KDE Plasma 5.22 Beta Ready For Testing With Much Better Wayland Experience

[Awesomeness]

I've been using Linux since the late 90s, I saw a transition from XFree86 to X.org, I witnessed the progress behind the development of KDE 2.0 and Gnome 2.0 and I don't remember this "We've made changes to better support XFree/X.org" ever. For some reasons the support for Xorg came naturally without any friction while the road to Wayland is not just bumpy, it's almost hellish.

X.org is a fork of XFree86. There was nothing to migrate. That's like complementing LibreOffice for their excellent OpenOffice document compatibility.🙄 For someone insulting others' intelligence by assessing brain damage, your memory seems to be in surprisingly bad condition (it's either that or you're lying on purpose – neither makes you look good).

[spstarr]

Should see even more fixes coming.. after testing patches from Kwin developers multi-GPU support is coming.. they fixed my bug. My laptop has an iGPU and AMD dGPU and it turns on my external DisplayPort now, before I had the laptop screen turn on only or fail to display anything, random I'm unsure when this will land in 5.22 or 5.23, we'll have to see.

[theriddick]

No mention of a Xgamma kgamma equivalent I see. Something I need due to my monitor having limited gamma adjustments. (and also some games needing more/less gamma then desktop).

PS. Not same thing as colour temperature shifting. Gamma is a RGB way to control brightness and contrast at same time I believe and is more effective for me when changing those separately.

[MadCatX]

MadCatX

I don't understand your first point at all. Despite Windows transitioning from GDI to Direct2D/DirectWrite (and changing the driver model as well, so two huge API changes), most old low level Windows 95 GDI applications continue to work fine in Windows 10, yet not even low-level X11 applications are 100% unusable under Wayland, e.g. those using screen capture.

I still fail to see the relevance of this simile. Microsoft didn't remove GDI from WinAPI. They had no reason to do so. Having support for the legacy GDI doesn't prevent the newer APIs from being added. Mesa also didn't remove OpenGL support when they introduced Vulkan. On the other hand, Wayland and X11 are two conceptually different technologies that can't easily coexist - you're running either one or the other.

As to your second point: it leads to a duplication of effort FFS. GTK/Qt/EFL and all other toolkit developers must write a ton of code to to simply output something on the screen.

How would an addition of yet another rendering API help the matters? Most UI toolkits have been doing their own rendering long before Wayland came along. For instance, Qt had its QPainter since Qt1.

And what if I don't want to use these three toolkits? Am I royally screwed?

There is Cairo, there is Skia, there is SDL, there is Kivy and probably a few dozen more.

Wayland also requires you to duplicate a ton of code to write a compositor. Xorg on the other hand provides basic window management by itself, so full-featured WMs/compositors can be written relatively easily and we have a ton of DEs and WMs for Xorg/X11 which all work nearly perfectly

In the Wayland land you can base your compositor on wlroots to avoid having to write most of the boiler-plate code yourself.

and have minimum complexity and duplication of code.

Well, KWin, for example, is a very complex codebase and most of this complexity comes from X11 code. Roman Gilg wrote a few interesting blogposts about this.

Lastly Wayland compositors offer different APIs for certain functions, so good luck writing an app which supports all of them and doesn't break down the road because those APIs are not part of any standard. Mutter APIs are not a standard.

There are XDG-shell specifications compositors are supposed to follow and these are considered to be the standard.

[StarterX4]

Cool, 5.21.90 is finally flawlessly usable under Wayland (even as for beta), so i can painlessly switch to Wayland now.

[StarterX4]

what does birdie want ? i don't know

maybe someone can meet him in real world drink a beer with him and ask: Birdie what do you want ?...

All the grumbler wants is to grumble.

[oiaohm]

Windows 95 and Windows Vista are all completely different protocols: GDI and Direct2D/DirectWrite. Various low-level utilities which target GDI from Windows 95 still work just fine in Windows 10, e.g. applications which change desktop resolution, capture the screen, etc. etc. etc.

https://www.dummies.com/computers/operating-systems/windows-7/how-to-make-an-older-program-run-in-windows-7/

That is not in fact true. There are many programs that don't work under Windows Vista and Windows 7 from Windows 95. Yes this is why Microsoft provided a Windows XP virtual machine in professional version. There is a compatibility layer in Windows Vista and 7 around older GDI applications just like Xwayland not all functional is there. Yes there are particular ways of screen capture a Windows 95 application could do by GDI that don't work with the migration to windows 2000 and never work on XP.

There is a long list of features GDI if a Windows 95 application uses them they either don't work at the change to Windows 2000 or don't work at the change to Vista.

The change from X11 bare metal to XWayland is basically like all fall out from the change to NT(the 2000 change) and the XP-Vista change in one hit not spread out.

[Kver]

I need to admit, I'm a little bit charmed that Birdie is holding up Window Vista of all things as an example of compatibility. Until multiple service patches managed to duct-tape the thing together huge numbers of applications broke outright. Not even applications as far back as '95. Even with Microsoft bending over backwards to build in compatibility layers. Microsoft even released "compatibility updates" which specifically fixed applications which Vista broke. We're not talking about obscure programs either, Vista broke such things as the Adobe suite, Age of Empires, Realplayer, Premiere, Java 5, and most importantly the Dora the Explorer game.

The fact is, when things change stuff will probably break.

[bple2137]

While I disagree with most birdie's arguments (like the one with migration between different X11 implementations or strict comparison to commercial systems), I think that some of the Wayland criticism is actually legit. I wouldn't go into technical details, because I don't have any power to judge design of something as low level as this, but I think that Wayland is not designed with users - and how they tend to interact with desktop Linux - in mind. There was higher priority of decreasing responsibility of display protocol (going from extreme to the opposite extreme?) and making it secure, rather than creating solution that is reasonably easy to implement/adopt, that fits the ecosystem (multiple DEs, fragmented pieces everywhere). It might not be any issue with how the protocol works at all. Actually I'm typing it on Wayland session on a PC that I use day-to-day for gaming, internet, multimedia and some amateur audio production. It's pretty snappy and while dealing with 4K screen it works way better than X11 session. The thing is, there are those little caveats here and there and it takes a lot of time to solve them all. I've spent quite some time (out of curiosity) to check out the development process reading through GitLab pages and mailing lists. The biggest issue here is to coordinate all those people with different points of view and make them came out with common ideas. Those are not technical difficulties and rather political ones.

Anyway, Wayland seems to be the nearest future of desktop Linux whether we like it or not, Birdie. After all a lot of good things have happened to Linux in order to make Wayland possible and the graphical stack is not so heavily dependent on a single, monolithic server anymore. If not Wayland then something else sooner or later.

[smitty3268]

Wayland is a serious brain trauma in progress which wastes people's time and energy:

I find this highly amusing, given that you repeat the same arguments in every new Wayland topic that gets posted here. I thought you enjoyed wasting people's time and energy in pointless repetitive tasks.

I've been using Linux since the late 90s, I saw a transition from XFree86 to X.org, I witnessed the progress behind the development of KDE 2.0 and Gnome 2.0 and I don't remember this "We've made changes to better support XFree/X.org" ever. For some reasons the support for Xorg came naturally without any friction while the road to Wayland is not just bumpy, it's almost hellish.

Well, yeah.

• Welcome to the modern era of constant online news, social media, manufactured outrage, etc.
• That was mostly just bolting on stuff on top of existing code. Wayland is a complete break and rewrite. That was intentional. People knew it would be difficult, that's why they put it off for so long
• We now have 35 years of userspace code that relied on X and it's various API's and libraries that all had to be ported over or otherwise supported one way or another. There was a lot less legacy UI apps to worry about in the past, and the userbase was much more willing to just abandon a lot of mostly unused stuff than people are today, or else they would channel that passion into coding fixes themselves. Now large numbers of people who are passionate about such things just rant on the internet.