X.Org's Xvidtune Sees New Release After A Decade

[Quackdoc]

Surely that's not something X11 can relate to. ^^

The reality is that the Linux (or Unix) display ecosystem was always lacking features, barely keeping up with what is necessary.
X11 itself is only an accumulation of extensions, because the core protocol is broken and lacks pretty much everything.

So from that point of view, you could say that Wayland really does follow the footsteps of its daddy.
Well, ok, at least the protocol extensions are now handled a bit saner, but the basic problem isn't really that different in that regard.

Ah sure, missing features, causing fragmentation. Lets talk about that, what features is a protocol missing that handles the communication between clients and a compositor?
For the last 25 years, that was quite normal. Metacity vs Compiz, xfwm4 against compton etc.

yes, fragmentation has always been an issue on linux, but the goal should be to make it better not worse. you could easily make an OSK on X11 and have it support all the major compositors, even this isn't possible on wayland in it's current state. to make an effective OSK you need compositor specific features. one of the key parts of accsessibility is being able to pick and choose the parts an individual needs, wayland fails hard here.

Wayland is an active regression in this regard. the proper screen recording tools are often either barebones cross platform via flat+pipewire, or implement compositor specific functions. with docks they are limited to the platform they are made for, docks that were once at least working across multiple DEs were DE specific. wayland is a massive regression in this regard. and while yes, it does show some signs of improving, it's very slow. and in other regards, almost zero progress is being made.

[Weasel]

Following your train of thought, lets remove any restrictions, lets log in everyone as root, and problem solved, since security is paranoia we don't need it, right?

If you want to pick extremes, then logging in as root is definitely infinitely better than not having any input, internet connection, or personal files at all. Sure, it sucks, it's insecure and a huge risk, but at least you can get some job done. The alternative is not having a PC at all, which is infinitely worse.

So even in extremes, lack of security is better than full-on paranoia. Functionality always trumps security.

That said, the best approach is a middle ground. Now I fully admit that X11 lacks security in some cases, and it's not ideal, but as I exemplified above, lack of security is infinitely better than paranoia. So while X11 could improve a bit, it's still miles better than Wayland's approach.

Here's what an ideal security would look like on a display server/compositor: Allow full access like X11 between apps running under the same user or when both are unsandboxed, but not different users. Querying windows positions between apps of different users can be disabled/blocked, that's fine, or between sandboxed apps.

The point here is that when you run two apps with same user or unsandboxed, you trust your god damn apps. Period. I don't need to be babysit by a pathetic display server protocol paranoia and hysteria.

Why would I even give a shit about querying my window positions if the app, running under the same user, can access ALL OF MY FILES?

NOTE: I do sandbox my browser for example. It can't access my personal files because it's untrusted. So in this case, it would be totally fine if X11 blocked its access to apps running under my main user account. Most apps that have access to the internet should be sandboxed.

Wayland just goes too far and it's crippled by design. Insanity.

[andreduartesp]

If you want to pick extremes, then logging in as root is definitely infinitely better than not having any input, internet connection, or personal files at all. Sure, it sucks, it's insecure and a huge risk, but at least you can get some job done. The alternative is not having a PC at all, which is infinitely worse.

So even in extremes, lack of security is better than full-on paranoia. Functionality always trumps security.

That said, the best approach is a middle ground. Now I fully admit that X11 lacks security in some cases, and it's not ideal, but as I exemplified above, lack of security is infinitely better than paranoia. So while X11 could improve a bit, it's still miles better than Wayland's approach.

Here's what an ideal security would look like on a display server/compositor: Allow full access like X11 between apps running under the same user or when both are unsandboxed, but not different users. Querying windows positions between apps of different users can be disabled/blocked, that's fine, or between sandboxed apps.

The point here is that when you run two apps with same user or unsandboxed, you trust your god damn apps. Period. I don't need to be babysit by a pathetic display server protocol paranoia and hysteria.

Why would I even give a shit about querying my window positions if the app, running under the same user, can access ALL OF MY FILES?

NOTE: I do sandbox my browser for example. It can't access my personal files because it's untrusted. So in this case, it would be totally fine if X11 blocked its access to apps running under my main user account. Most apps that have access to the internet should be sandboxed.

Wayland just goes too far and it's crippled by design. Insanity.

I was asking for a middle ground, privileged access to my keystrokes, but you said a keylogger without privilege to be ok (this is the way wine does things in X, on macOS they use a different approach and will use another approach on Wayland), so I just followed your thought, if a keylogger with no privilege is ok, a keyboard driver is ok too. Another thing, if restricted access to the print screen is ok then access to the video driver is ok too.

Wayland is just a video protocol, and one part of modern Linux desktops, your files can be preserved using a sandboxed approach like flatpack. But I think any password you are typing, or sensible data will not be preserved as is on your disk.

Wayland was created with sandboxed apps in mind, so if you are using it without a sandbox you are losing security, it's your choice as someone who knows the difference between a sandboxed app and unsandboxed, but the majority of users don't know, they just want to use the computer with their data as safe as possible

I don't know if you are a software developer, but any platform will give you tools to solve problems, with a different approach, you can do the same thing on osX, Windows, Android, x server or Wayland, but using different approaches, one platform doesn't need to be attached the way other works to be a good platform, it just needs to give you tools to solve the problems, not tool to mimic another platform, this is why it's not X12. And Wayland is an open platform, if you think you need a protocol to know the position of a window, you can propose the protocol, and if accepted all windowed compositors (yes the basic protocol does not require an environment with windows, the protocol was designed for mobile as well) will implement it, maybe XDG is the right place for a protocol like this.

And finally, Wayland prevents wine (or anyone else) to put a keylogger on my machine, don't prevent wine to register a global shortcut, Wayland just changed the way wine developers will do things, using a more secure way, if they use it correctly, from an end-user perspective it will seem that nothing has changed, or better, it will be possible to control any global shortcut in one place, instead to use 5 different software to configure global shortcut from every framework possible

[ssokolow]

And finally, Wayland prevents wine (or anyone else) to put a keylogger on my machine, don't prevent wine to register a global shortcut, Wayland just changed the way wine developers will do things, using a more secure way, if they use it correctly, from an end-user perspective it will seem that nothing has changed, or better, it will be possible to control any global shortcut in one place, instead to use 5 different software to configure global shortcut from every framework possible

Wasn't the Win32 API way to do global shortcuts something like "Set a global event hook and filter out the keys you want?" ...as in more XGrabKeyboard and less XGrabKey as far as retrofitting sandboxing and user permission prompts goes? (tracing back to its being designed as a familiar upgrade for Win16 developers.)

I seem to remember reading something to that effect.

[mangeek]

...any platform will give you tools to solve problems, with a different approach, you can do the same thing on osX, Windows, Android, x server or Wayland, but using different approaches...

This is how I see it. I'm not a software developer, I'm a tech person daily-driving GNOME on Wayland for years just fine, and it works better than X11 for some very practical things for me. I think there's a communication problem here; some people think Wayland is broken because it doesn't do some things, and others (myself included) think those things are out of scope for what Wayland should do. The accessibility example is a great one, but I don't think the problem is the display protocol or compositor, I think it's the lack of a reference API for accessibility that developers can target from their apps regardless of whether they're targeting X11, Wayland, GNOME, KDE, IceWM, Android, macos, or Windows. The compositors could all use that API, but the accessibility API shouldn't be part of the display protocol.

[Berniyh]

I was asking for a middle ground, privileged access to my keystrokes, but you said a keylogger without privilege to be ok (this is the way wine does things in X, on macOS they use a different approach and will use another approach on Wayland), so I just followed your thought, if a keylogger with no privilege is ok, a keyboard driver is ok too. Another thing, if restricted access to the print screen is ok then access to the video driver is ok too.

Are global hotkeys actually supported in wine? It's hard to find documentation about this, but it definitely doesn't work over here (on x11).

[andreduartesp]

Wasn't the Win32 API way to do global shortcuts something like "Set a global event hook and filter out the keys you want?" ...as in more XGrabKeyboard and less XGrabKey as far as retrofitting sandboxing and user permission prompts goes? (tracing back to its being designed as a familiar upgrade for Win16 developers.)

I seem to remember reading something to that effect.

This is how an active app gets the keyboard, but windows software (not only windows software, but for wine that's the case) can create a global shortcut to be activated while inactive and the way to do it on X is basically adding a keylogger, it has access to all keystrokes, wine is good software and discards all they don't use, but bad software can use the same technic to store all your keystrokes and send to a server, or something like that. At Wayland, the app needs to register a global shortcut on the compositor, and only with that sequence the app will receive the input, any other sequence will be filtered by the compositor (and it opens some doors, you can change that sequence at the compositor, and the app will never know, so you can have a custom global shortcut for any app at the same config page)

PS. I'm using wine as an example, but any global shortcut on X works the same way (Plasma Global Shortcuts, probably Gnome global shortcuts too)

[andreduartesp]

Are global hotkeys actually supported in wine? It's hard to find documentation about this, but it definitely doesn't work over here (on x11).

yes, they have, and that was the biggest problem to port wine to Wayland for a long time (the global shortcut protocol at Wayland is very new)

[Weasel]

Wayland is just a video protocol

Display protocol*.

A protocol that refuses to give you window positions on the screen.

An unusable display protocol for power users.

Wayland was created with sandboxed apps in mind

So why does it enforce its bullshit on every app, even non-sandboxed ones?!? That's the problem.

If I don't sandbox an app, it's because I trust it and don't want to. It's that simple. It's not up to those idiots to tell me what I need or what to do.

[RealNC]

Maybe try the latest beta. I always went back to X11 after trying Plasma Wayland due to kwin crashing (which unlike X11 is fatal to your session), but with the newest beta, kwin has not crashed once over here.
plasmashell is a different story, but plasmashell crashing or freezing is not such a big deal. (Still annoying, though. Hope they get that fixed for the final.)

Well, right now I don't even have a reason to use KDE Wayland betas. KDE X11 non-beta is solid. It works, does what I want it to, no issues here. It's kinda hard to even find a reason to enter the Wayland experiment. I'm sure there's plenty of people who have use-cases that Wayland does better for them. But for me, if ain't broke, don't fix it. Even if KWin is more stable in the latest git master or whatever, if it crashing means the whole sessions goes down... That's just not something I'm looking for in my Linux desktop, let's put it this way.

Maybe the experience is way better for people using something else instead of KDE, and some non-NVidia GPU. But for me, 10 years more for X11 seems very likely. And I don't even mind. "It just works" in my case.