compositing tech last year.

[M@GOid]

https://www.gamingonlinux.com/render...801&type=stats
Nvidia is 60% of gpu marketshare.
I anticipate the question - who cares? Roman's employer cares.

That is Linux gaming market share. A very small one if you think about it.
The truth is the biggest GPU vendor by far is Intel. It dwarfs both AMD and Nvidia, since the majority of users out there are not dedicated gamers. And on those machines, Wayland works already.

Nvidia had put their users on a awkward position, so is their responsibility to resolve the mess they created. I would take this on consideration on the next GPU purchase. If you want a good Linux desktop experience, opensource drivers are the only way to go. If you think otherwise, don't come belly-aching for things people already warned you not to do.

[coats]

But will it support xrandr --panning ? Failure to support it disqualifies the current KDE immediately, as far as I am concerned. Thank God (and the developers) for TDE...

See https://bugs.kde.org/show_bug.cgi?id=415876 --

[andre30correia]

That is Linux gaming market share. A very small one if you think about it.
The truth is the biggest GPU vendor by far is Intel. It dwarfs both AMD and Nvidia, since the majority of users out there are not dedicated gamers. And on those machines, Wayland works already.

Nvidia had put their users on a awkward position, so is their responsibility to resolve the mess they created. I would take this on consideration on the next GPU purchase. If you want a good Linux desktop experience, opensource drivers are the only way to go. If you think otherwise, don't come belly-aching for things people already warned you not to do.

in some countries most of laptops come with dual graphics intel/nvidia, amd does not exist in laptop in a lot of countries, dont come with this stupid arguments and where is your data about most of ppl don't game? if steam usage is 1% and linux desktop is about 2% maybe some millions play games after all. Nvidia have some troubles but for years the only usable driver in linux/unix ecosystem was nvidia one and continue working quite well in heavy jobs.

[andre30correia]

Maybe KWin support for Wayland wasn't production ready, but Wayland itself was fine. ANd yes, Mir is just unnecessary NIH.

If Canonical had just quietly gone and done their own thing because they wanted to, then great no problem. Instead they wanted everyone else to adopt it, for no good reason and then spouted utter lies and bullshit about Mir vs Wayland to justify why everyone else should do what Canonical wants.

the big problem here is simple, Mir was working much better than wayland before they give up and one question why should everyone use Red Hat software? Same thing, mir was a great project with big and fast improvements and oriented to the end user not like Wayland or X

[miabrahams]

Best of luck to romangg for this project. I hope it is a refreshing change of pace and that it brings us closer to KDE usability on Wayland.

[STiAT]

While I don't exactly like forks, work of this as soon as stabilized enough could make it's way upstream or the projects could merge in the end again (or KWin get obsolete in the end?).

Who knows, for now, for a easier development process the fork is probably the way to go. I hope we won't see kwinft and kwin live on next to each other. Basically, while having currently a different focus (one on features and the other on stability to not impact their existing users), they try to solve exactly the same issues.

We'll see where it ends up, it's the freedom of open source, and maybe something better comes out of this in the end.