<gchristensen>
sorry dmj`, but that attitude is't appreciated and won't get you very far.
greymalkin has joined #nixos
<dmj`>
gchristensen: well I got your attention so it partially worked I guess, here's the issue https://github.com/nix-community/yarn2nix/issues/143. Trying to package facebook's docusaurus and be careful, this derivation will freeze your computer. nix-build at your own risk.
<dmj`>
gchristensen: I've spent maybe 4 days on it
<dmj`>
black smoke, but no build
<dmj`>
lerna and even typescript have been known to descend into circular import hell, along with the package.json copying that enables this as well in yarn2nix itself
hlisp has quit [Remote host closed the connection]
<maralorn>
I must say that I also had bad luck with trying to package yarn stuff in the last days. (although I gave up much quicker.)
<gchristensen>
dmj`: no, I don't respond to help-vampire tactics.
<dmj`>
gchristensen: don't feel obligated to
<dmj`>
gchristensen: it's not a fixable issue afaict
<dmj`>
haskellPackages has spoiled everybody
<clever>
dmj`: haskell isnt as terrible as js :P
<dmj`>
At least go had static linking in mind, so we can fetch w/o building. They got one thing right.
<pistache>
pjt_014: so that "system" attribute needs to be passed to nixpkgs itself, not to the derivation itself (see https://termbin.com/xzmo)
<maralorn>
dmj: How do you mean that spoiling?
<dmj`>
maralorn: haskellPackages is probably the best language package ecosystem nixpkgs has
h0m1 has quit [Quit: WeeChat 2.8]
<dmj`>
gchristensen: I wouldn't want to crash your CPU anyways
h0m1 has joined #nixos
<dmj`>
I have tried, node2nix and npmfromnix too
<simpson>
dmj`: Do you have a reproducible test case for the bricking? What exactly gets bricked?
<danderson>
weirdly, nix and npm not playing well just endears nix to me more.
<maralorn>
dmj: And still a lot of people (including me a few weeks ago) are very unsatisfied with the haskell+nix story.
gustavderdrache has joined #nixos
<dmj`>
simpson: yep, if you nix-build the derivation in said comment https://github.com/nix-community/yarn2nix/issues/143 your machine will freeze. When you nix-build --keep-failed and cd into the failed build, yarn build will still hang. Note, building outside of nix-build with yarn solo works fine.
<{^_^}>
[nixpkgs] @matthewbauer closed pull request #86437 → nixos/geoclue2: add network-online.target as wanted by geoclue2 → https://git.io/JfOPD
<dmj`>
maralorn: I /love/ nixpkgs haskell and I think it saved Haskell from its own build system and its own community drama, and wouldn't mind helping you with anything related to it, I use a lot work and fun.
<dmj`>
maralorn: makes builds even more granular for Haskell, better cross-compilation, can dontCheck on an entire dependency chain
<dmj`>
maralorn: I use mostly nixpkgs haskell vanilla and its great, what part don't you like
<notgne2>
dmj`: have you tried pnpm2nix
<simpson>
dmj`: Oh, so not a bricking. Yeah, that sucks, but don't mischaracterize software. We can both be honest about software, *and* software can be bad.
<dmj`>
notgne2: sounds made up
<notgne2>
it's the only thing that has been able to build all of the projects I've tried
<dmj`>
simpson: I'm honest about the fact it's bad, or my understanding of it is at least bad.
<dmj`>
notgne2: for js?
<notgne2>
yep
<{^_^}>
[nixpkgs] @aszlig opened pull request #87708 → firefox: Add patch to fix AES GCM IV bit size → https://git.io/Jf8by
<pistache>
is it possible in Nix to create an attrset where fetching any of its subattributes return null ? I suppose not, but just to be sure
teto has quit [Quit: WeeChat 2.8]
<maralorn>
dmj: Well, I am right now very satisfied with haskellPackages.
maddo has quit [Quit: See ya]
<simpson>
dmj`: I'd suggest that using the word "bricked" and the phrase "black smoke" are explicit dishonesties. Anyway, do you need to develop with npm, or just install some Node packages? If it's the latter, then altering your local nixpkgs to have the desired packages is relatively reliable.
<dmj`>
maralorn: as am I. pythonPackages is /ok/, C and C++ projects are workable. rust seems to have a good story now, the rest is very rough though
<maralorn>
But I was very thrown back a few months ago by repeatedly finding broken haskellPackages.
<dmj`>
simpson: I'm being very honest when I say that calling nix-build on that derivation has frozen my computer and caused a kubernetes cluster to fail. I am in now way exaggerating. One machine had 32 cores
<dmj`>
no*
<dmj`>
simpson: I'm being completely truthful and honest
<maralorn>
And I was made to believe e.g. by the existence of haskell.nix, people who don‘t understand nix/nixpkgs very good and other factors, that getting something to build with hackagePackages is hard and unlikely.
<dmj`>
simpson: I need to call lerna and typescript to generate ./lib directories that otherwise yarn2nix does not generate because it itself never calls yarn build
<simpson>
dmj`: If it boots, it ain't bricked. If it computes, then the magic smoke's still inside. Please chill out a little. (Although it's awesome that a k8s cluster crashed. Sounds like an opportunity for a postmortem analysis!)
<pistache>
dmj`: that seems like bad resource limiting on that cluster, it's not supposed to happen
<dmj`>
simpson: wouldn't be the first time a kubernetes cluster crashed ;)
lord| has quit [Ping timeout: 256 seconds]
<danderson>
kubernetes clusters exist in non-crashed states?
<maralorn>
What doesn‘t help is that I am relying on reflex-platform a lot and that uses an old nixpkgs which leads to a lot of necessary overrides if you want newer deps, which was hard to do.
<danderson>
TIL.
<dmj`>
maralorn: that's just propaganda, haskell packages in nixpkgs is mature
<dmj`>
danderson: :P
<dmj`>
maralorn: you need to get off of reflex-platform, switch to miso and I can set you up with a /great/ understandable nix haskell infrastructure for ghcjs, with incremental builds, release builds and deploy via nixops
<dmj`>
maralorn: try now, or your money back
lord| has joined #nixos
<iqubic>
dmj`: Why do you think miso is better than reflex?
<maralorn>
dmj: That would amount to completely rewriting a large project.
piezoid has quit [Remote host closed the connection]
<infinisil>
pistache: Nope
<infinisil>
(regarding your question about attrsets)
<dmj`>
iqubic: reflex itself is fine, but reflex-dom has no respect for the DOM
<iqubic>
Why do you say that?
<dmj`>
maralorn: that's ok, it will need a rewrite eventually
<infinisil>
pistache: Though I wish this would be supported. The main problem however is that `builtins.attrNames attrs` would be hard to define if that was supported
<infinisil>
where hard = impossible
<dmj`>
iqubic: it comes down to speed, the FFI, WASM not being a replacement for the DOM itself (yet).
<iqubic>
Ah. I see.
<iqubic>
Isn't Mozilla working on that last point?
<iqubic>
Also, this is getting a bit off topic.
<dmj`>
iqubic: if your framework isn't doing /efficient/ dom manipulation for you, you should find a new framework
<maralorn>
dmj: Does miso have a story for desktop and mobile-apps?
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/Jf8Nm
<dmj`>
maralorn: you can cross-compile miso to iOS yes. and I have, and also MacOS desktop (someone else is doing that)
<maralorn>
I really like the jsaddle+webkit-gtk-target of reflex.
<dmj`>
maralorn: the point is, nobody wants to use a website as a mobile app, they actually want to use native components from iOS itself, which nobody has done yet
<maralorn>
dmj: Oh, I don‘t care about apple products.^^
<dmj`>
maralorn: well a billion people have iPhones so, if you start a business you will
<maralorn>
I care about an Android app and I don‘t need it to look native.
<dmj`>
android is very insecure and the GC means worse battery life
<maralorn>
dmj: Not starting a business.^^
<dmj`>
maralorn: you'll be working for one that probably will care then
cr4y1 has quit [Ping timeout: 258 seconds]
<maralorn>
Well, this is getting very fundamental know, but I will not switch to a closed source mobile OS …
<maralorn>
Back to topic.
<dmj`>
maralorn: M$ acquired Github, we are controlled opposition
paraseba has quit [Remote host closed the connection]
<dmj`>
the good news is that Apple is stuck with UIKit for a long time
<maralorn>
dmj: And I really don‘t like the dependence of nixos on github.
<dmj`>
SwiftUI is closed source, and so is Xamarin
<maralorn>
But what can I do …
<dmj`>
there's always gitlab
<qyliss>
a billion people have iPhones?
<maralorn>
I can find out what the desktop app story for miso is.
<dmj`>
maralorn: just join the slack
<dmj`>
qyliss: no, but over 100M probably
<dmj`>
qyliss: rich people who fund your company have an iPhone, lets put it that way
m0rphism has quit [Ping timeout: 264 seconds]
thc202 has quit [Ping timeout: 260 seconds]
<maralorn>
dmj: I have never joined a slack.^^
<dmj`>
maralorn: slack is just IRC with a GUI
sigmundv has quit [Ping timeout: 256 seconds]
proofofkeags has quit [Remote host closed the connection]
proofofkeags has joined #nixos
<maralorn>
dmj: Is there a usecase for compiling a miso app without ghcjs?
<arianvp[m]>
And slack is down :)
<dmj`>
maralorn: there's an IRC #haskell-miso
<dmj`>
arianvp[m]: o/
<dmj`>
maralorn: sure, miso architecture is portable to /any/ language. It can be used standalone from JS, or easily ported to Rust / Reason
<dmj`>
maralorn: and I mean they would FFI directly into it w/o much abstraction in their host language required
proofofkeags has quit [Ping timeout: 256 seconds]
<dmj`>
rust and reason seem to work well with nix too
Darkmatter66 has joined #nixos
Darkmatter66_ has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @numkem opened pull request #87710 → the-way: init at 0.2.3 → https://git.io/Jf8AZ
<mitchellh>
simple question here, using `nix-shell` I accidentally created a derivation that `mv` from an input instead of `cp`. This broke it on subsequent runs (obviously). But I couldn't figure out how to get `nix-shell` to rerun/rebuild that dependent derivation. What am I missing?
<clever>
mitchellh: run `nix-store --verify --check-contents`
jgeerds has quit [Ping timeout: 256 seconds]
<gchristensen>
I won't dare suggest it seriously since I've never tried it on Catalina, but the multi-user nix-daemon installation on macOS makes that impossible
codygman has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @veprbl pushed commit from @rtldg to master « vscode: remove unused option (#87703) »: https://git.io/Jf8A5
hlisp has quit [Ping timeout: 246 seconds]
<hpfr[m]>
I’m about to start using NixOps to manage a few VM’s on my home server. I’m confused about handling the state. Should I deploy from my desktop or from my server? Does it really matter if I have convenient access to both? Any suggestions
<hpfr[m]>
I’m thinking server so I can manage it identically from my laptop or desktop over ssh, I guess?
<dmj`>
mitchellh: nix-shell won't have the same purity guarantees as a nix-build, in general its more useful for incremental development than building packages IME. nix-build will mv the source to /tmp and build in an isolated environment (like a docker container).
morgrimm has joined #nixos
<clever>
dmj`: but if your using a single-user nix, poorly written derivations can still break /nix/store/
<dmj`>
clever: shouldn't the default nix installation on darwin be multi-user these days?
<clever>
dmj`: Catalina says otherwise
<dmj`>
clever: that's why I stay on High Sierra :)
leungbk has joined #nixos
<dmj`>
you just need a retina screen and SSH into some NixOS box (bare metal)
<{^_^}>
[nixpkgs] @aanderse opened pull request #87712 → zabbix: init at 5.0 → https://git.io/Jf8xv
<epl692>
I am trying to setup a reverse proxy for a https service not on port 443, but I'm confused about how I set that up on the configuration file. https://nixos.org/nixos/options.html#services.nginx.virtualhosts.%3Cname%3E.listen is where I'm at but I don't seem to grasp the syntax
<gchristensen>
I'm starting sway like this at the console: dbus-run-session sway -c /etc/sway/config -d and sometimes when I run systemd-run commands I get Failed to start transient service unit: The name org.freedesktop.systemd1 was not provided by any .service files which seems to indicate some problem with how I start my session. any ideas?
<mitchellh>
dmj`: yeah this shell.nix is much more complicated than it has to be currently just because I was experimenting with something.
<clever>
mitchellh: run `nix-store --verify --check-contents`
<mitchellh>
clever: I saw that, let me give it a go
<mitchellh>
That did it. Thank you. Time to read some man pages.
<dmj`>
mitchellh: feel free to paste code, clever is an expert in all things NixOS
steshaw has joined #nixos
<clever>
mitchellh: if your using single-user nix, then anything nix builds can freely delete things in /nix/store and corrrupt paths
<freeman42x[m]1>
is there anything Nix related (or maybe just a Nix package) which could be used to restore the operating system at a previous time?
<simpson>
On non-NixOS?
<emily>
freeman42x: sure, nixos lets you boot arbitrary past generations, or what do you mean?
hlisp has joined #nixos
<freeman42x[m]1>
on NixOS since everything is done using Nix including the OS it is practically very hard to break the OS and not be able to roll-back
<freeman42x[m]1>
emily: yes, I am asking on non-NixOS. For any non-NixOS Linux distribution to be more precise
<emily>
I think there's a limited appetite for that because it'd end up just reimplementing NixOS, but maybe home-manager would be of interest to you?
<emily>
ZFS snapshots / boot environments / ... are designed for this kind of thing
<emily>
some distros have integration with them
<freeman42x[m]1>
cole-h: I want something like: D-VCS but for the entire operating system. So if that it breaks I can just reasily roll back
felixfoertsch23 has joined #nixos
<emily>
are you sure you don't just want NixOS? :P
<freeman42x[m]1>
my use case is the following: I start in some OS state and then do some changes to install the Haskell ecosystem. I want then to be able to easily revert those changes no matter what they were
felixfoertsch has quit [Ping timeout: 260 seconds]
<freeman42x[m]1>
while I am developing this project I would like an easy way to roll-back after I do some code changes, no matter how dangerous they are
hlisp has quit [Ping timeout: 260 seconds]
<dmj`>
mitchellh can I ask what your shell.nix is trying to do. Would love to help. I use nixpkgs w/ Go, Haskell, Python, and NixOS testing framework w/ Qemu and NixOps at work.
<dmj`>
mitchellh: * ^
<freeman42x[m]1>
<emily "are you sure you don't just want"> no, I just explained my use-case, so, no, I do not want NixOS, that is my OS of choice but I am testing stuff for other users which can choose to use other OSs
hmpffff has joined #nixos
<emily>
some things that may be solutions: zfs/btrfs/... snapshots and rollbacks; Docker containers; ...
<freeman42x[m]1>
emily: changing the file system is not a solution. The user choses his own FS.
<mitchellh>
But in that, I accidentally did a `mv` instead of a cp` on line uhhh 61.
<mitchellh>
And like I said, its a hacky thing, in reality I'll probably build that from source. I was just doing the fastest possible thing to mess around
leungbk has quit [Read error: Connection reset by peer]
lord| has joined #nixos
<gchristensen>
mitchellh: it is cool to have you here, welcome :)
<dmj`>
mitchellh: yes welcome :) A hashicorp section of nixpkgs would be nice. I've been packaging up a lot of stripe-mock and stripe-cli packages recently, it's almost trivial since go statically links everything, nix can just fetch them as source.
<mitchellh>
gchristensen: dmj` hello hello. :D we have teams working on system packages now (targeting deb/rpm first along with Docker, brew, etc.) but maybe we can do Nix stuff too.
shafox has joined #nixos
gustavderdrache has quit [Quit: Leaving.]
<gchristensen>
that might be interesting. iirc we have a good bit of the tools packaged already, and fleshing that out further would be good. also, having upstream involvement is always useful :)
<pjt_014>
pistache: thanks for the example you provided eariler, I'm trying to build it rn
<pjt_014>
oh hey it just finished too
<pjt_014>
only took 90 minutes of building gcc and crap >_<
hlisp has joined #nixos
turona has quit [Ping timeout: 265 seconds]
<taktoa[c]>
dmj`, mitchellh: I believe there are derivations in nixpkgs for nomad, terraform, consul, and vault
<mitchellh>
gchristensen: yep for sure. I meant it more as in helping with maintenance, the nixpkgs for our tools was one of the first thigns I looked at :D
turona has joined #nixos
<taktoa[c]>
oh woops graham said the same thing
<gchristensen>
cool :)
<clever>
> vmTools.buildRPM
<{^_^}>
<LAMBDA>
<dmj`>
mitchellh: sounds great, have used patchELF before to build rpm/debs, clever tells me you can use an FHS environment in a VM to accomplish similar. Here's the beginnings stripe's go pkgs I plan on using in a NixOS testing enviroment https://gist.github.com/dmjio/93f0a6f78079486f16c699d0feb51c59
hlisp has quit [Ping timeout: 258 seconds]
<mitchellh>
Nice
<unclechu>
hey, i want to build very specific executable and change permissions for it but `chmod` fails with "operation is not permitted" but i run `nixos-rebuild` with `sudo`.
<unclechu>
how should i do it?
<unclechu>
how do i `chmod` inside `nixpkgs.runCommand`?
<clever>
unclechu: which permissions are you trying to add?
<unclechu>
clever: 4550 and also i'm trying to set some ACL flag
<clever>
unclechu: you cant put setuid or acl stuff in /nix/store/
<clever>
unclechu: the only permission bit you can set is +x/-x, you have no choice over any other bit
<unclechu>
i wonder how `chromium-suid-sandbox` works
<keithy[m]>
Hi Guys I am looking for a project collaborator/coach to work on a (relatively simple) distributed backend filesystem (and binary cache) project. One of the features will be api compatibility with a well known and unbelievably (for what it provides) expensive commercial product. I have a nickname for this project, I call it "fartifactory". If you would be interested please get in touch keithy@consultant.com
aleph- has quit [Read error: Connection reset by peer]
<__red__>
Greetings - I'm trying to use elm under nixos and when I start my session with: "nix-shell -p elmPackages.elm elmPackages.elm-format" and attempt to use the REPL - I get CA issues
<__red__>
anyone seen that before?
<__red__>
(and by CA issues, I mean: "InternalException (HandshakeFailed (Error_Protocol ("certificate has unknown CA",True,UnknownCa)))" as it tries to download some metadata)
<unclechu>
clever: but it seems there's nothing to handle ACL records in that wrappers api, right?
<clever>
unclechu: i dont think there is anything for that, why do you need ACL's?
hlisp has quit [Ping timeout: 256 seconds]
<unclechu>
clever: my initial intention was to forbid to do anything with that binary by anyone except root and particular users (via ACL), i think a custom group might work too in this case
<clever>
unclechu: nix doesnt allow uid or gid in /nix/store/, everything must be root:root
<clever>
unclechu: nothing in /nix/store can be a secret from other users
<__red__>
I guess I need to dig in as to why the CA certificates aren't being recognized by this application
hlisp has joined #nixos
Supersonic112 has joined #nixos
Supersonic has quit [Disconnected by services]
Supersonic112 is now known as Supersonic
<keithy[m]>
Re: Fartifactory - We have a discord room open , or can use telegram if preferred. https://discord.gg/eXGHwB
rogue_koder_ has quit [Ping timeout: 256 seconds]
rogue_koder_ has joined #nixos
pca has joined #nixos
Darkmatter66_ has joined #nixos
hlisp has quit [Ping timeout: 256 seconds]
<pca>
Hi! I am new to NixOS, I am currently experiencing problems with nvidia optimus driver with GDM. When nvidia optimus is enabled,
<pca>
the login screen is replaced with an error screen saying something has gone wrong
<keithy[m]>
Hi Guys I am looking for a project collaborator/coach to work on a (relatively simple) distributed backend filesystem (and binary cache) project. One of the features will be api compatibility with a well known and unbelievably (for what it provides) expensive commercial product. I have a nickname for this project, I call it "fartifactory". If you would be interested please get in touch keithy@consultant.com
<simpson>
Mature and on-topic~
<clever>
keithy[m]: have you seen nix-serve ?
<keithy[m]>
One of you guys could get this up and running an an hour or two, so I am looking for a little help. Its all nixos.
rogue_koder_ has quit [Ping timeout: 264 seconds]
<emily>
I think you'll find that "relatively simple" and "distributed backend filesystem" are mutually exclusive
pca has quit [Ping timeout: 245 seconds]
pca60 has quit [Remote host closed the connection]
<abathur>
emily I guess it depends what it's relative to? :D
<keithy[m]>
Afaik NixServe is a standalone, this project aims to support distributed app development for applications using the hedera consensus service, so we need a set of mirrors and need to think in terms of hundreds of clients.
<emily>
this is really off-topic, I'd say to take it to #nixos-chat but tbh I don't think random job postings are appropriate there either
<abathur>
is it off-topic on the discourse?
<simpson>
I'm not sure if it's even a job posting so much as a plea for some sort of software to pop out of the void and manifest on-disk.
<abathur>
(might be a better venue regardless, if not)
hlisp has joined #nixos
<simpson>
keithy[m]: Let's start with what's on-topic. What pain points are you currently suffering from with your existing binary cache? Are you using nix-serve already?
<emily>
"get this running in an hour or two" × "using [some blockchain thing], so we need a set of mirrors and need to think in terms of hundreds of clients" is a pretty optimistic combination too
<keithy[m]>
watch this space!
ntropix has quit [Remote host closed the connection]
<keithy[m]>
Not using an existing binary cache
<keithy[m]>
the problem is not really about binary caches per se.
hlisp has quit [Ping timeout: 256 seconds]
rogue_koder_ has joined #nixos
<keithy[m]>
actually nix-serve might be a part of this!
<keithy[m]>
nix-serve looks a little nix focussed.
<keithy[m]>
my next problem will be compiling openresty without sse4.2
<colemickens>
what even is the binary cache protocol exactly? I swear I had it running out of a plain old Azure storage bucket thing a while back.
<simpson>
keithy[m]: Okay, good luck.
drakonis has quit [Quit: WeeChat 2.8]
morgrimm has quit [Ping timeout: 246 seconds]
<colemickens>
oh, I get it, nix-serve just serves requested paths out of hte store as nar/narinfos, that's kind of neat actually. not what I'd do for an actual cache though
<clever>
colemickens: you can use `nix copy --to file:///path` to make a dir of .narinfo and .nar.xz files, then just upload it to any static http server
hlisp has joined #nixos
<colemickens>
yeah, I remembered, thats exactly what I was doing
<colemickens>
complete with a script to diff the listings because the 'az' client would blanket reupload everything :S
<keithy[m]>
simpson: thanks!
<clever>
colemickens: there is also `nix copy --to s3://bucket` for backends nix itself supports
<__red__>
Okay - I'm at a complete loss. elm tries to download some metadata but complains about an unknown CA... yet I can use wget to pull the same file.
<__red__>
doing a systemcall trace on elm shows that it's reading /etc/ssl/*
<__red__>
what can I be missing?
<mitchellh>
if I install the Go pkgs (nixpkgs.go), it doesn't work out of the box with a lot of things cause it needs special CFLAGS env vars. It works if I do a `nix-shell -p go` it works fine. See error: https://gist.github.com/mitchellh/593f9cc012fac345ae2edefae45268fc
<mitchellh>
Question is: any guidance on getting whatever it requires in my normal shell? I'm not super sure where its coming from.
<clever>
mitchellh: most compilers wont work if installed with nix-env, only nix-shell and nix-build can make a compiler work
<mitchellh>
what's the standard expected practice there? Right now I've just been using direnv to hop into a shell with Go.
<mitchellh>
but its a bit clunky and I don't want to force a .envrc on every project :P
<clever>
mitchellh: i only ever compile things in nix-shell, but some people use direnv to automate entering the shell
drakonis_ has quit [Quit: WeeChat 2.7]
drakonis has joined #nixos
Darkmatter66 has joined #nixos
rickhull has joined #nixos
Darkmatter66_ has quit [Ping timeout: 256 seconds]
marcusr has quit [Remote host closed the connection]
proofofkeags has quit [Remote host closed the connection]
proofofkeags has joined #nixos
marcusr has joined #nixos
<cole-h>
mitchellh: You might also be interested in lorri -- it's like `nix-shell` without having to run `nix-shell` (when paired with direnv): https://github.com/target/lorri
proofofkeags has quit [Ping timeout: 256 seconds]
<clever>
cole-h: lorri is mostly just a cache for direnv, and deals with async building
vika_nezrimaya has joined #nixos
<vika_nezrimaya>
How to make Exim able to read /etc/shadow (PAM auth) with minimal security compromises?
<energizer>
you sometimes need to run `lorri shell`
<vika_nezrimaya>
I assumed exim listens as root so it'll be able to authenticate, but, alas, it doesn't
<cole-h>
Maybe I'm just lucky, but I've never needed to run `lorri shell`
hlisp has quit [Remote host closed the connection]
hlisp has joined #nixos
<{^_^}>
[nixpkgs] @gnidorah opened pull request #87715 → maxx: drop → https://git.io/Jf4fw
<clever>
vika_nezrimaya: i think the pam library can run a setuid helper, to handle pw validation
<clever>
vika_nezrimaya: and that setuid helper will use /etc/pam.d/ to enforce what its parent proc can actually ask of it
<mitchellh>
cole-h: thanks yeah I have my eye on it, just slowly adding pieces as I learn more
<vika_nezrimaya>
mmhm, except this setuid helper, unix_chkpwd, only seems to allow to login as current user if you aren't root (which exim is not)
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « skrooge: 2.21.1 -> 2.22.1 »: https://git.io/Jf4JS
<bqv>
I don't think it has a nix package
<bqv>
But cool!
<vika_nezrimaya>
bqv: sad
<vika_nezrimaya>
that would make for a neat recursion
<bqv>
Lol
<pxc>
Guix has a lot of strengths imo. The documentation is very nice, and it got a really lovely command-line interface early on. Grafts are pretty cool, too
<bqv>
Hey maybe I could make the guix package and module into a standalone flake
<pxc>
that would be a great use of flakes
<vika_nezrimaya>
Grafts?
<bqv>
Totally
<abathur>
just waiting for gix
<pxc>
vika_nezrimaya: grafts are a Guix feature for making post-hoc security hotfixes to existing packages without rebuilding
<CMCDragonkai>
What's the best way to use `confluent-hub install`
<pxc>
I think it's pretty neat that Guix uses the same language for package recipes as it does for its builders (Guile for both), too
<colemickens>
is it primarily about security fixes? I thought it was more of a temporary development backdoor sort of thing
<colemickens>
like, does Guix actually push security updates via grafts?
<pxc>
I think the motivation is security updates
<colemickens>
interesting
<pxc>
Guix has also centralized their equivalents of the various xyz2nix type tools and presents them to users with a uniform command-line interface. That's pretty nice
<pxc>
since they both build packages that are stored in the mutually desired way and behave in the right way, why not though? Nix packages keep the promises Guix packages expect of their dependencies and vice-versa!
morgrimm has joined #nixos
<pxc>
A S S I M I L A T E
<puffnfresh>
Guix used to use the nix-daemon and create normal derivations
<puffnfresh>
doesn't any longer, right?
<drakonis>
yes
<drakonis>
ish
<drakonis>
it has a stripped down daemon and is currently replacing the remaining parts with guile
<pxc>
are they switching to a different derivation format, or just the same derivation format but with Guile builders instead of bash builders
<drakonis>
they still use current format
<pxc>
neat
<drakonis>
the
alp has joined #nixos
<drakonis>
the daemon is incompatible
user_0x58 has quit [Quit: Leaving]
palo has quit [Ping timeout: 272 seconds]
palo1 is now known as palo
<drakonis>
drvs are currently still the same but i think it may no longer be compatible in the future
morgrimm has quit [Ping timeout: 246 seconds]
<drakonis>
since they want to go guile all the way
<pxc>
makes sense, I think that's a cool goal for the project
<pxc>
I noticed that your guix module that you linked to from the original PR was under profiles/
<pxc>
which is specifically what profiles/ is not forrrrrrr lmao
<bqv>
Ah, balls, yeah
<ashkitten>
drakonis: that seems like it'd be better with syntax highlighting
<bqv>
I mean it was a WIP
<drakonis>
knowing how to read scheme helps
<drakonis>
quite
<drakonis>
they all use emacs
<pxc>
bqv: I'm only teasing, it's great
<bqv>
:p
<ar>
/34
<pjt_014>
what's up with nix store paths that have url parameters?
<pjt_014>
like /nix/store/6afjbhgxgl3jrwiwackzslkjcn0bhd1z-?id=14adc898a36948267bfe5c63b399996879e94c98
hlisp has joined #nixos
<pxc>
bqv: I actually still haven't switched over to a nixflk-like setup on this system yet. I spent some time today reading through nixrdp's flake.nix and yours trying to decide what I wanted to reproduce and what I didn't
<kalbasit>
pxc: we’re planning a rewrite with lessons learned if you’re interested
<ninjin>
I still run monolithic `configuration.nix` files to make it easy to shift them between hosts. How do people manage them when you break them down into smaller shareable pieces? Do you really shift the whole tree of files to each host?
<pxc>
ninjin: I'm not sure what you mean by 'shift'. But I use the same repo for multiple hosts. As I've done it before, I make /etc/nixos/configuration.nix a symlink to a hosts/some-hostname/default.nix file
<pistache>
that works too
<pxc>
my old nix config was crufty and dumb, though. I included entire checkouts of nixpkgs as submodules. Decided it's time to redo it since there are some nice tools and conventions around pinning and otherwise referring to external repos that have emerged in the community
rauno has joined #nixos
<kalbasit>
pxc: and now it’s flakes time. hopefully it’ll help us with some of the problems with externals
<pxc>
I think my goals are similar to shabka's (I want Nix config for multiple platforms in one repo) but I like that nixflk is playing with that new machinery
<kalbasit>
well then come join us, we’re trying to achieve the same thing
<ninjin>
Thanks pistache and pxc, I guess I meant “transfer” by “shift”. I guess my dream would be to collapse the tree into a single host-specific configuration on my local box and then transfer it, but I may very well be missing something elegant as I am still a bit of a beginner.
<ninjin>
Equally plausible: I am overthinking it. ;P
<srhb>
ninjin: I mean, that's what git repos are for :)
<srhb>
ninjin: I want the version history anyway.
<pxc>
ninjin: having a single configuration.nix file is nice for readability for sure. Is that why you like the singular configuration.nix file?
<srhb>
ninjin: For large scale things (10s to 100s of machines) I prefer the local checkout and I push the finished system instead of config files.
<srhb>
ninjin: Still in a git repo.
<ninjin>
srhb: Sure, but having a bleedin’ git repo shifted to a server and then set to be owned by root feels so odd. '^^
<srhb>
ninjin: I don't have mine owned by root.
<pxc>
I give myself ownership of it even though it lives in /etc, like a total heathen
<srhb>
Likewise. Well, mostly, /etc/nixos is a symlink into my repo.
<pxc>
on some systems it doesn't live under /etc, though. I just clone the same repo somewhere into ~
<ninjin>
pxc: I guess the way I currently see it is that the modularity is nice for sharing, but that the final configuration itself is nice as a single file as it becomes the “ground truth” rather than the repo itself.
<pxc>
or I used to; I don't use any non-NixOS machines with my personal config atm, but I'm planning to again soon
rickhull has quit [Ping timeout: 245 seconds]
<ninjin>
srhb: By “finished system” you mean an image as in cloud-style?
<pxc>
ninjin: Hm. I feel like I'll kind of always want a whole hierarchy of files because I want somewhere nice to stage packages and modules alongside my configuration.nix
drakonis_ has joined #nixos
<srhb>
ninjin: The /run/current-system "system" derivation in its entirety
<srhb>
ninjin: (as in nix copy somesystemderivation/nix-copy-closure somesystemderivation)
<ninjin>
pxc: Yeah, you can kind of do that inside a single file but it turns really ugly, real fast.
<srhb>
ninjin: You can transfer entire system trees like that, and switch them remotely. The target never has to se the config.
drakonis has quit [Read error: Connection reset by peer]
drakonis1 has joined #nixos
<srhb>
ninjin: That's how tools like NixOps work
<ninjin>
srhb: Now that sounds awesome, I think I have to dig into the manual again.
<pxc>
ninjin: if NixOS had a neat little command for pretty-printing the configuration that you're actually using (as Nix, not as a derivation), would that satisfy the same desire you're describing?
<srhb>
ninjin: I very much like it for large scale deployments because I can build the system locally and inspect, say, its /etc/ structure and all systemd units before deploying it.
<bqv>
only testing so far is that it passes "flake check" and starts to build, but looks ok otherwise
<bqv>
i'll switch my config to it too
<pxc>
awesome
chloekek has quit [Ping timeout: 256 seconds]
<pxc>
it's almost midnight here so I'd better get to bed soon, but I'll do some real playing with it tomorrow. I didn't actually set up the daemon tonight lol
<bqv>
heh
<ninjin>
pxc: Yes, I think it would. Although I have to admit that srhb’s suggestion of building locally and then pushing is very tempting.
drakonis_ has quit [Ping timeout: 265 seconds]
<srhb>
It's nice when you have one "dev environment" that you like to stay in, at least. Which for me is my work laptop.
<ninjin>
pxc: Sleep is important, with a baby in the house you learn to appreciate it thoroughly. ;)
maxter has quit [Ping timeout: 240 seconds]
<pxc>
ninjin: the only creature in my care is a 17-year-old chihuahua, and happily she's a champion sleeper
<pxc>
I'm not sure she's capable of human baby levels of noise, lol
<pxc>
g'night all!
<ninjin>
Sweet dreams! Thanks a ton for the feedback.
<fps>
hm, is this a stupid way to get a shell for my local package development: nix-shell -E 'with import <nixpkgs> {}; callPackage ./default.nix {}' --pure
<{^_^}>
[nixpkgs] @peti pushed 614 commits to haskell-updates: https://git.io/Jf4Tb
<fps>
hmm, weird. i removed the src = fetchurl and replaced it with fetchFromGitHub and it still downloads the gnu hello source ;)
<fps>
ok, nix-collect-garbage showed me the way :)
asheshambasta has joined #nixos
<srhb>
fps: Presumably you didn't change the sha?
opthomasprime has joined #nixos
opthomasprime has left #nixos [#nixos]
oborot has quit [Ping timeout: 264 seconds]
orivej_ has joined #nixos
orivej has quit [Ping timeout: 272 seconds]
Acou_Bass has quit [Ping timeout: 258 seconds]
<fps>
srhb: yeah, that was it :)
<fps>
i stumble over that again and again ;)
lemsip has joined #nixos
<srhb>
fps: Yeah, it can be confusing. It helps to remember that Nix would have to download every source of every thing every time you change a version of anything if it had to verify the sha _after_ the fact. :)
<fps>
i would like to iterate over some things in the src without committing and redownloading again and again. i tried this: nix-build -E 'with import <nixpkgs> {}; (callPackage ./default.nix {}).override { src = ./. ; }'
<fps>
hmmm
zupo has joined #nixos
<fps>
oh override is just for arguments, not for attributes i guess
<fps>
i have a simple makefile that expects the env var PREFIX to be set. the standard builder passes $out to --prefix in the configure call. that doesn't help me here though..
<DamienCassou>
thank you etu. In the context of NixOS, does it mean that this function creates some kind of virtual filesystem that looks like a more standard OS?
cr4y1 has joined #nixos
<etu>
DamienCassou: A chroot I think yes
<makefu>
DamienCassou: the function populates a folder with all the "default" paths (/usr/lib etc) and then chroots into said system
<makefu>
it is /very/ inefficient, almost as bad as docker containers
<makefu>
in terms of disk usage
zeenk has joined #nixos
<DamienCassou>
thank you both
<DamienCassou>
can anyone tell me why nixpkgs-review considers lutris as being broken? Where is this package marked as broken?
cinimod has joined #nixos
<sphalerite>
DamienCassou: probably because it depends on p7zip which is marked as insecure
<cinimod>
builtins.trace "${reverseDepsList}" reverseDepsList; gives error: cannot coerce a list to a string - what should I do to see what the value of reverseDepsList is?
<yorick>
`phases
<{^_^}>
[nixpkgs] @Mic92 merged pull request #87723 → IPFS,Fix: Require the ipfs-migrator package for handling upgrades → https://git.io/Jf4Iy
<pistache>
hmm.. after overriding a NixOS module from my configuration, nixos-rebuild restarted all the services of the machine
<pistache>
this gave me an error about D-Bus, "Failed to create bus connection: Connection refused", which cleared itself when running nixos-rebuild again
maxter has joined #nixos
<pistache>
however, now I have something very weird: two systemd units are failing because they can not find "/sbin/getty"
anderslundstedt has joined #nixos
<pistache>
the generated service files (for getty@.service and serial-getty@.service) are indeed pointing to /sbin/getty
iqubic has quit [Ping timeout: 272 seconds]
<pistache>
the path for getty@.service is /nix/store/...-systemd-243.7/example/systemd/system/serial-getty@.service
zupo has joined #nixos
wnklmnn has joined #nixos
<{^_^}>
[nixpkgs] @mnacamura opened pull request #87727 → rounded-mgenplus: replace p7zip with libarchive → https://git.io/Jf4Yi
pjt_014 has quit [Ping timeout: 240 seconds]
vidbina has joined #nixos
<pistache>
oh right... doing mkForce on systemd.services is not a good idea.. I'm dumb.
rauno has quit [Remote host closed the connection]
<sshow>
has anyone had problems starting deluge in deamon mode with remote access? Packets are arriving, but no connection with deamon, and no logging whatsoever.
<fps>
in my naive mind i would just copy all the python parts over to the store in the makefile as well, but then there's the questions 1] how to make the python program find the executables in the store? and 2] how to make the ogfx-ui.py "executable"?
<fps>
i guess for 1] i could patch the paths to subprocess.popen etc in the code
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<sphalerite>
(the answer is yes, but you should really have a good reason)
<bkv>
because if i try what i want to try without that, i'll end up with a nix store where nothing ever gets GC'd, which is definitely something i'd like to avoid :)
<bkv>
so how could i achieve that?
<sphalerite>
o.O
<sphalerite>
well, unsafeDiscardStringContext, but caveat emptor
<bkv>
ah, yes, awesome
o1lo01ol1o has joined #nixos
leotaku_ has joined #nixos
leotaku has quit [Ping timeout: 264 seconds]
maddo has joined #nixos
atopuzov[m] has joined #nixos
agsdheidjd has joined #nixos
<{^_^}>
[hydra] @basvandijk merged pull request #756 → GitInput: only convert integer option values to int → https://git.io/Jf43W
<{^_^}>
[nixpkgs] @jtojnar opened pull request #87731 → maintainers/scripts/update.nix: various fixes and clean-ups → https://git.io/Jf4st
<{^_^}>
[nixpkgs] @markuskowa opened pull request #87732 → welle-io: fix, add qtgraphicaleffects to inputs → https://git.io/Jf4sm
amfl_ is now known as amfl
h0m1 has quit [Ping timeout: 244 seconds]
<wnklmnn>
Hi, to get started with nix derivation I tried to package the smallest java application I could find wich is not in nixpkgs. Can someone have a look at my derivation and tell me if I they would change something? https://gist.github.com/wnklmnn/8cb8e9bfba36ac103303c7b6aa6d2710
<{^_^}>
[nixops] @edolstra closed pull request #1342 → Use niv to pin nixpkgs channel for nix shell → https://git.io/Jf8R3
Peter_Storm has joined #nixos
wnklmnn has quit [Read error: Connection reset by peer]
<bkv>
well, i would have, but he's gone now
h0m1 has joined #nixos
cosimone has joined #nixos
<bkv>
damn, what i was planning wouldn't work anyway because of pure evaluation mode
wnklmnn has joined #nixos
opticnerve has quit [Quit: bye]
<bkv>
wnklmnn: your derivation is missing a 'version' (use the same one from src)
<bkv>
other than that and a few syntactical quirks, looks good
cfricke has joined #nixos
<wnklmnn>
bkv thanks :)
fendor_ has joined #nixos
fendor has quit [Quit: Leaving]
fendor_ is now known as fendor
noudle has joined #nixos
<betaboon>
is using `python3.pkgs.*` prefered over `python3Packages.*` ?
karantan has joined #nixos
karantan has quit [Client Quit]
iyzsong has joined #nixos
hlisp has quit [Remote host closed the connection]
<freeman42x[m]1>
fps: how would you use a docker though? is that not a bit meta? I need to install a .deb for controlling a remote Amazon Workspaces VM
<{^_^}>
[nixos-homepage] @garbas pushed 13 commits to landing-page: https://git.io/Jf4ZF
<freeman42x[m]1>
simpson: if it would be run inside a docker it would be an extra level of indirection which would make it slower and maybe cause issues
<dejanr>
anyone know how to find out who is using p7zip, this error is not very usefull "p7zip is abandoned and may not receive important security fixes"
<simpson>
freeman42x[m]1: I don't know how to respond to "should" with anything other than no, you shouldn't use AWS; no, you shouldn't use Docker; no, you shouldn't use third party debballs shipped by disreputable publishers; no, you shouldn't put up with any of this silliness.
<simpson>
But, if you want to run this package, then I'm not sure what's wrong with the FHS approach.
<freeman42x[m]1>
simpson: I am required by my work to use AWS VMs ...
<freeman42x[m]1>
what I should is irrelevant as long as it is not illegal, immoral or fattens me
<simpson>
dejanr: I can't find where that message comes from; it doesn't appear to be in nixpkgs. I also searched with a search engine for "p7zip is abandoned" and found nothing. Do you have more information?
fusion809 has quit [Remote host closed the connection]
<simpson>
freeman42x[m]1: I am not receiving a paycheck from your employer, and thus have the freedom to criticize their poor decisions. I think that AWS customers are immoral for putting cash into Amazon's hands, including your employer.
<niksnut>
simpson: this channel is not the right place for that kind of discussion
<simpson>
niksnut: Understood.
piezoid has joined #nixos
<infinisil>
dejanr: Does the error not explain how to use it anyways?
<infinisil>
It sure does for me
maddo has quit [Read error: Connection reset by peer]
<infinisil>
dejanr: Ahh I misunderstood your question. I thought you were asking how to use it
<infinisil>
But you're asking where it's used instead
<infinisil>
dejanr: Does this happen for `nixos-rebuild switch`?
dejanr has quit [Ping timeout: 264 seconds]
piezoid has quit [Quit: leaving...]
hmpffff has joined #nixos
dejanr has joined #nixos
<infinisil>
dejanr: (See the channel logs for what you missed if anything)
proofofkeags has joined #nixos
piezoid has joined #nixos
<tomturbo>
hello, I can't for my life get pre-commit to work with python hooks. whenever I try to run "pre-commit install-hooks" I get the error message ".../python3.7: No module named virtualenv". Furthermore, I can easily reproduce this behavior by creating an empty git repository and adding this file: https://pastebin.com/9r3hhxbu
piezoid has quit [Client Quit]
hlisp has joined #nixos
<tomturbo>
I have installed pre-commit via environment.systemPackages = [ ... gitAndTools.pre-commit ... ]
o1lo01ol1o has quit [Remote host closed the connection]
<tomturbo>
infinisil: thanks, that worked. does that mean that it will just take some time until it works in stable as well?
<kazzimazzi[m]>
Hi, how do I override alacritty derivation in my configuration.nix to use master branch?
o1lo01ol1o has joined #nixos
<infinisil>
tomturbo: Does it not work if you change it to `channel:nixos-20.03`?
<infinisil>
(I tried this myself, but I'm not sure how to reset the state for it to try again)
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
<infinisil>
timCF: (replied there)
ilikeheaps has quit [Ping timeout: 260 seconds]
<gchristensen>
I don't have a org.freedesktop.systemd1, and this is causing me trouble. I do have a session, and I start sway at a tty with dbus-run-session sway. I do run multiple user services, systemd-run --user true works fine. systemd-run --user --shell results in Failed to start transient service unit: The name org.freedesktop.systemd1 was not provided by any .service files. any ideas? here is some more
plutes has quit [Read error: Connection reset by peer]
leotaku has joined #nixos
<tomturbo>
inifinisil: hm, it actually did. I asssumed that the reason why it worked before was that the command explicitly uses the unstable channel as opposed to 20.03 (which is what my system is on). why do these commands work but the normal way doesn't?
<{^_^}>
#87502 (by DamienCassou, 2 days ago, open): Wrongly configured org.freedesktop.systemd1 service on user's dbus session with i3
<gchristensen>
atemu12[m]: interesting
cosimone has joined #nixos
fendor_ has joined #nixos
hlisp has joined #nixos
<infinisil>
tomturbo: Not sure, could be that your system uses an older version yet
<infinisil>
Oh, probably some python env shenanigans
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<{^_^}>
[nixpkgs] @orivej opened pull request #87737 → gdb: fix cross compiling for android → https://git.io/Jf4lm
fendor has quit [Ping timeout: 272 seconds]
<infinisil>
tomturbo: Probably good practice to install your project dependencies like pre-commit from the project's shell.nix
<infinisil>
Problem shouldn't happend with that
morgrimm has joined #nixos
wolke has joined #nixos
buggymcbugfix has quit [Excess Flood]
buggymcbugfix has joined #nixos
<tomturbo>
infinisil: the version returned by "pre-commit --version" is 1.21.0, the same as the one installed on my system. that's a good idea, I'll try to create one
curious92[m] has joined #nixos
curious92[m] is now known as rigola[m]
noudle has quit []
<{^_^}>
[nixpkgs] @cw789 opened pull request #87738 → erlangR23: init at 23.0 → https://git.io/Jf4lr
ilikeheaps has joined #nixos
<JJJollyjim>
hmm, can i reference turn a nix derivation expression into the path of a .drv file?
timCF has quit [Remote host closed the connection]
<pistache>
why does (lib.mergeAttrsConcatenateValues {} {x=1;}).x evaluate to [ 1 1 ] ?
<pistache>
that doesn't match the description of the function
<s1341>
hi all.
<pistache>
is it because it's a bug that was never fixed because the function is deprecated ?
<asheshambasta>
I seem to have stumbled upon another thing I'm having trouble understanding:https://gist.github.com/asheshambasta/82be17b90fa0ee17af80e725d644544f#file-release-nix
<s1341>
How can I compile something with an older glibc?
<s1341>
specifically, i need to target glibc 2.27
<s1341>
but on my nix box I have glibc 2.29
<asheshambasta>
tl;dr: I'm applying an overlay to a "pinned" nixpkgs fetch, vs applying it via import <nixpkgs> { inherit overlays; }, the former case leads to an inf. recursion. The latter case doesn't.
<evax>
s1341: you'll need to pin nixpkgs to an older version, containing the glibc version you're targetting
<s1341>
evax is there a way to do that in a nix-shell specification?
<evax>
s1341: if you click the hash link for glibc 2.27 you'll see a code snippet that should help you get started
<infinisil>
evax: Oh that's a neat site!
<s1341>
thanks evax. but I need aflplusplus too... and that's not in that revision of nixpkgs. Is there any way to specify _two_ nixpkgs on nix-shell?
gxt has quit [Remote host closed the connection]
gxt has joined #nixos
infandum has joined #nixos
<infandum>
With dockerTools, I want to essentially put my default.nix derivation as a docker image. Everything seems to build fine, with the resulting binary copied over to the image, but the runtime dependencies aren't there. When building the default.nix, stuff like R and graphviz binaries are generated and correctly linked with the binary, but with the docker they aren't copied into the image. Why is this?
<infandum>
My contents = the package from the default.nix, but even with [pkg pkgs.graphviz] the graphviz binaries still aren't copied.
<evax>
infinisil: yes, I just found it actually, that was a pleasant surprise, I was searching for a script from some time ago allowing to install older versions of packages from older releases
<betaboon>
anyone know why nixos+arch+debian are compiling CONFIG_BLK_DEV_NBD as a module ? oO
<lewo>
infandum: do you see graphviz and R when you do something such as `nix-build -A default.nix | xargs nix-store -qR`?
<evax>
s1341: you could have multiple versions assigned to different variables in your let block
dnlkrgr has quit [Ping timeout: 252 seconds]
<s1341>
ok. so i need to do a script for nix-shell.
dermetfan has joined #nixos
morr has quit [Quit: WeeChat 2.7.1]
morr has joined #nixos
dnlkrgr has joined #nixos
<lewo>
infandum: and if you build and image with `contents = [graphviz]`, graphviz binaries should be in the /bin of the image
<asheshambasta>
I think this could be a result of a regression/change in nixpkgs upstream.
<infandum>
lewo: I just realized that all of my dependencies for the default.nix are specified as libraries, not runtime binary dependencies. How do I do that?
<infandum>
Like, the program calls twopi which is a binary, not a library
<infandum>
so I need graphviz as a binary runtime dependency, not library
<infandum>
The way I was doing it was a hacky changing of paths in wrapProgram
<teto>
with the latest nixFlake, whatever I do (nix run XXX), I end up with error: value does not have type 'app', at /nix/store/azbg4gcdpix6gh9xn1pmm358gh2wzp05-nix-2.4pre20200501_941f952/share/nix/corepkgs/derivation.nix:20:9 ? is that normal ?
opticnerve has joined #nixos
<manveru>
teto: yes
<manveru>
use `nix run flakename#packagename` instead
LysergicDreams has joined #nixos
<bkv>
(if you're running latest nix, it's now `nix shell`
<bkv>
not `nix run`)
<bkv>
but i don't know how many people use the nix flake. i didn't until this week
<infandum>
lewo: In answer to your original question, I do see the stores for graphviz and R in the default.nix command you gave me. However, as I said after, I realized that the binaries are never put in a path for my original program to see it.
<teto>
nix shell nixpkgs#chromium works but breaks once I specify my nixos-usntable checkout. Can't unstable be used as a flake yet ?
<manveru>
channels != flakes
<lewo>
infandum: you then have to wrap your program and set the PATH variable
<manveru>
`nix flake list` shows you which ones you can use
hmpffff has quit [Quit: nchrrrr…]
hax404 has quit [Remote host closed the connection]
<lewo>
infandum: you should find a lot of examples in nixpkgs if you grep wrapProgram
hax404 has joined #nixos
<infandum>
lewo: Okay, that's what I've been doing. So the issue was I forgot one of the programs (graphviz) to wrap into the path as I always had it installed to the system so I never noticed it was missing.
<infandum>
lewo: That's good to know that wrapProgram is the only way to solve the binary path issue.
<{^_^}>
[nixpkgs] @offlinehacker pushed 3 commits to master: https://git.io/Jf4B3
knupfer1 is now known as knupfer
<adisbladis>
manveru: Haha :D
LysergicDreams has quit [Ping timeout: 256 seconds]
<{^_^}>
[hydra] @grahamc opened pull request #757 → schema/Builds: use jobset_id instead of jobset name matches → https://git.io/Jf4BK
<morgrimm>
2 wildly different questions for people - 1) anyone running nixos on laptops, what're you using to hibernate? and 2) anyone using direnv/lorri, is it possible to cover subdirectories with one .envrc (or have direnv search upwards?)
acarrico has joined #nixos
<srid>
How dangerous is it to override `nix.package` to use 2.4 (or whatever newer version)?
<srid>
(on nixos)
<morgrimm>
Actually, source_up or source_env might for direnv
<srid>
My laptop hardly lasts one hour on battery (it is a P71), I don't even try to configure hibernate :)
<lunik1>
I don't remember setting up anything special for hibernate to work
o1lo01ol1o has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @flokli opened pull request #87742 → nixos/binfmt: move systemd-binfmt.service to binfmt module → https://git.io/Jf4Ru
o1lo01ol1o has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
LysergicDreams has joined #nixos
o1lo01ol1o has joined #nixos
<srhb>
srid: Pretty safe in my experience. If you keep a stable nix around (root it) I've never been unable to recover from catastrophic failures.
<srhb>
Because you can always get back.
<zie>
so the default installed kernel != the default ISO kernel. how can I install the default 20.03 ISO kernel?
<srid>
srhb: root it?
o1lo01ol1o has quit [Remote host closed the connection]
<srhb>
srid: As in make a gc root for it, so you don't accidentally lose it in a garbage collection.
waleee-cl has joined #nixos
o1lo01ol1o has joined #nixos
<srhb>
(You'll still be able to recover, it's just more of a hassle to bootstrap nix from a broken nix)
o1lo01ol1o has quit [Read error: Connection reset by peer]
o1lo01ol1o has joined #nixos
mallox has quit [Quit: WeeChat 2.8]
eoli3n_ has quit [Ping timeout: 240 seconds]
mallox has joined #nixos
paws has quit [Read error: Connection reset by peer]
eoli3n_ has joined #nixos
<{^_^}>
[hydra] @edolstra merged pull request #757 → schema/Builds: use jobset_id instead of jobset name matches → https://git.io/Jf4BK
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
Darkmatter66 has quit [Ping timeout: 240 seconds]
commandocrypto[m has joined #nixos
<Woutifier>
I have a nodejs project that I ran node2nix on, that gave me a couple of nix files. But I prefer that it fetches the source of the nodejs project from github during the build instead of me providing it in the local filesystem. What would be the way to do that?
<srhb>
Woutifier: Find wherever it refers to your src (likely ./.) and replace that with a fetchFromGitHub, presumably.
somerand0m[m] has joined #nixos
<Woutifier>
right, so just a overrideAttrs and that would be it
<srhb>
Woutifier: That'll likely also work, but you could also just change it in the nix expressions directly.
<srhb>
Woutifier: (The expression has to import the source to the store _somewhere_ in order to build it, so there must be a reference)
<Woutifier>
my idea behind that is to not modify the files generated by node2nix, so I don't have to do any changes to that part when I rerun it as versions change.
<{^_^}>
[nixpkgs] @badmutex opened pull request #87743 → backport: visidata is supported on darwin + add setuptools → https://git.io/Jf4Rp
<srhb>
Woutifier: Then your overrideAttrs idea sounds like the most logical solution indeed.
eoli3n_ has quit [Remote host closed the connection]
<Woutifier>
thanks, on a side note: relevant answers to questions are very quick on here
o1lo01ol1o has quit [Ping timeout: 256 seconds]
eoli3n_ has joined #nixos
eoli3n_ has quit [Remote host closed the connection]
eoli3n_ has joined #nixos
sridIrcCloud has joined #nixos
<srhb>
#nixos is a pretty great channel. :)
eoli3n_ has quit [Read error: Connection reset by peer]
eoli3n_ has joined #nixos
<visl>
silly question i'm sure: is nixos-rebuild switch == apt-get update && apt-get upgrade where as nixos-rebuild --upgrade switch == do_release_upgrade? -- how often/when should i be adding the --upgrade flag?
<srhb>
visl: No, not at all. I think it's best not to think too much in terms of other package managers
<srhb>
visl: switch is essentially a no-op if you do not update your channels
<srhb>
visl: It more or less means "immediately bring the system to the state defined in the configuration and the current channel (/state of nixpkgs)"
<visl>
so if i add a package to my config, is that "updating a channel" or is that 20.03 -> 20.10 or whatever
<srhb>
--upgrade on the other hand says "update the channels/state of nixpkgs"
<srhb>
visl: Updating a channel usually means "20.03 -> 20.03 (but newer)"
<srhb>
So, security fixes, new packages, *some* more impactful changes.
<srhb>
visl: Updating your channel (fetching a newer version of it) does not mean upgrading to a new stable release.
<srhb>
visl: If you look in nix-channel --list you'll see the version is hardcoded for each name.
<srhb>
visl: (It actually corresponds to a specific release branch on github)
<visl>
ok so `nixos-rebuild switch` is sync the system with the config; and --upgrade is pull new files and sync the system
<srhb>
visl: Yeah, --upgrade is "nix-channel --update && nixos-rebuild ..."
<visl>
nix-channel --list doesn't print anything for me...
<srhb>
visl: Root probably owns your channels, try with sudo.
<visl>
sure enough
<srhb>
visl: So, conversely (and to clarify, hopefully) upgrading to a newer _release_ of NixOS requires modifying your actual channel source.
<srhb>
visl: There's no nixos-rebuild switch for that.
<{^_^}>
[nixpkgs] @jtojnar opened pull request #87744 → chrome-gnome-shell: Fix missing introspection data → https://git.io/Jf40d
<visl>
ok, so i should occasionally nixos-rebuild --upgrade switch to get bug fixes
andreas303 has quit [Remote host closed the connection]
MichaelRaskin has quit [Ping timeout: 264 seconds]
<srhb>
visl: Yes, that's a good idea.
<srhb>
And bianually you should change your channel to the new release
eoli3n_ has quit [Ping timeout: 256 seconds]
<visl>
and if i want to upgrade "major" version of the os, i update teh version in nix-channel and then --upgrade switch?
proofofkeags has quit [Remote host closed the connection]
teto has quit [Ping timeout: 240 seconds]
andreas303 has joined #nixos
ddellacosta has joined #nixos
dingenskirchen has joined #nixos
<visl>
srhb: thanks. just to clarify since i'll wonder later `--add` is updating my channel, not adding/appeneding to it. there's only 1 active channel correct?
teto has joined #nixos
proofofkeags has joined #nixos
<srhb>
visl: Not quite. nixos-rebuild will _default_ to the channel named nixos.
<srhb>
visl: You can have as many as you like.
chloekek has joined #nixos
<visl>
srhb: hah ok. something to explore another day
eoli3n_ has joined #nixos
<srhb>
Lots of it. ;-)
<visl>
indeed. it's quite a different paradigm but one i'm enjoying
<srhb>
visl: Glad to hear it :)
proofofk_ has joined #nixos
proofofk_ has quit [Read error: Connection reset by peer]
MichaelRaskin has joined #nixos
proofofkeags has quit [Read error: Connection reset by peer]
fendor has joined #nixos
proofofk_ has joined #nixos
fendor has quit [Client Quit]
mallox has quit [Quit: WeeChat 2.8]
fendor has joined #nixos
mallox has joined #nixos
proofofk_ has quit [Remote host closed the connection]
proofofkeags has joined #nixos
mallox has quit [Client Quit]
mallox has joined #nixos
o1lo01ol1o has joined #nixos
shafox has quit [Remote host closed the connection]
erasmas has joined #nixos
veleiro has joined #nixos
gentauro has quit [Read error: Connection reset by peer]
mallox has quit [Quit: WeeChat 2.8]
Rusty1 has joined #nixos
<emily>
is it better to depend on makeDerivation or stdenv in a nixpkgs derivation?
<emily>
if you only use stdenv.mkDerivation
<qyliss>
I don't think you can just depend on mkDerivation, can you?
<bkv>
it works for certain scopes, like qt and python
gentauro has joined #nixos
<bkv>
but not generally
<emily>
ah, right
<emily>
the overridden mkDerivation there always catches me out :/
eoli3n___ has joined #nixos
eoli3n_ has quit [Read error: Connection reset by peer]
CMCDragonkai1 has quit [Remote host closed the connection]
<commandocrypto[m>
bkv: if you need it for a project, i can help you package it. looks like a good thing to have around - i guess very little software uses it as a dependancy until now?
<bkv>
i'm fully capable of packaging it :p just surprised it hasn't been
<commandocrypto[m>
* bkv: if you need it for a project, i can help you write a nix expression for it. looks like a good thing to have around - i guess very little software uses it as a dependancy until now?
<bkv>
that said, i've quite formally given up submitting stuff to nixpkgs, and i don't actually need it anyway
<commandocrypto[m>
lol. ya same
proofofkeags has quit [Remote host closed the connection]
proofofkeags has joined #nixos
<zie>
how would I start stage 2's init manuallly?
fendor_ is now known as fendor
<fendor>
Hello! Finished installation of nixos and get a lot of lines such as: "AER: PCIe Bus Error: severty=Corrected, type=Physical"
proofofkeags has quit [Read error: Connection reset by peer]
proofofk_ has joined #nixos
<fendor>
I was able to avoid it by setting `pci=noear` in the kernel, I am wondering if that is normal and I should make this option permanent?
<commandocrypto[m>
bkv: it doesn't look like it's used for anything... there are 2 projects listed in the "who uses PBC" and one is totally dead and the other i'm 90% sure is totally dead
<homtiri>
like to have a declarative way to build and deploy a database and any sort of
<homtiri>
webserver. Kinda like having a kubernetes setup that I can run locally or push somewhere, I'd like to have be able to
<homtiri>
<homtiri>
- Use nix-shell to develop individual services
<homtiri>
- Run the entire thing in a QEMU VM (or maybe like docker-compose) with a single command
<homtiri>
- Reproduce my VM setup in a real server with something like krops
<homtiri>
<homtiri>
I'm struggling to piece this together, does anyone have an example they can point me towards?
<homtiri>
<homtiri>
Thanks in advance!
user_0x58 has joined #nixos
drakonis_ has joined #nixos
drakonis has quit [Ping timeout: 244 seconds]
o1lo01ol1o has joined #nixos
seku has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
LysergicDreams has quit [Ping timeout: 240 seconds]
<pistache>
homtiri: the idea would be writing a NixOS derivation that gets your application's backend code/assets/dependencies
worldofpeace_ has joined #nixos
<pistache>
and using that NixOS derivation in a custom NixOS module that configures the webserver and related services (database, cache, SSL certificates, etc)
orivej_ has quit [Ping timeout: 256 seconds]
LysergicDreams has joined #nixos
<pistache>
once you've got these two things running, getting them to run in a QEMU VM is dead-simple: "nixos-rebuild build-vm"
dxtr has joined #nixos
<pistache>
if it's not a single application that you want to deploy but a set of services, you may want to look at NixOps
<homtiri>
I've made some good progress there and been able to build and run some of my apps with nix, but I'm not on NixOS at the moment. Would you suggest booting up NixOS instead of trying to reproduce this on another linux distro?
<pistache>
if you're already building your apps from Nix (with no out-of-store dependencies), I'd say you've already done the "hard" parts
<Yaniel>
somewhat humorously "Copyright (c) 2020 Advanced Micro Devices, Inc. All rights reserved."
<homtiri>
@pistache Right now my struggle is getting, let's say, nginx + postgres + my own service with a systemd definition up and running in a network like I'd do with docker-compose or kubernetes yaml, and ship it to a VM and single machine
rickhull has joined #nixos
<Yaniel>
but then deep in the source dirs it contains ARM's ASTC compressor code that is nonfree
<homtiri>
@colemickens yes, I think so. I'd like to have the infra with my apps up and running with a decent feedback loop as I iterate. And when I'm happy ship it to my server running NixOS
<pistache>
homtiri: indeed if it's a single VM, NixOs itself should be enough, but it doesn't change much anyway, most of the work you want to do is in NixOS itself
nbrspo^ has joined #nixos
<Gaelan>
fwiw, nixos-rebuild can deploy over SSH by itself
<homtiri>
I guess for this I should start a qemu or virtualbox VM manually on Arch Linux and use something like NixOps to deploy it? Is it possible to run nixos-rebuild without NixOS (in another distro with Nix installed)?
travelion78 has quit [Ping timeout: 272 seconds]
<homtiri>
nixos-rebuild build-vm sounds awesome
travelion78 has joined #nixos
<colemickens>
It sounds like you just want to run a NixOS VM and then deploy to it with NixOps?
<colemickens>
That gets you your immediate feedback loop, there's no need to build a raw VM image if you're going to manage it with NixOps anyway. (Unless this is at scale and you want a good gold image to start from, or whatnot)
<adisbladis>
homtiri: If you're using nixops you don't need to run nixos-rebuild
cinimod` has quit [Ping timeout: 240 seconds]
<Gaelan>
internally, nixos-rebuild is a shell script something along the lines of "nix-build ... && result/switch-to-configuration"
<homtiri>
yes, let's say I locally develop an OCaml app with Nix, and deploy it to a VM for integration tests, and then ship it to a NixOS server
waleee-cl has quit [Quit: Connection closed for inactivity]
<colemickens>
That should be fine. You could even use nix/nixos's vm test driver stuff to automate your integration testing
<homtiri>
Ok thanks, I'll get a VM with nixos & ssh access up and running then
o1lo01ol1o has joined #nixos
<colemickens>
Oh yeah, for Azure at least, back in the day of NixOps, it had its own bootstrap image.
<asymmetric>
can someone explain _module.args to me?
homtiri has quit [Remote host closed the connection]
<colemickens>
do we publish a bare minimal nixops bootstrap image for qemu?
<colemickens>
(w/ sshd enabled, w/o the installer, for example)
<lordcirth>
colemickens, I'm not aware of one
user_0x58 has quit [Remote host closed the connection]
user_0x58 has joined #nixos
o1lo01ol1o has quit [Read error: Connection reset by peer]
<camsbury>
seems fine I guess - no sign of windowsc
<{^_^}>
[nixpkgs] @siriobalmelli opened pull request #87755 → dpkg: fix build error on Darwin by patching missing dependency on xlocale.h → https://git.io/Jf4oS
<T0pH4t>
hello, I'm trying to install an unfree package on nixos during rebuild so i added this option 'nixpkgs.config.allowUnfree = true;' too configuration.nix but rebuild doesn't see to use it
<T0pH4t>
was this option changed at some point?
<clever>
T0pH4t: are you doing `import <nixpkgs> {}` at any point?
mallox has joined #nixos
<T0pH4t>
clever: no, just sudo nixos-rebuild switch
<T0pH4t>
which then invokes the configuration.nix i guess?
<{^_^}>
[nixpkgs] @vcunat pushed commit from @adisbladis to release-20.03 « kdeApplications.kdegraphics-thumbnailers: Fix 404ing patch url »: https://git.io/Jf46s
<clever>
T0pH4t: can you pastebin the configuration.nix?
eoli3n_ has quit [Remote host closed the connection]
<T0pH4t>
clever: thats a sub config which is imported by main config
orivej has quit [Ping timeout: 256 seconds]
<evanjs>
Is there anything like --keep-failed but like --keep-latest?
<evanjs>
That sounds like it would be very useful for nix build and etc
gulplante has quit [Quit: WeeChat 2.7.1]
<T0pH4t>
i've set this up so that i am using vagrant to boot nixos which then has a default config that loads a configuration.nix out of a shared folder, thats the part in the shared folder
nixbitcoindev has quit [Ping timeout: 240 seconds]
<clever>
T0pH4t: its nixpkgs.config.allowUnfree
<clever>
T0pH4t: you have the nixpkgs. missing on line 3
<T0pH4t>
wwosp
<T0pH4t>
yeah sory miss paste
<T0pH4t>
still get teh same error :/
codygman has quit [Ping timeout: 256 seconds]
iqubic has quit [Ping timeout: 244 seconds]
<pikajude>
what's the simplest way to create a tap device for my qemu guest
<clever>
T0pH4t: are you sure its actually using that config file? that typo should have given a totally different error
<pikajude>
which can be connected to the outside world
<pikajude>
i'm really not good with networking
Fare has joined #nixos
<T0pH4t>
i fixed the typo, yeah its using the file
<DigitalKiwi>
does anyone know how to get tty1 to work with nvidia driver?
<OmnipotentEntity>
works for me?
<OmnipotentEntity>
I don't think I did anything special to make it work
<Gaelan>
I have a flake that depends on nixpkgs. How to I make it point at a local git clone instead of github so I can test something?
<Gaelan>
How do I*
<bkv>
file://
<Gaelan>
right now my flake.nix doesn't specify a source for the nixpkgs input at all—it's just an argument to the outputs function that automagically shows up
<Gaelan>
do I need to add an inputs section then?
<{^_^}>
[nixpkgs] @emilazy opened pull request #87760 → freeimage: mark as broken on aarch64 → https://git.io/Jf4iJ
<bkv>
inputs.nixpkgs.url = "file://...";
cinimod` has joined #nixos
alp has quit [Ping timeout: 272 seconds]
<Gaelan>
error: file:// URL 'file://home/gaelan/src/nixpkgs' has unexpected authority 'home'
<clever>
pikajude: create a bridge that both eth0 and tap0 are members of
<pikajude>
yeah, if i do that i lose all internet connectivity
<clever>
pikajude: then packets can just flow between the 2 like a regular switch
<pikajude>
that's the part that's giving me trouble
<clever>
pikajude: you must then move all of your network config to br0, not eth0
mallox has quit [Quit: WeeChat 2.8]
<pikajude>
oh
<clever>
configure br0 like you used to configure eth0
<pikajude>
so if i set br0.useDHCP = true it might work
<pikajude>
i see
<clever>
yeah
<{^_^}>
[nixpkgs] @vbgl opened pull request #87761 → ocamlPackages.textutils: remove at 112.17.00 → https://git.io/Jf4i8
<OmnipotentEntity>
clever++
<{^_^}>
clever's karma got increased to 417
<pikajude>
well that's a bit confusing but i'll try it out
mallox has joined #nixos
cinimod` has quit [Ping timeout: 265 seconds]
<pikajude>
wow, i think it might actually be working
<DigitalKiwi>
omnipotententity: oh. it may be a macbook pro only problem :(
<pikajude>
clever++
<{^_^}>
clever's karma got increased to 418
<pikajude>
you are my hero
<Gaelan>
if I start one nix build, change a nix file and then start another build in parallel, will that break anything?
<clever>
Gaelan: nope, but when each one finishes, it will overwrite the result symlink
<clever>
Gaelan: depending on the order they finish in, your result may be pointing to the "wrong" thing
<Gaelan>
i expect neither to succeed tbh
<pikajude>
i thought it would create result-n symlinks though
<Gaelan>
i just want to see how each one breaks next
<clever>
pikajude: thats if you point a single nix-build to several drvs
<pikajude>
oh it chooses at the beginning?
<pikajude>
ok
<pikajude>
anyway Gaelan if you issue a nix-build it will create the .drv file and build based on that, so you can change nix files to your heart's content
bqv has joined #nixos
zeta_0 has joined #nixos
<Gaelan>
but yeah i'm doing some fairly niche cross things that evidently nobody's tried since 2019
bkv has quit [Ping timeout: 256 seconds]
<Gaelan>
because it's broken in all sorts of weird ways
<fresheyeball>
but does not appear to evaluate the reflex stuff at all
<infinisil>
Pretty sure that haskell.overlays is not a nixpkgs thing
<infinisil>
Seems like they just use it for them
alp has joined #nixos
mallox has quit [Quit: WeeChat 2.8]
ryantm_ is now known as ryantm
sasyctu is now known as sasyctu[m]
<fresheyeball>
infinisil: any advice on how I can integrate?
mallox has joined #nixos
oborot has quit [Ping timeout: 265 seconds]
<infinisil>
Don't know anything about reflex or ghcjs, so nope
leah2 has quit [Ping timeout: 260 seconds]
n3t has joined #nixos
n3t has left #nixos ["WeeChat 2.3"]
<DigitalKiwi>
anyone know how to contact jonringer?
<cole-h>
I think he has an account on Discourse. Don't know his IRC handle (if he has one) though.
<cole-h>
Worst case, you send an email to his git email.
<bqv>
or make a poorly written PR
fresheyeball has quit [Quit: WeeChat 2.7.1]
<worldofpeace>
DigitalKiwi: discord
<DigitalKiwi>
oh yeah git has emails...
<{^_^}>
[nixos-homepage] @garbas pushed to landing-page « missed closing tag »: https://git.io/Jf4Pq
fabianhjr has quit [Ping timeout: 272 seconds]
agsdheidjd has joined #nixos
<DigitalKiwi>
bqv: yeah that was my last resort he tests/merges a lot of my packages for me which is why i wanted to contact him lol so i could make an out of place comment on any of those...
<clever>
pikajude: tap0 doesnt need an ip assigned
<pikajude>
ok
<pikajude>
i wonder if i have to explicitly turn off dhcp for it
vidbina has quit [Ping timeout: 256 seconds]
<pikajude>
no, that didn't help either
ebopp has quit [Quit: leaving]
erasmas has quit [Quit: leaving]
<kaliumxyz>
I'm trying to get the logical discovery 2 to work on nixos
<{^_^}>
[nixpkgs] @badmutex opened pull request #87767 → backport to 19.09: python: pytz: 2019.1 -> 2019.3 → https://git.io/Jf41f
selfsymmetric-mu has joined #nixos
rickhull has quit [Remote host closed the connection]
MmeQuignon has quit [Quit: WeeChat 2.8]
eoli3n__ has joined #nixos
remirol is now known as lorimer
<selfsymmetric-mu>
Hey folks. I'd like a streamlined customize-install-and-contribute workflow. I think a reasonable way to do this is to fork nixpkgs…and then at some point call `fetchTarball`? I'm missing the middle step.
<selfsymmetric-mu>
I'm imagining using my fork as a channel. Is that right?
<selfsymmetric-mu>
I don't want to need to have it locally, since I want to be able to host something for all my machines, y'know?
<hr[m]>
Could you give a bit more detail, I'm not sure what your first sentence means
<selfsymmetric-mu>
Okay let's say I fork nixpkgs, make a feature branch, customize a package. I want to merge my feature into my master and start using it in my NixOS configuration. MEANWHILE I want to make a PR from my feature into the upstream nixpkgs master. See what I'm getting at?
Peter_Storm has joined #nixos
erasmas has joined #nixos
<selfsymmetric-mu>
Now I know how to do all that EXCEPT I don't know how to import my fork as a channel.
<cole-h>
Why not just set `NIX_PATH=nixpkgs=/path/to/your/local/checkout`?
<bqv>
back before i used flakes, i had a solution for that
<selfsymmetric-mu>
cole-h: Because I'd rather not locally check it out on every computer that uses my version-controlled configuration.nix
<clever>
selfsymmetric-mu: i think you can do /archive/pull/42/head.tar.gz, but ive not tried that
<bqv>
that does work, yes
<selfsymmetric-mu>
Beautiful, thanks. bqv++
<bqv>
but obviously you're limiting yourself to one PR at a time
<{^_^}>
bqv's karma got increased to 9, it's a crit!
<selfsymmetric-mu>
Wait why?
<bqv>
if you fetch a tarball, you then use that tarball as one nixpkgs. if you want to apply more than one PR, you have to merge them, which is why i came up with that solution
<selfsymmetric-mu>
Oh I see. I just stick them under my `pkgs`.
<selfsymmetric-mu>
So I have `pkgs`, `pkgs.unstable`, `pkgs.nur`…
<bqv>
you'll have trouble if you ever want to use a module from a PR
<selfsymmetric-mu>
And now I'll have `pkgs.pr87766` or whatever.
<selfsymmetric-mu>
Well I'll give it a shot, let's see…
opticnerve has quit [Ping timeout: 260 seconds]
<pikajude>
oh, i may be getting firewalled, so i'm going to check on that
<pikajude>
nope, that wasn't it either
joehh1 has quit [Ping timeout: 244 seconds]
joehh1 has joined #nixos
leah2 has quit [Ping timeout: 272 seconds]
pbb has quit [Remote host closed the connection]
mmlb has quit [Read error: Connection reset by peer]
<pikajude>
tcpdump explicitly actually does not support multiple interfaces and it has no way to print the interface name for each packet so `-i any` is actually useless
mmlb has joined #nixos
<pikajude>
the recommended method is to start one instance for each interface lol
<clever>
pikajude: yep, thats what i said to do earlier
<pikajude>
you said you can pass multiple interfaces to tcpdump, which is what confused me
<thommey>
their current development version does, but there's no release with that feature
<thommey>
and there hasn't been a release in years
<pikajude>
could it be a problem with the mac address?
<clever>
pikajude: ah, i can see how you could misread the msg
pbb has joined #nixos
<clever>
pikajude: the guest inside the vm should have a unique mac, and the eth card must support mac spoofing i believe
<pikajude>
does the mac address reported by the qemu macos need to match the one on tap0?
<pikajude>
oh, well, i don't know anything about that
<clever>
pikajude: no, the mac's need to all be unique
<pikajude>
how do i know if mac spoofing works
<clever>
not sure
<pikajude>
cool
lukegb has quit [Ping timeout: 258 seconds]
<acowley>
Something in the manual tripped me up, but I'm not confident enough to fix it myself: the nix manual suggests that the `-I` option gets its semantics from those of `NIX_PATH`, which it links to in the manual. But when I tried setting a colon-separated list of paths via -I, it did not work.
<clever>
acowley: -I cant take a : seperated list, you need multiple -I's
<clever>
acowley: -I foo=/foo -I bar=/bar
<acowley>
clever: thanks, that's where I eventually came to in figuring out my problem. I think the manual could use a slight rewording.
<acowley>
clever: Yes, that's how I fixed my problem.
fusion809 has joined #nixos
<pikajude>
i can freely choose whatever mac i want in the qemu command invocation
<T0pH4t>
so looking for some advice as to setting up containers in nix configuration programatically. https://pastebin.com/NJWzM6z9 <- example. Trying to declare multiple instance of a service, but I can't seem to figure out how to merge the containers problem into the main config
<clever>
pikajude: what mac did you last try?
<pikajude>
52:54:00:c9:18:27
<T0pH4t>
properly into main config*
<pikajude>
i don't think it matters which one it is, does it?
<pikajude>
it just has to be unique
<clever>
pikajude: thats probably fine, as long as its unique
lukegb has joined #nixos
<clever>
pikajude: there is a special bit in the first byte, that marks an addr as a broadcast one, which you shouldnt use
<pikajude>
ok
<pikajude>
well, this one is certainly unique and it's still not getting passed to enp8s0
<clever>
pikajude: it would probably be simpler to just use the nat code nixos/iohk use, you already have an example of how to configure everything
<bqv>
by the way, nixos-containers work poorly with flakes
<bqv>
because they just behave as though you're importing nixpkgs and nothing else
<bqv>
all your specialArgs and extra modules are gone
<bqv>
it's distressing
<pikajude>
yeah
<pikajude>
i have my nat already configured for my nixos-container so it'll be a pain to figure out how to merge the two
knupfer has quit [Quit: knupfer]
knupfer1 has joined #nixos
knupfer1 has quit [Client Quit]
knupfer has joined #nixos
<pikajude>
clever: can you link it to me again, the link is on my laptop which isn't here
<c4rc4s>
I've just installed kmail, but it seems that it can't find one of its dependencies (kmail-account-wizard). I see that kmail-account-wizard is pulled in as a build input, and it exists under /nix/store ... any advice for digging deeper?
aw has joined #nixos
<clever>
DigitalKiwi: that would be the problem then
<morgrimm>
Anyone know what package the Iris openGL drivers are in off the top of their head?
<morgrimm>
Actually, I should probably use nouveau
Edes has joined #nixos
pbb has joined #nixos
camlriot42 has left #nixos ["User left"]
<Edes>
Hi, I have a question about nix, if I'm trying to import an external file through `builtins.readFile`, can I have an antiquotation (the `${foo}` things), if not, is there any way to put the values of nix variables in this file?
<clever>
Edes: if you make that file a nix expr, you can just run `import` on it directly
agsdheidjd has quit [Ping timeout: 272 seconds]
agsdheidjd has joined #nixos
__monty__ has quit [Quit: leaving]
fabianhjr has quit [Ping timeout: 264 seconds]
<Edes>
it's not a nix expression though, it's supposed to be an xmonad config
<`_>
Edes: There's also `pkgs.substituteAll` however your file that you're attempting to import won't be valid nix until it's substituted.
dnlkrgr has quit [Quit: WeeChat 2.7.1]
<Edes>
I just want to point to the path of a package
dermetfan has quit [Ping timeout: 272 seconds]
asymptotically has quit [Quit: Leaving]
dnlkrgr has joined #nixos
alp has joined #nixos
<`_>
Edes: The way I deal with something like that is I have a function `writeNixScript` that extends `writeTextFile` with `nix-instantiate --parse` as the `checkPhase`, so I pass the nix that I put through `substituteAll` and it still gets syntax checked.
<`_>
However I don't import these files, so YMMV
<morgrimm>
Edes: If you use home-manager, you can have your cake and eat it too. I basically generate my dotfiles in a Nix context, and home-manager takes care of putting the files in place
<morgrimm>
I installed mesa in my system packages, I expected to get the opengl drivers in /run
<morgrimm>
But looks like nothing's there
<clever>
morgrimm: how did you install mesa?
<Yaniel>
you also need an actual graphics driver
<clever>
morgrimm: and you only get drivers in /run if you set hardware.opengl.enable = true;
<morgrimm>
I enabled the i915 kernel module for my iGPU, and mesa in my host systemPackages - but that hardware option is definitely the one I'm missing
sigmundv has quit [Read error: Connection reset by peer]
<morgrimm>
I thought it'd be set by home-manager's wayland module
<morgrimm>
er, sway module
sigmundv has joined #nixos
Edes has quit [Quit: Lost terminal]
<morgrimm>
Yup, that does the trick - thanks all!
<morgrimm>
Really appreciate the help
eoli3n_ has quit [Ping timeout: 240 seconds]
hexagoxel has quit [Quit: ZNC 1.6.5+deb1+deb9u1 - http://znc.in]
eoli3n__ has quit [Ping timeout: 256 seconds]
<clever>
morgrimm: home-manager cant touch /run
<clever>
morgrimm: home-manager can only manage $HOME
<morgrimm>
No, I know, but can't its modules enable nixos options?
hexagoxel has joined #nixos
<ldlework>
eadwu[m]: heyyyyy
<commandocrypto[m>
<bkv "presumably the old version works"> wait what version of nix received flakiness?
<ldlework>
eadwu[m]: Got a clue how you would get a steam game, that's running via Proton, the wine thing provided by Steam's 'platform support', working via nvidia-offload?
<bqv>
no idea
<{^_^}>
[nixos-homepage] @garbas pushed 5 commits to quickstart-guide: https://git.io/Jf4Qk
misasaki has joined #nixos
<{^_^}>
[nixpkgs] @Mic92 pushed commit from @mmlb to staging « go: stop setting GOPATH »: https://git.io/Jf4Qt
misasaki has quit [Client Quit]
misasaki has joined #nixos
<commandocrypto[m>
do you have to be running off the flakes branch of the nix repo?
<morgrimm>
Is the flakes branch kept up to date with master by the way? I'm using pkgs.nixFlakes in a nix-shell to get support
<morgrimm>
but if I can pin mine to nixpkgs/flakes and throw the need for the shell out, that'd be good
<commandocrypto[m>
> This branch is 668 commits ahead, 30 commits behind master.
<{^_^}>
error: syntax error, unexpected ',', expecting ')', at (string):310:33
<commandocrypto[m>
shouldn't the bot assess bangs instead of blockquotes or am i holding it wrongly
<commandocrypto[m>
> help
<commandocrypto[m>
it = matrix
<cole-h>
Bot uses > for evaluating arbitrary nix expressions
<commandocrypto[m>
but ya morgrimm it looks like the flakes branch is fairly regularly updated from master, so i'm probably going to do what you're doing (using pkgs.nixFlakes in a nix-shell)
<morgrimm>
If it's fairly updated, you might be able to just pin the nixpkgs input to the flakes branch
<commandocrypto[m>
<morgrimm "but if I can pin mine to nixpkgs"> actually i spoke before i really knew what the alternative was - how would this work. i don't understand what you'd be pinning
<morgrimm>
then I don't think you *need* the shell
philr has joined #nixos
<morgrimm>
Like in your flake inputs, you can pin a branch of a repo - so you could pin the nixpkgs input to nixos/nixpkgs/flakes
<morgrimm>
Then you'd only get package updates as they came into the flakes branch ofc, but you could do flake operations anywhere without the need for a shell
<{^_^}>
[nixpkgs] @andir opened pull request #87773 → [20.03] firefox: Add patch to fix AES GCM IV bit size → https://git.io/Jf4QF
<morgrimm>
Then again, if you're doing work on your flake, you'll probably be in your flake dir anyway, so
<morgrimm>
personal preference
rotaerk has joined #nixos
jumper149 has quit [Quit: WeeChat 2.8]
CMCDragonkai1 has joined #nixos
<bqv>
there is no need to use the flakes branch of nixpkgs at this point in time or anytime soon
<bqv>
however if you use the nix repo, it is a requirement that you use the flakes branch
<{^_^}>
[nixos-homepage] @garbas merged pull request #413 → Make call more visible on NixCon Team page → https://git.io/JfL9V
<{^_^}>
[nixos-homepage] @garbas pushed 2 commits to master: https://git.io/Jf47z
<{^_^}>
[nixos-homepage] @garbas pushed 0 commits to nixcon: https://git.io/Jf47g
CMCDragonkai has quit [Read error: Connection reset by peer]
CMCDragonkai1 is now known as CMCDragonkai
<morgrimm>
bqv: What do you mean?
ambro718 has joined #nixos
kvda has joined #nixos
<ambro718>
How do I ssh to a nixops machine without using "nixops ssh", when another tool invokes ssh (e.g. rsync, scp, git)?
Pot2Dash has joined #nixos
<bqv>
morgrimm: the nixpkgs flake is merged into master
<bqv>
because it's not destructive
gustavderdrache has joined #nixos
<morgrimm>
But only as the nixFlakes package, right?
<kvda>
Is there any special about having `-devel` packages in nix? I'm looking for libsqlite3 devel for static linking but it doesn't seem to come with it, nix search doesnt seem to show any of these for other programs. Is there another channel perhaps or is it just not there?
philr has quit [Ping timeout: 256 seconds]
<julm>
ambro718: you can add another ssh key of your own to the machine, or sqlite3 $NIXOPS_STATE .dump to get the private part of nixops' ssh key
<AmandaC>
is the unstable channel in the middle of a massive resturcture or something? I'm doing my weekly update, and I just saw three different builds of openssl 1.1.1g come down, with three differeent hashes (both -bin and -dev )
<ambro718>
julm: thanks, I suppose adding a key is easier
shibboleth has quit [Quit: shibboleth]
dnlkrgr has joined #nixos
<Gaelan>
kvda: try [packagename].dev
o1lo01ol1o has quit [Remote host closed the connection]
morgrimm has quit [Ping timeout: 256 seconds]
<kvda>
Thanks @Gaelan, nix seems to find the package, but after running `pkg-config --static --libs sqlite3` there are still no .a files
ambro718 has quit [Quit: Konversation terminated!]
<{^_^}>
[nixpkgs] @7c6f434c merged pull request #87743 → backport: visidata is supported on darwin + add setuptools → https://git.io/Jf4Rp
<{^_^}>
[nixpkgs] @7c6f434c pushed 3 commits to release-20.03: https://git.io/Jf45w
<clever>
kvda: this will build all packages staticly with musl
justanotheruser has quit [Ping timeout: 258 seconds]
philr has joined #nixos
<kvda>
clever as a side effect there will be .a files?
<clever>
kvda: yes
<commandocrypto[m>
<morgrimm "Like in your flake inputs, you c"> oh see. i was thinking about the nixos/nix repo. what you're saying makes sense now - nixpkgs must have a flakes branch that points to the flakes branch in nix. and as you say, nixpkgs wouldn't be destructive (nix does appear to be slightly more problematic as it appears they rebase from master regularly but master isn't getting as much from the flakes branch)
<kvda>
clever looks promising, libtool: link: ar cru .libs/libsqlite3.a
<Gaelan>
kvda just make sure you're building everything against musl, then—I assume having two libc's in your binary would be a recipe for a Bad Time
<clever>
Gaelan: yeah
<kvda>
@Gaelan yep was thinking the same, thanks :)
<kvda>
pretty sure ocaml can be switched to musl
<clever>
kvda: youll want to build your stuff from pkgsStatic.stdenv
<kvda>
do i need to add that to buildInput in the shell file or do you specify it elsewhere?
<clever>
kvda: the shell file must use pkgsStatic.stdenv instead of the normal stdenv
<clever>
kvda: and get all deps from pkgsStatic instead of pkgs
<kvda>
clever oh i see pkgStatic.stdend.mkDerivatin ..
<bqv>
pkgsStatic?
<kalbasit>
Setting up a new laptop with 4k display :( Can someone remind me how to change the console font while installing so I can see what I'm doing?
<clever>
kvda: yes
<cole-h>
kalbasit: Are you looking for `console.font` for your configuration.nix?
mbrgm_ has joined #nixos
ottidmes has quit [Ping timeout: 264 seconds]
johnjay has quit [Quit: WeeChat 2.3]
piezoid has quit [Remote host closed the connection]
mbrgm has quit [Ping timeout: 246 seconds]
mbrgm_ is now known as mbrgm
ottidmes has joined #nixos
<kalbasit>
cole-h: yes, but I can't type that if I can't see, it's way too small. I was hoping for something ala `loadkeys colemak`