<anderson_>
multun: yes, I think it can be a really good use case after all.
drakonis2 has joined #nixos
pamplemousse has quit [Ping timeout: 260 seconds]
<adyatlov>
How about this folks export NIX_REDIRECTS='/nix/store/...-crossover-19.0.0/crossover/bin/wineloader=/nix/store/...-crossover-19.0.0/crossover/bin/.wineloader-wrapped'
<{^_^}>
mercode-org/meros-nix#60 (by Yutyo, 23 hours ago, open): The Problematic of Public Branding
tilcreator has joined #nixos
<energizer>
there's a discourse thread somewhere that gave me the impression nixos doesnt really want to be easy to install lest the number of beginner-users grow too much for experts to handle
<multun>
I can understand that
<adyatlov>
multun: What is that about
<colemickens>
Hm. I don't recall reading that out of the thread
<colemickens>
That sounds kinda gate-keepy?
<quinn>
i was thinking the same thing while reading that whole interaction. the more popular & inviting nixos gets, the more dissappointed users it will create
<multun>
colemickens: I think nixos is a super hard distro
<colemickens>
Is it a desire for it to be that way, or concern that it is that way?
<multun>
colemickens: making it more easy ot install is a bit of a trap
<colemickens>
aha, ok, thanks for the nuance, I think I get the point
<quinn>
colemickens: i mean if the distro is way easier to install than doing anything else users expect to be able to do, you set them up for a lot of confusion and frustration once they get past the first step. it'd be like having the tutorial level super easy, level 1 really hard
<colemickens>
ack
<qyliss>
You only get one chance to make a first impression
<multun>
^
<adyatlov>
Back in 2004 installing Mandrake didn't teach me anything about Linux. The Gentoo installation manual did
<energizer>
yeah yall are phrasing it more preceisely than i remembered
<quinn>
multun: have you looked through any of the mercode stuff yet? i'm trying to figure out what it was actually supposed to be, and it's hard with their website down
<MichaelRaskin>
quinn++
<{^_^}>
quinn's karma got increased to 4
<samueldr>
a linux distro, if vaguely unspecific
<gchristensen>
the hard thing about growing is sharing and communicating the culture and ideas
<adisbladis>
Mer means more in Swedish, and code is self explanatory. So I interpretet it as MoreCode
<gchristensen>
for example, a common thing for new people is "can't you please just add libc at a normal place?" completely missing the point
<energizer>
adisbladis: well that clears that up :D
<adisbladis>
This is obviously wrong but I'm sticking to it :)
<MichaelRaskin>
colemickens: I think it is not gatekeeping, but basically if our install procedure is hard then it is likely the user doesn't feel the problems NixOS solves
<samueldr>
(and annoyingly, naming stepping on toes of maemo mer, mer project, another linux distro thingy, broadly said)
<adyatlov>
gchristensen: It's like "why can't I assign variables" in a functional language. One has to accept the paradigm
<gustavderdrache>
"Mercode is a free libre open source community which gathers coders, hackers, activists, makers and designers worldwide. Also it is the Turkish name of planet Mercury, the first planet of the solar system." - https://github.com/mercode-org/mercode-pug/blob/master/about.pug
<gchristensen>
adyatlov: right
<MichaelRaskin>
I think we have recently tried to understand what in the world this project actually is
<quinn>
adisbladis: well if they took 8 months and 1k commits to figure out they didn't like the os, it's reasonable to expect they would have needed a LOT of code to get to a working product.
<MichaelRaskin>
Probably in a discussion whether some of their packages is worth trying to package
thc202 has quit [Ping timeout: 244 seconds]
<MichaelRaskin>
quinn: it looks like they do not have an actual consensus
rodrigo_ has joined #nixos
<multun>
quinn: the guy that doesn't like it wasn't the one doing all the work
<adisbladis>
gustavderdrache: That just makes it less clear =)
<quinn>
multun: makes sense
chpatrick has joined #nixos
drakonis_ has joined #nixos
<gustavderdrache>
i think the answer to the question of "what is this?" is "an ambitious undertaking"
<MichaelRaskin>
I guess before trying to find out their spread of opinions on Nix, we should all bring forward _our_ spread of opinions on doc formats
<ashkitten>
i agree with people who say nixos is a difficult distro, and i've talked at some length with gchristensen about some of the issues it has but i don't think the installation process is the biggest issue
<gchristensen>
<3 gustavderdrache
<{^_^}>
gustavderdrache's karma got increased to 6
<quinn>
MichaelRaskin: mercode-org/meros-nix/pkgs seems to have a bunch of themes and an idle game written in electron. and deezloader
<multun>
gustavderdrache++
<{^_^}>
gustavderdrache's karma got increased to 7
<ashkitten>
the thing is, install is hard yes but the difficulty goes up from there
drakonis has joined #nixos
<multun>
installing nixos is actually pretty easy I think
<multun>
I really don't like the nix-stuff interfaces
<quinn>
yeah, i installed in an hour or two the first time and now it's like under an hour easily.
<MichaelRaskin>
colemickens: it's not a pure laugh, though!
<quinn>
(those times being generally lower than arch or gentoo, and not much worse than debian)
drakonis2 has quit [Ping timeout: 244 seconds]
orivej has quit [Ping timeout: 246 seconds]
<samueldr>
everyone experiences it differently due to, non-coincidentally, their experiences!
<gustavderdrache>
speaking as someone who installed nixos and gave up (went back to pop os), it's definitely a lot of work to get right
<samueldr>
so saying it's easy/hard is unfair, as broadly speaking, it's both
<adisbladis>
Coming from Gentoo I was like "whoa, is it this easy?!"
<ashkitten>
honestly, it would be really cool to have a graphical nix configuration manager. we've got all these options, all with descriptions of what they do, grouped nicely, and no connection between that and the configuration itself
orivej has joined #nixos
<gustavderdrache>
and i've found a much happier balance using nix on my machine while still having imperative stuff because i'm lazy
<adisbladis>
But I understand that this is not the perspective everyone is coming from
<multun>
gustavderdrache: oh interesting
<infinisil>
ashkitten: Problem is that you'd need something like an AST editor to fully represent configs
<infinisil>
Because they can be arbitrarily complex
<samueldr>
ugh, not that discussion again :/
<MichaelRaskin>
My OS is managed with Nix, and I am not using NixOS, and my partition setup is described as a shell script
<qyliss>
the subset supported by the graphical editor doesn't have to me though
<qyliss>
*have to be
<ashkitten>
^
<samueldr>
there can be different levels of support
<quinn>
infinisil: on the other hand you could have a few checkboxes to support common use cases and leave text install for anything fancy
<MichaelRaskin>
Yeah, that seems to be done very right in that latest GUI configuration tool
<qyliss>
For Spectrum I'd like to have a graphical configuration editor that generates an XML file or something that is then imported by a human-writeable .nix file
<MichaelRaskin>
It edits a module to import
<infinisil>
quinn: Yeah that might be the best option
<adisbladis>
quinn: This is in line with my thoughts around a UI
<colemickens>
hm, "GUI for common stuff + drop to edit Nix for advanced topics" is still better than "GUI for common stuff + tons of random files|syntaxes for advanced topics"
<ashkitten>
i think it would be good to allow users to install nixos by checking boxes rather than writing their own configs
<infinisil>
qyliss: Why not JSON?
<multun>
infinisil: that's the "or something" part I guess
drakonis_ has quit [Ping timeout: 260 seconds]
<qyliss>
infinisil: because I think XML would be more suited to the sort of configuration system I'd like to have
drakonis2 has joined #nixos
<MichaelRaskin>
infinisil: because the notion of valid XML is kind of defined, and JSON…
<infinisil>
I guess it wouldn't matter if it's generated
<qyliss>
XML is very good at representing trees and bad at everything else, but what I want is basically trees
<qyliss>
For NixOS, XML probably wouldn't be the right choice, but I don't want to have NixOS configurations
<ashkitten>
i would've considered myself a pretty good linux sysadmin before i started using nix, but the first install process took me hours because i had to figure out what all these options meant before i could get a working system
<qyliss>
The principle could still work though
<infinisil>
For Nix at least there's builtins.fromJSON already
<gchristensen>
gustavderdrache: we'll get you one day
<qyliss>
there's builtins.fromXML too!
<infinisil>
Whaaa
<qyliss>
I think?
<qyliss>
> builtins.fromXML
<{^_^}>
attribute 'fromXML' missing, at (string):312:1
<qyliss>
> builtins.toXML
<infinisil>
Not in my version, I just checked :)
<{^_^}>
<PRIMOP>
<qyliss>
Oh, just toXML
<qyliss>
my mistake
<joepie91>
ashkitten: a big risk there is people starting to ask for arbitrarily complex boxes
<qyliss>
maybe it'll be JSON after all :)
<adisbladis>
toXML is a bit.. Let's say odd..
<joepie91>
you need to have your scope very well-defined before attempting something like that, IMO
<infinisil>
> builtins.fromTOML
<{^_^}>
<PRIMOP>
<infinisil>
qyliss: TOML is your other option :P
<joepie91>
so that you have a clearly communicable reason to say yes or no to certain requests
<qyliss>
TOML trees would be *horrible* :P
<adisbladis>
What? Unpossible.
<colemickens>
we all know yaml is the answer
<ashkitten>
joepie91: i don't see why scope creep is a problem, honestly, if it's hidden behind an "advanced" button
<qyliss>
maintainability
<ashkitten>
sure, but we can make it maintainable
<qyliss>
easier said than done, but maybe!
selfsymmetric-mu has joined #nixos
<adisbladis>
JSON with schema maybe?
<ashkitten>
this is obviously not a thing that exists yet, anyway
<joepie91>
ashkitten: because if you don't have a well-defined scope, you will quickly end up with the same unruly complexity there is in the *non-graphical* interface now, but with checkboxes instead of attributes
drakonis has quit [Ping timeout: 256 seconds]
<ashkitten>
and i won't be the one to make it, for many reasons
<infinisil>
ashkitten: joepie91: Oh! How about an option set `graphical-configurator.*`, and all options under that set are exposed to the graphical configurator
<joepie91>
especially considering that UI design is difficult and not all of the possible NixOS configs may even be representable within a graphical UI
<infinisil>
And they could have annotations that they mirror other options
<joepie91>
reasonably anyway
<infinisil>
And you could restrict option types under that set to be of e.g. booleans only
<MichaelRaskin>
Plain strings sound acceptable
<ashkitten>
i don't really see an issue with exposing chunks of raw nix code in the ui
<ashkitten>
if that's what the user wants
<joepie91>
like, if the goal is to have a simplified interface then you need to think carefully about where that simplicity actually is supposed to come from, where the line with the more complex usecases is drawn, and how to have a migration path between the two
<MichaelRaskin>
And it feels natural to be able to set the hostname
<infinisil>
Oh and with graphical-configurator.*, you can have option set nesting for different configuration tabs/pages
<joepie91>
you can't just put the original complexity in a different mold and expect it to be simpler
<MichaelRaskin>
ashkitten: weren't you formulating the task as exactly helping the users who are currently busy finding out what they want
<ashkitten>
anyway i won't bikeshed more about this thing that hasn't even reached a full idea yet and that i won't be building
<ashkitten>
MichaelRaskin: i said behind an "advanced" setting
<ashkitten>
there's a good chance people would want to use it after they've installed, too
<ashkitten>
and that they'll want more advanced features later
<selfsymmetric-mu>
When I have a pull request that's been approved, what happens then? Just wait? Ask a bot to do something?
<joepie91>
(the experience I'm drawing from here is mainly that with monolithic kitchensink frameworks and generic CMSes of various kinds, which try to attempt this exact same thing and invariably turn into a mess of incoherent featuresets)
<ashkitten>
but again, i'm not going to participate in this bikeshedding session anymore
<adisbladis>
selfsymmetric-mu: Which PR?
user_0x58 has quit [Quit: Leaving]
andreoss has quit [Ping timeout: 256 seconds]
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/Jf2jR
<adyatlov>
It appears that my executable is immune to libredirect.so
<selfsymmetric-mu>
tobeportable++
<{^_^}>
tobeportable's karma got increased to 1
<quinn>
selfsymmetric-mu: dwarf fortress is definitely important.
<selfsymmetric-mu>
Thanks quinn. :)
<selfsymmetric-mu>
I understand that for unreviewed pull requests the squeaky-wheel-gets-the-grease method may be acceptable for the long-term.
<selfsymmetric-mu>
But I don't understand approved-but-not-merged; my question is: what factors prevent approved pull requests from being merged?
morgrimm has joined #nixos
<selfsymmetric-mu>
A silly way of asking that question is: if we were to merge all approved pull requests this minute, what would befall nixpkgs?
<adyatlov>
OMG the executable that does the open() doesn't do dynamic linking so it is immune to LD_PRELOAD libredirect.so GRRRRR
<selfsymmetric-mu>
Thanks for the merge, adisbladis++
<{^_^}>
adisbladis's karma got increased to 0b1010011
<adyatlov>
That's a large number
<selfsymmetric-mu>
Hahahaha, whoever added these cheeky variations to nixbot did great
<infinisil>
:D
<adisbladis>
selfsymmetric-mu: Anyone can approve a PR, but you need to have commit access to nixpkgs which far fever people have
<multun>
selfsymmetric-mu: anybody can approve a PR, it may not be relevant. if the PR triggers a big rebuild, it also shouldn't go directly to master
<infinisil>
It's very easy to approve a PR, but by merging it you're pretty much responsible for any breakages it causes, so people won't merge if they're rather sure it won't cause trouble
morgrimm has quit [Ping timeout: 256 seconds]
<infinisil>
That last if should be s/if/unless
<infinisil>
And since nixpkgs is so big, with thousands of packages and modules, it's very hard to know whether something will cause problems
<infinisil>
Or it *can* be very hard for some PRs at least
<{^_^}>
[nixpkgs] @cdepillabout pushed 2 commits to haskell-updates: https://git.io/Jfaeq
LevelChart8 has joined #nixos
<LevelChart8>
(I know this is technically the wrong room, but I have not had success getting results elsewhere.) I have nixpkgs on OSX and I need to set allowBroken = true to ~/.config/nixpkgs/config.nix but I don’t know where to find this file.
<LevelChart8>
Where can I find this?
<multun>
what do you think of "maintainers can merge stuff for their own packages" ?
<infinisil>
LevelChart8: At the path you just mentioned?
<LevelChart8>
so it sits in my OS root
<LevelChart8>
and not the repo?
<infinisil>
multun: The RFC was just closed due to time constraints by the author, it's not rejected or anything, and anybody is free to pick it up again
<infinisil>
LevelChart8: Yes, the ~ there means your home directory
o1lo01ol1o has joined #nixos
o1lo01ol1o has quit [Read error: Connection reset by peer]
<multun>
infinisil: I was reading through to and starting to get to that conclusion
<selfsymmetric-mu>
Thanks for the context around merging adisbladis multun infinisil.
njha has joined #nixos
<selfsymmetric-mu>
So review approvals in this case are merely a weak signal. They only show that two people agree with the changes, rather than one.
<multun>
I sometimes review stuff to help the maintainers with the easy ones
Vikingman has quit [Ping timeout: 245 seconds]
<njha>
So I'm trying out NixOS right now, and I want to store a secret (WireGuard Private Key) declaratively. Putting it in /etc/nixos/ seems like a bad idea, but the issue for storing secrets is still open (#24288).
<infinisil>
Hmm okay so that makes me think: In my https://github.com/Infinisil/nixus (formerly nixoses), the whole deployment is script built with Nix itself
<infinisil>
And with that, it's impossible to support a `secret.text` option for using a string secret
<infinisil>
Without putting the secret to the store
jluttine has joined #nixos
fabianhjr has quit [Ping timeout: 260 seconds]
<infinisil>
s/to/into
artistsvoid has quit [Ping timeout: 258 seconds]
<njha>
is there a log somewhere so I can see why one derivation failed
<infinisil>
Nixops on the other hand doesn't have a problem with that, since it can just pass the secret text to the deployment script/binary without going through Nix
<njha>
did a nix-build for an arm sbc and it has many thousands of lines of output
<infinisil>
njha: Do another build with `--max-jobs 1`. This will limit it to one derivation only, so at least output won't be overlapped
<emily>
what's the correct way to get the system identifier for the system to be built in a NixOS config? nixpkgs.crossSystem isn't always set, nixpkgs.system/nixpkgs.initialSystem seem deprecated or underdocumented, are you meant to get it from stdenv?
noudle has quit []
<cole-h>
infinisil: You finally changed the name!!! :D
<infinisil>
cole-h: Just now :3
<infinisil>
emily: Possibly through `pkgs.hostPlatform.system`
<emily>
infinisil: how does that relate to pkgs.stdenv.{host,build}Platform.system? :s
<emily>
why are there like 5 ways to look at the same thing? ;_;
<infinisil>
emily: Most likely the same, and pkgs.host/buildPlatform just mirrors the ones from stdenv
<njha>
it apparently builds for other people so I might be doing something obviously wrong
anderson_ has quit [Ping timeout: 260 seconds]
<infinisil>
njha: The project doesn't have a pinned nixpkgs, so there's really no guarantees what might happen for different people (which all could have different nixpkgs')
<infinisil>
"Builds on the nix release-20.03 branch." is in the readme I guess
<njha>
hmm I should be on 20.03
<njha>
can I force the build to run in qemu or something
<njha>
(to maybe solve the exec format error)
morgrimm has quit [Ping timeout: 240 seconds]
user_0x58 has joined #nixos
<infinisil>
Wouldn't know how, if at all possible
<ldlework>
infinisil: I think you would dig it.
<infinisil>
njha: Ah this is your repo, didn't even notice :P
<infinisil>
ldlework: Hmm not sure! I do like Nix for my builds, but it's really not the nicest language in general imo
<infinisil>
I haven't given it a try though, so maybe I'll like it after all :)
<infinisil>
njha: If you got it building at a previous nixpkgs version, just try to find that again, then you can bisect
jluttine has quit [Ping timeout: 265 seconds]
<njha>
👍 thanks
marusich has quit [Quit: Leaving]
<{^_^}>
[nixpkgs] @wchresta opened pull request #88645 → idris2: init at version 0.2.0-840e020 → https://git.io/Jfafq
anderson_ has joined #nixos
jluttine has joined #nixos
marusich has joined #nixos
h0m1 has quit [Ping timeout: 260 seconds]
<adyatlov>
Well people! I've got Crossover 19.0.0 working
<{^_^}>
[mobile-nixos] @samueldr opened pull request #149 → boot/gui: Use libffi rather than bespoke bindings → https://git.io/JfafB
<anderson_>
ldlework: as far as I understand, org is supported officialy only inside emacs. There are other implementations, but none of them have the same status...
<anderson_>
But there was a guy trying to port it to styx
<ldlework>
anderson_: huh?
<ldlework>
no, the guy is using emacs to parse org files
<ldlework>
it's a nix-based build system
<anderson_>
ldlework: oh, well... that's fine to me :)
<ldlework>
pulling in emacs, and running a bit of lisp is no problem for nix :)
<anderson_>
It can even be done in parallel, as some form of "emacs styx mode"!
<anderson_>
s/parallel/tandem/
<quinn>
a kernelPatch i'm adding right now requires a change to modDirVersion, is there a nice way to change that in the patch attrset?
<quinn>
ldlework: nix *can* do it, but basic org syntax is actually quite simple, it would probably be almost as easy to write a basic parser that discards the complicated stuff (most of which is probably not useful for styx) than get emacs to reliably do what you expect it to do, quickly.
<ldlework>
quinn: not sure what you mean by nix can do it, emacs is doing it, nix is just making it easy to use emacs as a tool
<quinn>
ldlework: i mean that having used emacs as a text editor, i would not trust it as an elisp runtime.
<ldlework>
uh ok
o1lo01ol1o has joined #nixos
eyenx has quit [Quit: bye]
eyenx has joined #nixos
eyenx has joined #nixos
eyenx has quit [Changing host]
LevelChart8 has quit [Quit: LevelChart8]
spacefrogg has quit [Quit: Gone.]
aw has quit [Quit: Quitting.]
aw has joined #nixos
spacefrogg has joined #nixos
<emily>
the claim "it would be easier to reimplement org-mode in Nix than to get the existing org-mode implementation to work" seems... strong...?
<emily>
org-mode is pretty big
o1lo01ol1o has quit [Ping timeout: 246 seconds]
<emily>
I wouldn't want to do even basic lightweight markup parsing in Nix frankly
<emily>
elisp might not be a great language but it's better for that than Nix
<ldlework>
the brilliant thing about using nix for your site generator
<ldlework>
is that you can bring in literally *any tool* to get whatever job you need done
<ldlework>
lol
<bqv>
Yeah… I'm on the use emacs side
<ldlework>
i think the org-mode work needs more work though, i don't think we should merge it tomorrow
<ldlework>
we should get it into a branch and someone will need to clean it up (imo)
<ldlework>
it's a good start though
felixfoertsch23 has joined #nixos
jybs_ has quit [Remote host closed the connection]
felixfoertsch has quit [Ping timeout: 272 seconds]
<anderson_>
quinn: maybe the guys on remacs aren't doing a better work for the foreseeable future?
<anderson_>
But truth to be said, Emacs is very clunky. A monster of this time.
<quinn>
i haven't used remacs, would you recommend it? (btw i'm not trying to start an editor war, i think emacs is great as an editor. just wary of it as a runtime)
cole-h has joined #nixos
<anderson_>
quinn: NO
<anderson_>
It is in a very alpha stage today.
avn has quit [Ping timeout: 272 seconds]
<quinn>
it's definitely very clunky, but i'm only frustrated about it being clunky because it's so amazing and if it performed like neovim i would be so so so happy to use it again.
<anderson_>
quinn: remacs is Emacs (re)written in rust.
<quinn>
i've heard of it, but i had the same understanding as you: that it's pre 1.0, which is all i need to know to not want to use it yet.
<cole-h>
But with cachix/install-nix-action, you get a full nix install, basically.
<superbaloo>
cole-h: I might steal some of that from you :)
<colemickens>
Oops. I think graham tweeted about it and posted a link to the toot here and I saw it.
<colemickens>
anyway, it's closer to what I hoped/expected for nix-shell :)
<superbaloo>
well, laurn has some more limitation than nix-shell
<cole-h>
superbaloo: Feel free :^) I like the idea of laurn as well, but haven't had a chance to check it out (mostly because I'm still building my NixOS system hehe)
<superbaloo>
cole-h: I should probably throw that in nixpkgs at some point in the future for people to experiment more easily
<superbaloo>
but there is still some more things I'd like to implement before
<superbaloo>
like I implemented something like direnv (hooks into bash and launch laurn when you enter a dir)
<superbaloo>
but I haven't implemented a whitelist yet
<cole-h>
But also keep in mind it'll probably be growing for a while. At some point, you gotta know when to say "yeah, this is OK for now" and cut a release or something ;^)
<superbaloo>
which might be useful :D
<superbaloo>
cole-h: yeah that's for sure, I'm just trying to remove some of the sharp edges before
chpatrick has quit [Quit: Connection closed for inactivity]
pamplemousse has joined #nixos
<hyper_ch>
nixos-unstable isn't building anymore :(
pamplemousse has quit [Ping timeout: 246 seconds]
proofofkeags has quit [Remote host closed the connection]
<quinn>
so i'm trying to write an overlay that provides a patched linux kernel, and i get a bunch of failed assertions (CONFIG_DEVTMPFS, CGROUPS, et cetera) when i try to apply it. https://hastebin.com/uxokubodiy.http (i have looked at linux/kernel/generic.nix, but i don't really see what i'm doing wrong)
<colemickens>
Is there a --keep-failed but for success? I want to see what else was in the tree aftre I built. My chromium is missing its desktop file.
<quinn>
colemickens: if you're making it with nix-build, it should be in ./result
armin has quit [Remote host closed the connection]
<colemickens>
hm, I thought that was just $(out) ?
<quinn>
oh, i think that is just $out.
<quinn>
honestly, i would probably add something to the last phase that is a guaranteed error
proofofkeags has joined #nixos
<colemickens>
heh, I'll just take a stab at where the desktop file is, I'll surely be wrong and we'll get a failure, iterate from there. good thing chromium builds fast :P
<quinn>
if you added it to current-system, it should be under /run/current-system/sw/share/applications
proofofkeags has quit [Ping timeout: 265 seconds]
<colemickens>
no I'm literally building chromium
<colemickens>
and need to copy it from the build output to $(out) in the derivation
lopsided98 has quit [Remote host closed the connection]
<quinn>
colemickens: honestly, i would add `find` to the buildInputs, then run find at the final phase and copy the output into ${out}.
<colemickens>
oooh :) I like that
<emily>
colemickens: chromium-git is documented as missing the desktop file in the PR fwiw
<emily>
(I was just looking at nixpkgs-chromium earlier)
<colemickens>
by me ? :P
<emily>
by volth
<colemickens>
I mean, I'm pretty sure its in the build output
jasongro` has joined #nixos
<colemickens>
volth's definition lacks the bits from the chromium derivation that copies the desktop file. I'm assuming it's in there, I just need to copy it.
<colemickens>
I could be wrong of course.
lopsided98 has joined #nixos
<emily>
right
<colemickens>
although, now that you say that, I remember the comment mentioned Selenium and I did notice when using it that Google acted weird about the UA...
* colemickens
mutters about it being a matter of some gn flag somewhere
<eadwu[m]>
quinn: Override the erroring options in `extraConfig` and that should work. You probably disabled something the default config set as `kernel.yes` but the option doesn't exist anymore because you changed the option it depended on.
<quinn>
eadwu[m]: i will register that i can do that, but it looks like common-config is just not getting pulled in at all, and i'd like to figure out why (rather than just monkey-patching the problem with my module), because i get this error even if i don't provide a structuredExtraConfig argument.
sarcasticadmin has joined #nixos
justanotheruser has quit [Ping timeout: 260 seconds]
<eadwu[m]>
The default kernel?
<quinn>
eadwu[m]: i mean i'm just running overrideAttrs on linux_5_6, so yes?
<quinn>
eadwu[m]: the error persists even if i comment out everything inside of overrideAttrs. if i have audio_linux_5_6 = unstable.linux_5_6.overrideAttrs (o: {}); and audioPackages = super.recurseIntoAttrs (super.linuxPackagesFor self.audio_linux_5_6);, i will still get the same error
est31 has quit [Remote host closed the connection]
user_0x58 has quit [Ping timeout: 260 seconds]
est31 has joined #nixos
<quinn>
eadwu[m]: well, it's too early to say, but it looks like switching to .override might have fixed my problem. i omitted my structuredExtraConfig (but the error was showing up without that anyways). i would still really appreciate knowing why this made a difference
drakonis_ has joined #nixos
<ldlework>
gchristensen: you around?
<ldlework>
hi drakonis
drakonis has quit [Read error: Connection reset by peer]
<quinn>
ldlework: you should be able to just callPackage to a nix expression, why are you putting an import in there?
<simpson>
emily: IOW (and sorry if this has already been covered) a flake is meant to "cut across concerns", and not just be part of a composite ports tree, but a collection of ports tree transformations.
mumuluxi has joined #nixos
<emily>
simpson: which is all well and good, but kind of weird when you in practice have to depend on some fixed nixpkgs branch and are expected to also export a bunch of leaf packages via the packages attribute,
<emily>
it is just weird that the intended way to use flakes has a lot of friction with the actual design of flakes
<simpson>
emily: Sure. I'm suggesting that there will *always* be friction. WP: "These concerns often cannot be cleanly decomposed from the rest of the system in both the design and implementation, and can result in either scattering (code duplication), tangling (significant dependencies between systems), or both." https://en.wikipedia.org/wiki/Cross-cutting_concern
wucke13 has joined #nixos
<ldlework>
quinn: OK I removed that, styx = super.callPackage /home/ldlework/ext/styx { };
<emily>
(and similarly, it's somewhat awkward to export your system's "nixpkgs + overlays" in a way that's accessible via the CLI, so either you do more hacky fuss, or you have no visibility into/access to your overlayed package set)
<ldlework>
quinn: i switched, but still nix search shows 0.7.1 instead of 0.7.2
<quinn>
mac10688: if you use home-manager (not a pat of nixos, but a related tool), it's under home-manager.users.<you>.xsession, but if you don't use it then i'm not sure.
Negher has quit [Ping timeout: 256 seconds]
<mac10688>
oh ok, never heard of home-manager, i'll give it a google
<mac10688>
thanks quinn
<quinn>
mac10688: FYI, i use wayland so there could be an xsession thing in normal nixos, i'm just not familiar
<quinn>
ldlework: want to hastebin the whole thing and i'll take a gander?
<mac10688>
seems like home-manager is for defining a configuration file for the specified user
<mac10688>
I kind of don't want to deal with another package right now, I feel overwhelmed already lol
<mac10688>
i think another package will complicate things and make me sad and want to go to sleep
<energizer>
home-manager adds some extra configuration options that make sense at the user level
zebrag has quit [Quit: Konversation terminated!]
<quinn>
mac10688: i don't want to be rude, but it's fair to say that nixos is on the hard end as distros go, if you're just learning linux it might end up being more limiting than helpful
<quinn>
i'm not saying you shouldn't do it, just something to be aware of
<energizer>
otoh sometimes starting out with the more restrictive tool is easier than having too much freedom
<mac10688>
thanks, you are probably right. But I just love the idea of defining a configuration file for my environment too much to stop. I'll just have to deal with it
jasongro` has quit [Remote host closed the connection]
cole-h has quit [Quit: Goodbye]
<quinn>
energizer: hmm. perhaps
LevelChart8 has quit [Quit: LevelChart8]
<energizer>
¯\_(ツ)_/¯
<quinn>
mac10688: fair enough. i understand the appeal, and share you zeal. that being the case, i would recommend you try editing the xmonad.config option earlier. (though actually IMO xmonad is also on the very hard end of window managers)
cole-h has joined #nixos
<quinn>
if you know haskell it's pretty nice, but still challenging even then, and if you don't it can be kind of unpleasant
zebrag has joined #nixos
zebrag has quit [Client Quit]
<mac10688>
I got into this whole thing to learn haskell and fp. XMonad is a fun challenge and hobby as of late
<{^_^}>
[mobile-nixos] @samueldr opened pull request #151 → initrd: Remove leftover hack from when isContainer was used → https://git.io/JfaId
jasongrossman has joined #nixos
alp has joined #nixos
<quinn>
oh well that's great. being a haskell guy myself it was still frustrating, but felt very satisfying and open-ended if buggy. i assume you're familiar with how to configure your background from your xmonad rc?
drakonis1 has quit [Read error: Connection reset by peer]
<mac10688>
no
butcher has quit [Ping timeout: 260 seconds]
<mac10688>
the xmonad website told me to look for the .xsessionrc file
<quinn>
mac10688: well i thought there was a way to do it in xmonad, but it looks like there isn't. i would just do as energizer said and make a ~/.xsessionrc. you can always manage it with home-manager later if you want.
<mac10688>
thanks, trying that now :)
mac10688 has quit [Quit: WeeChat 2.7.1]
Negher has joined #nixos
mac10688 has joined #nixos
cole-h has quit [Ping timeout: 256 seconds]
<calbrecht>
is it prefered to clear the __pycache__ on pythonPackages to have a deterministic output or to keep the __pycache__ for faster execution times?
mac10688 has quit [Client Quit]
mac10688 has joined #nixos
<energizer>
i think the point of that issue is pip will fix that nondeterminism shortly.
<energizer>
anyway the compilation time on normal size python packages is pretty minimal. you can benchmark it if you want
<energizer>
i have PYTHONDONTWRITEBYTECODE=1 in my profile personally
<calbrecht>
thanks energizer. ok so, i don't know the implications for the determinism thingy either. i get it, this is good for cache size on hydra, right? is it low in general or does it not matter so much?
smatting has joined #nixos
<calbrecht>
also i noticed in the case of psutils, when i try to mkShell the servo project, the mozilla mach system wants to rebuild psutils anyway...
<energizer>
it's probably quite small, but i'd guess the nixpkgs folks value determinism over all else anyway
<calbrecht>
ok, so i'll remove the __pycache__ in the virtualenv package, while at it.
<sphalerite>
I've put boot.debug1mounts in my kernel commandline in order to get a shell in the initramfs, but it's not working — when I press i on the serial console in hopes of getting an interactive shell, it prints "Starting interactive shell" and then nothing more happens. What might be amiss here?
<quinn>
sorry, i have no idea how to debug initramfs stuff. i would really appreciate a sanity check on that derivation i posted, i feel like i'm going crazy or something
<sphalerite>
quinn: that works for me
<sphalerite>
quinn: which nixpkgs version are you using?
<quinn>
sphalerite: i do, but it uses a pretty outdated kernel and is just a little too complicated for my liking.
<sphalerite>
ah ok
<quinn>
sphalerite: are you sure that you're actually evaluating the with super.lib.kernel? if i turn the option off that sets boot.kernelPackages to audioPackages it works fine, but i get 'attribute kernel missing'
<quinn>
(if i turn that option on)
<{^_^}>
[nixpkgs] @Mic92 closed pull request #88594 → PR to test CI error messages → https://git.io/Jf2EX
<quinn>
sphalerite: so it's available in nix repl if i import nixpkgs-unstable, but...the file is still present in nixos-20.03 which is what i use. i can try to just steal it from unstable but that's kind of a messy solution. do you have any ideas what's going on?
<sphalerite>
quinn: aah, lib.kernel was introduced in afa0e02d64b1 and said commit is not in 20.03
<sphalerite>
quinn: you can use the unstable one or import kernel.nix manually. I think the former is nicer
meh` has joined #nixos
justanotheruser has quit [Ping timeout: 244 seconds]
<quinn>
sphalerite: well i'm kind of mystified by the file's presence if it wasn't in the namespace, but i think it's already occupied enough of your time and i need to sleep, so i'll just say thank you Very much, this was making me really crazy
<quinn>
sphalerite++
<{^_^}>
sphalerite's karma got decreased to 88
<{^_^}>
Wait no, it got *increased* to 90
<sphalerite>
quinn: it was previously just imported manually wherever it was used
lord| has quit [Ping timeout: 260 seconds]
<sphalerite>
e.g. pkgs/os-specific/linux/kernel/common-config.nix: with import ../../../../lib/kernel.nix { inherit (stdenv) lib; inherit version; };
<quinn>
haha, what the hell? i considered it but that seemed so clearly deficient that i immmediately discounted the thought
lord| has joined #nixos
<sphalerite>
hehe
<sphalerite>
good thing it's been improved!
jco has joined #nixos
<sphalerite>
teto++
<{^_^}>
teto's karma got increased to 7
<sphalerite>
(in teto's absence)
quinn has quit [Ping timeout: 265 seconds]
obadz has quit [Quit: WeeChat 2.8]
<{^_^}>
[nixpkgs] @Mic92 opened pull request #88669 → nixos/high-dpi: Reasonable defaults for high-density displays → https://git.io/JfamG
pjt_014 has quit [Ping timeout: 258 seconds]
agsdheidjd has joined #nixos
butcher has quit [Ping timeout: 260 seconds]
<{^_^}>
[nixpkgs] @vcunat pushed 3 commits to release-20.03: https://git.io/Jfaml
<yutyo[m]>
<quinn "colemickens: i mean if the distr"> Installing seems easy after seeing its hardness to use it.
<yutyo[m]>
<multun "^"> Weird enough, my first impression while installing was good. It was bad when driver problems came in, that I have spent a month to solve.
<adisbladis>
hyper_ch: I'll just ask ofborg to run the tests before merging
<rsoeldner>
Can someone provide a simple example using `nginx` with letsencrypt ? I added `enableACME=true;` to the `services.nginx` service but to my understanding this is a placeholder cert and will be automatically replaced, which never happend so far
<hyper_ch>
adisbladis: no worries
orivej has joined #nixos
<hyper_ch>
I know some people who got bitten by the vfs_io_urin corruption
<bqv>
emily: oh, no, for containers I just make the config inside the conatiner as thin as possible, for now. Can't remember if it's potentially fixable by overriding the submodule
hmpffff has quit [Quit: nchrrrr…]
orivej_ has quit [Quit: No Ping reply in 180 seconds.]
<Taneb>
I'm trying to modify the derivation for vbam to enable the wx GUI. I've added wxGTK to the build inputs but it's failing to find the wx headers.
orivej has quit [Ping timeout: 260 seconds]
orivej has joined #nixos
<Taneb>
The error message gives me a lot of paths, which include /nix/store/blah-wxGTK-2.8.12.1 but not /nix/store/blah-wxGTK-2.8.12.1/include, where the file in question can be found
aveltras has joined #nixos
<Taneb>
Any idea what I could be doing wrong?
<Taneb>
(it is looking in, for example, /nix/store/blah-zip-3.0/include)
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
<jakobrs>
now I just did nix-env -iA nixos.plasma-browser-integration, and for some reason, winebrowser is the default browser now, despite not having wine installed
<jakobrs>
And nearly everything (xhtml+xml, plaintext, etc) opens in libreoffice
hmpffff has joined #nixos
quinn has quit [Ping timeout: 260 seconds]
<jakobrs>
Setting mime types manually using xdg-mime does not work.
butcher has quit [Ping timeout: 260 seconds]
orivej_ has quit [Ping timeout: 265 seconds]
orivej has joined #nixos
quinn has joined #nixos
<jakobrs>
Is there any documentation on MIME in NixOS anywhere?
<Taneb>
Hmm, I think I'm seeing weirdness with CMAKE_PREFIX_PATH
<jakobrs>
I thought I should point out that nix flake clone fails if you do not have SSH keys set up.
<NobbZ[m]>
I'm playing around with `dockerTools.buildLayeredImage`, how can I add current nixchannel of the buildhost (or a given specific commit) to the container?
quinn has quit [Ping timeout: 265 seconds]
phnom has quit [Ping timeout: 256 seconds]
<bqv>
Presumably most people using nix hate docker.
<NobbZ[m]>
Docker is currently the only way I could use to sneak `nix` to my workplace…
<bqv>
NobbZ[m]: it builds whatever you tell it to. Channels are out of scope by then
<calbrecht>
beeing calbrecht: start building rust servo, get sidetracked over mozillas python mach buildsystem and end up fixing virtualenv *sigh lol
Maxdamantus has quit [Ping timeout: 260 seconds]
<NobbZ[m]>
I'm trying to build a container for now that builds my `home-manager` config in CI. I need the channels for that. Of course I could add and update in the entrypoint, though I'd prefer to have them baked in and "pinned"
Maxdamantus has joined #nixos
teto has joined #nixos
o1lo01ol1o has joined #nixos
<calbrecht>
i mean, i could need some help. because i don't even know if i am on the right track now with this problem #66366
<{^_^}>
channels redefined, was defined as Largest Nix-related channels: #nixos, #nixos-dev, #nix-darwin, #nixos-chat, #nixos-aarch64, #nixos-wiki, #nixos-borg, #nixos-security, #nix-core
<adisbladis>
Building docker containers with Nix inside is a bit more involved than just using buildLayeredImage
ent has quit [Ping timeout: 272 seconds]
<gchristensen>
we're gonna need to paginate soon, or justredirect to a wiki page
datakurre has quit [Ping timeout: 260 seconds]
<NobbZ[m]>
Yeah, maybe its the wrong tool. As I said, playing around to learn, though I'm not sure if `buildImage` would work either.
jesushelpthisqwe has joined #nixos
<adisbladis>
NobbZ[m]: I did some work a few weeks back towards replacing the official Docker image with one produced by buildLayeredImage https://github.com/adisbladis/nix-docker
<adisbladis>
That one includes the channel that was used to build the image
<sshow>
can anyone point me to a package with a systemd service using a config file and systemd runtime directory?
<{^_^}>
[nixpkgs] @orivej-nixos pushed commit from @r-ryantm to master « psi-plus: 1.4.1159 -> 1.4.1231 (#88674) »: https://git.io/JfaWC
<infinisil>
Not sure what you mean, "unstable" in the bot here just points at an unstable nixpkgs
<Tritlo>
I mean, how did you know it was under haskell.compiler?
<infinisil>
Ah, just knew about it from working with it before
<Tritlo>
ah okok
<Tritlo>
Is there any way to figure it out from looking at the repo? I tried unstable.haskellPackages.compiler, since the file is called haskell-packages, but that didn't work
waleee-cl has joined #nixos
proofofkeags has joined #nixos
mallox has quit [Quit: WeeChat 2.8]
<infinisil>
Tritlo: If you search through nixpkgs for "haskell-packages.nix", you'll find that it's defined in all-packages.nix
<infinisil>
And that the haskell attribute points to it
<numkem>
I have hydra setup in my environment and it's working well. I import the channel and I can use it for one project. For the other when I try to use a package from that channel I get the error path `/nix/store/[...] does not exist and cannot be created` what could be different to create this problem? I have nix-serve with signing working on the hydra host as well.
<clever>
numkem: if nix cant find something in a binary cache (hydra/nix-serve), it wont bother checking again for an hour
nek0 has quit [Remote host closed the connection]
<clever>
numkem: `nix show-config | grep narinfo` you can adjust these options with --option
nek0 has joined #nixos
andreoss has joined #nixos
slack1256 has joined #nixos
proofofkeags has quit [Ping timeout: 256 seconds]
<{^_^}>
[nixpkgs] @Aver1y opened pull request #88692 → libusb1: Fix for pkgsMusl → https://git.io/Jfa89
Jackneill has quit [Ping timeout: 264 seconds]
<andreoss>
why nixos-generate-config can only be run from the live-cd, and not from a foreign distro which happens to have nix installed?
<clever>
andreoss: you can forcibly install it with nix-env, but a lot of the install stuff assumes certain kernel features you may not have
Jackneill has joined #nixos
dingenskirchen has quit [Remote host closed the connection]
dingenskirchen has joined #nixos
proofofkeags has joined #nixos
asbachb has joined #nixos
<asbachb>
Hi. Someone has a working configuration for a systemd restart when touching a file?
<pistache>
how can I make a wrapper for a program that needs to needs to exec(argv[0]) ?
<pistache>
should I use security.wrappers for that ?
<kloenk>
does nixos need the systemd-remount-fs.service unit? because if I see it correctly nixos mounts / the right way via the fileSystems module, which are then later written into /etc/fstab
<{^_^}>
[nixpkgs] @doronbehar opened pull request #88693 → gnuradio: format arguments and inputs → https://git.io/JfaBL
<cole-h>
pistache: Do you mean like it checks the name of the binary for different features or something?
fendor_ has quit [Quit: Leaving]
fendor__ has quit [Quit: Leaving]
abathur has quit [Quit: abathur]
<pistache>
cole-h: yes, it uses argv[0] to re-execute it self on reload, which allows seamless reloads, but that doesn't work on NixOS since the new binary will be at another path
<pistache>
so I'd like to execute it from a symlink to the real binary, but I'm not sure if I should make my own activationScript of there is already something in NixOS that I can use for that
<pistache>
security.wrappers seems to be only for setuid/capabilities wrappers
veleiro has joined #nixos
<kloenk>
pistache: im not sure, but maybe sway does something like it, but on there own. sway does not start if you have /var mounted as a tmpfs becaus of the execute bit in tmpfses
asymptotically has quit [Remote host closed the connection]
<symphorien>
pistache: looks like a good candidate for a patch...
<pistache>
symphorien: for a patch to what ?
<symphorien>
make it exec its name from PATH instead of argv[0]
<pistache>
well it's not in PATH anyway as it's used in a NixOS module
<symphorien>
ah right
<pistache>
the program is haproxy, which can handle keeping connections open while reloading
<pistache>
I think I could use a wrapper from security.wrappers and set capabilities to "", but that feels a bit hacky
<cole-h>
pistache: `makeWrapper` is probably what you want, but it's not declarative -- it has to be done at build time.
<asbachb>
eadwu[m]: Is there documenatation on "X-Restart-Triggers"?
<pistache>
cole-h: I don't understand how could makeWrapper help ? the wrapper's path would still change, no ?
veleiro` has joined #nixos
<cole-h>
How do I get `man 2 fork` to work someone please help
<cole-h>
pistache: Yeah, you're right.
<cole-h>
A tricky problem...
veleiro has quit [Ping timeout: 260 seconds]
<{^_^}>
[nixpkgs] @AndersonTorres opened pull request #88694 → lnch: init at 2017-02-16 → https://git.io/JfaB5
<pistache>
ah, using security.wrppers with capablities = "" doesn't even work, since that makes pcap fail
jhillyerd has quit [Quit: WeeChat 2.7.1]
<symphorien>
if you need a global fixed path, use /run/current-system/sw/bin/haproxy
alp has joined #nixos
user_0x58 has joined #nixos
<pistache>
symphorien: but that implies adding the package to environment.systemPackages, which doesn't feel right for a package used in a module
<symphorien>
well if you think of it, the idea of "security wrappers without caps" boils down to the same.
<symphorien>
just with /run/wrappers/bin instead
<pistache>
yes, it's true
<pistache>
I could write my own activationScript, but then I'm not sure where would I put that global path
<symphorien>
if really you don't feel comfortable with it, make a symlink in /var/lib/haproxy
<symphorien>
and then the reload dance would be: symlink the new haproxy in /var/lib/haproxy and then have the old one exec it.
<symphorien>
I doubt you need an activation script, systemd tmpfiles should be enough
veleiro`` has joined #nixos
<pistache>
ah yes I don't need an activation script, I can just build the symlink in the service itself
<pistache>
still not sure where to put it, /var/lib/haproxy could be a problem if /var is mounted noexec
veleiro` has quit [Ping timeout: 256 seconds]
<symphorien>
/run maybe ?
<pistache>
ah right in /run/haproxy/, that seems like a good compromise
<{^_^}>
[nixpkgs] @AndersonTorres pushed 2 commits to master: https://git.io/JfaRE
<niso>
by chance someone knows of an example using runInWindowsVM?
KeiraT has quit [Ping timeout: 240 seconds]
drakonis1 has joined #nixos
asymptotically has joined #nixos
drakonis_ has quit [Ping timeout: 246 seconds]
KeiraT has joined #nixos
o1lo01ol1o has joined #nixos
drakonis_ has joined #nixos
eoli3n_ has quit [Remote host closed the connection]
eoli3n_ has joined #nixos
<eyJhb>
talyz: will check on reboot if it works :p
<eyJhb>
Any good way to disable the .sudo_as_admin_successful file in NixOS?
abathur has joined #nixos
quinn has joined #nixos
drakonis1 has quit [Ping timeout: 260 seconds]
symphorien[m] has joined #nixos
xd1le has quit [Ping timeout: 256 seconds]
CMCDragonkai1 has quit [Ping timeout: 265 seconds]
quinn has quit [Ping timeout: 272 seconds]
rsoeldner has joined #nixos
xd1le has joined #nixos
gustavderdrache has joined #nixos
<rsoeldner>
Hi, can anyone provide a simple example with a machine configuration (for nixops) with nginx and tls\ssl cert. I have set `enableACM = true;`. It looks like a default cert is generated but to my understanding letsencrypt will generate one and a service is later exchanging these certs ?
<asbachb>
Is x-restart-triggers a nixos construct? I don't find systemd documentation.
<pistache>
asbachb: yes
<pistache>
asbachb: it is handled by NixOS's activation scripts, to restart the service when the triggers change
<kloenk>
didn't found any file beside the path itself and the .drv file
xd1le has quit [Ping timeout: 260 seconds]
fmcs has joined #nixos
proofofkeags has quit [Remote host closed the connection]
xd1le has joined #nixos
proofofkeags has joined #nixos
LevelChart8 has joined #nixos
shafox has quit [Remote host closed the connection]
<hyperfekt>
the file for fetchzip mentions that the checksums for github archives can change, is this still true?? :o
fusion809 has quit [Remote host closed the connection]
codygman has joined #nixos
cr4y1 has quit [Ping timeout: 246 seconds]
fendor has quit [Ping timeout: 265 seconds]
proofofkeags has quit [Ping timeout: 256 seconds]
Habbie has joined #nixos
<infinisil>
kloenk: With the experimental `nix` command there's `nix path-info --sigs`
<Habbie>
hello! I just installed Nix on Debian Buster, and I'm trying `nix-shell '<nixpkgs>' -A powerdns` as suggested by the docs, but i don't get a fresh shell, there is no output (except the first time when it downloads a lot of things), and it exits 1
<Habbie>
how can i debug this?
xd1le has quit [Read error: No route to host]
zebrag has quit [Quit: Konversation terminated!]
<jakobrs>
Hm, it seems the rust code was committed directly to master
<jakobrs>
at least I can't find any relevant branches or prs
zebrag has joined #nixos
codygman has quit [Ping timeout: 256 seconds]
<jakobrs>
Habbie: try nix-build '<nixpkgs>' -A powerdns
<jakobrs>
no wait
<Habbie>
also, what does '<nixpkgs>' mean there?
<jakobrs>
That's an ... interesting error
<jakobrs>
'<nixpkgs>' is the standard package repository
xd1le has joined #nixos
<jakobrs>
`nixpkgs` is the channel name
<jakobrs>
Are you sure it exits 1? Also, what shell do you use?
<Habbie>
my shell is bash
<Habbie>
i'm sure it exits 1
<infinisil>
jakobrs: nix-shell always uses bash
<Habbie>
if i run it under strace it behaves differently in a way i cannot immediately explain
<jakobrs>
infinisil: I meant what shell nix-shell was launched *from*
<Habbie>
this process exists when running nix-shell under strace
<jakobrs>
This is definitely abnormal behaviour, `nix-shell '<nixpkgs>' -A powerdns` works fine for me.
<Habbie>
and ctrl-D stops strace and that process
dingenskirchen has quit [Remote host closed the connection]
dingenskirchen has joined #nixos
<jakobrs>
Does nix work right otherwise? Are you able to do, say, `nix-shell -p vim`?
<Habbie>
same - first time a download, then exit 1. second time, silence, and exit 1
<jakobrs>
... does nix-build work? or nix-env?
<jakobrs>
or `nix run`?
<Habbie>
i don't yet know how to use those commands, sorry
<jakobrs>
Try:
<jakobrs>
(one of)
<jakobrs>
nix run nixpkgs.vim # should start a shell with vim available
<jakobrs>
nix build nixpkgs.vim # should download vim
<jakobrs>
nix-build '<nixpkgs>' -A vim # the same
<jakobrs>
nix-env -iA nixpkgs.vim # install vim
<Habbie>
which vim -> /nix/store/p7w8v85cqs4p9a5b3a2hbli54lhy2jfv-vim-8.2.0510/bin/vim
<jakobrs>
Erm, I'm sorry, but I don't think I have the time to help right now, sorry. Try to get help from someone else here, or possibly open an issue, because this is *definitely* not intentional behaviour.
<Habbie>
'nxi run nixpkgs.vim' makes vim available, so that works - something is preventing nix from showing [nix-shell] in my shell prompt, it appears
jhillyerd has quit [Client Quit]
<jakobrs>
`nix run` is essentially a reimplementation of `nix-shell -p`, so it seems likely to me that it's a bug with nix-shell (and the other "old" nix commands)
<jakobrs>
Based on that I'd probably open an issue.
<Habbie>
i killed off liquidprompt in my .bashrc and now 'nix-shell -p bash' shows [nix-shell]
jakobrs has left #nixos ["WeeChat 2.7.1"]
<Habbie>
ok, i'm pretty sure liquidprompt is causing this pain
<fresheyeball>
but it says plugin's is an unexpected arguement
<cole-h>
`rofi.override`, not overrides (unless that was a typo)
<fresheyeball>
oh
<cole-h>
adisbladis: 👀
<fresheyeball>
let me try that
<clever>
error: Package ‘HTF-0.14.0.3’ in /home/clever/apps/nixpkgs-older/pkgs/development/haskell-modules/hackage-packages.nix:9295 is marked as broken, refusing to evaluate.
<{^_^}>
anonymous function at /var/lib/nixbot/nixpkgs/nixos-19.09/repo/pkgs/applications/misc/rofi/wrapper.nix:1:1 called with unexpected argument 'plugins', at /var/lib/nixbot/nixpkgs/nixos-19.09/repo/lib/cu...
<adisbladis>
cole-h: I've wanted transmission-qt in nixpkgs for years
<{^_^}>
anonymous function at /var/lib/nixbot/nixpkgs/nixos-20.03/repo/pkgs/applications/misc/rofi/wrapper.nix:1:1 called with unexpected argument 'plugins', at /var/lib/nixbot/nixpkgs/nixos-20.03/repo/lib/cu...
<cole-h>
Huh
<infinisil>
Also fails with 20.03, so that plugins argument seems to be an unstable thing :)
veleiro`` is now known as veleiro
<cole-h>
Seems like it didn't make it into 20.03, then :)
<fresheyeball>
cole-h: ok I am trying rofi with plugins against nixpkgs master
<fresheyeball>
that worked
<cole-h>
:)
tobeportable has joined #nixos
<fresheyeball>
cole-h++
<{^_^}>
cole-h's karma got increased to 59
orivej_ has quit [Ping timeout: 258 seconds]
orivej has joined #nixos
jklmh has joined #nixos
raiaq has joined #nixos
agsdheidjd has joined #nixos
anderson_ has joined #nixos
fresheyeball has quit [Quit: WeeChat 2.7.1]
<numkem>
clever: I've come back to it after a much longer than the negative timeout and it's still happening. Is it possible that nixops does something different and bypasses the binary caches?
<{^_^}>
nix#3431 (by pmiddend, 9 weeks ago, merged): installer: also test for xz to unpack
andreoss has quit [Ping timeout: 258 seconds]
<numkem>
clever: when doing the deploy itself. But trying to get the outPath variable from the package in nix repl does the same thing. You can tell it's trying to download something but it fails. I guess it's unrelated to nixops
<clever>
numkem: what is the error it gives in repl?
craftyguy has joined #nixos
<bqv>
> pkgs.python3Packages.pexpect.version
<{^_^}>
"4.7.0"
<numkem>
clever: path does not exists and cannot be created
<Henson>
I'm trying to use "nix-store --import" and I'm getting a "path .... is not valid" error. It works on 2 other machines, but not on a third. Any suggestions?
<numkem>
clever: only happens with one of the project in my hydra
<clever>
numkem: and which path are you trying to build?
<numkem>
clever: does that help? when I try to do nix-env -iA <package> it says: warning: name collision in input Nix expressions, skipping...
<clever>
numkem: thats a different error, about you having colliding channel names between your user and root
agsdheidjd has quit [Ping timeout: 260 seconds]
<numkem>
clever: the path should already be built by hydra and in the channel. It is present in hydra's nix store
<numkem>
clever: in this case /nix/store/c4iqxx69kv5ln5b9jw84fmg710f6cwy1-testsuite-0.0.1
<clever>
numkem: and does `nix show-config | grep substituters` show the hydra?
drakonis_ has quit [Read error: Connection reset by peer]
<numkem>
clever: no... wtf
hmpffff has joined #nixos
<clever>
numkem: then you simply never added the hydra to nix.conf
<numkem>
clever: and cache.nixos.org is there twice
<gordon>
hi
<numkem>
clever: I did do it in configuration.nix as nix.binaryCaches
<Henson>
what happens if I try to nix-store --import something that is not a complete closure? Can I import arbitrary store paths, or do they need to have all of their dependencies?
<cole-h>
clever: Do you happen to know the difference between `nix.binaryCaches` and `nix.trustedBinaryCaches`?
<cole-h>
Is `trustedBinaryCaches` used for everybody, while `binaryCaches` is just for trusted users?
<clever>
cole-h: trustedBinaryCaches lets a user opt-in by using `--option substituters` without that user being trusted
<clever>
cole-h: binaryCaches is the list of default caches, which are trusted automatically
<gordon>
i’ve written a derivation for borgmatic, a kind of front-end for borgbackup. Borgmatic doesn’t directly depends on borgbackup, but it will call it. I’ve put borgbackup as a dependency, but the borg command isn’t in my path (because it’s an indirect pull in my env, i think). But in my borgmatic conf file, i’d have to put the path to the borg command, what’s the best way to do this properly?
<cole-h>
clever: Gotcha, thanks. I'll put my personal cachix into `binaryCaches` then :^)
<Henson>
clever: I'm deploying to a remote machine. Some of those paths come from the nix cache. Other paths are built on my computer. I tried copying the paths not in the nix cache into a NAR file that I can download from a fast server and import, instead of my local computer sending it over my slow internet. Importing the NAR file first was the problem, because dependencies had to be downloaded...
<Henson>
clever: from the nix cache first. After downloading them I was able to import the NAR file and skip sending the stuff over my slow internet connection.
<clever>
Henson: then you want `nix copy --to local?root=/mnt/` or `nix copy --to file:///mnt/`
dxtr[m] has joined #nixos
<clever>
Henson: both of those will create a directory containing a closure, which you can then --from on the other end
orivej_ has quit [Ping timeout: 256 seconds]
<Henson>
clever: will that have the same missing dependencies problem as nix-store --import? Or will it just copy the store paths over regardless?
orivej has joined #nixos
<clever>
Henson: it will always copy every dependency
<Henson>
clever: right, but I don't want it to copy every dependency. I want some to be downloaded from the nix cache, and the ones not in the nix cache to come from the NAR file.
<clever>
Henson: that gets complex real fast because you dont know what paths the machine is lacking, and what paths may be on the cache
<Henson>
clever: true, but I'm making certain assumptions. I don't mind if my computer has to upload some extra that are missing, but I know the bulk of them (about 750 MB) are going to be needed. Instead of uploading 750 MB from my computer at 100 kB/s I want to download the NAR file at 2 MB/s and then upload whatever little bits were missing.
<clever>
Henson: you can just give nix copy the path to the start of that 750mb closure
<numkem>
clever: ok, I'm getting a different behavior when adding my binary cache to extra-substituers... Now i have 2 questions. Why is binaryCaches not affecting it but works during a nixos-rebuild swich? And any examples on how to not bork the current trusted keys (got an error saying it wasn't valid key)?
<clever>
and just exclude the other things
<clever>
numkem: the cache.nixos.org key is in the default list, so you loose it upon setting any extra keys
<clever>
numkem: you need to lookup that default key, and add it back yourself
<Henson>
man, being able to do "nixos-rebuild test" is so awesome. I can park a "sleep 10m;reboot" in a screen session on a remote computer, then atomically reconfigure its networking, then test to make sure it works before cancelling the reboot and running "nixos-rebuild switch". So awesome.
<pistache>
Henson: you can also use --rollback to avoid the reboot
<clever>
Henson: you can also just resolve the /run/current-system symlink before you test, then run $THATLINK/bin/switch-to-configuration test, to switch back to whatever it was before
<clever>
--rollback can have unpredictable affects sometimes when ran repeatedly
<Henson>
pistache: the reboot is there in case the "nixos-rebuild test" renders the system unreachable. Then it will reboot into the previous reachable state.
<clever>
Henson: but rollback can revert it back to the previous state as well
<energizer>
clever: is there a rebuild command where i can just pass it a hash or a generation number that doesn't require me to go looking in /nix/var/?
<bqv>
well if you didn't note down which generation was what, then obviously you're not going to be able to
<numkem>
clever: what if I have a port? I've tried putting the same key I gave in nix.binaryCachePublicKeys but it's not accepting it
<numkem>
clever: the cache is working if I disable key checks
<superbaloo>
I guess nix-instanciate -v would do
<clever>
numkem: the port isnt part of the key
<clever>
numkem: the name in the key, must match the name in the signatures
orivej has quit [Ping timeout: 240 seconds]
<numkem>
clever: in my case it's hydra, but it's not happy with it somehow
<leungbk>
I'm having trouble using nodePackages.create-react-app. Would anyone happen to know of any quirks with the package on NixOS? Here's the error I'm getting: http://paste.debian.net/1148471
<energizer>
there was a project for tightly integrating nixos with git, anybody remember what it's called?
<clever>
numkem: so the public key must begin with cache.nixos.org-1:
<numkem>
clever: ok, it is hydra if I swap that with mine
<clever>
numkem: the name comes from when you ran nix-store --generate-something
<ldlework>
It would be cool if there was a way to load strings from a file, such that embedded nix expressions get interpolated/evaluated as if they were lexically locally to where that text would be loaded from
<ldlework>
like as if you had a string literal instead
<clever>
ldlework: builtins.scopedImport
<ldlework>
:O
<numkem>
clever: the signing works just fine with nix.BinaryCaches (only by nixos-rebuild?)
<clever>
ldlework: its like import, but it can temporarily mutate the globals for the imported thing
<pistache>
does anyone know if the functions in the Python test driver are thread-safe ?
<clever>
ldlework: this example overrides the __mul function, causing `1 * 2` to return boo
<ldlework>
clever: ah but you still have to respecify all thvariables
<ldlework>
clever: nix really promotes long swaths of bash just stuck in the code as a literal string because it's more conveinent to get interpolated values into the string that way
<cole-h>
Does anybody know how to prevent kernel messages from bleeding into my TTYs?
<bqv>
yes
<clever>
ldlework: there is also substututeAll and pkgs.substituteAll
<ldlework>
clever: the basic problem is you're in some function right, there's a bunch of local variables, then there's a string with 200 lines of bash, which have embedded nix expressions with all the variables in scope etc
<cole-h>
bqv: :(
<bqv>
(was searching my shell history)
<bqv>
i remember a command to fix that
<ldlework>
there's no way to load a string "as if" you typed it literally, without actually typing it literally
<cole-h>
Was it a sysctl one?
<bqv>
edit a file in /sys to be "4 3 4 8" or something
<bqv>
yeah
<bqv>
that!
<bqv>
whatever that is
<ldlework>
everything will have you pass all the variables into whatever is doing the loading
<numkem>
clever: I've tested through configuration.nix to import my channel and install something in systemPackages and it works as expected. Is there a way to pass a binary cache to nixops if nix.binaryCaches aren't used by default?
<ldlework>
which is typical, but i just notice that it results in lots of long bash string literals
<ldlework>
anderson_: however, since there are so many repos, I'm wondering if it makes sense to have a separate organization, but I'm not sure.
<clever>
numkem: is the remote machine using any caches at all?
<ldlework>
clever: yes, the issue is having to respecify all the variables. with a literal string, they're already local, so you just write the local literal string.
<{^_^}>
[nixpkgs] @orivej-nixos pushed 3 commits to staging: https://git.io/JfagI
<ldlework>
with anything else, you'll have to pass all the variables that should be replaced, etc
<anderson_>
ldlework: I think not. There are only two relevant ones as far as I looked: the website and the software itself.
<numkem>
clever: none specified no. I'm trying to deploy using nixops to the hydra machine itself
<clever>
numkem: i think you want to adjust deployment.hasFastConnection
<cole-h>
Is there a way to remove the "run `nixos-help for the NixOS manual" message from the TTYs? I just want the login prompt and the NixOS version header stuff there.
LevelChart8 has quit [Quit: LevelChart8]
<anderson_>
ldlework: The other repos are just styles and themes, I think they can be "merged".
<ldlework>
a magic function that let you load a string from disk while interpolating based on the lexical scope of the call site would obviate whatever motivates so much long bash string literals
<clever>
numkem: ah, i would expect the hydra machine to already have the paths in question?
<clever>
ldlework: yeah, thats not really possible with current nix
<ldlework>
anderson_: it means they cannot be managed and released independently
<numkem>
clever: indeed, and they are there. So it's my local machine that is complaining because it doesn't have those paths
<ldlework>
clever: yeah i know, i was more observing that since it can't, we got lots of really long bash string literals all over :)
<clever>
numkem: yeah, the nixops machine will want the full closure on itself before it tries to deploy
<clever>
numkem: so you need to setup nix.conf on that local machine first, to be able to pull everything from hydra
<numkem>
clever: which is currently failing because of the weird problem with the trusted-public-keys
<clever>
numkem: if you curl the narinfo like in the example above, what does the signature say?
<bqv>
,locate bin tree
<{^_^}>
Found in packages: tree
<bqv>
..oh
<ldlework>
anderson_: btw, I create #nixos-styx for us
<numkem>
clever: Sig: hydra:[...] (longer than the public key)
<anderson_>
ldlework: joining!
<clever>
numkem: yeah, so your pubkey in nix.conf has to be in the form of hydra:<base64>
<clever>
numkem: the command that generated the secret should also have made a public at the same time
<numkem>
clever: I'm copying the same value as I put in nix.binaryCachePublicKeys
<numkem>
clever: using that key through configuration.nix downloads the paths from the cache just fine
<clever>
numkem: what is the exact error your getting?
<cole-h>
That help message shows up due to `documentation.doc.enable = true;` -- just gotta `services.mingetty.helpLine = lib.mkForce "";`
<numkem>
clever: substituter 'http://172.31.99.3:5000' does not have a valid signature for path '/nix/store/c4iqxx69kv5ln5b9jw84fmg710f6cwy1-testsuite-0.0.1'
<clever>
numkem: that implies that you got the public key wrong, any change you used the name hydra for 2 different keys?
<clever>
numkem: also, hydra has its own nix-serve built in, on the same port as the UI, you dont need nix-serve running seperately
<numkem>
clever: it's a copy-paste straight. I only have that one key
<numkem>
clever: ha! well I'll take that out. Where is that public-key set?
<craftyguy>
I created a simple package that uses meson/ninja to build and install, among other things, a conf file under sysconfdir (e.g. /etc on other *nix systems). I can see it exists under /nix/store/<hash>/etc/<file>, but when I install with nix-env, it doesn't show up under /etc. what am I missing?
<craftyguy>
other files are installed where they should be, e.g. the binary, manpages
<clever>
craftyguy: nix-env can only ever modify ~/.nix-profile/
<clever>
craftyguy: for things to get into /etc/, it must be done from a nixos module like configuration.nix
<numkem>
clever: and you've put that file by hand there?
<craftyguy>
clever: I see. so I need to figure out how to make my package installable by configuration.nix
<clever>
numkem: `nix-store --generate-binary-cache-key` was just ran directly on hydra, that is the output path it was given
<craftyguy>
or, submit a MR to add this to the nixpkgs repo
<numkem>
clever: as root or as they hydra user?
<cole-h>
Is there a way to do `systemctl --user=someuser restart unit` and restart a certain user's unit, *without* switching user with something like doas or sudo?
<clever>
numkem: probably as the hydra user
<numkem>
clever: that's what I did previously, let's see if I could re-use the same key
<craftyguy>
clever: I thought (probably wrongly) that ~/.nix-profile would be an overlay onto /etc for the user
<craftyguy>
what's the point of installing to ~/.nix-profile/etc if most programs don't know that sysconfdir is ~/.nix-profile when built with nix-build?
<clever>
craftyguy: for config stuff, you usually run the program with `--config-file /path/to/config`
<craftyguy>
that's obviously application-dependent. and many apps look to /etc for some default config if nothing exists in, e.g. XDG_CONFIG_HOME
<clever>
craftyguy: for defaults, you usually patch it to read from $out/etc/, the same place it installed to
<clever>
craftyguy: that often happens automatically, because the sysconfdir you passed to it, gets baked into the code
<craftyguy>
maybe that's what I'm missing then. I need to tell meson somehow in the default.nix that sysconfdir is '$out/etc'?
<clever>
craftyguy: it usually happens via the same --prefix=$out logic
<clever>
that would have put things at /usr/local/etc/ on another distro
<craftyguy>
right. I was using the 'defaults' taht nix has for running meson, which I guess sets it to /etc
<clever>
mac10688: it will source ~/.xprofile if it exists, and then exec ~/.xession if it exists (which blocks the remaining normal logic)
<mac10688>
ok i'll check that out, thanks clever
<clever>
numkem: extraOptions wont merge duplicates correctly, look at the resulting /etc/nix/nix.conf
lorimer has quit [Ping timeout: 246 seconds]
wnklmnn has quit [Ping timeout: 260 seconds]
<clever>
numkem: you must use nix.binaryCachePublicKeys for it to merge right
<numkem>
clever: yet my key is already in there
<clever>
numkem: do you have duplicate trusted-public lines in nix.conf ?
<craftyguy>
I tried adding mesonFlags=["-Dsysconfdif=$(out)/etc"], but it just installs to a literal "~/.nix-profile/$(out)/etc", lol
<eyJhb>
Since #home-manager seems dead to this question, maybe someone here knows it
<eyJhb>
Currently getting the error `Error installing file '.config/git/config' outside $HOME`, with a somewhat basic config here - https://termbin.com/ihx9
<clever>
craftyguy: thats where you want ${builtins.placeholder "out"}
<numkem>
clever: nope one line
<clever>
numkem: is there another nix.conf in $HOME ?
<craftyguy>
clever: ohh, let me try that
<clever>
numkem: .config/nix/nix.conf
<numkem>
clever: ugh.. yes
rsoeldner has quit [Remote host closed the connection]
<cole-h>
Do .config/nix/nix.conf and /etc/nix/nix.conf get merged?
<numkem>
so I'd be better off putting it in configuration.nix or ~/.config/nix/nix.conf?
<clever>
cole-h: somewhat, but each key just overwrites the other
<clever>
numkem: i would just put it all in configuration.nix
<rycee>
eyJhb: At the moment HM only manages files inside your home directory, and I imagine `/persistent/home/eyjhb` is not the same as `$HOME`.
<rycee>
eyJhb: If you want HM to consider `/persistent/home/eyjhb` your home directory then it should work to set `home.homeDirectory = "/persistent/home/eyjhb"` in your configuration.
CMCDragonkai has quit [Remote host closed the connection]
CMCDragonkai has joined #nixos
wavirc22 has quit [Quit: ZNC 1.7.2+deb3 - https://znc.in]
<numkem>
clever: what do you know, it's fine now my key is in nix.conf global... that is some weird behavior no? is it becuse my user is a trusted use that it reads my file?
drakonis_ has joined #nixos
<clever>
numkem: not entirely sure how the per-user nix.conf plays a role when using nix-daemon, things get a bit complex there
<rycee>
eyJhb: Although, I have no idea what `home.persistence` is doing.
wavirc22 has joined #nixos
<eyJhb>
rycee: I wouldn't have imagined that was the issue, since it was complaining bout git. But I see, it is because it symlinks the config from the persistent
<eyJhb>
THanks rycee ! ;) Savior of my night
<eyJhb>
rycee++
<{^_^}>
rycee's karma got increased to 27
remirol has quit [Ping timeout: 265 seconds]
<numkem>
clever: and it works just perfectly now... that was a crazy ride. Massive thanks to you! clever++
<eyJhb>
Seems nice so far, when I don't srew it up
<ldlework>
adisbladis: hey i seriously don't mean to bug you about this, but anderson_ and I are just wondering if you'll get to the repo today, as we are compiling some issues to file and stuff
<energizer>
you could make the repo and transfer it later
<cole-h>
I'm losing my mind with nginx. I just want `localhost/torrents` to be the same as `localhost:9091/transmission/web/` without needing all that other crap >:(
<cole-h>
infinisil: Same one as last time x) `services.nginx.virtualHosts."localhost".locations."/sonarr/" = { proxyPass = "http://127.0.0.1:8989/"; };` gives me a busted webpage because it's trying to access relative files
<infinisil>
8989 is 9091?
<cole-h>
Sorry that was for sonarr
inkbottle has joined #nixos
lorimer has joined #nixos
<cole-h>
Either way, both don't work ;D
<clever>
cole-h: depending on if the proxyPass ends with a / or not, it will either direct you to /sonaar/ or / directly, i think
<infinisil>
Did you include the full location in the torrent redirect?
m0rphism has quit [Quit: WeeChat 2.7.1]
<clever>
cole-h: you will want to tcpdump port 8989 and confirm what exactly nginx did to your url
<vyorkin>
hi, I'm trying to create a dev shell, but using nix flakes. But I have no idea how to do this properly. Here is what I've tried so far ttp://ix.io/2n9O
<cole-h>
clever: They're all failing to get stuff like the stylesheet and scripts.
<clever>
cole-h: and what URL is it trying to load the stylesheet from?
<cole-h>
clever: localhost/Content (where it is actually located at ${pkgs.sonarr}/bin/UI/Content)
<cole-h>
Or rather, the files it's looking for are located at ${pkgs.sonarr}...
<clever>
cole-h: if you check the html source for the page, how did it reference the stylesheet?
<cole-h>
clever: with `href="/Content/bootstrap.css"`
<mac10688>
clever: Thanks for the link earlier about default display manager. I think I learned something valuable. I have sddm display manager enabled. It was default config but I think it won't do me any good because I'm trying to configure xmonad
<clever>
cole-h: then it expects /Content to be in the root of your http server
<mac10688>
it seems that if I have sddm enabled, it won't try to read my .xsessionrc file
<mac10688>
I'm about to test it
<clever>
cole-h: you will need to setup another entry in locations, to proxyPass /Content to /Content
medvid has quit [Read error: Connection reset by peer]
<clever>
mac10688: that default script is ran by all display managers, and the tail end is where it launches the chosen desktop manager
<mac10688>
oh
<cole-h>
clever: Since I also want to have `/torrents` accessible through the same way, is there any better way than a "/Content" proxypass?
<cole-h>
(e.g. what if /torrents has a conflicting `Content` directory it wants)
<clever>
mac10688: .xprofile lets you inject extra things, and .xsession lets you overirde that, and stop it from running the chosen desktop manager
<mac10688>
ok that's good information
<clever>
cole-h: patch the software to properly use a relative path
<cole-h>
clever: Yeah, figured. Thanks for the help!
<cole-h>
clever++
<{^_^}>
clever's karma got increased to 437
medvid has joined #nixos
<{^_^}>
[nixpkgs] @etu opened pull request #88715 → home-assistant: fix zha in home-assistant → https://git.io/Jfa2V
<cole-h>
etu: Got a question about your container setup with sonarr and friends. How do you access `series.lan`? Do you have a DNS entry or something for it?
<vyorkin>
ok, nvm, we've figured it out :) here is the working flakes.nix which allows to run the SHELL=bash nix dev-shell: http://ix.io/2n9R
orivej_ has joined #nixos
orivej has quit [Ping timeout: 272 seconds]
<etu>
cole-h: Yes, my router has a dns entry in the local dns
<cole-h>
etu: So basically it just looks like `192.168.1.69:8989 -> sonarr.lan`?
<etu>
cole-h: well, I have nginx listening on port 80 on for that domain name and then it proxies into my container on the right port. So I enter http://sonarr.lan in my browser and don't have to worry about ports
chloekek has quit [Ping timeout: 272 seconds]
<etu>
And yes I have configured that nginx host to not listen to the internet :p
<clever>
cole-h: what about using virtual hosts, so you effectively have 2 different / dirs
<cole-h>
Yeah, that sounds like what etu is doing (minus stuff being in containers)
<cole-h>
Then the part I'll probably have trouble with becomes "how do I make sonarr.lan actually go to sonarr.lan?" (for example)
<etu>
I'm off for tonight though
<cole-h>
etu: o/ Thanks for the pointers.
<etu>
cole-h: either /etc/hosts or your local dns server :)
<cole-h>
But does it just say "Hey, sonarr.lan is over here at 192.168.1.23", or should it say "Hey, sonarr.lan is over here at1 192.168.1.23:8989"
<cole-h>
Oh
<cole-h>
Uh
<cole-h>
That was easy
<cole-h>
etu++ clever++
<{^_^}>
etu's karma got increased to 17
<{^_^}>
clever's karma got increased to 0o666
<cole-h>
👀
<cole-h>
Nice karma, clever
<clever>
lol
spacefrogg has quit [Quit: Gone.]
aw has quit [Quit: Quitting.]
aw has joined #nixos
spacefrogg has joined #nixos
konobi has quit [Remote host closed the connection]
orivej_ has quit [Quit: No Ping reply in 180 seconds.]
orivej has quit [Quit: No Ping reply in 180 seconds.]
orivej has joined #nixos
<das_j>
is there an easy way to substitute a known path without having the nix expression?
leungbk has joined #nixos
<eyJhb>
talyz: sorry for throwing you under the buss :p
<leungbk>
I'm having trouble using nodePackages.create-react-app. Would anyone happen to know of any quirks with the package on NixOS? Here's the error I'm getting: http://paste.debian.net/1148471
<leungbk>
__monty__ has quit [Quit: leaving]
dingenskirchen has quit [Remote host closed the connection]
dingenskirchen has joined #nixos
dan224 has joined #nixos
mac10688 has quit [Quit: WeeChat 2.7.1]
orivej has quit [Quit: No Ping reply in 180 seconds.]
<anderson_>
Hello, people! A newbie question: how can I make a minimal (?) "nixos-rebuild build-vm" in order to test graphical (Wayland and X11) programs from an overlay I am hacking?
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<{^_^}>
[nixpkgs] @pstch closed pull request #88136 → Allow mpv to use youtube-dl provided at runtime → https://git.io/Jfuah
<cole-h>
Make a separate configuration.nix and point it at it: `nixos-rebuild build-vm -I nixos-config=./vm.nix`
<{^_^}>
[nixpkgs] @sheenobu opened pull request #88723 → fmt: fix cmake include path → https://git.io/JfaVi
mac10688 has joined #nixos
<mac10688>
hi all, I was messing around with my .xsession file and now I can't log into my nixos laptop. I gave my .xsession chmod +x and I guess it didn't like that. I was wondering if anyone can tell me how I can boot into terminal so I can login and fix it
<mac10688>
when I try to login now, it's just boots me back to the login screen
<{^_^}>
[nixpkgs] @adisbladis merged pull request #88717 → emacs: Move toolkit to function arguments → https://git.io/Jfaaz
<{^_^}>
#88718 (by adisbladis, 56 minutes ago, open): services.x11.videoDrivers: Don't include vmware driver on non-x86 platforms
<mac10688>
THANKS rycee
<ldlework>
If I have a fetchGitHub for some source, and I am writing a runCommand, how can I run ${pkgs.vgo2nix}/bin/vgo2nix against that source? just "cd ${the-source}" before running the command?
<ldlework>
if i do the-source = pkgs.fetchGitHub ... ?
<adisbladis>
ldlework: vgo2nix --dir
<ldlework>
adisbladis: i want to put the deps.go into $out
<adisbladis>
You also want --outfile
<ldlework>
:P
<adisbladis>
Um
<ldlework>
adisbladis++
<{^_^}>
adisbladis's karma got increased to 86
<adisbladis>
You can't
<ldlework>
what do you mean?
<adisbladis>
Are you trying to use vgo2nix inside the sandbox?
<maralorn>
ldlework: I guess you know, but so that you don‘t trip over it, it's: fetchFromGitHub
<adisbladis>
vgo2nix does network access
<ldlework>
I'm using nix to run vgo2nix + some source to generate the deps.go as the output,
<ldlework>
oh
<ldlework>
I see
<ldlework>
maralorn: hehe thx
<adisbladis>
go.sum hashes are non-standard go-specific hashes
<cole-h>
adisbladis: It looks like `default` doesn't like references to `pkgs`... Dunno how to get around that yet though :P
<adisbladis>
vgo2nix needs to run nix-prefetch-git over each input
<ldlework>
adisbladis: fine i'll just generate it offline and stick the deps.go file next to the nix :)
<ldlework>
adisbladis: thanks for the heads up
<ldlework>
"offline" i mean, manually
<adisbladis>
=)
mac10688 has quit [Remote host closed the connection]
<ldlework>
adisbladis: it's kind of a shame though because i can't write a standalone expression to fetch and build this one thing on github
<ldlework>
if the source updates, i'll have to manually do this again
<ldlework>
oh well
<adisbladis>
Yeah.. That's sadly a reality of the go lock format :/
<ldlework>
adisbladis: anyway to escape the sandbox?
<cole-h>
--option sandbox false
<ldlework>
we need rust's impure
<cole-h>
Why?
<ldlework>
but i don't want the sandbox off for the whole thing, just one derivation
<ldlework>
cole-h: we just discussed why!
<cole-h>
I don't see any Rust-related discussion :P
piezoid has quit [Remote host closed the connection]
mauli has joined #nixos
mmchen has joined #nixos
<energizer>
ldlework: i thought transferring a repo also transfers the issues
<cole-h>
Only if the owner of the repo does the transfer. From what I understand, the owner hasn't been seen nor heard from in 2 years, so in reality this is really a fork.
<ldlework>
^
<energizer>
i was suggesting ldlework could create his own fork and transfer it to nix-community later
<mmchen>
Does anyone have a setup of using coc.nvim with rust-analyzer in a nix-shell environment? I am using the mozilla overlay but the completions are incomplete. For example when I type `use std::`, i only get self or super, but on my non-nix setups, I get things in std. I feel like something is missing with rustc.
<{^_^}>
[nixpkgs] @aanderse opened pull request #88725 → nixos/networkmanager: apply --filter-policy=STRICT to modemmanager service → https://git.io/JfaVF
o1lo01ol1o has quit [Remote host closed the connection]
<jfhbrook>
hello
<jfhbrook>
I'm trying to learn/use nixos with a remote instance, and I've been told that I can run nixos-rebuild locally with a remote target of my DO instance
<jfhbrook>
which sounds great! but I'm running arch
<jfhbrook>
and so I've installed nix on arch
<jfhbrook>
but it doesn't come with nixos-rebuild
<jfhbrook>
and I'm trying to figure out how to fix that and I can't find anything about this online
reallymemorable has joined #nixos
<jfhbrook>
is this a thing people know about?
<jfhbrook>
currently copying files using ansible which seems less than ideal
<quinn>
jfhbrook: i use git, but it's kind of clunky.
aveltras has quit [Quit: Connection closed for inactivity]
o1lo01ol1o has joined #nixos
<cole-h>
energizer: Only admins can transfer repos, IIRC.
<jfhbrook>
🤔 can you elaborate on that?
mauli has quit [Ping timeout: 246 seconds]
<jfhbrook>
I'm pretty new to this and I don't understand most of the toolchain
<jfhbrook>
I need my hand to be held a little
<cole-h>
jfhbrook: Why not just ssh into the DO instance?
<jfhbrook>
I am - I'm using ansible to copy files over from my local machine and then running it on ssh
Thra11 has quit [Quit: WeeChat 2.8]
<energizer>
cole-h: if admins want to move it into nix-community, they can. if they don't, that's fine too. the point is to unblock ldlework from filing issues
<ldlework>
if I am doing buildGoPackage against some github fetched source, how can I run a replacement on a file in its source?
<cole-h>
energizer: I mean, ldlework isn't an admin. Which means he can't move it AFAIK.
<ldlework>
energizer: oh we started our own organization
<jfhbrook>
but it seems like I can improve on that workflow and it seems like I should be able to use local nix paths to build the remote thing - I think this will make it easier to install patched packages (want to issue some updates to mysql) and manage secrets
<ldlework>
energizer: thanks for looking out though :)
<cole-h>
ldlework: tbh you should have made it an actual fork
<energizer>
cole-h: yes the admins would need to consent of course
<ldlework>
cole-h: we did
<cole-h>
ldlework: Doesn't show as one
<ldlework>
cole-h: a disconnected fork, not a github linked one
<ldlework>
a github linked one means that the original author can showup and delete the original
<ldlework>
it can cause problems
<cole-h>
Wait, really?
<ldlework>
i've reached out to them via email
<jfhbrook>
I know that from experience
<jfhbrook>
it's not fun
<cole-h>
Ouch
<ldlework>
honestly, i would have no problem if they came back and wanted to continue to be the maintainer
<{^_^}>
[nixpkgs] @aanderse pushed 4 commits to release-20.03: https://git.io/Jfawe
<jfhbrook>
anyway - happy to abandon that strategy, though I guess this leaves me doubling down on copying files and running remote commands with ansible
<ldlework>
if I am doing buildGoPackage against some github fetched source, how can I run a replacement on a file in its source?
<jfhbrook>
which isn't *entirely* satisfying
<ldlework>
do I need to use patchPhase?
mmchen has quit [Remote host closed the connection]
<cole-h>
jfhbrook: Why do you need to build locally?
<ldlework>
I had to do this recently but I don't remember
<jfhbrook>
cole-h: I would like to have a git clone of nixpkgs locally with modifications and adjust my NIX_PATH locally and have secrets files, again locally, and use them to modify the remote system without cloning nixpkgs on the remote and without scp-ing secrets to the remote
<cole-h>
Well, either way, I think you should just be able to `nix-copy-closure` your build artifacts. Read the man page and see if that works for you?
knupfer has quit [Ping timeout: 260 seconds]
veleiro has quit [Ping timeout: 240 seconds]
<jfhbrook>
cole-h: like I said I'm happy to abandon that strategy - pursuing it because someone suggested it to me
<cole-h>
jfhbrook: I sure hope you aren't referring to your secrets by path (e.g. `asdf = ./secret/asdfg;`)
<jfhbrook>
I mean
<cole-h>
Because if you are, those "secrets" are in the world-readable store.
<jfhbrook>
right now they're inline - that's most of what I'm trying to fix
<cole-h>
That's also not very secret :P
<jfhbrook>
the documentation I read suggested that paths would keep them out of the store
<cole-h>
No, definitely not. Paths get copied *into* the store.
<cole-h>
Unless you run `toString` on them, which then just gives you `"/some/path/here"`
<cole-h>
jfhbrook: From what you're describing, it sounds kinda like you might be interested in NixOps.
<{^_^}>
[nixpkgs] @adisbladis opened pull request #88727 → slack: Return a dummy derivation on unsupported platforms → https://git.io/JfawT
<jfhbrook>
I'm somewhat interested but I want the greater power/flexibility of terraform
<cole-h>
jfhbrook: `dbpassFile` is a string, not a path. So `dbpassFile = "/run/keys/asdfg";` is safe, because it doesn't actually copy the file into the store.
<jfhbrook>
and I'm worried that nixops is less mature than nixos
<jfhbrook>
yes ok, that was my impression - but then that file needs to exist somewhere, yes?
<jfhbrook>
I would like to not have to scp that file to the remote
<cole-h>
You'll have to.
<jfhbrook>
ok
<jfhbrook>
that's unfortunate but understandable
<jfhbrook>
can I delete the file after the derivation does its thing?
<jfhbrook>
should be able to, yeah?
<cole-h>
Depends on what you expect "deleting the file" to do
<jfhbrook>
is there a common place to put such secrets?
<jfhbrook>
I expect deleting the file to delete the file
<jfhbrook>
ie I don't want a file called /home/myuser/statesecrets.txt on my remote
<cole-h>
Yes, `rm` will indeed `rm` the file. But if it's in the store, it's in the store.
leungbk has quit [Ping timeout: 260 seconds]
<cole-h>
(btw, NixOps has been a thing since 2012 -- definitely not immature, IMO)
<jfhbrook>
ok but we've already established that I'm using dbpassFile which doesn't put such a file in the store
<jfhbrook>
or that at least this is the goal
<cole-h>
Yes indeed
<cole-h>
So if you remove that pass file, it will no longer exist and anything that wants to read from it will break.
<jfhbrook>
ok, but nix itself won't want to read from it unless I change the derivation, yes?
<cole-h>
No, nix itself won't read from it ever, since it's a string.
<jfhbrook>
so I can upload the file, run rebuild, and delete it, and as long as the actual application isn't using the secret I should be good?
<jfhbrook>
this line right here does something with the file - saying that it's not nix but a bash script generated by nix reading it strikes me as pedantry
<cole-h>
Yeah, Nix isn't reading from it, nextcloud will (or more specifically, the "nextcloud-setup" service that sets up everything for nextcloud to function properly)
<cole-h>
Nix won't check if it exists, that systemd unit will.
<jfhbrook>
ok but this seems to confirm my suspicion
<adisbladis>
Where the {} are derivation attributes on the runCommand derivation
<ldlework>
oh I thought they were environment variables for the builder
<jfhbrook>
ok so it sounds like barring my friend sitting down to fully explain how I would run nixos-rebuild locally on a non-nixos machine that I should stay the course with ansible; that I should give nixops a second look and that I need to find an alternate strategy for secrets management because my intended one doesn't work the way I was hoping
<jfhbrook>
this seems fine
<jfhbrook>
alright, thanks
<infinisil>
Raito_Bezarius: Ideally You'd use a GHC from nixpkgs, one with a matching glibc version
<pistache>
is there some way to force the evaluation of an expression at build-time, even if I'm not using that expression at all ?
civodul has quit [Quit: ERC (IRC client for Emacs 26.3)]
<infinisil>
Raito_Bezarius: Instead of however else you get GHC
<{^_^}>
[nixpkgs] @Ericson2314 pushed 6 commits to staging: https://git.io/Jfawj
<adisbladis>
Just as you would `pkgs.runCommand "somename" { meta = {}; } "touch $out"` is roughly equivalent to `stdenv.mkDerivation { name = "somename"; src = null; meta = {}; dontConfigure = true; dontBuild = true; installPhase = "touch $out"; }`
<adisbladis>
There is a bit more to it, but not that much
<ldlework>
Anyone got information about defining your own systemd services?
mbrgm_ has joined #nixos
<NobbZ[m]>
Is there some lorri related channel or can someone help me figuring out why lorri behaves like this?
mbrgm has quit [Ping timeout: 256 seconds]
mbrgm_ is now known as mbrgm
<ldlework>
Oh I see.
<NobbZ[m]>
I have lorri as a demon + direnv. When in a "direnv+lorri" environment, lorri uses my systems nixpkgs channel, though when I do `lorri shell` then the environment is correctly built against my users `nixpkgs` channel.
<ldlework>
Can I do `config.system.services = generateServices { ... };` to generate multiple systemd services?
iMatejC has joined #nixos
<FireFly>
ldlework: in general for things like that, I find it easiest to just look at what existing nixos modules are doing to get an idea of the exposed interface (and complement with reading configuration.nix(5))
<FireFly>
I mean wrt defining your own systemd services
<FireFly>
not sure how generateServices works though..
<FireFly>
or what it is
<ldlework>
config.systemd*
<ldlework>
I'm guessing so right?
<FireFly>
I believe so, I don't see why you wouldn't