[ Xorg vs Wayland Thread ] "Improve the crap" versus "Create a new Crap" ?

Re: Gaming on Linux - I'm a very avid gamer. I was gaming on Linux back in Loki days on Slackware 10. It used to be mainly for excitement, brain tease puzzles and such and sometimes for benchmarking but since my stroke damaged my left side, it has become physical therapy, especially on somewhat older games that I'd played for years. The fast action on some games and the precision required on them as well as some puzzle games triggers muscle memory and counters the natural instinct to favor damaged limbs.

Granted the newest games I am currently playing are from ~2020 but they are AAA titles and about half of them have no Linux version and must be played in Steam or WINE. Probably because I build custom kernels on top notch hardware in Linux and because my Main PC is multiboot, for a time, I installed games on an NTFS partition (slightly favoring Windows and by extension, DirectX) so I could run gaming benchmarks to compare results on Linux vs/ Windows. In every case Linux native versions tested much higher, often as much as 18%-20% higher. "Windows-only" games via Steam or WINE test as good or better (12% higher average FPS) than on Windows. There's no way to objectively test how much real world gain can be achieved by running the games from ext4 vs/ ntfs partitions, but they feel noticeably better than on ntfs.

Bottom line is Valve/Steam changed everything in both a sudden quantum leap and a long, still growing, upward trajectory. The success of the handheld SteamDeck seems to show even those who run Windows as their Main on PC, don't tend to switch the default Linux OpSys on SteamDeck even though that is possible and allowed. I don't know that much about SteamDeck other than demos and reviews because I don't own one. I'm avid but not THAT avid and being unusual for me, not likely especially helpful as therapy.
It isn't DirectX that props up Windoews only gaming, it's some game developers and Valve is nibbling away at their profit motives.

Re: Xorg vs/ Wayland - Being aware that Xorg has evolved into a spaghetti code, patchwork quilt I was interested from the moment I read about Wayland. I sought out videos of symposium talks by developers who often apologized and begged for patience (because "it's really hard!") and it obviously is. It has been calculated that if the Linux kernel had instead been developed by any corporate team of developers, it would have cost well over 20 Billion dollars. Since IBM paid 34 Billion dollars for RedHat that seems a rather accurate assessment. I don't know if (or even if it is possible) to calculate the vale of Steve Jobs initiating the adoption of the Linux/BSD environment to create OSX but I'm confident it's absolutely staggering.

Despite the fact that I seriously doubt I will even be very interested in trying out Wayland for a few years yet, due to it's current restrictions and faults, I do think it's develoment is important. There is nothing quite like intense competition to negate the complacency that fosters resting on laurels.

i welcome wayland if it benefits my games, and yeah, i am a gamer also.
gaming these days with Linux is good, and i have observed better framerates at linux, just like enorbet has.

i have to say i dont know much about wayland, ill have to about it.

Originally Posted by enorbet Granted the newest games I am currently playing are from ~2020 but they are AAA titles and about half of them have no Linux version and must be played in Steam or WINE. Probably because I build custom kernels on top notch hardware in Linux and because my Main PC is multiboot, for a time, I installed games on an NTFS partition (slightly favoring Windows and by extension, DirectX) so I could run gaming benchmarks to compare results on Linux vs/ Windows. In every case Linux native versions tested much higher, often as much as 18%-20% higher. "Windows-only" games via Steam or WINE test as good or better (12% higher average FPS) than on Windows. There's no way to objectively test how much real world gain can be achieved by running the games from ext4 vs/ ntfs partitions, but they feel noticeably better than on ntfs.

One of the most important things for gaming on GNU/Linux also, is the ability to remove clutter and set up a pure gaming environment. You could run a GNU/Linux distro only for gaming, that shuts down everything else that is not needed. furthermore you could run a desktop environment or window manager with only gaming relevant stuff on and nothing else. You can't do this on Windows. This also has great potential to optimize gaming performance, including RAM and CPU.

I think, in regards to Xorg and Wayland, one of the issues with Xorg is the graphics stack switch if you want to game. Games don't run natively through Xorg, but switches display. This is not a huge issue however, but it might be an issue if you want to game and also use the window manager/desktop for some gaming related things (like looking at performance data for your computer etc, or looking at a browser window for some game guides, information or whatever). I don't really know enough about it, but since Wayland is EGL which is OpenGL etc, this should probably work in a better way. Games and desktop integration could/should probably work better than on Xorg.

In any case, I'm certain that the potential for games is better on GNU/Linux than on Windows, as Windows can't be customized to the degree GNU/Linux can, including removing non-gaming clutter and overhead from other running programs and stuff. And that's just the obvious. I think also things like drivers etc, frameworks and so fourth has the potential to run games better on GNU/Linux than on Windows, and I suppose some part of that is the display manager.

Ps. Ofcourse, a good solid Wayland basic window manager wouldn't hurt either. Or even if the time comes, a gaming specific window manager.

Cool post, zeebra, but I find that a custom kernel that fits hand-in-glove with your hardware and tweaks like setting vm_swappiness much higher than the common default of "10" makes differences that swamp any WM/DE difference, given decent hardware and that's the major difference to Windows. Microsoft ( or TinyLimp ) simply increases hardware requirements while it drops support for others, and makes unalterable default setting designed for the lowest common denominator supported hardware as it minimizes unpaid tech service. Linux allows a user to break stuff but when you know (or are willing to test) your hardware's capabilities you can max it out.

Additionally in my case, my common workflow stays largely involved in some sort of MultiMedia so I have no need for a "gaming specific" OpSys. What works for games also works for DAW and also in Video playing, transcoding and editing. For example my GPU is so old that when I run the Shadow of the Tomb Raider Benchmark it tests and shows that game is literally 99% GPU bound and that's with an i5-10600K with the game at mostly Ultra settings. At 2560x1440 near full Ultra I consistently get 98+% FPS even inside full KDE Plasma. I get around 101 in Fluxbox and routinely under 84-85 in Windows.

Originally Posted by enorbet Cool post, zeebra, but I find that a custom kernel that fits hand-in-glove with your hardware and tweaks like setting vm_swappiness much higher than the common default of "10" makes differences that swamp any WM/DE difference, given decent hardware and that's the major difference to Windows. Microsoft ( or TinyLimp ) simply increases hardware requirements while it drops support for others, and makes unalterable default setting designed for the lowest common denominator supported hardware as it minimizes unpaid tech service. Linux allows a user to break stuff but when you know (or are willing to test) your hardware's capabilities you can max it out. Additionally in my case, my common workflow stays largely involved in some sort of MultiMedia so I have no need for a "gaming specific" OpSys. What works for games also works for DAW and also in Video playing, transcoding and editing. For example my GPU is so old that when I run the Shadow of the Tomb Raider Benchmark it tests and shows that game is literally 99% GPU bound and that's with an i5-10600K with the game at mostly Ultra settings. At 2560x1440 near full Ultra I consistently get 98+% FPS even inside full KDE Plasma. I get around 101 in Fluxbox and routinely under 84-85 in Windows.

Well, pretty much anything that people have made a large effort with in GNU/Linux outperforms the Window counterpart. I'm pretty confident that would apply to gaming as well. I hate to add graphics stack to that, considering the giant effort that is MESA3d, but well, the graphics drivers with the Linux Kernel still have some way to go. But part of this is also the display server and things like game/desktop integration and high performance/quality graphics.

Personally I think if these things got some traction, gaming under GNU/Linux would be a far better place for gamers than Windows. It really needs some momentum somehow.. While Windows is a general purpose OS, GNU/Linux variants can also be made into a game specific OS (distro). Just that alone will guarantee better performance.

Add to that, gamers are above average incentivized and able computer users, switching from Windows to GNU/Linux as the main PC gaming platform really should be a no-brainer. It's kind of the scenario of the chicken and the egg though...

On gaming, surely you can learn from the OS choices made by the manufacturers of gaming consoles. I'm no gamer but there's the Playstations, Xboxes & competitors. What do they teach you?

Originally Posted by business_kid On gaming, surely you can learn from the OS choices made by the manufacturers of gaming consoles. I'm no gamer but there's the Playstations, Xboxes & competitors. What do they teach you?

what operating systems game consoles use? Modern game consoles generally use custom operating systems designed specifically for their hardware and gaming purposes. Here's a breakdown of the major ones: Nintendo Nintendo Switch: Uses a custom operating system called Horizon, which is based on the FreeBSD kernel. Sony PlayStation 4: Uses a custom operating system called Orbis OS, also based on FreeBSD. PlayStation 5: Employs a customized OS likely building upon the foundation of Orbis OS. Microsoft Xbox Series X/S: Utilizes a modified version of Windows 10 Core tailored for gaming and console features. This provides strong app compatibility and ties into the broader Xbox ecosystem. Xbox One: Initially ran a modified Windows 8 Core, but this was transitioned to a customized Windows 10 Core system after updates. Older & Retro Consoles Many older consoles had more rudimentary systems designed purely for running games rather than the feature-rich operating systems we see today. Often, they'd use simple firmware to execute game code directly. Key Reasons for Custom Operating Systems Optimization: Custom OSes allow console manufacturers to fine-tune them for their specific hardware, maximizing performance and resource utilization within the limitations of the console. Security: Closed software environments provided by custom OSes help limit piracy and unauthorized modifications. Unified Experience: These systems enable a consistent user interface and feature set across all games on the platform. Let me know if you'd like more details about a specific console's operating system!

thats what google gemini says about consoles

It stands to reason. The GPL is an impediment to proprietary software, but BSD isn't hampered that way. Earlier Sinclair ones I knew about used some programs from CP/M. The Sega Megadrive had a 68000 chip, which would have handled a BSD kernel. Nintendo were inclined to be behind cpu-wise in their technical spec.

https://www.osnews.com/story/138565/...one-paragraph/

The state of X.org and Wayland in one paragraph Wayland and X.org are both part of freedesktop. Whatever maintenance is still happening on X.org is mostly being done by people who primarily work on Wayland. There isn’t some kind of holy war going on between The Wayland Developers who want to kill X.org, and The X.org Developers who believe it is great and want to keep it. They’re nearly all the same people, and they all want X.org to die. AFAIK there isn’t anybody who is actually clamoring to *do the work of maintaining X.org upstream*. There are people who don’t want it to die because Wayland doesn’t yet have the features they need or the NVIDIA proprietary driver doesn’t work well on Wayland or whatever, but AFAIK, none of those people is actually volunteering to maintain X.org long-term. ↫ Adam Williamson There’s really no clearer summary of the current state of affairs than this.

I've had somewhat limited exposure to both codebases (more so X11) and I can tell at a glance why Wayland was created. It's basically a successor to X11 that had to be written from scratch - their APIs are basically the same, only Wayland doesn't make you want to slam your head through your keyboard every time you use it. Ok, not really, but if you're mediocre at X11, you'll be great at Wayland - which tells you how full of quirks and unexpected interactions X11 is.

In other words, X11 can't be improved. Some fundamental design decision were made at the beginning which basically necessitate a rewrite:
- A lot of features X11 treats as luxury are assumed to be supported on most modern systems (like transparency) - this makes it so you have to write much more code in order to add special support for something that's never not going to work.
- X11 has 2 different input systems and event types - in fact you can represent mouse scroll in 3 different ways.
- X11 documentation is probably the worst series of texts ever created in the human history - like variables having names opposite to what they should be doing or do, and descriptions which only make sense if you started your development journey with punch cards.
- X11 is trust based, which means any client (program) can make any other appear buggy and crash, which then requires coordinated effort to figure out "whose fault it is".

I could go on, these are just from my own bad experience with X11, there's a bunch more online. Trying out Wayland felt like a vacation after working on a larger X11 feature. Sure, there's a bunch of places it could still improve, but it's mostly here. The things that are implemented work orders of magnitude better than they do in X11 with far less code.

I'm adding mouse events to conky - it took me ~50LoC to add them to Wayland, I wrote over 200LoC for X11, it's still not fully implemented, a whole lot more complicated, and there's bugs I don't even know of yet (Wayland just worked as expected with minimal effort & code). Higher productivity means developers will be able to add much more features quicker which will greatly accelerate addition of new features.

Welcome to LQ, @caellian.

Your post took me back to the 'bad old days' of Xfree86 vs Xorg, early drivers, ISA buses and the sort of messing we all did to get X starting , never mind running well. For some hardware reason, ISA buses couldn't provide any answers. Hardware didn't exist unless it was in Xfree86.conf or Xorg.conf. The 16 interrupts were fiercely contested. Scripts were in bash or Perl. We crafted Modelines to help our massive tiny-screened monitors land their low-res images on the visible screen section. /sys, dbus Plug & Play, and things happening automagically were yet future.

It's nearly inconceivable that any code base could come from there to here without suffering in the process.

Originally Posted by business_kid Welcome to LQ, @caellian.

Thank you!

Originally Posted by business_kid It's nearly inconceivable that any code base could come from there to here without suffering in the process.

That's a huge thing Wayland got right I think - every protocol is versioned (including the core ones) and can "shed off" legacy features when the time comes, in a way that allows both apps and compositors to support a range of versions. Some parts of X11 do this, but the core protocol which is responsible for majority of the session doesn't.

I got into Linux relatively recently (6/7 years), so thankfully, things mostly just worked for me (unless I broke them myself).

My X11 seems to work. I have run it on the BSD's, Slackware and Debian for many years - no issues at all. I also remember the old days of XFree86 vs X.org and writing xorg.conf files. But for quite a few years now, the xserver has started up at the correct native resolution, with hardware acceleration, without me having to configure anything at all.

In my view, X.org is being talked into obsolescence by those following a corporate agenda, as was the case with sysvinit. These people managed to convince the majority that the situation had been intolerable for years and this new fangled thing is the answer. This is the danger in blindly following freedesktop.org.

If you want to focus on a huge chunk of bloated, obsolete and poorly audited code, then look no further than the Linux kernel itself. Or gcc/g++, or the new fangled systemd, which is becoming more bloated and full of feature creep and holes as time goes on.

Before you look at those however, you should probably pay more attention to the massive codebase of the chromium or firefox browsers and study their structure and implementations as well. You may also want to read up on glibc vs musl and you will come to realise that, though glibc is bloated, musl has its own limitations.

Doing some research on the Wayland protocol and compositors will be similarly enlightening.

I'm with you 100% on systemd.

My Xorg works too, and from what I read, Wayland has not implemented stuff or done it poorly. That said, I do accept that Xorg is clunky code. But I expect to move to Wayland when necessary.

But coders write code. What they do very poorly and reluctantly is edit code, remove or alter features that somebody else wrote, decide what is unnecessary and junk it. It is an absolute minefield for them. Better to rewrite it. How many bugs have been patched in X? Or found in testing so they didn't need patching? That has to be rediscovered.

The Unix philosophy is "Do one thing, and do it well." X, Firefox, & chrome are 3 programs that fall at that hurdle. Systemd fails spectacularly. It would be better to section them, as otherwise, this is going to keep happening, as patch creep is inevitable.

Getting deja vu with this Wayland debate: https://www.linuxquestions.org/quest...ml#post6469000

What many perhaps missed in that last debate, is that Wayland adoption is in fact being forced (by the planned obsolescence of xorg and by way of dependency of desktop environments such as e.g. gnome, with distributions providing a systemd -> wayland -> gnome dependency chain as the "default" or only option - which was a big part of systemd proliferation, and Wayland is in fact corporate backed, as with systemd.

In terms of X.org forks, to continue maintenance - that may well happen once X.org completely abandon it. So despite the talk about no one "stepping up", which is the usual propaganda (no one was "stepping up" to maintain sysvinit either, except they did "step up" and it is being maintained by Jesse Smith (of Distrowatch) and others) you may well see X.org forks in the future.

To quote Theo de Raadt:

The writing has been on the wall a very long time that some people believe their role in the ecosystem is to reduce software choice and push everyone into vertical software monocultures

https://marc.info/?l=openbsd-ports&m...4852204712&w=2