Yeah that stuff I myself got rid of back in the Gnome 1 days when you could still customize all that. I appreciate that the Gnome devs came to the same conclusions as me.

• Desktop icons are usually hidden behind windows. It makes more sense to put launch icons in front of everything else, like Gnome does when you hit the Win/Super key.
• Minimize is unnecessary when you have workspaces and efficient window switching methods. Same for the task bar.
• A long time ago, when I was young, I played around with themes and shit, when that was still supported. There were always problems. Theming any complex set of applications yourself is basically impossible without causing issues, and themes made by others also exhibited this problem. You need a concerted long-term effort, plus buy-in from basically all app development teams, to make a comprehensive theme. Note how Microsoft/Apple don’t want to support theming either.
• Systray was always some excuse for apps to just keep running in the background while not taking up taskbar space. Steam can go fuck itself, just please exit when I ask you to, and don’t treat the close button as “minimize”. If I wanted Steam running in the background I would just put it on another workspace.

I have this working on Debian like how you have it set up, everything on an encrypted lvm volume, except for boot and efi. Just one disk though. When waking up, it asks for the password like it does during normal bootup. It then restores RAM from the encrypted swap after you type the password. I think it worked out of the box, but it has been a while, so not 100% sure if I had to enable this somehow. Anyway this looks good to me.

If you manually run systemctl hibernate, does that work? Assuming this also does not work, you need to look at the logs during the failed hibernate attempt. Probably something like sudo journalctl -f and/or sudo dmesg -wH (for kernel logs). Open this up in two terminals, run systemctl hibernate and observe any errors or warnings.

It’s possible this is a hardware/driver issue, i.e. a driver prevents hibernation or fails at it. You may be able to figure out which driver/device is responsible by looking at the logs.

Interesting, thanks.

The tokens can be worth different amounts. There can be 1 ¢ tokens, 10 ¢ tokens, 1 € euro tokens, 10 € tokens, whatever. Your wallet app will, when withdrawing, generate various tokens worth different amounts.

Using those tokens will be somewhat like paying in cash at a store that does not return any change. You gotta pay the exact amount. In order to facilitate that, you’d withdraw tokens worth only small amounts. There wouldn’t be like be any 50 € tokens in practice. If you wanted to withdraw 50 €, you’d get 1000 1 ¢ tokens, 100 10 ¢ tokens, and the rest 1 € tokens or something like that, to make sure you always have the exact amount ready to pay for anything.

I looked at this a looong time ago, but the basic idea is that the tokens (equivalent to cash coins/banknotes) are generated on the end user’s device, through some public-key cryptographic back-and-forth protocol. The issuer (bank/central bank/payment provider) does not see these tokens (they’re only on the end users device), but can verify that they’re legit (i.e. issued by them) somehow.

You can take one of these tokens to them, and deposit it in an account. They won’t know who it’s from but they know it was legitimately issued by them. Depositing a token is also supposed to be the only way of figuring out if it is a legit token, the bank will not tell you if a token is legit unless you deposit it.

When someone pays with these tokens in a shop, the shop will want to immediately (during checkout) deposit them, to make sure they’re legit, and also to make sure the token hasn’t been double spent. A shop that doesn’t do that makes itself vulnerable to fraud. This means shops will have a hard time hiding their revenue (to dodge taxes) compared to cash.

If someone you trust gives you a token (birthday money from your grandma, say), you don’t have to immediately deposit said token, since presumably you trust your grandma to not give you fake or double-spent tokens. Since you trust you grandma, there is no need to deposit the token and involve the bank, and that transfer would be untraceable (it’s literally just copying a number from her phone to yours).

The idea is that shop owners would have a hard time dodging taxes without opening themselves up to fraudsters using fake tokens, while the customer cannot be identified. You’d also be able to exchange tokens with family and friends in a way that isn’t traceable, as long as you trust them to not screw you over.

Ok so it’s not the uinput permissions.

When you use lightdm, do you use it to log into a Gnome wayland session, same as with GDM? Or is there any other difference between using GDM and LightDM? What exactly is/isn’t working, and how can you tell it is related to the XTEST extension?

Also I can use the PS5 controller as a mouse just fine.

Ok, this could maybe be the kernel driver creating a “real” touchpad device. Steam is able to create fake/virtual mouse and keyboard devices, I suspect that’s what steam uses XTEST and/or uinput for. So even though your PS5 touchpad works, that does not invalidate my theory that steam wants XTEST for mouse/keyboard input fakery, because that’s what exactly what XTEST is for.

So technically, Steam wouldn’t require XTEST for controller input, it would require XTEST to map controller inputs to fake mouse or keyboard inputs. I don’t know what exactly steam does if XTEST isn’t present, like what exactly doesn’t work?

And XTEST is an X11 protocol extension, it probably doesn’t work properly under Wayland anyway.

The null here is concerning to me.

Dunno, ‘(null)’ might just refer to the default connection (or server or screen or whatever X11 object this refers to), so not sure if this something to be concerned about. I presume X11 clients in general do work? Like xterm or xeyes work, right?

I am having the same bug as in here :/

The fact that you cannot log in may seem/look like that exact issue you linked to, but that’s from 2019 and may have a different cause. In general, GDM will start some executable, (i.e. gnome-session or something like that, or at least it used to be gnome-session, haven’t used gnome in a while) which if it exits/crashes this will kick you back to GDM. It could crash for many reasons.

You should check the logs for sure for both issues. X11 session logs (relating to Gnome X11 not starting), should be in ~/.xsession-errors (or ~/.xsession-errors.old for the previous session, I think). There may be a /var/log/Xorg.0.log for xorg, which I guess could also be the thing crashing (maybe). There may be other things logged with journald that could be relevant. journalctl --user should show all the logs for the user session. Logs from Gnome, since it runs as your user, would presumably show up there. My guess is all the relevant logs for a Gnome wayland session would also be there, as well as (hopefully) Xwayland errors/warnings, since Xwayland is actually the thing that would report it doesn’t support XTEST.

Note that journalctl likes to show the oldest logs first, so look at the timestamps. You can press G to scroll to the end. There are various ways to filter messages, look at journalctl --help. You might want to use --grep to look for anything related to XTEST or xwayland.

Lightdm and GDM are systemd units (systemctl list-units), the logs would show with something like sudo journalctl -u lightdm.

Interesting. I will say I suspect still this is to make the controller act as a mouse or keyboard somehow, which steam can do.

One interesting thing is that it says in the bug report, that steam also requires uinput, which is used to create virtual input devices from userspace. And maybe (speculating) you need that for XTEST to work on wayland or something, or maybe steam can use uinput instead of XTEST if it’s not available.

Or maybe it’s security-related somehow (Regular programs being able to create virtual keyboards is kind of a security issue).

For me the output of getfacl /dev/uinput will list my username with read/write permissions. I have no idea what is responsible for giving me this permission but there it is.

Can you check with getfacl in both sessions?

I have never heard about this, can you link to your source for this? Also what’s the symptom here exactly?

AFAIK steam and games directly use the kernel devices for controller input, not any X11 extensions.

Edit: Wait!? Do you want to use the controller as keyboard + mouse? Inside of GDM? Because that might use XTEST after reading about it, since XTEST seems to be some kind of testing protocol that allows one to emulate a mouse.

GUI toolkits like Qt and Gtk. I can’t tell you how to do it better, but something is definitely wrong with the standard class hierarchy framework model these things adhere to. Someday someone will figure out a better way to write GUIs (or maybe that already exists and I’m unaware) and that new approach will take over eventually, and all the GUI toolkits will have to be scrapped or rewritten completely.