gemini - kennedy.gemi.dev

💾 Archived View for gemini.ctrl-c.club › ~phoebos › logs › kisslinux-2022-05-05.txt captured on 2024-05-12 at 16:02:49.
-=-=-=-=-=-=-
[2022-05-05T00:04:36Z] <acheam_> nice
[2022-05-05T00:04:40Z] <acheam_> chimera is really cool
[2022-05-05T00:05:00Z] <acheam_> and is pretty close to what my final kiss setup was like
[2022-05-05T00:05:54Z] <acheam_> (apart from the really cool bubblewrap build system and apk packages ofc)
[2022-05-05T00:20:16Z] <noocsharp> are you still on fedora?
[2022-05-05T00:20:33Z] <acheam_> yes
[2022-05-05T00:20:40Z] <acheam_> its treating me nicely as a daily driver
[2022-05-05T00:21:17Z] <acheam_> over the summer I might play around with things again though
[2022-05-05T00:21:20Z] <acheam_> but have no time at the moment
[2022-05-05T01:01:40Z] <illiliti> does chimera enforce logind, dbus, udev, pam, ...?
[2022-05-05T01:01:57Z] <illiliti> the things that kiss explicitly excludes
[2022-05-05T05:36:24Z] <testuser[m]> Hi
[2022-05-05T06:37:44Z] <travankor> does chimera enforce logind, dbus, udev, pam, ...? -> yes
[2022-05-05T06:38:18Z] <travankor> and webkitgtk is the recommended browser engine
[2022-05-05T06:41:09Z] <travankor> and gnome is the recommended de (lol)
[2022-05-05T08:07:19Z] <testuser[m]> https://termbin.com/flr8
[2022-05-05T08:07:20Z] <testuser[m]> so bad
[2022-05-05T08:29:14Z] <testuser[m]> https://github.com/mozilla/gecko-dev/blob/master/third_party/rust/audio_thread_priority/Cargo.toml#L37 found it
[2022-05-05T17:12:16Z] <dilyn> konimex: your icon on that artix link looks familiar
[2022-05-05T17:12:38Z] <dilyn> were you old enough in 2008 to be on the internet
[2022-05-05T17:18:18Z] <testuser[m]> 2008?
[2022-05-05T17:18:34Z] <dilyn> yeah you know
[2022-05-05T17:18:36Z] <dilyn> the dark ages
[2022-05-05T17:18:53Z] <dilyn> I knew a fellow on a forum way back when who always used a very similarly styled art
[2022-05-05T20:10:44Z] <greg904> illiliti_: I'm running your fork of wlroots that makes udev optional and I was wondering what NETLINK_BITMASK is? On my machine, the default value of 4 doesn't seem to work. I could not find documentation, but looking at some kernel code, I tried 1 and that works, but I don't know what I'm doing. Do you have any pointers?
[2022-05-05T20:29:22Z] <illiliti_> NETLINK_BITMASK specifies which bitmask libudev-zero will use to receive uevents
[2022-05-05T20:30:09Z] <illiliti_> by default, libudev-zero listens for uevents with 4 bitmask. you need to configure your device manager to send uevents to that bitmask for hotplugging
[2022-05-05T20:30:38Z] <illiliti_> 1 used by kernel, 2 used by udev for example
[2022-05-05T20:32:22Z] <illiliti_> netlink is quite underdocumented, especially uevents
[2022-05-05T20:33:54Z] <greg904> ok that makes sense, thanks
[2022-05-05T20:40:13Z] <illiliti_> oh i think i misread your text a bit. what i said basically means same thing for udev-less libinput, so just replace libudev-zero with libinput in my messages
[2022-05-05T20:41:14Z] <illiliti_> you use this https://gitlab.freedesktop.org/wlroots/wlroots/-/merge_requests/3490, right?
[2022-05-05T20:42:04Z] <greg904> yes
[2022-05-05T20:50:27Z] <illiliti_> what enum-backend are you using for libinput? netlink or demi?
[2022-05-05T20:50:47Z] <greg904> netlink
[2022-05-05T21:04:43Z] <illiliti_> should be okay. if you use device manager, just make sure it is capable of rebroadcasting uevents to NETLINK_BITMASK in order to make input hotplugging(if you need one) work
[2022-05-05T21:04:58Z] <illiliti_> for example, mdevd can do this natively via -O option
[2022-05-05T21:05:08Z] <illiliti_> busybox mdev needs helper. check contrib/mdev.conf of libudev-zero project for more info
[2022-05-05T21:05:33Z] <illiliti_> although, as you already discovered, it is possible to make input hotplugging work without device manager. just set NETLINK_BITMASK to 1
[2022-05-05T21:06:17Z] <illiliti_> you will have to manually set proper permissions for /dev/input/* though
[2022-05-05T21:10:44Z] <greg904> I think wlroots should be able to use seatd to open /dev/input/*
[2022-05-05T21:15:30Z] <greg904> but yeah you're right I will have to set permissions if I don't use a device manager
[2022-05-05T21:16:50Z] <illiliti_> i'm pretty sure you have these questions: why there's NETLINK_BITMASK? why libinput can't simply listen for kernel uevents with 1 bitmask?
[2022-05-05T21:17:13Z] <illiliti_> well, because nasty race condition may occur when device manager hasn't had enough time to set permissions for /dev/input/* which can lead to breakage
[2022-05-05T21:17:39Z] <illiliti_> in order to resolve this race bug we have to rebroadcast uevents: kernels uevents -> device manager -> application
[2022-05-05T21:22:56Z] <greg904> yeah I didn't know that you could send netlink from userspace to userspace, I just thought libudev used a UNIX socket or something like that, but then now I wonder why they didn't make libudev more generic like libdemi
[2022-05-05T21:23:57Z] <greg904> if it just reads uevent netlink that is the same from all device managers
[2022-05-05T21:28:49Z] <greg904> so that it could work with every device manager
[2022-05-05T21:34:09Z] <illiliti_> because they tend to over-engineer things and enforce architecture
[2022-05-05T21:34:36Z] <illiliti_> systemd's libudev uses eBFP under the hood for example
[2022-05-05T21:34:57Z] <illiliti_> systemd devs also plan to deprecate libudev in favor of libsystemd
[2022-05-05T21:35:30Z] <illiliti_> which enforces architecture even more