Casilda Is A New Project As A GTK4 Wayland Compositor Widget

[smitty3268]

So I'm asking you again: what industry ships new products with X11?

I can only guess he's referring to the AI/HPC space, where nvidia CUDA is standard, and is only now getting into decent wayland shape. And other stuff running RHEL. So they're probably still on X for a bit longer before they get updated to the current linux tech stack.

I highly doubt they're using xembed for anything though.

[access]

Why'd you need any display server at all on compute nodes? Most DC compute cards don't even have any display connectors (why would they).

[oleid]

I can only guess he's referring to the AI/HPC space, where nvidia CUDA is standard, and is only now getting into decent wayland shape.

Even they stuff doesn't need any X11, as it can be used headless for a long time now. Expect maybe very specialised things. The next iteration will be X11 free.

And other stuff running RHEL. So they're probably still on X for a bit longer before they get updated to the current linux tech stack.

Sure, nobody is going to dispose working servers only because the next version is going to be released in a few years.

I highly doubt they're using xembed for anything though.

I agree.

[intelfx]

God, Wayland sucks. How sad that you need to run a whole extra compositor just to some of the functionality of XEmbed?

Those who do not understand Wayland are condemned to whine about it on the Internet (c) Henry Spencer, probably

[Britoid]

webOS (LG TVs) uses Wayland too IIRC. RHEL 10 is probably going Wayland only.

RHEL going Wayland only doesn't really matter much because the only groups that use RHEL on desktop are film studios etc who will probably stick on older versions for as long as they can.

[emblemparade]

God, Wayland sucks.

I know you're a troll, but still:

XEmbed is almost unbelievably insecure, as really all of X is. Yes, it "just works", as do a lot of very insecure programs.

We all know that Wayland still does not match X in features. The community is evolving it gradually and carefully. But I guess for you gradual progress means "it sucks".

Trolling also sucks. It degrades the conversation for everybody.

[oiaohm]

Indeed. Because the industry is happy using X11 for every usecase (in this particular case, XEmbed).

https://github.com/kovidgoyal/kitty/issues/7083

There is 1 big problem with XEmbed what do you do when the developer of application you want to embed says bugger off.

Yes Nvidia under X11 means that a nested X11 like Xephyr and Xnest don't work with accelerated graphics.

Yes as stupid as sounds you will have on Nvidia hardware under X11 better chance that application will embed using Wayland compositor Widget with Xwayland for a X11 application than attempt to use xembed due to how few applications in fact support xembed and how many using accelerated graphics.

What about applications that you xembed and decide to clam the short cut keys your application code wants to use.. There are a long list of xembed problems.

Embedded has been going stacked wayland compositors for a while. Mostly because this avoid the input key problems and you are not needing application developer to include custom code so you can embed the application. Yes overhead has not turned out to be a major problem. Stacked wayland compositors Stability has been the selling point over Xembed and it has proven this in many different car entertainment system and the like use cases. There are a lot of ways Xembed can cause hard to diagnose issues including causing x11 server stalling.

[stormcrow]

... stuff...
16 years of this Wayland crap being pushed onto everyone and it's still not good enough.

16 years may seem like a lot of time for many people. It's really not in the open source world.

Do you know how long it took to get a relatively useful Xserver for Linux that wasn't hanging and crashing at random? It wasn't 16 years. It was closer to 20+. Rose colored glasses at this point. XFree86, later on X.org, was every bit as bad as Wayland/Weston/wlroots/etc etc is right now. That's why people began working on replacements to begin with. It was utter crap compared to the commercial rock solid and reliable X servers on the commercial Unices that could measure uptime in years.

Not a Wayland proponent by any stretch, but I'm getting tired of people that have never written a line of code more complex than an Arduino or a few small automation scripts think they know howlong it takes to get a complex piece of software, especially something as incredibly hardware dependent as a display stack to a usable state. The whole thing is being coordinated across a global volunteer network with disparate opinions on how to proceed it's lucky anything gets released at all!

[Quackdoc]

16 years may seem like a lot of time for many people. It's really not in the open source world.

Do you know how long it took to get a relatively useful Xserver for Linux that wasn't hanging and crashing at random? It wasn't 16 years. It was closer to 20+. Rose colored glasses at this point. XFree86, later on X.org, was every bit as bad as Wayland/Weston/wlroots/etc etc is right now. That's why people began working on replacements to begin with. It was utter crap compared to the commercial rock solid and reliable X servers on the commercial Unices that could measure uptime in years.

Not a Wayland proponent by any stretch, but I'm getting tired of people that have never written a line of code more complex than an Arduino or a few small automation scripts think they know howlong it takes to get a complex piece of software, especially something as incredibly hardware dependent as a display stack to a usable state. The whole thing is being coordinated across a global volunteer network with disparate opinions on how to proceed it's lucky anything gets released at all!

this is no where close to the same situation. I mean, lets say perhaps we learned nothing from the past decade or two of development. the perhaps we could say "well take a look at how long x took". but that's crap. we have learned lots, and we should have learned more. Let's be clear here. The reason why wayland is so stagnant is because everything regarding wayland wayland is a poorly designed. It can take literal years to add something simple as there is an extremely large amount of NIH syndrome going on. We still don't even have a even somewhat universal way to configure libinput for anything that isn't covered by udev rules. Between the fragmentation of portals, protocols, and everything else. Wayland is one of the most poorly conceived ideas that we are now effectively stuck with.

[intelfx]

The reason why wayland is so stagnant is because everything regarding wayland wayland is a poorly designed <...>

Ah yes, the world-class software architects are visiting Phoronix again to spread their wisdom