Hacker News Reader

The trouble with your guake bug is it might be Gnome + Wayland specific. With the separation between the server and composer, bugs might need to be fixed in multiple places. Some window managers use wlroots, so maybe some bugs can be fixed there, but others have their own forks and implementations.

The hotkey thing is big and it's annoying because it's another thing that might need to be implemented/fixed in each and every composer and environment (and it could be different in every environment).

I've seen the X11/Wayland talks and I agree Xorg has tons of old crufty garbage in it, and screen locking in Xorg is not very secure. But the Wayland team seems to have made little effort in addressing even the most basic things like hotkeys, screenshots, etc.

I'm not sure if "user hateful" is the right term, but they don't seem to be prioritizing the most basic things people are asking for.

It's not that they're user-hateful it's just that they're (I think rightfully) being aggressive about scoping Wayland to things related to drawing surfaces and handling input events. If people started treating as one domain specific protocol that display servers use for drawing stuff on the screen and not "everything a desktop app might want to do" protocol there would be less conflict.

I'm glad they're leaving all the "desktop" stuff out and letting desktops agree on dbus interfaces. I doubt people writing little CLI utilities want to have to create dummy Wayland surfaces to do interact with the display server which is how wl-clipboard works.

So what does a Wayland-only future look like? Your clipboard, keyboard hotkey daemon, app runner, etc. all communicating happily over D-bus?

Clipboard is handled by Wayland. Global hotkeys also needs to be handled by the Wayland compositor in some way since only the compositor has access to the full stream of input events. The compositor might choose to offload input events for known global hotkeys to some helper application, like kglobalacceld on Plasma.

Exactly. What we’re really seeing here is technical debt from the fact that we have so many different desktop environments. Much of the cost of having Gnome and KDE was hidden by the fact that a lot of the job of putting pixels on the screen and handling keyboard and mouse events was done (very badly) by X in one way or another.

Now the compositor/screen manager isn’t doing that anymore because it never should have and it was a terrible division or responsibility in the first place. But we have devs pushing forward on multiple different fronts with different DEs and development is painfully slow.

I mean, from my perspective, they're literally prioritizing the things that I want my display manager to do well

- Handle rendering things to screen

- Handle user input

My issue is that while X has a LOT of other things it also happens to do, it just doesn't do those two primary things very well on modern hardware.

As an aside -

This reminds me of the systemd arguments all over again, but the same crowd that was ready to crucify systemd for all the things it does are now bemoaning all the things Wayland doesn't do.

But really, I think there are just a lot of folks who aren't willing to try something new, or to re-evaluate some of the toolchains they've made for themselves.

Now - I'm not going to blame them for that, having a working system change under you isn't fun, and it eats up time and resources some people don't have. But it also doesn't stop the new thing from replacing the old thing.

Particularly if the new things happens to be genuinely better in many respects.

And while I can certainly understand the pain point of missing features in Wayland, The way it gets user input and display output right are just delightful.

So delightful that I actually prefer it to my work macbook, which is not something I could ever say about X.

> the same crowd

It might not be the same crowd. There are many crowds on Hacker News, and the voting mechanism tends to select for whichever crowd is angry at any given time.

Which is, like, super frustrating. I want to feel like I have a relationship with the people I'm talking to, but I don't really. It's always different people, and they're not consistent with each other.

Entirely fair.

I almost wrote out an aside in the above comment, because the person I was replying to certainly didn't mention systemd.

I do think there are some interesting echoes between the two conversations, though. Namely that the "angry" that happens to be getting selected for has nothing to do with the merits of either piece of software, and a lot more to do with the fear of having to learn something new to replace something familiar.

Namely that the "angry" that happens to be getting selected for has nothing to do with the merits of either piece of software,

What, you don't think people have legitimate criticisms of Systemd, or Wayland?

No, I think there are plenty of fair criticisms of both. Just like there are plenty of fair criticisms of init and X.

But fair criticisms aren't really the comments I tend to see in these threads. Instead I see fear/resentment/anger expressed in the form of "How dare you choose to use X, when it doesn't solve my problem Z? Good ol' Y never lets me down on Z, you should just keep using that. And no, I don't care that you have problems A, B, and C with Y".

Which... looking back at that sentence... I sure used a lot of letters, but I think it captures the point.

It's not fear, it's resentment: the old ways worked, and often worked quite well, and everyone has mostly settled on reasonable ways to work around where the old ways were broken; the new ways quite simply don't work in ways that the old ones continue to work just fine, and we resent being forced to lose functionality we appreciate having.

Systemd does offer some nice functionality, but in return it makes certain unreasonable demands (.ini files everywhere, binary logs, systemd-{resolved,homedir,consoled,all-the-things}. Wayland does offer some nice functionality, but it is not ready for prime time yet. It is nowhere near ready for prime time yet. It cannot replace X11, because it utterly fails to replace key X11 functionality today.

Next year, or the year after, it might be different. It probably will be. I genuinely look forward to adopting Wayland when it upgrades X11. But I would utterly resent being forced to use it today, and I am very worried that I may soon be.

> But I would utterly resent being forced to use it today, and I am very worried that I may soon be.

Imagine world where there is no Wayland, how is it better? I fail to understand who forces you. Maybe default choice of distribution? Or that people don't want to maintain X.Org?

People maintained other init systems in Arch Linux, they are gone now. Too much work probably? People created Duvian [1]. If systemd is such an awful choice it should be amazingly popular.

[1] https://devuan.org/

> Imagine world where there is no Wayland, how is it better? I fail to understand who forces you.

You answered your own question. It starts with distros defaulting to Wayland, despite it completely missing features I currently rely on in X11. No big deal, I can always manually switch to X. It's only a small pain! Then more and more software starts being Wayland-only, because most people are using it by default. Not a huge deal, I can stick to older versions of software. Then drivers are only released for Wayland, because at this point the only people left running X are me and a few other folks like me. Meanwhile, the features I consider absolutely necessary are still missing, because the primary source of funding is no longer hobbyists but corporate vendors.

I have seen it happen before, with GNOME. A decade or more after the cascade of attention-deficit teenagers started stripping out features it is still missing plenty.

> People created Duvian. If systemd is such an awful choice it should be amazingly popular.

You miss the corrupting influence of money. Money means resources, and resources mean being able to extend tentacles into every project. Resources mean being able to present an offer to other projects that they can't refuse: patches that implement needed functionality and, oh yeah, also mandate systemWaylanD.

No, without Wayland there would be stagnation. Less users, less corporate sponsorship, less developers.

You and me (I use XOrg too, no xmonad on Wayland) enjoy status quo for free. We consume work that maintainers do not enjoy.

> and I am completely burnt out on that on its own merits [1]

[1] https://ajaxnwnk.blogspot.com/2020/10/on-abandoning-x-server...

---

> attention-deficit teenagers

JWZ cliches are worn out, marking group you don't agree with disorder regresses discussion. Someone would call you old whiner and he would be right. I would recommend you join BSD camp.

> You miss the corrupting influence of money.

I missed nothing. Same money brought what we have today. A lot of people work on Open Source for free.

> mandate systemWaylanD.

Straight from ajaxnwnk:

> X11 clients are eternal, there's just way too many of them out there that are line-of-business and will never be ported to anything else [2].

[2] https://news.ycombinator.com/item?id=24925459

---

I feel sick seeing such attitude. I would reply anymore.

You bring up a good point, it's something to think about.

I try to address this with completely transparent and public voting, and also non-numerical voting, meaning every vote has to also be a "tag", like on Slashdot: insightful, interesting, troll, flamebait, offtopic, etc.

But not only that, you also get to see who tagged you, and this tells you whether it's someone you know or just some random.

What is interesting, is that you can tag people friends or foes. When you friend someone, suddenly you can see icons appear on other accounts: friends-of-friend (green/red pill), and foes-of-friend (green/green pill).

I never got the impression that the tagging system is widely used, however.

I can work with a system running on systemd, upstart or old good init but I cannot work without screen sharing especially now that everybody is working remotely. I can probably survive with all the other missing features but not this one. I'll consider Wayland only when all the different softwares my customers and friends use will work on it as they do on X.

But remote desktop just barely works on Linux. X forwarding is slow, VNC is buggy especially when certain applications packages expect hardware acceleration, and don't even get me started with all the issues with the keyboard mapping that can arise application to application. Meanwhile in Windows the remote application is close to flawless.

Maybe Wayland is solving that, I really don't know. However I don't need them. What I need is screen sharing in Slack, Meet and TeamViewer and that works with X. I also occasionally need Skype, Webex, Zoom, Jitsi and Teams. They also work with X. I'm confident that any random system my customers throw at me will work with X if it's advertised to work on Linux. Not so much if I were running Wayland.

If X would stop working this morning I'm afraid I'll have to go back to Windows, after 11 years, and use WSL2. Not a future I'm looking forward to and not something I'd thank the Wayland community for.

Screen Sharing works fine for browser based videoconferencing like Jitsi.

Some apps like AnyDesk just haven't invested in support for Wayland, while shovelware like Zoom tries to force you to install their client.

> Some apps like AnyDesk just haven't invested in support for Wayland

That they have to invest work for simple functionality to work is the problem. Until Xwayland can support this, Wayland is broken by design.

I've been using NoMachine on my local network for a while, it's fast enough and works on Windows, OS and Linux (you can just about play video over it even).

I haven't looked into iOS and Android.

I'm really going to have to explore this option. Does it have encryption capabilities?

> This reminds me of the systemd arguments all over again

I picked up on a completely different common theme: people are upset because major distros replaced a thing that worked with something else, and the new thing has a lot of problems that the old thing didn't.

A few years ago I had a bug like this under X + Noveau + GNOME, where Windows were reporting the wrong coordinates.

It was fairly easy to find an example on stackoverflow that created a Window using Gtk, and reported it's position.

Once something like that is on the bug report things tend to get fixed pretty quickly.

Report the bug on the outer-most thing first, e.g. GNOME shell in this case.

> Hell, I'd be happy if there was just fewer bugs.

There are strategic and tactical issues here.

Tactical issues are obvious. Wayland's problem with screenshots and all the bugs from not having a decades long lineage.

Strategic issues are less obvious. As this blog pos alludes to the standard X server is reaching a maintainability crisis where long term devs don't think the project is a good idea.

I think Wayland is a tactical mis-step myself for reasons you and GP mention, but I think now that the bad has been covered someone needs to mention the good:

The reference X server cannot practically be replaced. The reference Wayland compositor quite possibly isn't ever going to get established before being replaced. This is an excellent thing - it beings evolutionary pressure to the display system.

Wayland has problems, but unlike the current X ecosystem the problems can be fixed by replacing bits of software. The Wayland protocol seems light enough that Wayland+1 can implement it as a compatibility layer. When the dust settles, leaving X will have been a good idea.

I think pipewire broke my audio when I upgraded to Kubuntu 20.10. Removing it fixed it and seems to have had no ill effects.

Pulseaudio used to break everything, and I used to delete it to fix them. Gstreamer used to crash a lot.

Now they work all the time. Pipewire appears to be a better architecture than Pulseaudio. We have to hope it un-bugs faster than PA did.

Pulseaudio doesn't crash any more, true. But it still uses gobs of cpu to play audio and has other idiosyncracies that I have no idea how to work around. For example, I have no volume control in the GUI for the headphone port on the front of my computer, but I can control it just fine using alsamixer. After hours of trying various suggestions I've just decided to control my volume from the terminal and move on with my life.

Yeah, I'm not sure why the entire X protocol had to be thrown away. I suspect for a while the shiny new desktop stack will just be X on Wayland until someone writes a Wayland compositor that is also an X server just without the bits that can't work. One of the reasons for Wayland was fast compositor updates, but I still don't see why the compositor and window manager have to be one and the same, you could delegate management to a wm and present the full window. Heck, you could also replace Xlib and xcb with versions that know when they're running locally and skip the network protocol for most operations.

Sadly I don't have time to work on that, which really sounds like a full time job for a team of at least 3 developers, and from the looks of it, the people in charge don't share my vision for integrated coexistence of old and new.