00:05 <ifreund> leon-p: finally put up a PR for the riverctl river-options implementation though I have one more yak to shave before merging

00:05 <ifreund> let me know if you think the CLI interface does/doesn't make sense

00:06 <ifreund> we also have a small, zero-allocation arg parser thing now

00:10 <g_w1> hello, I tried to build on nixos using the provided shell.nix in the wiki, but failed with a linker error: https://cdn.discordapp.com/attachments/606246764651085829/800513514732519433/message.txt

00:12 <g_w1> should I not be using latest master?

00:12 <ifreund> g_w1: you're using wlroots master, you need wlroots 0.12.0

00:12 <g_w1> ah, thx

00:12 <ifreund> np

00:13 <g_w1> seems like nixos has 0.12.0? https://github.com/NixOS/nixpkgs/blob/master/pkgs/development/libraries/wlroots/default.nix

00:14 <ifreund> Idk how nixos works, but I removed that function from wlroots master

00:14 <ifreund> and it's definitely present on 0.12.0

00:14 <g_w1> ok

00:15 <g_w1> is there a way to check what version of wlroots is installed?

00:16 <ifreund> normally I'd say ask your package manager but if we aren't trusting it you could read the wlroots headers I guess

00:17 <g_w1> oh nixos has 0.11.0, nvm. nixos master is different than 20.09

00:17 <g_w1> i will just use the master shell and ill update the wiki if it works

00:18 <ifreund> oh lol, I read the symbol name wrong. Had a couple people report that building failed with the one that got removed on master

00:18 <ifreund> the layer surface one you hit is one I added in 0.12.0 :D

00:19 <ifreund> why the hell does nixos still have wlroots 0.11.0 though? I thought they were kinda up to date?

00:19 <ifreund> or maybe that's just the unstable channel

00:20 <ifreund> and yeah feel free to update that wiki page however you see fit :)

00:20 <g_w1> its unstable, I could use it if I wanted, but im on 20.09

00:24 <ifreund> I guess the point of nix is that you can use unstable just for dev environments and stuff

00:25 <ifreund> heck, I used it for a bit to get llvm11 binaries before they were packaged for void

00:25 <g_w1> ok nice it built thx

00:26 <g_w1> yeah it was fairly easy to just download the wlroots.nix file

00:29 <g_w1> ok updated https://github.com/ifreund/river/wiki/Building-on-NixOS

12:40 <leon-p> fwiw I also get the crash on exit on master, so it is not my code causing the issue.

12:44 <ifreund> leon-p: got a stacktrace for me?

12:45 <ifreund> piping stderr to a file can be helpful if it only happens on a tty

13:03 <leon-p> log is here: https://paste.rs/f4y

13:04 <leon-p> actually seems like the debug code of std is crashing?

13:05 <leon-p> although I have no idea how `return_address - 1` could trigger an overflow.

13:10 <leon-p> stack trace by gdb is here: https://paste.rs/jG9

13:10 <leon-p> ifreund: ^

13:36 <ifreund> leon-p: are you sure you're using wlroots 0.12.0?

13:38 <ifreund> there's also the fact that you seem to be using the logind wlroots backend while I use libseat

15:29 <ifreund> I think I need to put an actually secure screenlocker protocol on the TODO

15:35 <ifreund> or maybe what I really want is a screensaver protocol and have the locking part be handled by the compositor

15:35 <ifreund> context: https://www.jwz.org/blog/2021/01/i-told-you-so-2021-edition/

15:56 <ifreund> leon-p: debug symbols for libwayland might also shed some more light, though I assume something didn't remove its display_destroy listener before it was de-allocated

15:57 <ifreund> valgrind would probably tell us what that thing was

16:56 <leon-p> screenlockers in wlroots land currently use layer shell, don't they?

16:57 <leon-p> But yeah, something dedicated would be nice

16:57 <leon-p> I think hikari is already doing locking server-side

17:05 <ifreund> indeed, hikari handles locking server side

17:05 <ifreund> river currently has the same support for "screenlockers" as sway does, which is just layer-shell and input-inhibit

17:07 <ifreund> I think server-side is the simplest secure option, though a dedicated protocol may be able to offer the same level of security

17:09 <ifreund> I think a screensaver protocol leaving the locking itself up to the compositor is the way I probably want to go though

17:09 <ifreund> but that can come much later, the sway approach is OK enough for now