05:55 <larstiq_> ooh, moving views back to disconnected outputs looks nice

09:19 <leon-p> I have a touch screen laptop now and can reliably get crashes trying to interact with nautilus

09:20 <leon-p> I am stuck onbthe tablet commit until arch has the next wlroots patch version, but I dont think anything regarding touch changed recently

09:20 <leon-p> also massive +1 for getting rid of self

09:20 <leon-p> should have cold that patch egodeath

09:20 <leon-p> *called

09:30 <ifreund> hah, that's great

09:31 <ifreund> larstiq_: RE touch-related crashes, please get me an issue with a stack trace! I don't have any touchscreens to test with

09:31 <ifreund> oops, leon-p I mean

11:39 <leon-p> i'll do so as soon as I am back on my other laptop that has my keys and passwords

11:41 <ifreund> thanks :)

11:41 <leon-p> I am seeing some weird behaviour with the tablet tool here: it has no cursor until an application supporting the protocol is launched. after that it has the last cursor set by that on all other toplevels as well. Is there maybe no default cursor for tablet tools?

11:42 <leon-p> maybe pressing the button summons the cursor, needs mire testing

11:42 <leon-p> anyway, I wonder if there is a cursor more fitting for it than the default pointer image? I am pretty sure xcursor themes have a pen symbol

11:42 <ifreund> that's not intentional, I'll test that and have a look at the other tablet issue with the dropdown menus when I'm done the security contex impl

11:43 <ifreund> leon-p: feel free to suggest a better default cursor, I was just being lazy?

11:43 <leon-p> fair :]

11:43 <ifreund> s/?//

11:43 <leon-p> i think the issue is with my tablet pen: it turns itself off after some time and one needs to press a button to activate it again. duh

11:44 <leon-p> still not used to this, just bought my sisters old laptop off of her for notetaking since the pinenote does seem to be quite there yet

12:04 <leon-p> does anyone know if there is a way to make gdb cache downloaded debug info files? It's annoying having to wait a few minutes everytime I want to get a trace from a river dump

12:05 <ifreund> I think that's an arch problem, I just install the -dbg packages on void/alpine/chimera

12:06 <ifreund> relevant: https://wiki.archlinux.org/title/Debuginfod

12:07 <leon-p> brilliant, thanks

12:13 <leon-p> anyway, here are the debug logs and trace for the touch crash:

12:13 <leon-p> https://paste.sr.ht/~leon_plickat/0949af09d7d3c65f108627389af5230f59da52fa

12:14 <leon-p> funnily it's when trying to drag nautilus

12:18 <leon-p> also I think the cursor name we'll want to use for tablet tools by default is "pencil", I'll try to get a PR going

13:04 <itshog> leon-p: I also experienced a crash yesterday when using the laptop stylus in Xournal++

13:04 <itshog> I wasn't able to reproduce it though

13:06 <ifreund> itshog: do you use systemd or otherwise have coredumps configured on your system?

13:06 <itshog> I'm on Arch so yes I have systemd

13:07 <waleee> coredumpctl is your friend and freund

13:07 <ifreund> yes, please check if you have a river coredump in `coredumpctl list`

13:08 <itshog> Yes

13:08 <itshog> Should i upload the output of `coredumpctl info river`?

13:08 <ifreund> coredumpctl gdb, then bt to get a stack trace

13:09 <ifreund> if `info` includes a stack trace that's fine too

13:10 <itshog> `bt` or `bt full`?

13:11 <ifreund> bt full is nice :)

13:11 <itshog> Ok

13:16 <itshog> How can I save te output into a file?

13:16 <leon-p> itshog: btw, you might find rnote a better program for handwriting than xournal++. I find the UI a lot better optimized for touch and tablet input and it's GTK4 inistead of GTK3 like xournal so perhaps it wont crash for you as often.

13:17 <itshog> leon-p: Yeeah I was planning on trying it

13:18 <itshog> But tbh I really like Xournal++

13:20 <itshog> One thing I really like about Xournal++ is that exported pdfs have a considerably smaller size than those produced with other (proprietary) programs used by my professors

13:21 <itshog> Are rnote pdfs equally small?

13:23 <itshog> https://0x0.st/HF9p.txt

13:23 <itshog> Here's the output of `bt full` ifreund

13:26 <itshog> https://0x0.st/HF9V.log

13:27 <itshog> And here's river log from that crash

13:27 <leon-p> no idea about the PDF's rnote exports. Haven't really used it yet. I'll probably use it heavily next semester since I'll tutor a C program course.

13:28 <leon-p> ifreund: I am also getting crashes when trying to move any GTK4 program with the tablet tool. Do you also want a log and trace for that? I suspect it'll be pretty similar to the touch one

13:32 <ifreund> leon-p: I suspect it's the same bug, river just doesn't handle the client requesting to start a move while the cursor is hidden

14:30 <ifreund> leon-p: that bug should now be fixed :)

14:30 <leon-p> nice!

14:41 <leon-p> it's both kinda funny and sad that the best program for quick sketches and "casual" drawing is the oekaki uxn rom

14:41 <leon-p> I wonder how hard it would be to copy it as a native wayland client

14:42 <ifreund> is there not a native wayland uxn VM?

14:42 <leon-p> it's sdl, so it's technically native, but it behaves not like a native application

14:43 <leon-p> f.e. the window is not resizable, tablet tool is just client side pointer-emulation, etc.

14:43 <leon-p> and it sits at like 10% CPU usage when idle

14:44 <leon-p> it's a great little program, just has some rough edges

14:45 <leon-p> oh, and it also constantly has an audio port open, so it keeps pipewire active as well

14:47 <ifreund> yeah, sdl is pretty poorly suited to anything but games

16:28 <leon-p> so it's actually possible to have an almost flicker free boot from pressing the power button all the way straight to river with plymouth and greetd, however the one flicker I still see is rivers initial background colour before the init finished lol

16:29 <ifreund> leon-p: we'll solve that eventually, outputs need to be part of the transaction system

16:30 <ifreund> but that will probably wait untill wlroots 0.18 because then I can move to the backend-wide atomic API to modeset all outputs at once

16:30 <leon-p> no worries, I am surprised it even works this well. UEFI is magic

16:31 <leon-p> can you tell I've mostly used old computers or re-used installations made initially for old computers

18:13 <ifreund> itshog: the assertion failure you hit with touch should be fixed now as well :)

18:43 <leon-p> submitted a PR to use the "pencil" cursor shape for tablet tool devices

18:43 <leon-p> also finally figured out commit signing. the trick is to sign with your ssh key to avoid the gpg pain

18:44 <lordmzte> Having a bit of trouble here building the `river-git` AUR package. Seems like the zig-wayland submodule references a commit that doesn't exist. Log: https://paste.rs/0E8SG

18:49 <novakane> same error here

18:50 <ifreund> ah damn, forgot to push it

18:50 <ifreund> just a sec

18:53 <ifreund> done

19:08 <novakane> thanks!

21:32 <leon-p> ifreund: regarding the cursor image PR: since I think you are a bit more familiar with that right now, can we use wlr_xcursor_theme_get_cursor() to check whether "pencil" exists? I don't think we can use the returned object, but perhaps we can check on setting the xcursor theme and then set a global

22:55 <ifreund> leon-p: indeed, it looks like that should work. It's kinda lame that wlr_cursor_set_xcursor() doesn't return an error on failure but rather silently hides the cursor

23:32 <adamcstephens> oof