Firefox 75 On Wayland Now To Have Full WebGL, Working VA-API Acceleration

[Guest]

Strange times we live in, suddenly there's a lot of raging noobs on the internet opposing innovation and software development. Happy user of systemd, wayland and pulseaudio here. They work excellent for all my use cases. Most of these "opinions" are based on FUD.

Different in not innovation. Now it is time for big companies to monetize from the open source world. Today. They'll lose all tomorrow. It's a natural, human fact.

[arokh]

X.org works and it provides rich interfaces. Wayland works for special use cases and have a lot of limitations while providing almost no interfaces. Wayland would have been excusable, had it been a few years old with no major backers. But it's 11 effing years old and it's heavily backed by RedHat and Intel. I mean it took Microsoft a lot less to develop Windows GDI - basically it started with Windows 3 (1990) and was fully complete (and largely rewritten) with Windows NT 4.0, i.e. in five years. All Windows apps which have been written ever since use Win32 APIs which date back to Windows 95 days. And they preserved APIs while doing all of that - 16bit Windows 3 apps were trivial to port to Win32 and they even worked without porting.

I get that you have no idea what the Wayland protocol is, other than that I have no idea where you are going with this win32 analogy.

You're absolutely right about X.org and how complicated it has turned to be. Only its replacement is not a magical solution. It still feels as a half-assed solution nowhere near the level which is required from from a modern graphics subsystem (e.g. found in Android). I mean Google did it right the first time. Yeah, they had no compatibility to keep at all, but somehow Surface worked and provided almost everything (except windowing which is essential on desktop) required from the get go. Open Source developers are seemingly not interested in backward compatibility. The world meanwhile couldn't care less about this constant breakage as indicated by the Linux market share which has remained near zero (among general population) for the past 20 years. The 2% which people keep referring to mainly consist of IT professionals.

Is anyone's goal to take over the desktop with Linux? You seem confused about what Linux and open source development is in general. Looks like most of your complaints are about KDE's Wayland support, which says absolutely nothing about Wayland itself.

Wake me up when Wayland starts offering a GUI for modifying monitor modelines. In the world of Windows this feature has been available since Windows 95. It's year 2020 and people still often need to wrestle with GTF to get their monitors refresh right right in Linux. In Windows I go to NVIDIA control panel, add a custom refresh rate for my monitor, bam, DONE. Why is it that each Wayland compositor has its own hardware configuration which should in theory apply to all of them? Again, no APIs for that, each compositor has to reinvent the wheel. Super duper Wayland, super duper KMS, the super duper Linux kernel and we still have nothing alike. God, this cesspool of poorly designed and implemented technologies in Linux will probably never end. Everything is done to drive the average Joe away from Linux. And the poor sod had to understand why in his Fedora part of applications cannot share clipboard between them since some of them are native to Wayland and some run in Xwayland.

Wake up, Sway has this. Don't know about KDE/GNOME as I don't use either, but there's nothing stopping them from implementing it other than perhaps it's 2020 and there's EDID so it's not needed. Average Joe's computer is delivered with Windows on it, he doesn't have to worry about custom modelines at all.

https://blog.martin-graesslin.com/bl...and-clipboard/

Breakage on top of breakage. Yeah, I know it's an old post, the problem has probably long been solved but why Wayland DE's have to go through circles of hell to get everything right and they still cannot get some features right because "Effing modern light protocol"? Why not design a protocol which is rich out of the box and takes very little effort to write applications for? Where's GDI for Wayland? Why does each Wayland app have to render fonts, vector UI elements and everything else by itself? Why does we need to recreate graphics rendering in GTK, Qt, Enlightment, etc. etc. etc.? How can we tell all those disparate toolkits that they have to change their settings to look alike? Questions, question, questions.

These questions have been answered countless times and are easy to find. Your issue seems to be lack of understanding or lack of will to understand what Wayland or a toolkit is and who develops them.

In the end the user doesn't bloody care! His Windows 10 PC works perfectly out of the box. He doesn't have to understand how the system works because it works.

Why can't he stay on his Windows 10 PC? I don't get it.

Sorry, I'm still not convinced. Yes, X.org/xlib needs to be replaced. Wayland doesn't seem like a good solution overall.

Based on what you've posted it doesn't seem like you're trying very hard to understand the problems Wayland is solving. It's ok, because it's nobody's job to convince you.

[birdie]

Based on what you've posted it doesn't seem like you're trying very hard to understand the problems Wayland is solving. It's ok, because it's nobody's job to convince you.

"Wake up, Sway has this."

I don't bloody care about Sway!! I care about simple APIs being readily available for anyone who wants to write a window manager/compositor for Wayland. Right now each compositor has re-implement a ton of things which should have been part of API for f sake. You know the biggest strength of all successful operation systems? Bloody rich stable, supported APIs.

Whatever it's solving it's doing it extremely poorly while complicating a lot of things which should have been thought of from the get go.

Yeah, I don't understand anything. Yeah, I'm a poor sod still using X.org because I don't care about "modern" technologies - I want something which "just works" and, sorry, but Wayland is nowhere near it. Again, I don't see myself using Gnome, what a stupid idiot, right? Wayland replaces a whole lot of things while making it extremely difficult to reimplement them and making all desktop environments and window managers/compositors reimplement them. So, instead of providing good APIs not only for outputting raw pixels (what a great innovation, really), but everything which is required for a rich desktop experience it's giving us a bare skeleton. Wow. Great. 11 years old ... fast, lean and secure nothing.

I mean why do we need Wayland at all then? There are Linux applications which use KMS directly for f sake, e.g. mpv. Wayland doesn't provide almost anything on top of KMS for f sake. What a joke. A graphics subsystems. LMAO. I mean let's throw Wayland away as well and start using KMS? Right?

Innovation, my arse: "Let's remove almost everything from the graphics system and let everyone implement pretty much all the functions required for desktop", LMAO.

[birdie]

Just read this from a Wayland developer https://drewdevault.com/2019/02/10/W...-debunked.html

The core Wayland protocol does not define a mechanism for taking screenshots. Here’s another thing it doesn’t define: how to open application windows, like gedit and Firefox. The Wayland protocol is very conservative and general purpose, and is built with use-cases other than desktop systems in mind. To this end it only implements the lowest common denominator, and leaves the rest to protocol extensions. There is a process for defining, implementing, maturing, and standardizing these extensions, though the last part is in need of improvements - which are under discussion.

I mean the whole post is a series of facepalms. You have to be alternatively gifted to enjoy this hot mess of a brand new graphics subsystem.

[birdie]

And here's the wonderful Wayland architecture:



See? A Wayland compositor is basically talking to KMS/evdev directly. LMAO. Glorified KMS.

[browseria]

Just read this from a Wayland developer https://drewdevault.com/2019/02/10/W...-debunked.html

I think you missed this part, it is the second paragraph of this blog post:

The art of hating Wayland has become a cult affair. We don’t need to put ourselves into camps at war. Please try not to read this article through the lens of anger.

Can you give a non-angry reason why "a glorified KMS" is a bad idea? Or not a "simple APIs being readily available for anyone who wants to write a window manager/compositor". Or why wlroots doesn't fit this bill? Or Mir for that matter?

Finally the last paragraph of this blog post is worth re-reading again:

Another thing I want to note is that Xorg still works. If you find your needs aren’t met by Wayland, just keep using X! We won’t be offended. I’m not trying to force you to use it. Why you heff to be mad?

[birdie]

Can you give a non-angry reason why "a glorified KMS" is a bad idea? Or not a "simple APIs being readily available for anyone who wants to write a window manager/compositor". Or why wlroots doesn't fit this bill? Or Mir for that matter?

You offer wlroots as the counterargument that Wayland doesn't have good rich APIs for a graphics environment? LMAO. A project which has no relationship to Wayland and was created by the Sway developer. That's not a counter-argument, dude. You're proving me right.

This discussion is a poor joke.

[browseria]

A core tenant of good programming espoused by Ken Thompson is to isolate scope, to modularize, the unix credo "do one thing and do it well" comes from this. The main complaint of the X developers over the years is that X violates this principal. Wayland corrects that.

You are right that the surface area of Wayland isn't as big or as comprehensive as X, but it was never meant to be by design. Because it is good architecture. It is durable and flexible technology. Things that X are not and could never be. The things you are looking for are necessarily going to be at a higher layer of abstraction than the Wayland protocols, by design. The fact that people have decided to do multiple implementations independently rather than work on shared libraries is unfortunate, but also not atypical in linux development.

I understand that you're upset this is the case, and I even share your dismay, but what you are asking for isn't achievable, any more so than everyone dropping their independent implementations of package managers (or desktop environments, or windows managers, or compositors, or ...) and standardizing on just one. This is both the strength and weakness of the linux environment.

The test of a good project *is* that other people "that have no relationship to it [Wayland]" can take it and build on top of it in a meaningful way. Doesn't the very existence of wlroots, or the quick refactoring of Mir into a Wayland wrapper proof the pudding, so to speak?

[Alliancemd]

I have been waiting for a lifetime for VAAPI support...
But please don't call it "2020 - The Year of Linux on the Desktop".

[GrayShade]

You are right that the surface area of Wayland isn't as big or as comprehensive as X, but it was never meant to be by design. Because it is good architecture. It is durable and flexible technology. Things that X are not and could never be. The things you are looking for are necessarily going to be at a higher layer of abstraction than the Wayland protocols, by design. The fact that people have decided to do multiple implementations independently rather than work on shared libraries is unfortunate, but also not atypical in linux development.

That's a good way to put it. X11 is everything and the kitchen sync. Drawing API? Check. Fonts? Check. Settings? Check. Remoting? Check. Bitmaps? Check. OpenGL? Uh, sure. Tons of extensions for things that people expect from modern systems? Yeah, they're there.

Wayland just gives you a window on the screen, an OpenGL context and input events, then gets out of your way. It doesn't care about text rendering or program settings. That makes it more durable in the long run.

And it works. I've been using it for some years now. Does it have support for clipboard managers? No thanks, my DE has enough clipboard bugs as it is. Does it have built-in networking? No, but I'd rather use RDP, VNC, waypipe or even SSH X forwarding than have one implementation baked in.