>>101563606
Why would I explain it? It doesn't work.

>>101563606
Why? It's not a debate, just keep working on Wayland until it's accepted by everyone.

>>101563606
Complexity is... le bad. But unironically.

X already does this

>/g/ cant explain shit
All of you suck at this

>>101563606
It's bad cause every window manager has to write and implement its own compositor from scratch.

Who cares?

File: x-on-wayland.png (33 KB, 379x553)

33 KB PNG

>>101563606
Because you will never be in that situation.
<--- But always this.

>>101563606
Seems fine to me

>>101563691
Complexity is bad. Make something simpler than x not more complex and then we will talk.

File: x-architecture.png (36 KB, 533x507)

36 KB PNG

>>101563811
You had to do this with X too. Because X is a window manager not a compositor like mutter or kwin.

>>101563606
What is the use case for a windowing system which is completely useless without a ton of extensions handled by mostly the same people?

A couple of the core wayland devs were lazy fucks who argued for an useless wayland spec because simply they didn't want to handle shit like permissions. Which should be core.

>>101564023
It is simpler than X, it's simple to the point of uselessness.

>>101564083
>It is simpler than X, it's simple to the point of uselessness.
No it isn't look at >>101564013
and compare it with >>101564069

File: xwayland-clients.png (153 KB, 1272x769)

153 KB PNG

>>101564013
>always
What are you running that still needs X11?

>>101563606
A pixelbuffer API was ridiculously retrograde.

HMDs are showing the need for the need for resolution independence AND everything having a GPU makes the need for the efficiency of pixelbuffers redundant.
On top of that, Cloudflare remote browser isolation showed that pixel buffer based remote desktop is still shit and higher level network transparancy is still important.
They could have had a fastpath for 3D rendering, they should not have ditched network transparancy.

>>101564104
Half if not more of my linux software still require X meaning if I remove xwayland I won't be able to run it.
>How do you know
Well when these programs launch they show an X logo on the tittle bar.

>>101564124
It would probably be better to start over, or shave down X11 until it is clean enough.

>>101563606
requiring root to remap keys for all graphical windows

>>101564069
No, you didn't. X works without compositors and had its own working implementation so all devs had to do was use the default implementation and then any compositor could be used on top of that.

On Gayland each and every window manager has to reimplement the entire protocol which leads to massive duplication of work and WM-specific bugs. Meaning that the exact same code WILL work differently on different desktop environments for no reason.

Pair that with Gayland's retarded security theater that makes it unable to do shit Windows XP could and it's fucking sad.

>>101564174
>requiring root to remap keys for all graphical windows
That's inherently something which requires root, trivial to redirect keystrokes.

>switch to wayland session on plasma 6
>qt creator's code completion popups occasionally show centered on the screen and with full decorations
>switch back to x11
>no more glitches with any app

It's just not there yet.

>>101564236
i'm thinking of xmodmap or similar
if that usually requires root, it doesn't in my case, maybe because my X is started by the login user

>>101563606
you already made a Wayland shill thread yesterday

>>101564519
And the day before.

>>101563606
Explain to me why it's taken 15 years for it to become useable, and still requires Xwayland?

This thread still failed to give a good reason why it's bad

egaS

>>101564627
Asking why Xwayland is necessary is retarded. Wayland and X11 are two entirely incompatible protocols, so you need a compatibility layer like Xwayland to maintain backwards compatibility with legacy software that will never get updated for whatever reason.

PipeWire works the same way with regards to PulseAudio backwards compatibility.

File: 1720966921035925.png (244 KB, 1280x720)

244 KB PNG

>>101563606
>KMSaaaaaaaacck!!

>>101563606
it isn't.

>>101564124
X11 never had network transparency. Look up at the documentation. It never shows up, only appears in a handful of blog posts.

>>101565373
Okay. That doesn't answer the question.
If Wayland is so good, why is Xwayland still needed?

>>101564235
>X works without compositors
ya it works so well it tears so you retards add more layers of latency to fix it because x is fundamentally retarded.

>>101565502
he answered you already retarded shill.
it's the same reason all my applications think they're talking to pulseaudio even though I'm literally using pipewire.

grow a fucking brain, retard or go back.

>>101565501
when they mean network transparency they mean the woefully retarded baked in 2D software renderer in X.org.

>>101564236
No, that is inherently something which requires at least the privileges of the user whose session it is. In X, even when it's a service started by the system, key mapping can be done per-user.

>it's so good you need to emulate half your software on X11 anyway

File: 1685361878525712.png (325 KB, 2300x2100)

325 KB PNG

It is bad because it is a lie

Another nVidia FUD thread. Great.

>>101565584
>add boxes with every compositor name and one with a 'missing api' text because
>so you can add more arrows to make look it more complex
>see? Le wayland le bad wwww

>>101565620
they're still asshurt no one cared about an ecosystem they created while contributing zero code for it and are surprised Intel iGPU team pushed through GBM and other Linux DRM/KMS shit.

it's hilarious desu watching them kvetch at every fucking turn for things they could have prevented by contributing free and open source software, like Intel and AMD.

>>101565584
/thread

>10 years from now
>wayland is finally usable without caveats
>OP still makes the same thread

>>101565724
>but muh edge case
>muh 1980's motif app
>muh nvidia card
>muh le trannies develop le wayland

>>101565534
Just because X didn't do it perfectly isn't a reason to ditch it entirely.

Cloudflare remote browser isolation says pixels are shit too.

>>101565744
>Its not our fault it's bad... its y-your hardware!
>b-but all my gnome software just werks!
>steam is an edge case
>gaming is an edge case
>krita is an edge case
>veracrypt is an edge case
>synergy is an edge case
>matlab is an edge case
>java is an edge case
>wine is an edge case
>all electron software is an edge case
>virtual machines are an edge case

>>101565744
>autokey is an edge case

>>101565801
Wow! Software must be ported! Shocking!
>all electron software is an edge case
Electron supports Wayland since forever. You meant Discord and even that PoS supports Wayland.

>>101565840
Oh so it's no longer 1980s legacyware? Hmm.
>ignores everything except discord
as expected of a troon

>>101565832
Yes :^)
Have you thought in coding the patch to add the support for org.freedesktop.portal.GlobalShorcuts instead of crying about how your favorite does not work instead?

>>101565845
Wow, what a great and reasoned response. Typical novideo user.

>>101565913
>>101565801

>>101565845
Because Discord is brought up to say Wayland does not work with electron apps, even thought Discord uses a ancient Electron version that has initial support Wayland support.

>>101564235
They also wrote Wayland in C which in itself is a security risk.

>>101565923
>>101565620

>>101565937
Wayland is written in XML actually desu.

>>101565937
>C which in itself is a security risk
Learn to code properly.

>>101564130
A lot of programs actually work in Wayland but launch with X by default. Anything that uses relatively recent versions of SDL can be launched with SDL_VIDEODRIVER=wayland, and Qt apps can use QT_QPA_PLATFORM=wayland (or qwayland, don't remember)

Not all apps fully support Wayland yet, but it's worth trying

>Explain to me how this is somehow bad.
>ok
>NOOOOOOO

>>101565996
Arch Wiki has the magic words to make Discord work with Wayland too.
Project Wakefield isnt even complete and yet I have used DBeaver without Xwayland since months ago.

>>101566018
>expect explanations about tech
>in /g/
You're wrong, mister. This is /g/ - Transexuals.

>>101565800
>X didn't do it perfectly
it didn't do it at all.
if you weren't using xvnc, you were just jerking off anyway.

>>101565937
they wrote it expecting C-like ABI and software patterns.
there are Rust compositors that exist, e.g. COSMIC.

File: 1721915944603.gif (400 KB, 544x560)

400 KB GIF

>>101565506
Vsync always adds latency. There is no way to get around that except for Freesync and G-Sync (none of which are natively supported by Gayland.) In fact one of the biggest issues people have had with Gayland since release is that vsync cannot be turned off unless you're on KDE.

Not that it matters since Gayland devs render the cursor via software so there's always an extra layer of latency on Gayland.

>>101566137
Ummmmm sorry you're wrong okay

>>101566048
There's been plenty of explanations, and only screeching in response

>>101566137
Glad tearing protocol exists and Hyprland is based.

>>101566223
>>101566048

>/g/ shits in gaymen for years
>wayland gains traction
>suddenly gaming is important
Kek. All because trannies.

>>101566289
Don't worry, it works just fine on X :^)

>>101563606
The kernel shouldn't be suicidal.

>>101566095
>if you weren't using xvnc, you were just jerking off anyway.
I like jerking off, I used xpra. Still pushes pixels, but at least a window at a time. Nxproxy was a proper X11 proxy with re-attach though.

Cloudflare used skia to build a network transparent protocol. Apple is praised for it's smooth scaling of virtual monitors. Pixels are shit.

>>101566384
>xorg
>work
Nice meme

I don't care what's under the hood as long as my games work

>gayland
fags

>>101565506
If software calls drawing commands in the middle of a frame, then tearing is the correct result.

File: systemD.jpg (71 KB, 803x515)

71 KB JPG

>>101563606
>thread about irrelevant hyper-niche technology
>instigation of argument

You use windows

>>101569603
And you use troonix.
Next.

>>101564235
>On Gayland each and every window manager has to reimplement the entire protocol which leads to massive duplication of work
Not if you just use wlroots. Possibly aquamarine now as well? I'm not clear on whether they're intending for it to be portable or just do the minimum they need.

>>101563606
Because it doesn't work.

>>101570512
Works on my machine

Makes me want to KMS

>>101563606
> NOOOOO WE JUST TOOK AN ABSTRACTION OUT
sometimes an abstraction midway increases portability, is like pointing to a body with two arms and taking one off as a good thing
> IT'S THE SAME THING BUT CLEANER
except:
It's shit.
0. It's kiosk trooned to desktop (so we start from a rot foundation).
1. It's async oop in C with implicit sync (all implementations have exploits because of this)
2. It increases attack surface (communication is a workaround through dbus+portal+pipewire)
3. It has no security model (even xorg has one)
4. It has no protocol to ensure its "isolation" (that is simply not doing X). So it's trivial to bypass the "isolation".
5. It can suffer a DoS from moving the cursor too fast.
6. It obliterates portability among the same OS as every compositor is like a different OS.
7. Policy over mechanism
8. The core protocols are as capable, if not less, than curses library (the reason why everyone is doing their own extension for bare minimum things)
9. Literally everyone else (windows, mac, plan9, android) has a better display server. It's hard to find such a demented attempt even considering hobby OSes

File: Screenshot 2024-07-25 182155.png (3 KB, 131x88)

3 KB PNG

>>101563606

>>101572863
Nice copypasta. I bet you dont even understand shit of it, specially how 1 contradicts 9.

>>101572930
> you don't understand
then please explain wise anon
> 1 contradicts 9
how so?

>>101572986
isn't pretty much any display server apart from NeWS implemented with an async even loop?
By your definition, every display server has the dame vulns as Wayland, so it isnt special really.

>>101573037
the problem it's **object oriented like**, async with **implicit** sync in **C** (error prone as hell)
> xorg
nope?
> plan9
nope!
> android
doesn't surfaceflinger literally have a sync framework?
> mac
dunno
> windows
dunno (unlikely to be done the same way as it's done at kernel side)

>>101573149
So better no use C then, like hyprland (C++) or COSMIC (Rust).
I fail to understand what makes this pitfal so special in Wayland really.

>>101573181
I would guess C++ would suffer from this too (the majority don't program considering the latest specs), although I've not read hyprland code to say. For rust you might be right. Anyway it's true for the majority of implementations, and not only for a few, also there would still be 8 valid points.

>>101573282
I disagree completely with 0 (That's a genetic fallacy), 3 (Xorg security model is pretty much broken, that's why you don't see the X11 security extension widely employed), 6 is an exaggeration (There's no way hyprland would have gained traction if apps not made by hyprland didn't work. And let's face it: hyprland has a high chance of turning itself into its own display server with the level of autism the dev has), 7 reeks of Unix Philosophy, that I consider more a religious dogma rather than some kind of software engineering principle and 8 is basically 0 but explained.

>>101573372
> 3
For what I saw, most people just say that of xorg because of a too liberal communication, isn't that fixed by xace?
> 6
It's true as long as people tolerate gnome boycotting a common ground (for they have voice in the official protocols)

>>101573488
I swear if we didn't had to deal with GNOME and Nvidia bullshit this transition would have happened years ago.

>>101573488
Plus this:
https://www.x.org/releases/current/doc/man/man7/Xsecurity.7.xhtml

>>101573522
According to some comments on y combinator, apparently both Xsecurity and Xace broken.
https://news.ycombinator.com/item?id=29092612
The obvious solution for various X woes, like the independent DPI, the lack of security that works, etc, would be running nested X servers.
No idea if that its viable, really.

>>101573651
> nested
I would have thought of the same solution, building an abstraction layer to use them nested. Dunno about resources usage, but xorg doesn't eat much anyway and already have protocols for that.

>>101573759
Sounds like a fun, albeit nightmarish project to try.

>>101563606
Explain how it isn't by providing an implementation.
Hard mode: make it production-ready.
Ultra hard mode: take less than 15 years.
/thread

>>101573855
>kde, gnome, wlroots, hyprland/aquamarine, cosmic don't exists
k

>>101564007
me

>>101563606
utterly pointless. just have your de handle everything itself fuck the gayland spec.

Useless