Sorry, but X11 works on pretty much everything (except, of course, Microsoft Windows, quelle surprise) whereas Wayland is still in its infancy.

Come back to me when it has multiple decades of interoperability like X11 has (OK, I carry it over SSH tunnels but that's about it).

Come back to me when I can run it across a network like I do with Xorg every day.

X11 is awful and needs to die. It's a networked GUI that's terrible at being a networked GUI.

Running remote desktop over a 512KBit dsl connection via an SSH tunnel about 20 years ago was usable, the same SSH tunnel running an X app was unusable.

One of these is a system designed for networking and remote display from the get go, on hardware that was much much much slower than that of the time, the other is RDP, which was Citrix's bolting on networking to windows GDI.

RDP works well, X doesn't.

However, it's worth pointing out there are multiple X servers for windows. If you don't mind pissing around the the vanilla XOrg one works (or used to), otherwise get something like mobaxterm for better integration.

Yes, X11 remoting does not really work smoothly unless you are on the same LAN. Which was its original use case anyway. On the other hand, an X11-based desktop over TigerVNC works just fine over a VPN tunneled through residential DSL (sans any sound, but for my uses I don't even need it).

I have so far steered clear of Wayland, but one of these days I will have to bite the bullet. Hope by that time it has some working remote feature.

I must confess my hatred of X goes back a long way and may not be entirely rational. I just grew up with decent GUIs (Hi RISC OS) and using X always feels like using treacle for me, there seems to be a perceivable gap between any interaction and it's result, that I don't feel with other GUIs.

Wayland feels lovely in comparison.

Also, because X never actually had the decency to define a UI I can still experience the joys of Motif in 2023. I just discovered ddd the other day, lovely tool, but under wsl very 1970s motif'y.

You have this juxtaposition of a really powerful piece of software that looks old. Sure this doesn't really matter from a function point of view, but it's a jarring experience usability wise.

Had X had the decency to define (and render) the controls too then I suspect that the mishmash of horrible and still prevalent UIs would've been avoided.

Hope by that time it has some working remote feature.

Weston has RDP support. However [1]waypipe looks interesting too.

[1] https://gitlab.freedesktop.org/mstoeckl/waypipe/

Depends on the X app. If the app designer 'knew' the graphic card was local and he could wait for a response after drawing each pixel then network performance will be awful. If the app designer did some thinking in advance to send as many commands as possible before waiting for a response then network performance is fine.

I always assumed sending an entire window full of pixels would be slower than sending the commands that draw them. To be fair, I never tried because X was fast enough.

* firefox needs the --no-remote option when used remotely. That one caught me by surprise.

It's been some time since 'commands' were sent - i.e. drawing lines, placing fonts etc with complex drivers in the days of yore. Now it's more efficient to send pixel data.

You're also talking at a level below application development - no-one has talked raw X protocol for years, that's down to the underlying libraries.

> no-one has talked raw X protocol for years,

Probably almost never, because that is what XLib is for. Applications and toolkits run on top of it. I believe it by default tries to combine consecutive X11 protocol commands.

It is true newer X11 programs do the drawing and font rendering themselves and push pixels, and that really breaks the original idea. I first encountered it with some bloated Java program, that was totally impossible to remote the plain X11 way. The GUI library redrew the entire window for every change (not sure if Java still does this, it was 20 years ago). Using it with VNC server worked. It looks like a X11 server to the application, but sends only the changed pixels (or rather changed blocks) to the remote client.

> no-one has talked raw X protocol for years

No, but X11 has provided the infrastructure for NoMachine (currently the fastest modern competitor to RDP) and multi-session VNC to exist.

Without X11 we would be back to Windows 95 where the best you can do is screen scrape. Wayland is OK because it allows for multiple sessions in that regard. But unfortunately we are still back to "dumb" rasters, aka VNC. This will never be as efficient as utilizing a subset of an intelligent protocol.

If you have time; do compare NoMachine's X11 based offering against VNC or Waypipe. It is a little depressing as to the difference.

Also, the *fastest* solution we have, even beating Microsoft's annoyingly impressive RDP is X11 TCP/UDP (not tunneled through SSH) with an appropriate Xt-based toolkit like Motif. Looks like "old" but its performance far outpaces more modern solutions. For many use-cases, this really is the solution, still after all these years. And we are effectively going the wrong direction if we ever want to remedy that.

We might not care about remote desktop solutions personally (I prefer SSH); but someone who develops awesome software that we both use, just might need them... Hopefully Asahi Linux can one day fix this shortcoming.

Well - did you see any of the weaknesses noted in the article (or post)?

Are any of those important to you on apple hardware?

This article comes with a complementary bucket of free sand for all readers. Feel free to stick your head in it.

The Apple Silicon market is a closed one, without the free choice of mix-and-match GPUs and displays found in commodity PCs.

I'm not sure this is true is it? You can attach any display to an M1 mac as they have USB-C and HDMI (I have 2 dells attached to mine), and as we're talking about Linux surely any PCI-E GPU connected over thunderbolt would work if there are Linux drivers for it? (Mind you if it uses a binary blob you would need an ARM64 one).

Apple doesn't provide third party drivers for M1 silicon on mac os, but there's no reason one couldn't be written for Linux.

It basically boils down to 'Wayland is pretty much there, but you're going to have to do some work and may have to change your drivers'. It's not yet mature but without people attempting to switch it probably will never be there.

The problem is that if you have an edge case or complex configuration Wayland will not be a solution out of the box, and some of the commentators have a hard time dealing with real life constraints to ideological purity. For instance, the idea the resolution should not change, which doesn't sit well with games and full screen video playback.

Going to look at moving this FreeBSD box to Wayland - AMD and Intel should support it, and Nvidia support is now in testing for more recent cards (most cards younger than about 6-7 years old).

I do like X, but remoting can be a pain, and a non mainstream WM is used (I prefer cwm) it really does not play well with poorly written pieces of software that assume the existence of a full desktop environment, complete with dbus and APIs to add desktop icons.

There is, as ever, a 'Unix is not just Linux' problem. Wayland isn't in a bad shape in FreeBSD, is there to some extent in NetBSD, and due to lack of resource, not a priority for OpenBSD at the moment.

This guy is using his recent position of popularity to spread FUD about Xorg like a little child. He really is either a massive moron, or a snake pushing his agenda. These key points (echoed from my posts on Phoronix).

> "along came xf86-video-modesetting, which was supposed to be the future"

It is *still* the future. Most Wayland compositors use this exact Xorg driver as a starting point. They merely call it a "backend" rather than "driver".

> "but modesetting couldn't even handle tear-free output until earlier this year (2023)"

Uhh... so its still solved then... Perfect. Wayland is no longer needed. Problem solved.

> "That was in 2015. And here's the problem: X was already dying by then."

Is just nonsense weasel speech. What is dying? Linux? Grow up

> "Deal with it. The major players in desktop Linux have decided it's time to move on from Xorg"

No they haven't. A couple of desktop environment projects have perhaps to facilitate a tighter integration. Many Linux users don't use a desktop environment.

Take your Apple-style "Deal with it" nonsense and shut the fsck(8) up please.

Translating his whole drivel is simply:

> "Asahi Linux as a project doesn't have enough manpower to get the current industry standard display system working"

And that's fine. There are plenty of incomplete system ports in the same boat. You don't have to lie about your shortcomings. You are doing your best, people don't blame you for that.

@karlkarl

An industry standard in Linux? Really?

Haha, god no!

Established by UNIX but greatly benefitted from by Linux. That is why it would be so bizarre to think this level of industry standardization could happen again unaided, and within the Linux community.

They won't realize what they had until it is too late!

If Asahi Linux can't get Xorg working on recent Apple hardware, I know OpenBSD and Xenocara (X11) has made excellent progress. Perhaps migrate to that instead. Fantastic security to boot!

Quote from OA

"Xwayland will be supported, so that X.org apps will run under Wayland based desktops"

So when these Apple Silicon Macs start to appear on the second hand market, people will be able to run any of the software they need. Sounds fine as a starting point. I'm sure people will start tinkering...

Question: will this work on Asahi for the version of ARM that Apple have produced have a knock on to lesser kinds of ARM by generic laptop makers? I'd like some of this 15 hour battery life.

Icon: obvious really