gchristensen changed the topic of #nixos-chat to: NixOS but much less topical || https://logs.nix.samueldr.com/nixos-chat
<ashkitten> joepie91: sounds similar in parts to some of what i've experienced as an autistic person with adhd
<gchristensen> WORLDofPEACE: you changed color! (green to purple) and it suites you, but will require adjustment for me to recognize you :)
<WORLDofPEACE> I would also say that, I've had runins with that exact feeling and it was frightening to me, because I'm opposite in that analytical thinking had to be worked at. Once that aptitude was developed I was capable of this "switching" as an integrated aspect of my intelligence. And yet again, everyone's intelligences are different and don't develop the same
<gchristensen> ^.^
<ar> so, i'm playing around with a vm where i do the "nix-build '<nixpkgs/nixos>' -A system -I nixos-config=./configuration.nix" and nix-copy-closure dance, but …/switch-to-configuration boot and …/switch-to-configuration switch don't seem to modify /boot/loader entries
<ar> am i missing some important step here?
<infinisil> ar: Yeah, nix-env to switch the system generation
<infinisil> Check `vim $(which nixos-rebuild)`
<infinisil> (really a question for #nixos though :))
<joepie91> ashkitten: to what degree? if you feel comfortable saying
<ashkitten> i couldn't say, because my memory is too blurry
<ashkitten> i remember i used to think of my brain as a library, but i don't remember the rest of the metaphor
<ashkitten> all i know is that what you described sounds somewhat familiar
slack1256 has joined #nixos-chat
<joepie91> ashkitten: thanks :)
<joepie91> I have seen a few mentions of autism + ADHD now but it feels incomplete to me
<ar> infinisil: thanks!
<joepie91> I'm now talking to someone else who has my experience, and they also seem to have the feeling that it's an independent thing in some sense
drozdziak1 has joined #nixos-chat
supersandro2000 has quit [Disconnected by services]
supersandro2000 has joined #nixos-chat
rajivr has joined #nixos-chat
katrin has joined #nixos-chat
<ajs124> let it be known that I, ajs124, use non-ascii identifiers in nix attrsets and if they ever break, I'll be LongtimeUser4 from https://xkcd.com/1172/
<cole-h> 👀
<siraben> lol
Taneb has quit [Quit: I seem to have stopped.]
Taneb has joined #nixos-chat
<siraben> rmcgibbo: any adoption of Nix in the computational chemistry world? heh
<rmcgibbo[m]> i don't think so. my company has its own system that's not as neat as nix, and most opensource stuff is conda, it seems.
<abathur> I made the mistake of installing conda at some point
<siraben> conda is a package manager for python right?
<abathur> somewhere around the time when the number of pythons on my system tipped over into TooDamnHigh
<abathur> colloquially I think, though IIRC it's not technically python specific
<abathur> my complaints aren't really specific to python or conda, though
<abathur> I'm sure it's fine
<abathur> my complaints are just about naive users having 13 pythons installed as they follow various tutorials and having no idea what is what or why or when or where
<abathur> also fwiw, I'd rather be told to install conda than "just download and run this 2gb docker container because we've given up on documenting how to build our software and its dependencies"
<samueldr> uh
<samueldr> abathur
<samueldr> you should know that it's entirely documented into the dockerfile
<samueldr> ;)
<abathur> oh yes
<abathur> excuse me :D
* abathur has been evaluating binary analysis tools this weekend and has had to reinstall docker and download entirely too many containers
<siraben> purging /usr/local/bin was one of the best things I did to switch from homebrew to Nix
<siraben> though it's back again with some binaries (much less than before), since things like flutter are not packaged in nixpkgs for macOS.
<samueldr> (continuing from #nixos-dev)
<siraben> some people have the impression that higher cs education = software engineering which is not true
<samueldr> *one* particular individual started applying... "theoretical compsci" badly
<siraben> I would prefer if my CS program had more theoretical CS + mathy stuff but I know some people wouldn't be into that
<samueldr> e.g. instead of using language features, let's reimplement data structures on top!
* abathur throws down a flash-bang and appears in the chaos
<samueldr> when the task was a simple task
<samueldr> though maybe it's that individual who wanted to show off what they knew from the program
<abathur> at least they passed the whiteboard? :)
<siraben> in particular I think it would be tangential to teach devops in uni
<samueldr> no
<siraben> well maybe some knowledge
<siraben> or as a special topic
<samueldr> there was no whiteboard test for them, got hired because they knew the boss already
<samueldr> (and if there had been some whiteboard-like test, it would have been practical to the language, and not theoretical compsci)
<abathur> fair
<siraben> hm so this individual did theoretical compsci in their interview?
<samueldr> what I wanted to say is: it will be extremely YMMV unless there is a good concensus on practices
<samueldr> siraben: I'm not even sure there was an interview
<abathur> I think it's OK to have devops in the academy as long as they make them actually do the thing for real ops in the academy
<samueldr> (it was a small business)
<siraben> Ah.
<abathur> i.e., I'm sure there's some protein op in the biology dept. that could actually use an ops person
<samueldr> though, maybe it's more specific to the education system in my province, I think devops would be a collegial (CEGEP) thing
<abathur> but it's hard to imagine an *ops program that can properly prepare people if they aren't working on real projects with real needs?
<samueldr> vocational school
<siraben> reminds me of how we used git in a class but only clone/push/pull/commit were covered, granted it wasn't the focus but then you end up with people not knowing how to resolve merge conflicts etc.
<abathur> I guess a good referent is like
<abathur> I went to hs with a guy
<abathur> he did sound
<abathur> for band, drama, choir, church, anything
<abathur> fucked off in all of his actual coursework
<abathur> "high school"
<samueldr> here some jobs, like construction worker, need to do work as an apprentice before being able to work on their own
<samueldr> maybe that's what's needed?
<abathur> but he's also been touring nationally/internationally with reasonably well known bands for over a decade
<abathur> and yes, he did do a two-year degree in sound engineering in the middle of that
<abathur> but he had thousands of hours of experience on the soundboard for real performances by then
<abathur> samueldr: I guess it's just all about recognizing the mix for a topic and figuring out how to hit it
<siraben> maybe CS/software engineering should be split into two different degrees
<siraben> I've heard that CS curricula in Europe tends to be more theory focused compared to the US where it's like an engineering discipline.
<samueldr> hmm... I'm diving deeper into understanding wayland things and... I don't really like how each compositor has to implement getting the display and gpu things working
<samueldr> to me it feels like unneeded duplication of efforts that harms "composing" your compositor
<aleph-> Pretty much.
<aleph-> Off topic aside I hate unitaskers. But I love the idea of this https://i.imgur.com/WWriZru.gifv
<samueldr> well, unitaskers aren't necessarily bad, I hate frivolous unitaskers
<samueldr> those that do one task you can do well enough otherwise, or that are seldom used
<samueldr> e.g. a kitchen toaster is a unitasker
<aleph-> Hmm point
<samueldr> I don't eat toasted bread so I don't have one, but I do understand the place of it in a kitchen
<aleph-> I just have a toaster oven.
omnd has quit [Ping timeout: 256 seconds]
<samueldr> too bad that 1-slice sandwich toaster relies on a flame
<aleph-> There are apparently electric ones
<aleph-> I'll be honest I would totally eat hot sandwiches more if it was this easy
<samueldr> if I extrapolate from here, i have a "4-slice sandwich toaster"
<samueldr> electric
<samueldr> and it's not great because it's relatively slow, and hard to wash
<aleph-> Nod. Yeah I have this... whatever brand toaster oven that works and is decently big/powerful
<aleph-> Because my illegal apartment lacks a kitchen
<samueldr> if at least the trays were removable it would be less inconvenient
<samueldr> ah, though it's less "deep" than that gadget you linked
* abathur has one of those
<abathur> I don't think I've used it, personally
<abathur> though I do occasionally think of making a fluffernutter in it
<siraben> samueldr: I still need to set up screen sharing properly on wayland >.<
<DigitalKiwi> siraben: what i do when it's not packaged into nixpkgs but is in homebrew is to package it ;)
<samueldr> hmmmmm... am I a bad person for wanting to implement a DRM compatible shim through a FUSE filesystem?
<siraben> wayland at least makes it easy to have an external display with a lower DPI than my screen
<siraben> but man I use NixOS so I have a reliable system and don't need to be an expert at this stuff!
<{^_^}> #65775 (by Kiwi, 1 year ago, merged): ocrmypdf: init
<siraben> DigitalKiwi: hehe, yeah, guess I was too lazy to package flutter
<DigitalKiwi> 1 year!? that was 2019 :(
<lovesegfault> Just learned fortran has circular shifts of 2D arrays and my brain exploded a little bit
<siraben> DigitalKiwi: what's the context for that PR?
<aleph-> Yeah fortran is pretty great lovesegfault
<siraben> samueldr: I have one of those sandwich toasters, they're neat
<DigitalKiwi> fun fact: i've only ever needed ocrmypdf for one file and i did it on macos with the brew version while i was switching to nixos (i started off with nixos on a desktop and then i started using it in parallels as much as possible until i got comfortable enough that i didn't need macos anymore)
<DigitalKiwi> and then i spent like 6 months trying to get it packaged lol ;(
<abathur> in the future, developers of unbuildable software will just drop-ship you a cheap linux laptop containing only their software
<aleph-> Heh
<samueldr> laptoperization
<aleph-> abathur: So building packages at work a week or two back
<abathur> :}
<aleph-> Tag build fails.
<aleph-> Check the last master build that worked so I can start bisecting.
<aleph-> It's... the same commit id?
<abathur> ha
<abathur> oof
<DigitalKiwi> well at least that's a short bisect
<aleph-> Untagged, retagged and pushed. Fixed it fine
<abathur> hehe
<aleph-> Utterly mental
<samueldr> oh... I don't like the realization I just had
<samueldr> you couldn't "just" use software rendering with wayland
<samueldr> the compositor has to implement it if I understand things right
<aleph-> Iirc yes
<abathur> I've seen this GH tag page https://github.com/topics/fuzzing-management and now all I can think about is that it should just be emailing random requests/proposals all day to "see what sticks"
Fare has quit [Ping timeout: 264 seconds]
<aleph-> Okay so I hate that I'm so dang close to getting per app in-process wg tunnels working
<aleph-> Like millimeter from the finish line. >_>
<aleph-> I'd probably make more headway if I knew what IPC error code -71 meant
<samueldr> can you run a wayland compositor nested into another one? (ideally, arbitrarily)
<samueldr> (I don't think so)
<Ke> not arbitrarily, but it is just a protocol afaik
<samueldr> "just" a protocol :/
<samueldr> but interesting, though
<Ke> My view is that it's something like translator-multiplexer between kms and wayland
<Ke> compositor that is
<samueldr> I don't think so, considering I was running sway in weston, under X11
<samueldr> (I don't think KMS is at play at all here)
<Ke> you would need wayland-wayland multiplexer
<samueldr> well, with X11, but not with weston
<samueldr> btw, I don't have a particular goal (yet), just deepening my understanding
<Ke> sway on x11?
<samueldr> I wanted to make sure things wouldn't speak to DRM, DRI or KMS
<Ke> but how does that work?
<samueldr> sway wasn't aware of X11 (afaik) in that case
<samueldr> weston supports x11 as a backen
<samueldr> backend*
<samueldr> but sway does too!
<samueldr> you just run sway
<samueldr> and there's a window
<Ke> wow
<samueldr> now I'm looking to see if e.g. sway into cage works
<samueldr> I do have ulterior motives: running (barely, badly) wayland compositors where they can't
<samueldr> as a fallback
Fare has joined #nixos-chat
<Ke> I am planning to try isolated users that have access to sway unix socket
<Ke> I believe they don't have direct drm, opengl or anything
<samueldr> sway in cage on x11 works... except for spurious spam in the console... weston in cage in x11 also kind of works... but is weird
<samueldr> weston in cage assumes it needs to do CSD for itself
<samueldr> alright, I think I have a contingency plan with those shenanigans
aei has quit [Ping timeout: 246 seconds]
Fare has quit [Ping timeout: 264 seconds]
<ar> nixos with / on tmpfs is fun
<ar> i'm tempted to do this to my laptop
<ar> f0x: yup
<f0x> and yes, me too (but on my new server as nixos learning first)
<ar> f0x: /25
<ar> whoops
<Ke> "NetworkManager/system-connections".source = "/mnt/persist/nm-system-connections"; if you use nm
<Ke> where /mnt/persist is whereever you have persistent data
<etu> ar: Nixos on tmpfs is fun :)
<Ke> ":)" to emphasize effect
<Ke> but definitely I prefer it to making a huge effort with state version and risking unbootable system with rollbacks
waleee-cl has quit [Quit: Connection closed for inactivity]
<Ke> especially, when vast majority of the state has no value to be preserved
<DigitalKiwi> i know everyone hates signal here so i'm sure you'll love this https://mostlyabsurd.com/files/2021-03-08-070818_2880x1800_scrot.png
<Ke> is this your whole sales pitch?
<Ke> does loki support transparent offline message delivery and e2e encryption?
<Ke> was this the thing that was basically signal code, but with custom server
<DigitalKiwi> it's a fork of signal that runs on blockchain lol
<Ke> ah, I take it this is finally my ticket of getting rich fast without labour
<Ke> is this blockchain thing a jest or is it used for only something small scale?
<DigitalKiwi> idk
<Ke> I am fairly sure, not actual messages
<DigitalKiwi> there's a paper about what it does
<Ke> like you could use it to register user accounts
<Ke> eg. store pubkeys
<DigitalKiwi> someone made a PR for it but didn't want to be a maintainer so a few months ago i put it on my list of things to get around to lol finally updated it
<DigitalKiwi> i haven't even registered yet
<Ke> I know matrix is kind of shit, but I would hope that all GNU+people would get into it for the network effect to avoid even more horrible alternatives
<Ke> can you actually chat with this loki thing and does it support anything apart from android?
<DigitalKiwi> yes
<Ke> first one is a serious question, retroshare kind of had it all years ago, but did not work in practise
<Ke> and I really mean we could not get messages across
<Ke> it was fully distributed and had offline messaging etc.
katrin has quit [Quit: ZNC 1.8.1 - https://znc.in]
<DigitalKiwi> i just installed it on ios lol and that screenshot is the desktop app i'm about to PR into nixpkgs lolol
<Ke> I believe it would work, if you have efficient contact network
<Ke> and obviously retroshare was e2e only
<DigitalKiwi> haha who needs efficiency when i already stopped talking to so many people because of signal ;)
katrin has joined #nixos-chat
<Ke> quite
ixxie has joined #nixos-chat
<DigitalKiwi> the ios ap had 124 reviews lol
<Ke> well I think in retroshare you actually indeed would need a large amount of contacts
<DigitalKiwi> 4.4 stars
cole-h has quit [Ping timeout: 260 seconds]
<DigitalKiwi> i mostly am just amused by it because i keep thinking "what could be more annoying than signal? a fork of an old version of signal that adds all the joys of *BLOCKCHAIN*"
<DigitalKiwi> well it's not the worst app i've used anyway.
<DigitalKiwi> WORLDofPEACE: do you remember the fix for wire when it crashed when changing profile picture? because i have that on this one too lol
<DigitalKiwi> (session-desktop:4400): GLib-GIO-ERROR **: 07:58:05.520: No GSettings schemas are installed on the system
slack1256 has quit [Remote host closed the connection]
<supersandro2000> I really liked when my tmux turned into a christmas tree because systemctl wanted some credentials from me
<srk> DigitalKiwi: install gsettings-desktop-schemas ?
__monty__ has joined #nixos-chat
<Ke> install like systemPackages?
<Ke> so a hacker question, can someone find nixos tool to create gpt partition on block dev before sector 2048
<ar> hue: $ nixos-option services.xserver.autorun.declarations
<ar> This attribute set contains:
<ar> Segmentation fault (core dumped)
<Ke> s-dr knows a hack for an image file
<__monty__> Ke: Doesn't fdisk do the trick? Is it even possible?
<Ke> possible yes, and fdisk not in any way I know of
<Ke> value out of range etc.
<Ke> as in format makes it possible maybe not spec
<__monty__> I suspect it's not to spec.
<__monty__> Which means it's impossible.
<__monty__> What do you need it for?
<Ke> bootloader part
<__monty__> Hybrid MBR?
<Ke> well it's possible to extent I'm interested, linux will not care about spec
<Ke> protective mbr is practically always there
<__monty__> Not in the Hybrid MBR case.
<__monty__> Afaiui.
<Ke> well if you mean making partitions available in mbr, probably not
<Ke> never done that setup
<__monty__> It's only for the bootloader afaik.
<Ke> the limit is also in mbr partitiining, where it's definitely not the spec
<Ke> it's funny though how the experience seems to differ between file partitiining and blockdev partitiining
<Ke> where i is o or i
<Ke> cellphone while watching babby
<__monty__> File partitioning?
<Ke> babby tries to maul me with three(3) teeth thats more pressure per tooth
<Ke> like fdisk file.img
<__monty__> I don't see why anyone would want "GPT" partitions not to spec though. Btrfs and Zfs can use blockdevices without partition tables so I don't really see a use case.
<__monty__> Teething's hard on both watcher and watchee : s
<Ke> why would the spec matter
<Ke> iirc spec also mandated system part, which is there only when needed
<Ke> it's just numbers, not interpreting them by spec takes extra programmer effort
<Ke> or ignoring illegal partititions that is
<__monty__> Because if it's not to spec it's not a GPT partition. And many tools will probably fail to deal with it properly.
<Ke> don't believe that theory
<Ke> tools generally become permissive
<Ke> why not force esp then?
<Ke> note it takes extra effort to not work
<__monty__> Things not being aligned as expected makes things harder. I doubt programs looking for a GPT scan devices starting at the very first block.
<Ke> no you would scan the 128 partition descroption slots, like always, there are numbers there like start and len and uuid and label
<Ke> no extra work
<Ke> you always have to read them all
<__monty__> So it's just to avoid programs that expect an MBR mess with your device? Still sounds like terrible tradeoff.
<Ke> I am anyway more than 50% sure that the reason 2048 limit is there, because grub used to embed bootloader data after partition table
<Ke> http://rockchip.wikidot.com/partitions here claimed to be standard with partition at 64
<Ke> wikipedia image shows also partitions starting at lba 34
<Ke> If the block size is 512, the First Usable LBA must be greater than or equal to 34 (allowing 1 block for the Protective MBR, 1 block for the Partition Table Header, and 32 blocks for the GPT Partition Entry Array); if the logical block size is 4096, the First Useable LBA must be greater than or equal to 6 (allowing 1 block for the Protective MBR, 1 block for the GPT Header, and 4 blocks for the GPT Partition Entry Array).
<Ke> from spec
<Ke> so sector 34 is valid partition start in GPT
<Ke> there is though note To avoid the need to determine the physical block size and the optimal transfer length granularity, software may align GPT partitions at significantly larger boundaries. For example, assuming logical block 0 is aligned, it may use LBAs that are multiples of 2,048 to align to 1,048,576 byte (1 MiB) boundaries, which supports most common physical block sizes and RAID stripe sizes.
<Ke> there is requirement to align, but eg. cgdisk allows setting the alignment, but still does not allow partitions before the first partition
<Ke> GPT partitions should be aligned to the larger of:a the physical block boundary, if any b the optimal transfer length granularity, if any.
Fare has joined #nixos-chat
Dotz0cat has quit [Ping timeout: 245 seconds]
Hurttila has joined #nixos-chat
cosimone has joined #nixos-chat
omnd has joined #nixos-chat
dadada_ has quit [Quit: WeeChat 2.9]
dadada_ has joined #nixos-chat
monsieurp has quit [Remote host closed the connection]
Hurttila has quit [Ping timeout: 240 seconds]
<WORLDofPEACE> DigitalKiwi: I think it did that because it was missing settings schemas, maybe.
<gchristensen> man, `git am -3` is sooo much better than `git am`
<WORLDofPEACE> gchristensen: totallyyyyy
<WORLDofPEACE> DigitalKiwi: ehhh, I see everything that should be in the wrapper that is in there. IIRC u use plasma?
<__monty__> Ke: Thanks for looking it up and sharing the info.
monsieurp has joined #nixos-chat
<ar> ,launch Starship SN11 rolls out to the launch site in Boca Chica https://www.youtube.com/watch?v=UynhUVEkpg0
<{^_^}> Starship SN11 rolls out to the launch site in Boca Chica https://www.youtube.com/watch?v=UynhUVEkpg0: Ping for space stuff (edit this command to add yourself, see ",help"): infinisil Taneb ldlework etu philipp[m] eyJhb gchristensen __red__ red red[evilred] risson
<gchristensen> really impressive
<Taneb> How many of these do they have in warehouses or whatever
<infinisil> Just casually moving a rocket, no big deal
<philipp[m]1> That thing is not small.
<philipp[m]1> <Taneb "How many of these do they have i"> They pipelined the production a bit. While one is tested, the next few are being built.
<infinisil> It's impressive how fast they produce these
<philipp[m]1> Oh, forgot this room is bridged. i hope that doesn't look too bad in irc.
<joepie91> replies bridge poorly :P
<f0x> <joepie91 "replies bridge poorly :P"> yeah :^) (demo)
<joepie91> you should reply to a longer message
<joepie91> :P
<f0x> what's the point, it gets cut off anyways :P
<joepie91> exactly!
endformationage has joined #nixos-chat
cosimone has quit [Quit: cosimone]
rj has joined #nixos-chat
rj has quit [Ping timeout: 268 seconds]
rj has joined #nixos-chat
vulpine has joined #nixos-chat
rj has quit [Ping timeout: 268 seconds]
rj has joined #nixos-chat
rj has quit [Remote host closed the connection]
rj has joined #nixos-chat
<siraben> Anyone have some useful yasnippets? I have a few Nix ones: https://github.com/siraben/dotfiles/tree/master/emacs/.emacs.d/snippets/nix-mode
ixxie has quit [Quit: Lost terminal]
ixxie has joined #nixos-chat
rj has quit [Ping timeout: 268 seconds]
rj has joined #nixos-chat
cole-h has joined #nixos-chat
<veleiro> Nixpkgs teaches me linux, it just taught me networking
rj has quit [Ping timeout: 268 seconds]
rj has joined #nixos-chat
rj has quit [Ping timeout: 268 seconds]
ixxie_ has joined #nixos-chat
ixxie has quit [Ping timeout: 246 seconds]
<cole-h> one thing I realllly hate about zsh is that it doesn't have individual manpages for its builtins. I try to `man autoload` and "no entries for autoload". I have to `man zshbuiltins` and search for autoload, and THEN search for the flag I wanted to understand, and hope that my search doesn't overshoot the autoload section....
rajivr has quit [Quit: Connection closed for inactivity]
<__monty__> I seem to remember bash does something similarly annoying. Redirecting `man somebuiltin` to the "builtins" man page which only lists the builtins but not their documentation?
Fare has quit [Ping timeout: 264 seconds]
Fare has joined #nixos-chat
rj has joined #nixos-chat
Fare has quit [Ping timeout: 264 seconds]
Fare has joined #nixos-chat
waleee-cl has joined #nixos-chat
rj has quit [Ping timeout: 268 seconds]
rj has joined #nixos-chat
johanot has quit []
johanot has joined #nixos-chat
<gchristensen> iscsi is the worst thing
drakonis has quit [Remote host closed the connection]
drakonis has joined #nixos-chat
<ajs124> gchristensen: I'm sure there's worse. Maybe Ata over Ethernet?
<samueldr> ugh, software which don't describe their failures
<samueldr> describes it failed, but does no say why
<gchristensen> ajs124: can I persuade you to help? :P
<gchristensen> (not right this moment...)
<samueldr> worse than not describing the failure, is when the software re-uses the same exact wording at multiple places
<ajs124> gchristensen: maybe. which iscsi stack are you trying to use again?
<gchristensen> the module you handed me ;)
rj has quit [Ping timeout: 268 seconds]
<ajs124> right… that's a thing that happened. Although you might be more familiar with the code right now, if you've read it recently.
<ajs124> Ah, there's a pretty big bug in there, btw. Shutdown doesn't work, so I always use "sync; echo b > /proc/sysrq-trigger".
rj has joined #nixos-chat
<samueldr> ajs124: busybox's shutdown?
<samueldr> or is there a stage-2 running?
* samueldr tries to remember what was needed to fix busybox shutdown/reboot/poweroff commands
<samueldr> ah right, busybox shutdown requires you to use the busybox init
<ajs124> don't remeber, tbh. I just remember shutdown hanging. Might be some ordering thing, where it shuts down networking (why?) and only then tries to unmount filesystems.
<samueldr> (since it doesn't seem related) for Mobile NIxOS stage-1 the solution was to not use busybox shutdown/reboot/poweroff, but this https://github.com/mobile-nixos/hardshutdown
<clever> samueldr: i saw something about `shutdown -h --force --force` recently
<clever> the double force, tells it to ignore the rpc to init, and just kill it
<clever> ah, its in `man poweroff`
<clever> Force immediate halt, power-off, or reboot. When specified once, this results in an immediate but clean shutdown by the system manager. When specified twice, this results in an immediate shutdown
<clever> without contacting the system manager. See the description of --force in systemctl(1) for more details.
<samueldr> that's systemd's poweroff (and maybe others)
<samueldr> busybox's is really slim
<clever> busybox appears to signal pid1 via either kill or telinit
<clever> and yeah, cant initiate a shutdown on its own
* clever heads to bed
<samueldr> clever: if you look at my "hardshutdown" repo, it's based off a bit that is part of busybox already
<clever> that reminds me
<clever> kernel: Add support for 'tryboot' OS upgrade fallback mechanism on BCM2711
<clever> you can somehow pass strings to the kernel, when you hit the reboot syscall
<samueldr> ah!
<clever> and the kernel can then pass those on to the firmware
<clever> in this case, it changes what config files the firmware uses in the fat32 fs
<samueldr> that's how `reboot recovery` and `reboot bootloader` works on android-based devices
rj has quit [Ping timeout: 268 seconds]
<clever> ah
<samueldr> the implementation, kernel side, is extremely device dependent
<gchristensen> ajs124: I dunno, I never got it to finish booting -- it failed immediately after mounting the disk
<clever> and with `reboot tryboot` on an rpi, it will temporarily use different config files
<samueldr> clever: systemd's reboot already implements do_reboot2, so that's good
<clever> allowing you to test out firmware, without the risk of bricking the SD card
<samueldr> interesting
<samueldr> but only on the 4, right?
<clever> >> Add support for 'tryboot' OS upgrade fallback mechanism on BCM2711 #3937
<clever> i think so
<{^_^}> https://github.com/NixOS/nixpkgs/pull/3937 (by michelk, 6 years ago, closed): haskell-hindent: initial expression
<clever> it will use tryboot.txt instead of config.txt, i think thats all it does
<samueldr> yeah, a single bit
<clever> but from there, you can specify a different uboot name
<ajs124> gchristensen: _after_? That's interesting…
<samueldr> not enough info to reboot into a specific generation
<samueldr> but interesting still
<clever> samueldr: the noobs/pinn mechanism also uses this pathway
<clever> it tells the 1st stage (bootcode.bin) which fat32 partition to find start(4).elf on
<clever> so you can sorta get generations, by having duplicat fat32 partitions
<ajs124> gchristensen: I found something, wait
rj has joined #nixos-chat
<clever> the main firmware will be included in that generation as well!
<clever> so you have rollbacks for firmware too
ixxie_ has quit [Quit: Lost terminal]
<ajs124> gchristensen: https://gist.github.com/ajs124/a3dbaa7edd5282e28f3c5aafb7a3c357 maybe you need something like that?
<gchristensen> worth a go :)
<ajs124> We also only use loginAll.
<clever> to bed!, before i get distracted again! lol
<drakonis> https://typeclasses.com/ gosh i wish this wasnt so expensive for me lol
<gchristensen> after a "long time", initiatorRootTargetName # chmod: /mnt-root/tmp: No such file or directory ... then ... target # [ 138.435940] Did not receive response to NOPIN on CID: 0, failing connection for I_T Nexus iqn.2020-08.org.linux-iscsi.initiatorhost:example,i,0x00023d000001,iqn.2003-01.org.linux-iscsi.target.x8664:sn.acf8fd9c23af,t,0x01 ... then ... initiatorRootTargetName # [ 87.717506]
<gchristensen> connection1:0: detected conn error (1021)
<ajs124> gchristensen: I'll take a look later, but I'm not sure I have any good ideas, tbh
<gchristensen> I'm a super uninformed person trying to make the thing work, so I'd put the odds on you being able to figure it out higher than me :P if you don't, no worries, but if you do -- that'd be great
<lovesegfault> ashkitten: do we have a way to run pipewire with elevated priority yet?
<lovesegfault> in my attempt at using it I found it became unusable when I was compiling code
<lovesegfault> lots of stuttering, etc
<lovesegfault> I get around this with pulse by giving it a super low niceness
<ashkitten> lovesegfault: rtkit?
<lovesegfault> I use that with pulse, yeah
rj has quit [Ping timeout: 268 seconds]
<ashkitten> if rtkit is not helping then idk
<lovesegfault> Well, I didn't try rtkit with pipewire
<lovesegfault> do you have that setup?
<ashkitten> oh
<ashkitten> just enable rtkit?
<ashkitten> pipewire will automatically use it with the default config
<lovesegfault> Oh!
<lovesegfault> that's badass
* lovesegfault goes try
<ashkitten> fyi i had some issues with mumble's jack backend when rtkit was enabled
<ashkitten> mumble seems to be a badly behaved client and exceeds whatever realtime bounds expectations or something
<ashkitten> i talked to wtay about it, they said they could potentially make it so pipewire loads the config file <basename>.conf with a fallback to jack.conf, so mumble would load mumble.conf where rtkit would be disabled?
<ashkitten> but that is not in place right now
<ashkitten> mumble with pulse backend works fine though
rj has joined #nixos-chat
<lovesegfault> Let's see
<lovesegfault> playing music and running stress-ng
<lovesegfault> Nice, working perfectly
eta has quit [Ping timeout: 245 seconds]
<lovesegfault> ashkitten++
<{^_^}> ashkitten's karma got increased to 28
eta has joined #nixos-chat
rj has quit [Ping timeout: 268 seconds]
rj has joined #nixos-chat
<adisbladis> Is there some pasystray/pavucontrol equivalent for pipewire?
<adisbladis> I'm kinda tempted to sacrifice good bluetooth codec support (as I don't really use bluetooth audio a lot any more)
<adisbladis> But pasystray is just so good
<samueldr> I still haven't looked at pipewire, but isn't it supposed to work just like pulseaudio, and pavucontrol is the pavucontrol for pipewire?
<samueldr> (I would assume *then* that pasystray would be working still?)
<adisbladis> Hmm
<adisbladis> That would be great
<adisbladis> OTOH what practical advantages do you get from pipewire rn?
<samueldr> ¯\_(ツ)_/¯
<samueldr> (still haven't looked at it!)
<samueldr> I guess that as a drop-in replacement for pulseaudio, strictly, not much... but pipewire AFAIUI does more than allow your audio to work
<samueldr> IIRC, it also handles video streams
<samueldr> and is built more to "route" media
<hexa-> samueldr: enabling pulse support should be responsible for allowing pavucontrol et al to work
<hexa-> at least I enabled it and those tools are working for me
<hexa-> adisbladis even
<samueldr> :)
<hexa-> > services.pipewire.pulse.enable
<{^_^}> undefined variable 'services' at (string):489:1
<hexa-> also I saw an example where one needs to enable rtkit for realtime scheduling
<adisbladis> What about pipewires resampling methods?
<hexa-> during load I've experienced many artifacts
<hexa-> no clue
<adisbladis> The default pulseaudio resampling method _sucks_
<adisbladis> And introduces audible artifacts
__monty__ has quit [Quit: leaving]
<adisbladis> I've blind tested myself on it and I can hear the difference
Dotz0cat has joined #nixos-chat
spudly has quit [Ping timeout: 245 seconds]
rj has quit [Ping timeout: 268 seconds]
<samueldr> (in a VM) hah! bochs_drm has to be loaded before stage-2 starts for things to work
rj has joined #nixos-chat
<samueldr> without sddm, startplasma-wayland now doesn't simply hang doing nothing
<samueldr> oops
<ajs124> gchristensen: got it
<ajs124> boot.initrd.network.flushBeforeStage2 = false;
<ajs124> that's all you need to make the test succeed
<ajs124> I figured that out with boot.trace
rj has quit [Ping timeout: 268 seconds]
<ashkitten> adisbladis: iirc pipewire has only one resampling algorithm with a configurable quality parameter, chosen for performance and indistinguishable quality