gchristensen changed the topic of #nixos-chat to: NixOS but much less topical || https://logs.nix.samueldr.com/nixos-chat
<joepie91> samueldr: interesting legal conundrum: what if someone uses leaked driver source code to build a mainline driver?
<joepie91> considering that the leaked driver code should have been published as GPL
<ldlework> Speaking of AIs and stuff, I just finished "Level Six" the sequel to "Level Five" and find myself excited for the third book.
<samueldr> joepie91: IANAL, but AFAICT it's legally unclear
<samueldr> it's also legally unclear whether the GPL actually is enforceable
<samueldr> (I think)
<joepie91> well realistically almost all licenses are untested in court :P
<gchristensen> it turns out having unclear copyright ownership makes it difficult
<joepie91> so a certain amount of nonclarity is assumed baseline
<gchristensen> (or even clear, but highly distributed)
<abathur> laws are wild stuff
<abathur> should have some courts dedicated to legal fuzzing
<gchristensen> don't we though? :)
<samueldr> it's called: the courts
<abathur> no
<abathur> because those courts are setting precedent
<abathur> that's like letting the fuzzer write the docs
<cole-h> supersandro2000: I have a WIP patch for fixing that uutils issue with `tr`, if you wanna test it out
<cole-h> needs cleaning before I submit it upstream, but it works*
<samueldr> abathur: I'd gladly let a fuzzer write docs if it meant docs were written and truthful to the state of things
<abathur> it'd just flip the problem
<abathur> instead of users spending hours figuring out what is supposed to work the maintainers would be spending hours figuring out if the docs were describing a bug or a feature :)
<cole-h> one quirk I discovered is that, since rust is utf8-aware... `echo X | cargo run -- tr X '\377'` gives you ÿ (U+00FF), whereas `echo X | tr X '\377'` gives you the byte 0xff
<jtojnar> abathur is not this a thing? https://en.wikipedia.org/wiki/Declaratory_judgment
<abathur> I don't mean to suggest the courts don't have to disambiguate inputs
<aaronjanse> The bug vs feature decision is the job of legislature, right?
<abathur> but more that courts dedicated to legal fuzzing would be an arrangement where smart people are paid real money to treat the law like a piñata rather than us all having to sit around and wait for someone with the appropriate combination of money, personality, and legal representation necessary to actually force issues
<abathur> grey area probably
<abathur> depends on the law? :)
<abathur> for example, it's silly that we're this far in to OSS licensing and don't know the answers to some of these questions; I'm not if the outcomes, whether LARPed or real, would necessarily lead to any legislative decision
<abathur> *not sure
<abathur> they might, but it might also just mean people on all sides of GPL questions re-evaluate their risk tolerance?
<supersandro2000> cole-h: nice! Going to look at this probably tomorrow
<cole-h> sounds good. if you ping me before you test it out, I might have cleaned it up slightly
<supersandro2000> cole-h: in that gist or are you going to create a PR?
<supersandro2000> I also already patched autoconf to treat that issue correctly and fallback to bashism
<cole-h> depends on how happy I am with it :P
<cole-h> if I think it's "OK", then I'll probably PR it and ping you; otherwise, I'll update the gist. (but only on-demand, cuz, well, I'm iterating on it rapidly)
supersandro2000 has quit [Disconnected by services]
supersandro2000 has joined #nixos-chat
<supersandro2000> cole-h: I don't know rust and can't really tell if it is good code or not
<cole-h> I can :P
<supersandro2000> right now I just want a quick fix to make it ready for normal usage
<cole-h> I just don't want to fix a bug and worsen the code quality at the same time
<supersandro2000> or even worse add new bugs and regressions
* abathur blinks
<cole-h> I don't think I can regress the unescaping, but I can certainly add a new bug / inconsistenty with gnu coreutils
<supersandro2000> but we have a pretty giant test suite: building a complete OS with it
<cole-h> hehe, that's fair
* siraben catches up on messages
<siraben> aaronjanse: that's great!
<siraben> how did your BIOS settings get reset? lol
rajivr has joined #nixos-chat
tomberek has quit [Ping timeout: 240 seconds]
<supersandro2000> cole-h: why is ofborg so slow when you need it? https://github.com/NixOS/nixpkgs/pull/116366
<{^_^}> #116366 (by SuperSandro2000, 1 hour ago, open): cpython: allow $out reference again
<supersandro2000> I want to get this PR merged now...
slack1256 has quit [Remote host closed the connection]
<aaronjanse> No idea siraben. I think my hardware did it automatically after thinking I was stuck in a boot loop? (I was constantly rebooting because I was setting my tty colors and font)
<aaronjanse> *not "no idea" because I just stated an idea
<supersandro2000> cole-h: I found another potential task for you https://github.com/uutils/coreutils/issues/1822
<{^_^}> uutils/coreutils#1822 (by SuperSandro2000, 59 seconds ago, open): head: missing --zero-terminated
<supersandro2000> noticed it while building python that the set-source-date-epoch-to-latest.sh hook is broken
<supersandro2000> also somehow it thinks that my Lib is capitalized and install fails
<cole-h> supersandro2000: stdenv rebuilds take a while
<supersandro2000> cole-h: yeah true.. I just want the eval to pass finally
<cole-h> if staging things are missing from cache, ofborg will have to build them itself :P
<cole-h> and since they're from staging... lol
<cole-h> supersandro2000: any reason you're so eager to merge this PR? if it were me, I'd wait for fridh to chime in (or merge it, as they are wont to do)
<cole-h> (re: "I am waiting for ofborg, then pushing the change and then merging.")
<cole-h> ah, didn't read the PR body
<cole-h> disregard
<cole-h> sorry :)
<supersandro2000> cole-h: because it broke all PRs targeting staging.
<supersandro2000> I think based on the change here it is pretty clear what broke https://github.com/NixOS/nixpkgs/commit/738746a34b675297624338ab9aa3adb3d3cd4cf6
<supersandro2000> $out was forgotten to be moved to the new function
<supersandro2000> also I am working a lot from staging lately so I will probably rebase tomorrow forgotten that staging is broken
<supersandro2000> I also rebuild up to libxml2 which was broken beforehand and the issue was that the path to Python.h was filled with an eeeeeee path
<supersandro2000> that should be the sign of a removed reference
<supersandro2000> I now realize that I should have written that in the PR
<samueldr> in the commit message even
<supersandro2000> yeah, to late. try to remember that for the next time
<supersandro2000> cole-h: I just found another bug in uutils-coreutils which I cannot patch in autoconf
<supersandro2000> maybe you find some time the next days to look at this?
<{^_^}> uutils/coreutils#1823 (by SuperSandro2000, 2 minutes ago, open): install: does not remove path from files
<siraben> is it possible to reduce one's karma
<siraben> siraben--
<aaronjanse> siraben--
<aaronjanse> Aww
<siraben> siraben++
<{^_^}> siraben's karma got decreased to 6
<siraben> ope
<aaronjanse> :O
<aaronjanse> siraben++
<{^_^}> siraben's karma got increased to 7
<siraben> aaronjanse thanks
<aaronjanse> Np
<aaronjanse> I'm almost done with a MVP of a backup tool I'm building that's inspired by git-annex
<aaronjanse> Then I'll do root-on-tmpfs and have a very clean system for making sure everything is backed up
<siraben> ooh
<siraben> At the moment I'm using borg backup for backing up, the other alternatives seemed too complex as if I needed to be a sysadmin to learn to use it properly
<pie_> aaronjanse: youve seen restic?
<pie_> aaronjanse: show me youre thing when youre done \o/ im great at losing data
<aaronjanse> I used Restic for ~6 months
<aaronjanse> Will do
<aaronjanse> The main thing I want is a single view of all my important files, without having to have a local copy of them all
<aaronjanse> So my tool will let me easily pull/push individual large files that I rarely need, storing different parts of my backup repo in different places
<aaronjanse> Basically just git-annex but faster and written in Rust
tomberek has joined #nixos-chat
<pie_> I need to figure out udev one of these days. And file a bug report upstream asking for a pokey interface not dependent on files-on-disk -_-
<pie_> ive been wanting to do that for like a year now
waleee-cl has quit [Quit: Connection closed for inactivity]
<aaronjanse> https://xon.sh Hmmmmm
spudly has quit [Ping timeout: 246 seconds]
<elvishjerricco> I really hate how often I have to reboot just to use Bluetooth headphones on this system...
<pie_> elvishjerricco: VM_
<pie_> ?
<elvishjerricco> pie_: No. Native NixOS install
<ar> elvishjerricco: I didn't have to reboot in recent memory, but a few times i had to "pulseaudio -k"
<pie_> no i mean put it in a vm and reboot the vm ÉP
<pie_> * :P
<lovesegfault> I have a newer intel wifi/bt card and it sucks
<lovesegfault> the older one on my laptop was good
<elvishjerricco> pie_: It's not the time it takes to reboot. It's that I have to lose all my open applications
<ar> i have ax200 in my current laptop
<elvishjerricco> Sometimes I can find the right incantations of services and kernel modules to reload to fix everything but the same thing never works twice
evanjs has quit [Read error: Connection reset by peer]
<ar> it might help that i'm on nixos-unstable and linuxPackages_latest kernel
evanjs has joined #nixos-chat
<pie_> elvishjerricco: yeah thats why i said mazbe you can put the bluetooth stuff in a vm or container or something ÉP
<pie_> * :P
<elvishjerricco> That sounds complicated to say the least :P
<pie_> Getting real tired of this shit https://www.patrick-breyer.de/?page_id=594160&lang=en
cole-h has quit [Ping timeout: 245 seconds]
spudly has joined #nixos-chat
patagonicus has quit [Quit: The Lounge - https://thelounge.chat]
patagonicus has joined #nixos-chat
<ashkitten> yeah :/
<lunc> Morning
<aaronjanse> Good morning lunc
* lunc is looking forward to Friday
<aaronjanse> Me too
<aaronjanse> That's when my midterms are over
<lunc> heh
<AMG> I'm looking forward to my future vacation, which sadly remains TBD. Impossible to plan anything right now.
tomberek has quit [Quit: Connection closed]
__monty__ has joined #nixos-chat
ixxie has joined #nixos-chat
liebach has left #nixos-chat [#nixos-chat]
ixxie has quit [Quit: Lost terminal]
ixxie has joined #nixos-chat
lunc has quit [Read error: Connection reset by peer]
lunc has joined #nixos-chat
<jD91mZM2> aaronjanse: Sorry for missing your messages yesterday! I keep forgetting to check IRC
<jD91mZM2> You should definitely use home-manager, it's a life saver
<jD91mZM2> And yep, it's a workaround for SSBs. It will continue to work until they remove userChrome.css because it's probably also going to get called an "unused feature"
<jD91mZM2> You already have to enable an option with "legacy" in its name
<jD91mZM2> firefox is going down a bad road but hell if it isn't better than chrome regarding security
wmey1 has quit [Ping timeout: 272 seconds]
rj_ has joined #nixos-chat
Dotz0cat has quit [Ping timeout: 256 seconds]
tomberek has joined #nixos-chat
supersandro2000 has quit [Quit: The Lounge - https://thelounge.chat]
supersandro2000 has joined #nixos-chat
<gchristensen> I wonder how grub, ubuntu, rhel and stuff detect console parameters properly for bare metal
<f0x> do they :P
* f0x spent too long trying to figure out usb serial console :(
<__monty__> Does anyone grok how to navigate the Apple developer documentation website? I can only find things through the archive but that has admonitions all over the place the documents are no longer updated and I should use the documentation website, which seems more like an ad than usable documentation? : s
cole-h has joined #nixos-chat
<LinuxHackerman> I could imagine DTBs providing defaults in a lot of cases so no console parameter needs to be passed on the cmdline
pinage404[m]1 has quit [Quit: Idle for 30+ days]
<patagonicus> Learning how modules work was so worth it. Now managing all of the machines I have is so much nicer. :)
LinuxHackerman1 has joined #nixos-chat
ma27[m]1 has joined #nixos-chat
grahamc has joined #nixos-chat
ili1 has joined #nixos-chat
LinuxHackerman has quit [Ping timeout: 265 seconds]
ma27[m] has quit [Ping timeout: 265 seconds]
ili has quit [Ping timeout: 265 seconds]
immae has quit [Ping timeout: 265 seconds]
gchristensen has quit [Ping timeout: 265 seconds]
jess has quit [Quit: Reconnecting]
jess has joined #nixos-chat
grahamc is now known as gchristensen
immae1 has joined #nixos-chat
endformationage has joined #nixos-chat
rj_ has quit [Quit: rj_]
<cole-h> how so?
<patagonicus> Well, for one thing I've made a module for backups to my NAS. After I had that enabling backups for a new machine was basically just setting one option in that machines config + generating an encryption and SSH key. And if I notice that something's not working well I usually just need to change it once and then just pull the changes on the machines
<patagonicus> when I do updates.
<patagonicus> I plan on making the VMs and my desktop pull updates to the config automatically, although for some other machines I'd rather do it manually.
rj_ has joined #nixos-chat
slack1256 has joined #nixos-chat
rajivr has quit [Quit: Connection closed for inactivity]
siraben has quit [*.net *.split]
kloenk has quit [*.net *.split]
siraben has joined #nixos-chat
kloenk has joined #nixos-chat
siraben has quit [Max SendQ exceeded]
njha[m] has quit [Ping timeout: 244 seconds]
rycee has quit [Ping timeout: 240 seconds]
kalbasit[m] has quit [Ping timeout: 240 seconds]
aaronjanse has quit [Ping timeout: 244 seconds]
enick_4 has quit [Ping timeout: 244 seconds]
Guest68587 has quit [Ping timeout: 244 seconds]
jtojnar has quit [Ping timeout: 244 seconds]
kraem[m] has quit [Ping timeout: 244 seconds]
bbigras has quit [Ping timeout: 244 seconds]
joe[m]1 has quit [Ping timeout: 241 seconds]
mjlbach has quit [Ping timeout: 241 seconds]
philipp[m]1 has quit [Ping timeout: 241 seconds]
chvp has quit [Ping timeout: 241 seconds]
leons has quit [Ping timeout: 241 seconds]
crazazy[m] has quit [Ping timeout: 249 seconds]
manveru[m] has quit [Ping timeout: 249 seconds]
Ke has quit [Ping timeout: 249 seconds]
Ox4A6F has quit [Ping timeout: 249 seconds]
ma27[m]1 has quit [Ping timeout: 247 seconds]
LinuxHackerman1 has quit [Ping timeout: 270 seconds]
leonardp has quit [Ping timeout: 247 seconds]
emily has quit [Ping timeout: 247 seconds]
colemickens has quit [Ping timeout: 247 seconds]
kloenk has quit [Ping timeout: 265 seconds]
ili1 has quit [Ping timeout: 268 seconds]
puzzlewolf has quit [Ping timeout: 268 seconds]
noneucat has quit [Ping timeout: 268 seconds]
rmcgibbo[m] has quit [Ping timeout: 246 seconds]
DavHau[m] has quit [Ping timeout: 258 seconds]
immae1 has quit [Ping timeout: 258 seconds]
nicolas[m] has quit [Ping timeout: 244 seconds]
artturin has quit [Ping timeout: 244 seconds]
thefloweringash has quit [Ping timeout: 258 seconds]
danielrf[m] has quit [Ping timeout: 258 seconds]
veleiro has quit [Ping timeout: 240 seconds]
aanderse has quit [Ping timeout: 240 seconds]
lopsided98 has quit [Ping timeout: 264 seconds]
lopsided98 has joined #nixos-chat
aanderse has joined #nixos-chat
thefloweringash has joined #nixos-chat
manveru[m] has joined #nixos-chat
kloenk has joined #nixos-chat
joe[m]2 has joined #nixos-chat
colemickens has joined #nixos-chat
kalbasit[m] has joined #nixos-chat
crazazy[m] has joined #nixos-chat
Ke has joined #nixos-chat
leons has joined #nixos-chat
ili has joined #nixos-chat
rycee has joined #nixos-chat
danielrf[m] has joined #nixos-chat
noneucat has joined #nixos-chat
bbigras has joined #nixos-chat
siraben has joined #nixos-chat
immae has joined #nixos-chat
artturin has joined #nixos-chat
leonardp has joined #nixos-chat
jtojnar has joined #nixos-chat
LinuxHackerman has joined #nixos-chat
puzzlewolf has joined #nixos-chat
chvp has joined #nixos-chat
nicolas[m] has joined #nixos-chat
kraem[m] has joined #nixos-chat
mjlbach has joined #nixos-chat
rmcgibbo[m] has joined #nixos-chat
Guest57260 has joined #nixos-chat
veleiro has joined #nixos-chat
ma27[m] has joined #nixos-chat
emily has joined #nixos-chat
Ox4A6F has joined #nixos-chat
DavHau[m] has joined #nixos-chat
philipp[m]1 has joined #nixos-chat
enick_326 has joined #nixos-chat
aaronjanse has joined #nixos-chat
njha[m] has joined #nixos-chat
drakonis- has joined #nixos-chat
drakonis has quit [Ping timeout: 256 seconds]
drakonis- is now known as drakonis
ixxie has quit [Quit: Lost terminal]
lopsided98 has quit [Ping timeout: 240 seconds]
<abathur> anyone know how/why `nm --undefined` on a binary could show a symbol from GLIBC that I can't locate when reading it with xxd?
lopsided98 has joined #nixos-chat
<abathur> I haven't audited to see if I can find them all, but specifically I am looking at sudo on nixOS (inside `nix-shell -p sudo`) and nm says execl, execve, and fexecve are all present; I see execl and fexecve via xxd, but not execve
<gchristensen> I wish our systemd were stage-1
<gchristensen> erm
<gchristensen> I wish our stage-1 were systemd
<samueldr> paging elvishjerricco for collaboration purposes :)
<ar> that would require systemd to be present in initrd?
<samueldr> gchristensen: what feature is making you wish for it at the current time?
<abathur> execve isn't the only one, but the list is pretty short; execve, exit, getpgrp, sleep, and textdomain
<elvishjerricco> samueldr: Hm?
<elvishjerricco> oh
<samueldr> systemd!
<samueldr> ar: since no one answered... yes, it would
<elvishjerricco> I should really change up my code to be a very broken nixpkgs branch instead of a weird external bit.
<samueldr> and don't hesitate to ask for help where you need it!
<samueldr> but yeah, it wouldn't be that much more, considering udev is in stage-1 already, which brings a chunk of systemd libraries
<elvishjerricco> My initrd with this is like 20M :/
<elvishjerricco> I'm sure I could get it down, but I'm not sure where to start
<gchristensen> great so we're saving 2M and the only cost is miserably painful initrd scriptsn 18M ./rd0ykzgsi3w7nxykyigir63lg83g11l3-initrd-linux-5.10.15-initrd.efi
<elvishjerricco> One thing I did differently is I don't have that weird extrautils derivation. Instead, I just copy binaries and libraries into their usual locations so nothing needs to be patched.
<elvishjerricco> gchristensen: Oh the standard one is 22M? I thought it was like 12
<samueldr> gchristensen: your .efi file contains the kernel too :)
<gchristensen> nope
<samueldr> oh, sorry then
<gchristensen> I'm not using secureboot right now :)
<samueldr> the initrd can be a separate .efi file?
<elvishjerricco> Well... I don't feel so bad about 20M then :P
<gchristensen> yea
<samueldr> I thought it had to be combined with the kernel
<samueldr> (and it wouldn't matter for secureboot really... the bootloader and/or the kernel could/should verify it if loaded separately)
<gchristensen> the reason I'm bringing it up is I need networking, iscsid, multipathd, and other things running at the right time with the right options in initrd and I'm tired of shell scripts
<gchristensen> and having them pass on in to the right status in systemd afgter stage 1 would be nice
<gchristensen> so instead of making it work we're just giving up and going to run nixos in a vm
<samueldr> before investing more time in the custom mobile nixos init, I want to see and evaluate systemd in stage-1 properly
<elvishjerricco> One odd thing about systemd in stage 1 is that it won't pass status unless it thinks it's passing to another systemd binary. But stage 2 starts with the activation script. So I want to have a service in stage 1 that runs nixos-enter to activate the system in stage 1 before switching root and running the systemd binary directly
<gchristensen> well maybe we can replace activation scripts too ...
<samueldr> (mainly, if it can fail gracefully in a touchscreen-only environment)
<elvishjerricco> Well that's a whole other project :P
<samueldr> elvishjerricco: how does it determine it's another systemd binary?
<elvishjerricco> samueldr: I forget. But I don't think you can trick it without patching systemd
<samueldr> maybe we can alias the activation script to something like .../systemd-compat/systemd if it's relatively dumb
<elvishjerricco> It's pretty simple logic in the switch root code in systemd though
<elvishjerricco> I believe it checks if it's literally the exact same install path as the binary in stage 1
<elvishjerricco> as determined at build time
<elvishjerricco> Hm trying to find it in the systemd code... can't remember where exactly it is
<samueldr> since it closes the serialization bits afterwards
<gchristensen> I don't feel great about tricking it, anyway... I'd wager there is a way to make ~all activation scripts work at the right time without it
<samueldr> no need to trick it I think
<elvishjerricco> gchristensen: It'd be tough to do that in a way that both works at boot and when the user does nixos-rebuild switch
<samueldr> each initrd will have a specific generation
<samueldr> and the SYSTEMD_BINARY_PATH built-in will match that generation
<samueldr> no?
<samueldr> ah, it might be *another* systemd build though
waleee-cl has joined #nixos-chat
<elvishjerricco> I just wonder about things like updating etc
<samueldr> e.g. systemd-but-for-stage-1 (with fewer things enabled, smaller build size)
<samueldr> elvishjerricco: ?
<elvishjerricco> That's done during activation, no?
<samueldr> ah, I wasn't concerned about activation scripts
<samueldr> I was strictly thinking about handing off to the proper systemd so it deserializes state
<samueldr> and if it's the same derivation, it should already work
<elvishjerricco> Yea, initrd should only ever load a generation with the same systemd package
<elvishjerricco> so that's not a concern to me
<samueldr> I don't really know about activation scripts to know where there is an issue here
<elvishjerricco> But like etc is a particularly egregious example
<elvishjerricco> Because in order to make other activation scripts happen at the right time without activation scripts, you'd use systemd units. But updating etc in a systemd unit is a problem because etc is how we install our systemd units
<elvishjerricco> And then it's not exactly clear to me that our systemd unit restarting logic for nixos-rebuild switch would handle the other activation-scripts-turned-systemd-units correctly
<gchristensen> probably not
<elvishjerricco> So, in the meantime, `nixos-enter` from stage 1 is where I'm headed :P
cole-h has quit [Ping timeout: 256 seconds]
cole-h has joined #nixos-chat
<abathur> trying to run a GH action and it's been queued for 20m
<gchristensen> ouch
<abathur> tried to go to HN to see if there was a gripe thread, and HN doesn't want to load; obvious conclusion: HN is down, so GH can't keep up with the number of people actually working on code? :)
<gchristensen> lol
<gchristensen> maybe azure has no more servers
<gchristensen> that happens from time to time
<gchristensen> actually, https://status.azure.com/en-us/status
<abathur> nice
<gchristensen> elvishjerricco: what if I stage-3'd myself, and used systemd to pivot from one nixos generation to another with my iscsid servecise up and running :P
<gchristensen> "oh god" you say
<elvishjerricco> gchristensen: It's not far off from samueldr's stage-0 concept :P
<gchristensen> hmm can I bind-mount / to /mnt-root/ then my new /nix to /mnt-root/nix and then ....
<cole-h> gchristensen: Funny, I'm actually thinking about a similar-ish thing (to let us not need mountpoint=legacy for zfs)
<cole-h> just need to install a VM so I can start messing around
<elvishjerricco> cole-h: Why would such a thing be necessary for that? I'm pretty sure I know how to do it without any modification to nixos as it stands
<cole-h> oh?
<gchristensen> nice
<gchristensen> how ultra cursed is this idea elvishjerricco
<cole-h> I'm curious how you'd achieve that, elvishjerricco.
<elvishjerricco> Set mountpoint=/ for the root dataset. Make sure hardware-configuration.nix doesn't contain your automounted datasets. Since stage 1 imports with -N, we don't need to worry about spurious mounts. Manually add a filesystems entry for / and whatever other datasets, with options=["zfsutil"]; (might need a dash in there, I forget the option name). Then boot, and zfs-mount.service takes care of the rest
<elvishjerricco> So you should only need nix entries for / and /nix/store
<elvishjerricco> whatever datasets those are on
<elvishjerricco> But they can still use the automatic mountpoints, even if they're only to guide the mounpoints of children
<cole-h> Hmmmmmmmmmm, interesting...
<elvishjerricco> cole-h: Yea the magic is that zfsutil option. ZFS won't let you manually mount a dataset with a non-legacy mountpoint, but it will if you give that option
<elvishjerricco> specifically for this purpose
<elvishjerricco> and so that the zfs command can mount but whatever
<cole-h> "-o zfsutil This private flag indicates that mount(8) is being called by the zfs(8) command. "
<cole-h> I see, I see
<elvishjerricco> I want to move all my systems to work this way, using --no-filesystems when I regenerate my hardware-configuration.nix.
<elvishjerricco> But I haven't even tested it yet
<elvishjerricco> I just know that's how ubuntu and others do it
<cole-h> hm, I don't think you could do that, since there's an assertion for the presence of a root fs
<gchristensen> actually my idea would be ultra cursed since I'd have three /nix's and they'd all be out of sync
<elvishjerricco> cole-h: Right. You do have to have a root entry in your nix expression. Just needs to use options=["zfsutil"];
<elvishjerricco> Same for /nix/store if that's on a separate dataset
<elvishjerricco> But those are the only ones
<elvishjerricco> everything else will be properly automounted
<elvishjerricco> And even those two should still have a mountpoint in this setup; legacy not necessary
<cole-h> So then `--no-filesystems` wouldn't really be helpful, no? Or would you put a `fileSystems."/" = { ... }` in your configuration.nix instead?
<cole-h> elvishjerricco++ Either way, thanks for the info, this seems like a good direction to explore
<{^_^}> elvishjerricco's karma got increased to 19
<elvishjerricco> cole-h: --no-filesystems would be so that it doesn't generate entries for my dozen other datasets that SHOULDN'T have entries. I would manually keep the ones that necessarily aren't automounted, like /, /nix/store, and /boot
<elvishjerricco> Ideally nixos-generate-config just wouldn't generate entries for zfs datasets with mountpoints
<cole-h> ah ok
<clever> elvishjerricco: nixos-generate-config is also not aware of which block device zfs lives on
<clever> elvishjerricco: and if that is on luks, then the luks isnt auto-generated
<elvishjerricco> clever: Right, that's already something I have to manually include in my configuration.nix
<elvishjerricco> Though honestly I bet there's a way to figure that out
<clever> elvishjerricco: i think nixos relies on lsblk
<clever> when i run lsblk, i can see the LV's inside lvm that depend on the luks device, and thru to the raw block dev
<clever> but its not aware of / being on an LV
Dotz0cat has joined #nixos-chat
energizer has quit [*.net *.split]
kraem1 has quit [*.net *.split]
averell has quit [*.net *.split]
lovesegfault has quit [*.net *.split]
V has quit [*.net *.split]
blueberrypie has quit [*.net *.split]
kcalvinalvin has quit [*.net *.split]
kcalvinalvin has joined #nixos-chat
blueberrypie has joined #nixos-chat
lovesegfault has joined #nixos-chat
kraem1 has joined #nixos-chat
V has joined #nixos-chat
averell has joined #nixos-chat
kraem has joined #nixos-chat
<cole-h> elvishjerricco: Does `zfs mount` on a dataset also mount all its parent datasets?
kraem1 has quit [Read error: Connection reset by peer]
<elvishjerricco> cole-h: Dunno. I'd have to try it.
<elvishjerricco> but zfs-mount.service just does zfs mount -a
energizer has joined #nixos-chat
<cole-h> yeah, this is more for the installation process :P
<cole-h> documenting how I think it'll go
<elvishjerricco> I would wager it doesn't, but again, I'd have to try it
* cole-h goes to test
<cole-h> nope, doesn't.
slack1256 has quit [Ping timeout: 260 seconds]
blueberrypie has quit [Read error: Connection reset by peer]
blueberrypie has joined #nixos-chat
__monty__ has quit [Quit: leaving]
<f0x> isn't it great how my new server setup has infinitely many yaks to shave
<abathur> if there isn't a band called fractal yak...
<samueldr> ugh... code that silently ignores failure cases
<samueldr> one simple warning near the null check would have made it obvious where the issue was
<samueldr> instead of chasing through it through some scripting up into the native interface
<samueldr> and, more importantly, it's a symptom that things really aren't right if it gets to that place with a null (not) instance!
<supersandro2000> Is this where the pull request comes from https://git.wiki.kernel.org/index.php/Aliases#Serve_repo_on_the_spot ?
<supersandro2000> also why does git not have an interdiff by default!?
<samueldr> "pull request" the term comes from going on a mailing list and asking the maintainer to "pull from the branch" rather than applying patches
<samueldr> (if that's what you were asking)
* jess tips f0x over
<jess> :333
<joepie91> oh no, jess has found another target
* jess tips joepie91 over
<joepie91> :P
slack1256 has joined #nixos-chat
* joepie91 falls over
<jess> f0x is my fedi server admin (and they are very cool)
<joepie91> see now we're back to old times!
* jess catloafs on top of joepie91, victorious
<f0x> oh hey jess :)
<jess> hay \o
<ashkitten> 👀
<ashkitten> times have been fun in #nixos-chat lately huh
<jess> IM A CAT
<adisbladis> DONT START THE CAPS THING AGAIN, IM THE CAPS MAN
<ashkitten> cats good, ask how i know
<f0x> hmm maybe the 'ash' part in your name has something to do with it
<ashkitten> spot on
<ashkitten> ash is black and black cats are witches and witches ride broomsticks and broomsticks are like rakes and rakes are triangular and triangles are illuminati