[19:32:37] Sheila: Ermine: shouldn't seatd have an initscript with it so it can be running as a daemon? [19:33:08] `seatd -g video` in tmux is 'working' for me but it feels like a kludge and very not obvious [19:33:31] the group should probably be a conf.d setting as well. possibly. idk [19:33:39] [[nodiscard]]: thoughts? I think you run sway so you would know about seatd [19:40:18] yes, it should. [19:43:31] awilfox: https://git.alpinelinux.org/aports/tree/community/seatd [19:43:32] alpinelinux.org: seatd « community - aports - Alpine packages build scripts Assets: 0/2 [19:44:11] this looks horrible :) [19:44:44] seatd is actually dtrt here and logging to stderr and it's doing all this horrible shell goop to undo that [19:45:02] but, at least that confirms my suspicion that there should be an initscript [19:45:25] it also makes no sense to me to have a separate group for seats. the *only* thing seatd can do is things with dri cards, i.e. the point of the video group. [19:50:22] yeah. I linked it mainly to show support for having an initscript and how Alpine's doing it, not to endorse it. [19:50:49] > Why spend time isolating logind and keeping up with upstream when we could instead create something better with less work? [19:50:54] am i going to open the codebase of this [19:50:55] and laugh [19:51:13] "create something better with less work" is generally the calling card of the apocalyptically naive [19:52:35] actually seatd is good because it supports logind in addition to itself [19:52:50] it's just I am not running systemd on my nouveau testbox [19:52:56] the poor thing already has to deal with an nvidia card, and unstable kernel (6.13-next) [19:53:06] it really is aiming to reduce the amount of work that other people need to perform in order to get seat management [19:53:18] really don't want it to have to deal with systemd-musl too, yet [19:53:36] I can't speak to the code quality, tho. [19:53:41] every part of this that i'm reading [19:53:44] sounds like a suckless user [19:53:51] (though honestly, systemd-musl is running stable on my testbox..recently broke 90 days uptime on that laptop!) [19:54:07] > oh the api is bloated, "link" [19:54:07] > open link [19:54:07] > i can't find a single redundant interface [19:54:09] if it were suckless-adjacent it wouldn't be on sr.ht [19:54:16] Sheila: that's ground zero [19:54:20] are you joking? [19:54:23] sr.ht is kind of like "suckless without nazis" [19:54:38] which is at least slightly better [19:54:46] the second part isn't particularly material to the context in which i'm working [19:54:59] yes, they're pieces of shit humans, but i'm referring to the code and attitude to code here [19:55:21] both of which are, as far as i can tell, dunning-kruger-beginner level [19:55:27] -*- Sheila shrug. as long as fascists aren't involved, fae supports user choice. [19:55:34] reminder linux isn't about choice [19:55:41] https://www.freedesktop.org/software/systemd/man/latest/org.freedesktop.login1.html also, a lot of the power control things are redundant [19:55:45] 'interactive' should have been a flag instead of Reboot and RebootWithFlags; real win32 FooEx vibes lol [19:56:37] looking at the api it looks like they're for two different things [19:57:15] i'm just sick of people pointing to random oneperson hackscript shit as a viable alternative to the one everyone is actually using, and expecting people to "just" get used to it [19:57:19] people have never, and will never, "just" [19:57:27] there's also an argument you could make about having 'CanShutdown','CanReboot','CanHybridSleep', versus a single 'SupportsPowerThings' that either takes flags in or outputs a control word with all the various things supported.. [19:57:45] arcayr: wayfire and sway and most of the wayland compositors use seatd [19:57:48] arcayr: so we need to support it [19:57:50] and i don't use any of them [19:58:00] but i certainly watch them [19:58:04] and i stand by what i said [19:58:14] like, this isn't "some random thing mentions it in passing", this is hard dependencies for wayland compositors with a significant user share [19:58:28] what a wonderful state of that part of the ecosystem [19:58:30] and my goal is for us to enable wayland [19:58:31] fuck that, get me away [19:58:32] I am unfortunately in the position of "I'm doing Wayland support in Adélie", and that means that I have to provide what the ecosystem demands. [19:59:04] and wayfire is the simplest thing to actually build at all [19:59:09] X is dying, so it doesn't matter that I think the entire Wayland ecosystem is stupid and full of unnecessary extraneous trash. [19:59:16] ^ that [19:59:22] i'm not sure catering to fringe users who - by virtue of knowing about and wanting to use these - are totally capable of doing it themselves, instead of to people who just want to get shit done [19:59:25] is going to have the intended effect [19:59:30] i've been saying this to foxy for fucking years [19:59:57] and, I mean, personally I would just leave Adélie at supporting KDE and everyone else can pound sand [20:00:07] works for me [20:00:12] <-- skarnet (~ska-devel@Interlinked-0h0.7nh.mtp4o8.IP) has left #Adelie [20:00:15] t. someone who, if it doesn't work, puts it in the bin [20:00:29] but since I believe in user choice, I'm doing Mate, LXQt, and XFCE, and now Wayland support. [20:00:36] don't make me tap the sign [20:00:40] so. [20:00:42] KDE's wayland compositor pokes at GL things that even my GT 520 doesn't have, so we need something for downlevel hardware and that's going to be wayfire or whatever [20:01:26] there is actually a reason for supporting not-just-KDE [20:01:30] i'm unsure how that floods out into dealing with contrarianware. [20:01:37] I'm doing work that will convince people to come use our shit, so I'm not actually in the least sorry that it doesn't meet your approval. [20:01:46] Sheila: i am one of the people you need to convince [20:01:52] and i'm probably more invested than you realise [20:01:53] arcayr: seatd explicitly plugs into logind [20:02:02] how is that contrarianware [20:02:09] because i read the about page [20:02:18] just think of it is as 'the adaptor that wayfire uses to talk to logind' [20:02:35] because unfortunately, if you look at how the sausage is made, you are going to hate most of kde and qt too [20:02:59] wait until you see the things qt does to wayland to make it look more like x11 :> [20:03:04] i'm already well aware of how they work [20:03:12] (tbh most of wayland is contrarianware because they'd rather invent new shit than admit they fucked up years ago) [20:03:13] but the reason i find that "okay" is [20:03:37] (see e.g. the portal shit) [20:03:38] nobody has to swim upstream to do it, unlike, apparently, elogind [20:03:53] > elogind tries to isolate systemd-logind form systemd through brute-force. This requires actively fighting against upstream design decisions for deep integration, and the efforts must be repeated every time one syncs with upstream. And even after all this work, one is left with nothing but a hackjob. [20:04:00] > In order to not be part of the problem. We will not displace systemd-logind anytime soon [20:04:01] > any time soon [20:04:03] thinkies [20:04:27] one of my clients was building something on a platform that said it won't change "any time soon" [20:04:29] today, it broke [20:04:31] their goal, actually knowing the guy, is to make the seatd *API* displace raw-dogging logind [20:04:35] it's going to cost them over $10,000 in my time alone [20:04:38] and that is so that compositors can work on i.e. BSD as well [20:04:42] plus probably 30-50k of their vendor [20:05:04] awilfox: why [20:05:21] so that netflix engineers can have KDE too, I guess [20:05:28] does the existence of BSD really need to be justified? [20:05:32] yes [20:05:49] they use it on their edge nodes [20:05:56] kde, i would imagine, is not on there [20:06:05] I wasn't really prepared to have that sort of discussion rn, being knee-deep in nouveau hell, but okay [20:06:16] bsd doesn't actually have to be justified [20:06:18] do we really need to centralise everything to the point of monocultures so vast a CrowdStrike can kill commerce for a day due to shitty engineering? [20:06:27] cool argument [20:06:37] monoculture isn't the only argument [20:06:39] but also at a macro level exactly what i'm talking about [20:06:45] when i say contrarianware [20:06:53] "hyesuifthgbrseiugh its gonaenfusewifh weiolfhewifh vnendor coimmetricaal oh my god acpitalism" [20:06:57] like, ok fam [20:07:17] but the rest of the world has to make do and, like i said, will simply never "just" [20:07:27] linux kernel might actually die over rust forks and freebsd is the least-bad alternative (I typed "an attractive alternative" first, before remembering the absolute state of their wireless networking stack.) [20:07:45] freebsd is really nice and i'd actually put effort and money into it if they dropped the dogshit init system [20:08:01] there are movements and projects in there to replace it with something modern [20:08:02] i stopped writing shellscripts for init a long time ago and i'm not restarting now. [20:08:08] they are trying to learn the lessons of launchd-bsd [20:08:11] awilfox: hopefully it goes better than last time [20:08:12] yeah that one [20:08:17] although, personally, i did like that [20:08:30] ... I /contributed/ to it, so yeah [20:08:40] no wonder i liked it [20:08:45] but even if they end up with beastied or whatever, it likely won't have 1:1 with logind [20:08:58] and why should it? seatd allows that to not matter [20:09:26] this sort of indirection library is something I /really/ like and support because it means it is always possible to just go do something else later [20:09:40] not being tightly coupled to one solution that could go off the fucking rails at any moment [20:09:52] assuming of course [20:09:54] do I think logind will go off the rails? probably not now, not in 5 years, but what about 10 years from now [20:09:57] that it functions perfectly in a 1:1 manner [20:10:04] will IBM just sell red hat for parts and systemd gets abandoned? or etc. [20:10:05] that upstream doesn't change in a way that this bracket library now needs to change [20:10:12] that the maintainers of all of the component parts remain [20:10:15] or are replaced [20:10:33] awilfox: redhat will sell ibm for parts [20:10:38] redhat superpower 2030 [20:10:43] >3kb of C source [20:10:48] I mean, to be blunt, the entire Linux ecosystem is a fragile edifice balanced on the razor's edge of "gosh I hope somebody doesn't push this that way" [20:10:50] >all the maintainers of the component parts [20:10:51] lol [20:10:59] i mean feel free [20:11:03] i just think it's a massive waste [20:11:10] i genuinely don't actually care, and never will [20:11:15] i probably should've made that clear [20:11:16] if it was a complex thing, sure, it might be, and that's why we dropped consolekit [20:11:29] somehow it is maintained again [20:11:34] but it's trying to chase logind and there's no point; just use logind [20:12:19] https://github.com/ConsoleKit2/ConsoleKit2/ if you were wondering [20:12:35] sounds like a idevelopement framework [20:12:35] *this* is a massive waste [20:12:39] appkit consolekit [20:12:45] whatever other kits they have [20:12:46] arcayr: logind is a fork of consolekit actually [20:12:55] consolekit was the original fdo thing [20:12:57] oh cool so we've already gone through the process i was memeing about once before [20:12:58] fuck yeah [20:13:01] systemd ate it to become logind [20:13:10] ck2 is "don't systemd me bro" [20:13:19] *this* is where you want to aim your derision, not seatd which is just a wrapper to make things portable [20:15:23] also, UIKit, HealthKit, WebKit, StoreKit [20:15:32] feels like I'm missing one [20:15:39] we did waste time doing ck2, up until it became impossible to avoid (e)logind. [20:16:43] Sheila: sddm's bullshit code really forced my paw on that, they kept trying to claim "we still support ck2!!!" but then they prioritised login1 over consolekit for the dbus interface, and ck2 tries to be login1 too for "compatibility" but sddm needs more than "compatibility" [20:16:44] awilfox: storekit lmao [20:16:47] is that what they use in their retail [20:16:53] and it was just such a fucking waste of my time to try and make it work [20:16:57] when logind is right there and actually working [20:17:06] arcayr: that's in-app purchases [20:17:06] yeah. [20:17:24] o [20:17:29] StoreKit is how your microtransactions work and how u pay for ur spotifies (with 30% going to tim apple) [20:18:09] and patreons (also with 30% going to tim apple) [20:18:50] awilfox: AI assistanceware named FoxKit :V [20:19:00] SwiftUI at least isn't SwiftUIKit or SwiftKit [20:19:11] what really need is KitKit [20:19:59] right, now back to.. crying into a hungry-fox® dinner and staring really hard at drivers/gpu/drm/ and wondering where my life went wrong. [20:20:19] Sheila: ngl that is where the name came from [20:20:38] I was WolfKit in #iphonedev and then became FoxKit and that's where foxkit.us comes from [20:25:48] arcayr: anyway, there's no way we can bring wayland at all without 'simpler' compositors first, you have to crawl before you can walk before you can run before you can fly. and yes, that does mean sometimes you have to package things that are weird. they are, at the very least, useful debugging aids. [20:25:54] at least wayfire is telling me EGL is the error [20:26:02] kwin_wayland literally dies with "Could not start Wayland." [20:26:05] that is the only message even with --debug --verbose [20:26:28] fuck kde in particular for making wayland a lot harder than they needed to, but I digress, mostly because I still like most of the kde peeps, they just really really went overly opaque on this shit [20:26:56] what's wrong with weston? [20:27:02] the literal reference implementation of a wayland compositor [20:27:05] unmaintained mess [20:27:05] by the people who work on the wayland spec [20:27:12] unmaintained? [20:27:13] wtf since when? [20:27:21] doesn't even build last time I tried, had three outstanding MRs that would have fixed it [20:27:22] unmerged [20:28:56] it looks like two of them got merged finally [20:29:00] i'm not really seeing how that's really any different to the altware you're using [20:29:16] well, it builds, and it actually does run on x86 already [20:29:36] it probably would on arm if I could figure out how tf to get the vc2 in the pi to support gbm [20:29:56] oh look, there's even a weston MR to try and add support for the pi's stupidity regarding media playback :> [20:29:57] also unmerged rn [20:30:01] which [20:30:22] https://gitlab.freedesktop.org/wayland/weston/-/merge_requests/1640 [20:30:51] I remember watching this get developed in #dri-devel and #gentoo-arm - it's neat to see it make it to MR state [20:31:03] so what doesn't work [20:31:24] it's literally easier to get wlroots-based compositors to build/run than it is to get the reference one to. [20:31:32] right now, playing back media that does that plane surface thing, i.e. hardware mpeg decoding, vaapi, dvds, vpx in firefox [20:31:56] it's just black or green depending on which thing you are using [20:31:58] Sheila: please find me one (1) example to back that up [20:32:13] which is a pretty big thing considering youtube is now 100% vpx [20:32:51] like you do not seem to understand that I do have actual reasons and bug reports and experiences informing me of the state of things and why I'm doing what I'm doing [20:33:22] I would *rather* do weston if only because then I could run their stupid faces in the fact that their own damn product is broken, but right now I need to fix mesa and the kernel [20:33:41] (I would *rather* just keep X11 forever since it fucking works, and it took about, oh, idk, 35 YEARS to get to the point it worked?) [20:33:51] (I have HORROR stories from the late 90s and early 00s about how shit it was) [20:33:53] -*- Sheila gestures at wayfire… [20:33:59] (makes wayland look like a walk in the damn park) [20:34:22] oh man we're going back to that [20:34:25] categorically ceebs [20:34:35] i'm detaching this, i am not talking about fucking x11 ever again [20:34:53] well I mean it's dead so [20:35:15] my point is, it took X11 about 35 years to not make me want to set everything on fire [20:35:37] so wayland being idiotic right now is not only expected, it's probably better now than X11 was 10 years on [20:35:40] which would have been like 1996 [20:37:49] still mildly angry that literally everybody decided XZoom was a feature that nobody needed. [20:38:07] so, anyone else feel like having a go at the foxy punching bag before I really do tab back to Terminal and wonder why the hell I'm even working on nouveau when I'm never going to be able to make a single fucking person on earth happy because the linux graphics stack is such a wonderous disaster of fucking idiot architecture, lack of design, and egos over quality?