worldofpeace changed the topic of #nixos-dev to: NixOS Development (#nixos for questions) | NixOS 20.09 Nightingale ✨ | | | 20.09 RMs: worldofpeace, jonringer |
<hexa-> hope we don't have contributors in embargoed countries … rip
<hexa-> probably members, rather than contributors … idk
<hexa-> it sounds off
supersandro2000 has quit [Disconnected by services]
supersandro2000 has joined #nixos-dev
rajivr has joined #nixos-dev
teto has quit [Ping timeout: 258 seconds]
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos-dev
<ekleog> I'd guess they had a shared company account and it got blocked after being opened in an embargoed country, blocking a whole organization sounds very unlikely
<samueldr> hexa-: we maybe did and do and had
<samueldr> there are... missing... data in the nixos org coming from... unclear... circumstances
<ekleog> yeah :( I wonder when resentment towards github will push us out of the github potential well we're currently in
* cole-h looks at the uptick in GH Actions integration with the Nixpkgs repo
<samueldr> that's been concerning me a whole lot
<samueldr> actions in itself, not really... actions not relying strictly on running nix builds, yes
<abathur> are we being vague about the circumstances for Reasons, or? :)
<samueldr> *they* are
<abathur> *they* being github, or the user?
<abathur> or both?
<samueldr> yes
<abathur> is there some sort of prohibition on saying this user's name?
<samueldr> I don't know, but I prefer preserving their privacy
<abathur> oh, fair
<samueldr> and it's more than one
<samueldr> (plausibly for different reasons)
<supersandro2000> using GitHub actions for building nixpkgs is probably a bad idea
<supersandro2000> they are slow and take at least 2 to 3 minutes cloning the damn rpeo
<abathur> I'm pretty sure at least one of they posted on discourse the other day, and I was surprised to see it, but I've started to wonder if it is a forbidden topic/knowledge
<samueldr> not AFAIK, but I wouldn't go deciding about that
<samueldr> (since it was screamed in the bird-void, I guess it's okay to link to it)
<abathur> fair; (I noticed discussion about one instance, at least, so I'm not really in-the-dark here; just wanting to avoid quixotically blabbing if there are good reasons to put a sock in it). I suspected they had disappeared, but they've posted on discourse.
asbachb has quit [Remote host closed the connection]
asbachb has joined #nixos-dev
<asbachb> Any ideas where/how to put a thumbnailer config to nixpkgs?
hplar has quit [Ping timeout: 268 seconds]
hplar has joined #nixos-dev
cole-h has quit [Ping timeout: 260 seconds]
asbachb has quit [Remote host closed the connection]
mkaito has quit [Quit: WeeChat 3.0]
justanotheruser has quit [Ping timeout: 260 seconds]
tilpner_ has joined #nixos-dev
tilpner has quit [Ping timeout: 264 seconds]
tilpner_ is now known as tilpner
thibm_ has joined #nixos-dev
thibm has quit [Ping timeout: 265 seconds]
justanotheruser has joined #nixos-dev
supersandro2000 has quit [Read error: Connection reset by peer]
supersandro2000 has joined #nixos-dev
<lovesegfault> is `nixos/release.nix`'s test attr broken?
<lovesegfault> I thought nix-build ./nixos/release.nix -A tests would be the way to check all nixos tests
<lovesegfault> but seems like that doesn't work anymore
<lovesegfault> (on master)
<abathur> I think flakes weirded it, iirc, but I didn't really know how it was supposed to work
<ekleog> wait flakes went in nixpkgs‽
<abathur> oh, I'm sorry
<abathur> I'm not reading clearly
<lovesegfault> ekleog: are you still awake or are you already awake? :P
<ekleog> uhhhhhh lemme call my lawyer
<abathur> hehe
<lovesegfault> lol
<ekleog> lovesegfault: anyway, to answer your initial question, now I've checked nixos/release.nix isn't a flake, I personally never had enough RAM to actually run all the tests at once
<abathur> <3 ekleog
<{^_^}> ekleog's karma got increased to 10
<ekleog> <3 abathur
<lovesegfault> ekleog: I'm using that abandoned data center machine
<{^_^}> abathur's karma got increased!
<ekleog> that said, it looks like things are more broken than usual, as there's an eval error before I oom
<lovesegfault> trying to bump binutils to 2.35.1 now that the CURSED bump to 2.34 is through
<lovesegfault> Yeah, that's what I'm getting too
<lovesegfault> how the hell is hydra passing though? I thought it eval'd this file?
<ekleog> maybe it hasn't passed since this was borken?
<lovesegfault> maybe?
<lovesegfault> postgresql-wal-receiver.nix seems to be the culprit
<lovesegfault> eh, nvm
<ekleog> yeah, though it looks like there's some other error if you comment it out, but then `nix-build . -A nixosTests` does pass after that
<samueldr> hydra uses a different evaluator, if it matters here
<lovesegfault> huh, yeah, nixosTests does build
<lovesegfault> if I comment out that postgres thing
<lovesegfault> but if I try to build ./nixos/release.nix -A tests it still doesn't work
<lovesegfault> spooky
<ekleog> fwiw I'm on fixing postgresql-wal-receiver, feel free to look into the other one
<lovesegfault> I'm trying to see if nixosTests will actually fully eval and start building anything
<lovesegfault> argh, goddamn nixUnstable not respecting my allow unfree
<ekleog> FWIW, nixosTests will normally build more tests than hydra does
<ekleog> hydra only has a subset of all the nixos tests that is channel-blocking
<lovesegfault> Hm
<lovesegfault> I need to take the time to setup my own hydra
<ekleog> I've just pushed fe8ec6a07ff1597495204bdcc979b39b5aac3e5e that should fix postgresql-wal-receiver
<lovesegfault> pulling
<ekleog> (erratum: it only fixes eval of nixosTests, there's some more stuff to do to fix its own eval, this time I'm waiting for a successful build before calling it a day and pushing)
<lovesegfault> Yup, that first commit fixed nixosTests
<lovesegfault> started the build, going to eat dinner and brb
<ekleog> actually fixed the build with a7331d1403e7480b2a68700977cef290491ababc, checked with a full build
kalbasit has joined #nixos-dev
<ekleog> and sent in so this file sees some more love and doesn't start sulking next time
<{^_^}> #108449 (by Ekleog, 1 minute ago, open): postgresql: add passthru.tests
orivej has quit [Ping timeout: 246 seconds]
<clever> how do you un-mark a haskell package as broken? i cant find the input hackage2nix is using
<{^_^}> #108453 (by lovesegfault, 47 seconds ago, open): binutils: 2.34 -> 2.35.1
<lovesegfault> Could someone with hydra access help me create a jobset to test this?
<lovesegfault> Given our last binutils bump I think it's warranted :P
bgamari_ has quit [Ping timeout: 258 seconds]
kalbasit has quit [Ping timeout: 265 seconds]
Emantor_ has quit [Quit: ZNC -]
Emantor has joined #nixos-dev
cole-h has joined #nixos-dev
<qyliss> The NixOS module that configures agetty is called services.mingetty, because it used to use mingetty instead of agetty 10 years ago
<qyliss> I think it should be renamed to either services.getty or services.agetty
<qyliss> anybody have thoughts as to which of those options we should go with?
<cole-h> qyliss: I'd personally go with getty, since that's the target / service name/
<qyliss> good point
<cole-h> s|/$|.|
teto has joined #nixos-dev
jonringer has quit [Ping timeout: 264 seconds]
teto has quit [Quit: WeeChat 3.0]
teto has joined #nixos-dev
orivej has joined #nixos-dev
cole-h has quit [Ping timeout: 264 seconds]
greaka has quit [Ping timeout: 256 seconds]
kgz has quit [Ping timeout: 256 seconds]
__monty__ has joined #nixos-dev
greaka has joined #nixos-dev
kgz has joined #nixos-dev
<siraben> supersandro2000: requesting your review on a couple of PRs of mine that have gone unmerged
<siraben> 50K PR/issues in 2 years, my god
<supersandro2000> siraben: going to take a few days. I have many unreads and currently I am fixing all the google python modules
<siraben> supersandro2000: Ok, no problem.
AlwaysLivid has joined #nixos-dev
teto has quit [Ping timeout: 260 seconds]
AlwaysLivid has quit [Remote host closed the connection]
AlwaysLivid has joined #nixos-dev
greaka has quit [Ping timeout: 260 seconds]
greaka has joined #nixos-dev
WilliButz has quit [Ping timeout: 240 seconds]
WilliButz has joined #nixos-dev
andi- has quit [Ping timeout: 258 seconds]
andi- has joined #nixos-dev
bgamari has joined #nixos-dev
mkaito has joined #nixos-dev
mkaito has joined #nixos-dev
MichaelRaskin has quit [Ping timeout: 260 seconds]
jonringer has joined #nixos-dev
<{^_^}> firing: RootPartitionLowDiskSpace:
pmy_ has quit [Quit: WeeChat 3.0]
<worldofpeace> qyliss: merged your PR, hopefully without a hitch
<qyliss> worldofpeace: ty :)
<qyliss> weird that {^_^} didn't say anything about it
<worldofpeace> qyliss: trying to get back into nixy stuff while "shedding my skin"
<qyliss> worldofpeace: :)
<qyliss> ever figure out that info thing?
<worldofpeace> qyliss: this shit it actually crazy
<worldofpeace> It's not the only thing that suddenly happened. I have another issue with ninja that's reproducible back to 20.03, but I tried on a live usb's nix store and couldn't reproduce :(
<qyliss> yikes
<worldofpeace> I wonder if my store is fudged, but I think I tried to do the "repair" stuff and it didn't do anything. so weird
<qyliss> worldofpeace: did you try with --check-contents ?
<worldofpeace> qyliss: maybe. tbh I might just try it again
<worldofpeace> `sudo nix-store --verify --check-contents --repair`?
pmy has joined #nixos-dev
<worldofpeace> welp that did nothing 😒
<qyliss> :(
<worldofpeace> perhaps my motherboard is harboring some sort of spirit that wreaks havoc on all my debug attempts to evade discovery
<worldofpeace> might just go with that. issue #666666 nixos/nix "Nix gave my computer a nefarious spirit - what do I do now?"
<qyliss> I've had a problem like that before
<{^_^}> #84283 (by flokli, 39 weeks ago, open): firefox fails to build on latest master if compiled with a AMD EPYC 7401P CPU
xwvvvvwx has quit [Quit: ZNC 1.8.2 -]
xwvvvvwx has joined #nixos-dev
<arianvp> is there any way to speed up nixos eval speed? nixos-rebuild takes 10 seconds on my machine these days just to eval :/
<arianvp> is flakes fixing this to some extent?
<qyliss> flakes implements an evaluation cache
<qyliss> which it can do because flakes run in pure mode
<qyliss> I'm not sure if non-flake code can run in pure mode. I hope so.
zarel has quit [Ping timeout: 272 seconds]
zarel has joined #nixos-dev
teto has joined #nixos-dev
<niksnut> arianvp: maybe in the long term, since it allows modularizing nixpkgs/nixos
<niksnut> most nixos modules could be moved out of the nixpkgs repo
<arianvp> I'm not convinced that will improve speeds. From my experiments the long module list is only a small factor in perfomance issues
<qyliss> couldn't we just not import all modules by default?
<arianvp> It will improve it. just not as much as I want :)
<qyliss> no need to move them to another repo for that
<arianvp> anyhow I just moved to flakes and my dry-run speed is no sub-millisecond instead of 10s =)
<qyliss> (or flakes)
<arianvp> (So seems to help for my usecase somewhat)
<niksnut> otherwise we'll have to hope that nickel will save us
<arianvp> haven't looked at nickel yet. is there actual intent to turn it into a less adhoc replacement of the nix language?
<arianvp> that'd be cool.
obadz has joined #nixos-dev
<arianvp> is it this?
<niksnut> yes
<qyliss> that's exciting
<qyliss> niksnut: is that "yes" to "is there actual intent...?" or "is it this?"?
<niksnut> both
<gchristensen> spooky
<lukegb> 👻
<drakonis> there's an irc channel for nickel btw
<drakonis> its called #nickel
<drakonis> write that down
thibm_ has quit [Ping timeout: 246 seconds]
<worldofpeace> qyliss: actually figured something out
<worldofpeace> doesn't mean I'm closer, though.
<JJJollyjim> hmm
<JJJollyjim> how do you spell that?
Cale has quit [Read error: Connection reset by peer]
rajivr has quit [Quit: Connection closed for inactivity]
<qyliss> worldofpeace: maybe you can find a way to installinfo outside the sandbox? Then you could try gdb-ing it or something
<worldofpeace> qyliss: yep, sounds like an interesting enough idea. though, it would be funny if it worked outside the sandbox
<qyliss> if it does there's always breakpointHook
<qyliss> although I don't think I've ever got that to work myself
jonringer has quit [Quit: Leaving]
<worldofpeace> qyliss: if you could humor me for one second
* qyliss sits up
<worldofpeace> add breakpointHook to texinfo, and then edit it so it will fail pretty early. (maybe put buildPhase = ''false"?). when breakpoint hook gives you a command when it fails, copy that and open another terminal... `sudo -i` `nix-shell -p cntr` and then paste that command. lastly provide the output of `ls` in that `/var/lib/cntr` which you should have be cd`d too
<worldofpeace> (I hope this all makes sense...)
<qyliss> i forgot what cntr does
<worldofpeace> tldr mounts the sandbox filesystem etc.
<Mic92> Maybe I should improve the readme for this one
<qyliss> oh yeah cool
<Mic92> It has a video, but many people seem to ignore it.
<sphalerite> ain't nobody got time for videos :p
<qyliss> worldofpeace: so are you asking me if that seems like the right thing to do, or are you asking me to do that?
<worldofpeace> Mic92: oh yeah, right. I think the most wartorn nerds eyes subconsciously suppress anything userfriendly
<worldofpeace> qyliss: the latter, to do that and provide `ls`
<qyliss> alright
<srhb> Speaking of, I somehow broke it recently and now get "unable to move container mounts to new mountpoint: EOPNOTSUPP: Operation not supported on transport endpoint"
<qyliss> worldofpeace: want to give me a nixpkgs rev to make sure we're on the same page?
<worldofpeace> qyliss: I just tried off of ff97a25fc6da512edf8fb32cf27f001e0d7382c0
<qyliss> ack
<Mic92> srhb: mhm. I thought I fixed it, but is it possible that you are missing /var/lib/cntr?
<srhb> Mic92: I'm not.
AlwaysLivid has quit [Remote host closed the connection]
<srhb> But maybe I haven't updated recently.
<Mic92> srhb: oh hang on, Is anything mounted to that directory?
<srhb> Mic92: Nope :)
AlwaysLivid has joined #nixos-dev
<Mic92> srhb: mount | grep cntr
<srhb> Mic92: Yes, nothing.
<Mic92> or strace -f -e mount cntr
<Mic92> cntr attach ...
<qyliss> worldofpeace: bin build dev etc nix proc tmp var
<Mic92> qyliss: your stuff is in build
<worldofpeace> the fuckery. excuse me gchristensen
<worldofpeace> drumroll everyone
<worldofpeace> 🥁
<worldofpeace> on my cntr there is no /bin/sh
<qyliss> Mic92: worldofpeace asked for ls in /var/lib/cntr
<qyliss> aha!
<Mic92> worldofpeace: cntr exec sh
<qyliss> I have /bin/sh though
<worldofpeace> I mean... I was about to pull up my rainbow suspenders. but I'm just like... how?
<qyliss> puck: you might know the answer to this?
<Mic92> worldofpeace: usually nix would put a /bin/sh into the buildbox.
<Mic92> *build sandbox
<worldofpeace> mind you the build failed with me doing nothing "/nix/store/85qzvs8l6blvi4v85j2s3z0vga1d3wzf-bootstrap-stage3-stdenv-linux/setup: ./configure: /bin/sh: bad interpreter: No such file or directory
<worldofpeace> "
<worldofpeace> I just added breakpointHook
<qyliss> worldofpeace: Nix version?
<Mic92> worldofpeace: mhm. that should not depend on the derivation but on nix alone
<qyliss> puck: or at least be amused by it :P
<worldofpeace> nix (Nix) 2.4pre20201201_5a6ddb3
<worldofpeace> I'm using nix unstable. I think this explains a lot of things 😸
<qyliss> I'm on 2.3.6
<sphalerite> worldofpeace: what does `nix show-config | grep sandbox-paths` say?
<worldofpeace> sandbox-paths = . I've just looked at my branch and I think through a rebase it must've created some problem
<sphalerite> there's supposed to be a busybox bind-mounted to /bin/sh in the sandbox, but maybe that's missing due to different nix config?
<worldofpeace> yeah, I think if I drop that silly patch I added a while ago I should be good (think it was fixed on master?)
<sphalerite> lol It probably needs to be gated behind a version check since it's a backwards-incompatible change.
rmcgibbo has joined #nixos-dev
<worldofpeace> what's annoying is that it was this confusing to realize I didn't have /bin/sh
<qyliss> I'm sad the static busybox needs to be there at all
<sphalerite> ^
<qyliss> I don't really see why we'd need it
<rmcgibbo> Has anyone previously considered making (or already built) a github bot that runs nixpkgs-review every hour or something against open nixpkgs PRs using github actions?
<qyliss> except for useless claims about POSIX-compliance in the sandbox
<worldofpeace> qyliss: thanks ✨ I don't think I would've figured this out without u
<worldofpeace> it was seriously confusing
<qyliss> :)
<{^_^}> firing: RootPartitionLowDiskSpace:
<puck> <qyliss> I don't really see why we'd need it <- i believe some configure scripts assume it exists
<puck> i kinda want to see bind mount support for the derivations
<puck> so you can build slightly more arbitrary environments, and e.g. bind-mount /bin/sh in mkDerivation
<qyliss> couldn't we have some automatic patcher for those like patchShebangs?
<puck> i wonder if nixpkgs' GNU Make still depends on /bin/sh
<puck> haha, yep
<symphorien[m]> everything that uses system does depend on /bin/sh
<puck> right, that is a bit of a catch-22, because you need a libc to compile a shell, and a shell to compile a libc
evanjs has quit [Quit: ZNC 1.8.2 -]
evanjs has joined #nixos-dev
terrorjack has quit [Ping timeout: 268 seconds]
terrorjack has joined #nixos-dev
cole-h has joined #nixos-dev
thibm_ has joined #nixos-dev
mmlb22 is now known as mmlb
greaka has quit [Quit: ZNC 1.7.5 -]
AlwaysLivid has quit [Quit: We are a collection of 7 billion codependent atoms. Stop hating based on constructs and come along for the ride.]
greaka has joined #nixos-dev
obadz has quit [Ping timeout: 260 seconds]
sreybastien has joined #nixos-dev
<sreybastien> hi there, i search somme support to debug gpg behavior in initrd stage :
<{^_^}> #106563 (by reyman, 3 weeks ago, open): Failed to authenticate with gpg on ssh session with initrd gpgcard option activated
<sreybastien> Any help / highlight is welcome
<sreybastien> I found a weird way to resolve the problem
<sreybastien> It seems the gpg loaded try to run an inexistant gpg-agent and not the correct gpg-agent in \crypt-ramfs\.gnupg folder
sreybastien is now known as sreybastien_aw
__monty__ has quit [Quit: leaving]
justanotheruser has quit [Ping timeout: 272 seconds]
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos-dev
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos-dev
<{^_^}> resolved: RootPartitionLowDiskSpace:
<samueldr> with the NixOS modules system, following the doc about specialArgs, I thought it could be used to avoid infinite recs for paths in imports
<samueldr> though, at the call site of `eval-config.nix` it seems to break
* samueldr checks something
<ekleog> Ericson2314: Hey! Do you know how bad a change that makes stdenv add gcc and the like at the _end_ of the build input list instead of at the beginning would be? it'd allow obvious things like `nix-shell -p gcc_multi` to work, while currently they for totally counter-intuitive reasons don't
<ekleog> feels like it'd be a major backwards-compat change, but hopefully it wouldn't break too much?
<ekleog> (sure the fix is to use `gccMultiStdenv`, but if there was no fix it'd be much better)
<supersandro2000> I would appreciate some feedback on the python hook addition here
<{^_^}> #108529 (by SuperSandro2000, 2 minutes ago, open): pythonPackages.pytestCheckHook: Add disabledTestFiles option