<mek42_laptop>
I ordered a new laptop a couple days ago and am debating whether to use gentoo supplemented with nix package manager or just use nixos
<gchristensen>
probably should just use nixos
<gchristensen>
:P
<mek42_laptop>
i'm doing some nixos repo searches now to check on the state of things
<mek42_laptop>
how is nixos as a daily driver desktop?
<clever>
mek42_laptop: i use nixos on my desktop, laptop, nas, and router
<attente[m]>
does anyone use the nixpkgs-mozilla overlay to run firefox-nightly? i can't seem to find it even after adding it to my configuration. fwiw i'm on nixos-unstable
dermetfan has quit [Ping timeout: 240 seconds]
Vitalii has quit [Ping timeout: 256 seconds]
ifndef_define has quit [Ping timeout: 256 seconds]
CapsLock has quit [Changing host]
CapsLock has joined #nixos
<dramforever>
ifndef_define: did you enable the extension pack?
<dramforever>
eh, too late
dbmikus has quit [Ping timeout: 258 seconds]
eadwu has joined #nixos
lawlesseel has quit [Remote host closed the connection]
Ralith_ has quit [Read error: Connection reset by peer]
lawlesseel has joined #nixos
<flokli>
dramforever: i think vbox Version 5 is currently broken
<flokli>
I started reviewing the pull request for version 6, but got sidetracked
npmccallum has joined #nixos
<flokli>
There's something about the tests not working currently due to nested virtualization. I didn't run them so far...
stites has quit [Quit: ZNC 1.6.6+deb1ubuntu0.1 - http://znc.in]
stites has joined #nixos
ikitat has joined #nixos
rcshm has quit [Read error: Connection reset by peer]
rcshm has joined #nixos
ikitat has quit [Ping timeout: 240 seconds]
<q6AA4FD>
i have some services i'm looking to run on my nixos box, is there a canonical "easiest" way to do this? nixops with libvirt, nixos-container, or are all of these methods going to be much more difficult than configuration.nix?
rcshm has quit [Remote host closed the connection]
rcshm has joined #nixos
jluttine has quit [Ping timeout: 244 seconds]
<gchristensen>
yes
<gchristensen>
just use configuration.nix, it is what it is for
npmccallum has quit [Quit: npmccallum]
<q6AA4FD>
alright, thanks
mobile_c_ has joined #nixos
<ryantm>
Should `nix-build -A git` in a checkout of the nixpkgs-channels nixos-18.09 branch fetch git from cache.nixos.org? When I do that with commit 168cbb39691, it tries to build a bunch of stuff.
<ryantm>
Should `nix-build -A git` in a checkout of the nixpkgs-channels nixos-18.09 branch fetch git from cache.nixos.org? When I do that with commit 168cbb39691, it tries to build a bunch of stuff.
<ryantm>
sorry for the double post.
thc202 has quit [Ping timeout: 272 seconds]
<q6AA4FD>
mek42_laptop: the issue with mixing nixos and gentoo is that you probably need to provide anything that needs 3D acceleration with gentoo itself, which can make everything very confusing. you could probably hold it together by fixing the versions together, but that would be a lot of work i think
<q6AA4FD>
mek42_laptop: additionally, a bunch of stuff you wouldn't expect needs 3D acceleration and will break if it doesn't have it
jluttine has joined #nixos
<samueldr>
ryantm: didn't have to here; but maybe I did previously accidentally?
* samueldr
tries on another computer
stites has quit [Quit: ZNC 1.6.6+deb1ubuntu0.1 - http://znc.in]
<ryantm>
samueldr: I think we might have misconfigured our NixOS instance to not use cache.nixos.org. Thanks for giving us a sanity check. We'll double check here.
<alex_giusi_tiri>
would it be possible to have a var assigned a list but catted with itself, as in: `nix-repl> p = {a ? []}: rec {a=[0]++a;}`. calling the function `p {}` would give "error: infinite recursion encountered, at (string):1:24"
<dramforever>
ryantm: What does this output? nix eval -f path/to/checkout git.outPath
<dramforever>
alex_giusi_tiri: nix lists are strict in length (or 'spine'), unlike haskell
<dramforever>
so no infinite lists
<alex_giusi_tiri>
but, what does that mean? am i trying to make an infinite list? I am unsure I understand
<alex_giusi_tiri>
what I intend to do is have a default value for a var, and add somthing to it later
<dramforever>
wait, what do you expect as the result?
<dramforever>
'add something to it later' <- I'm afraid this isn't possible
<alex_giusi_tiri>
I would want to append to whatever it currently contains
<alex_giusi_tiri>
as I understood, "lists can be modified, but you would get a new list back"
<dramforever>
and this new list can't go back to the variable it came from
<samueldr>
attributes in a derivation will be made available as an environment variable*, thus I guess `inherit curl` is like `curl = [pkgs.]curl;`, which toStringed is the path to curl, making it available under $curl?
<samueldr>
* unless passAsFile is used, something relatively new
* samueldr
somehow skipped the already plenty good answer
silver has quit [Read error: Connection reset by peer]
stites has quit [Quit: ZNC 1.6.6+deb1ubuntu0.1 - http://znc.in]
rpg has joined #nixos
stites has joined #nixos
Supersonic has quit [Ping timeout: 250 seconds]
Supersonic has joined #nixos
<q6AA4FD>
i'm trying to configure prometheus, i'm getting a 'error: The option value `services.prometheus.scrapeConfigs.[definition 1-entry 1].static_configs.[definition 1-entry 1]' in `/etc/nixos/configuration.nix' is not of type `submodule'. here's my configuration.nix: https://ix.io/1AUG
<simpson>
q6AA4FD: I can't view your pastebin, SSL_ERROR_RX_RECORD_TOO_LONG. But I've experienced this sort of pain before. scrapeConfigs needs to be a list of attrsets. Same with static_configs.
<q6AA4FD>
simpson: when you say list of attrsets, you mean something like scrapeConfigs = [{ foo=bar;}; { foo=baz;}]; ? because i'm following this syntax, with no luck
<simpson>
q6AA4FD: Yeah. In particular, instead of just a hostname, try { targets = "localhost:9090"; labels = {}; }
mkoenig has quit [Ping timeout: 250 seconds]
doyougnu has joined #nixos
<q6AA4FD>
simpson: right now i have static_configs = [{ targets = "localhost:9090"; }]; and i'm getting the same error
<q6AA4FD>
i get the same error when it's just static_configs = { targets = "localhost:9090"; }; as well
rcshm has quit [Remote host closed the connection]
mkoenig has joined #nixos
rcshm has joined #nixos
<simpson>
q6AA4FD: Keep playing around with the arrangement of things, I guess. I could have sworn that the arrangement I showed you was working for me.
o1lo01ol1o has quit [Remote host closed the connection]
<q6AA4FD>
simpson: i just found the error! targets is supposed to be a list too... holy cow
ddellacosta has quit [Read error: Connection reset by peer]
ddellacosta has joined #nixos
<dramforever>
romanofski: no if you want it to be a file in /nix/store. All files there are set to 1970 *after* buildinh
<dramforever>
*building
<romanofski>
right drat
eadwu has quit [Quit: WeeChat 2.3]
<romanofski>
benkolera: explained to me that it's not something I want to change
<romanofski>
so seems like we need to look for alternatives
<romanofski>
thanks dramforever
rcshm has quit [Remote host closed the connection]
rcshm has joined #nixos
<dramforever>
romanofski: what exactly requires a non-1970 timestamp though?
<romanofski>
dramforever: ah in our Haskell project we're using dyre, which recompiles the configuration in a program if the executable or the config has changed
<romanofski>
but the time for the executable won't change since they stick with 1970 (the mtime that is)
eadwu has joined #nixos
<romanofski>
I think there are a few other Haskell projects which fall into the same ball park
<romanofski>
termonad is one I'm aware off
<dramforever>
do you need dyre for normal use and not just development?
<romanofski>
yes
<romanofski>
normal use
<romanofski>
since once installed and you provide upgrades, it should rebuild the executable
<dramforever>
I suppose you can just create the environment using nix-shell and run a normal build inside
<dramforever>
normal build -> non-nix build
<romanofski>
dramforever: we were thinking of adjusting the library to use something else than mtime determining the rebuild
<dramforever>
I am thinking of adjusting your haskell program to use something other than nix to build user code
<{^_^}>
[nixpkgs] @thoughtpolice opened pull request #55679 → keybase: on Linux, symlink kbfsfuse, git-remote-keybase into main package → https://git.io/fhQxr
<{^_^}>
[nix] @matthewbauer opened pull request #2677 → Support --disable-shared flag. → https://git.io/fhQxo
<hyper_ch2>
I have an issue with systemd.network to setup a bridge. It all works fine in 18.03 but not anymore in 18.09 or pre-19.03. https://paste.simplylinux.ch/view/raw/4532ad1e From what I can see is that in 18.09 and pre-19.03 eth0 also get a default gw entry while in 18.03 br0 has two defualt gw entries.
rauno has quit [Ping timeout: 259 seconds]
rauno has joined #nixos
zupo has quit [Ping timeout: 258 seconds]
alex``` has quit [Ping timeout: 268 seconds]
<hyper_ch2>
the problem seems to be that nixos for some reason adds two gws
xkapastel has quit [Quit: Connection closed for inactivity]
schjetne has joined #nixos
alex``` has joined #nixos
<{^_^}>
[nixpkgs] @vcunat pushed 91 commits to staging-18.09: https://git.io/fhQhU
<uvnikita[m]>
Does anyone know what is a proper way to install nix package manager in multi-user mode on centos while being root? Ideally I would like to have an rpm package, but it seems like hydra is not building rpms anymore (https://github.com/NixOS/nix/commit/94b2e4e1be4526f2f85efd29a52a63e1858ea414). `sh <(curl https://nixos.org/nix/install) --daemon` doesn't work either because it explicitly forbids to run the script as root
<uvnikita[m]>
The reason I'm asking is because I'm thinking to use nix as a package manager on our slurm cluster which is currently configured via puppet. So I'm planning to write a simple puppet module that will install and configure nix on the nodes.
ivan has joined #nixos
<manveru>
uvnikita[m]: easy workaround would be to just create a temp user and remove it after install is done?
<uvnikita[m]>
manveru: yeah, that's an option. Fills a bit clunky though
<manveru>
true, the proper way is to figure out why it doesn't want to run as root :)
gagbo has joined #nixos
<uvnikita[m]>
manveru: I guess :) Or to figure out why rpm can't be built anymore. Was hoping that I'm missing some obvious solution :)
<Taneb>
I've got a hydra build that seems to be stuck in a loop. How can I kill it hard?
<Taneb>
"cancel"ing it doesn't seem to actually stop it
<v0|d>
hey, is there a nix interface for machinectl & nixos-container?
<v0|d>
in which prob. I can create containers, add users, configure etc.
<teto>
running nix master, I see "operation 'addToStore' is not supported by store" (a nixops deployed remote builder). How can I solve this ?
balsoft has quit [Read error: Connection reset by peer]
<monotux>
I'm using incorrect search terms, but I'd like to abstract some options used in most of my virtualhost definitions in nginx, something like this: https://pastebin.com/zMqqnTru
<monotux>
any pointers?
<monotux>
obviously incorrect syntax :)
codezero has quit [Read error: Connection reset by peer]
codezero has joined #nixos
<{^_^}>
[nixpkgs] @dgarzon opened pull request #55689 → nodePackages_10_x.@angular/cli: init at 7.3.1 → https://git.io/fh7eY
winem_ has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @dywedir pushed commit from @steveeJ to master « rdedup: 3.0.1 -> 3.1.1 (#55530) »: https://git.io/fh7eV
rycwo has quit [Ping timeout: 240 seconds]
rycwo has joined #nixos
<qyliss^work>
monotux: delete the { on line 1
<monotux>
:D
<monotux>
thanks!
e-user has joined #nixos
<jD91mZM2>
If I want to submit a package A to nixpkgs, but it depends on package B which also isn't already there, do I submit them both as one PR or separate?
<etu>
jD91mZM2: You can do one PR
rycwo has quit [Ping timeout: 268 seconds]
knupfer has quit [Remote host closed the connection]
<jD91mZM2>
etu++ that's a fast answer!
<{^_^}>
etu's karma got increased to 5
knupfer has joined #nixos
<jD91mZM2>
oh no has the karma been reset recently?
<{^_^}>
[nixpkgs] @dywedir pushed to master « rdedup: mark as broken on darwin »: https://git.io/fh7eS
ckauhaus_ is now known as ckauhaus
<etu>
jD91mZM2: [Reset] Not that I know of, I just don't have that many magic internet points ;)
<jD91mZM2>
etu += 2147483642
<jD91mZM2>
real bummer that doesn't work :)
knupfer has quit [Ping timeout: 246 seconds]
<johanot>
jD91mZM2: be careful you don't overflow that
<johanot>
sphalerite: right. that was a saturating_add, how stupid of me. hehe
<sphalerite>
actually I think it's the type that panics :/
DrLambda has quit [Ping timeout: 246 seconds]
<monotux>
another silly question. I'd like to read a setting in a service definition in a nginx definition (webroot path for a package), what's the shorthand here?
<jD91mZM2>
johanot: That value is (max 32-bit signed int - 5)
<jD91mZM2>
sphalerite: Panics in debug, silently wraps in release :)
Makaveli7 has joined #nixos
__monty__ has joined #nixos
<johanot>
jD91mZM2: hehe. so you were careful after all :)
<petrkr>
some bug in new release-18.09 branch? I just switched to it and now getting for next switches "switch-to-configuration: line xx: use: command not found"
<sphalerite>
petrkr: that's a bug where switch-to-configuration is interpreted as a shell script rather than a perl script. Can't remember exactly what causes it though.
<sphalerite>
petrkr: does the interpreter mentioned in the shebang exist and run?
primeos_ has quit [Quit: WeeChat 2.3]
DrLambda has joined #nixos
<petrkr>
seams yes, but do not know if there can be space between #! and command
<qyliss^work>
There generally can be
<sphalerite>
yeah that's fine
<qyliss^work>
But there was a bug in a Linux 5.0 kernel beta about it IIRC
primeos has joined #nixos
<petrkr>
I tried older generation, there switch works, but when I will reboot to latest one, again gets that switch error
<qyliss^work>
So make sure you're not running that.
nikivi has joined #nixos
q6AA4FD has joined #nixos
<petrkr>
seems that is bigger issue, since now I try use some "command not found" and it say same :)
<petrkr>
well if I will run that script as parameter to that interpreter, then it works.
<sb0>
why does 'nix-shell -p python2' set up a shell with python2 available, but 'nix-shell -E "let pkgs = import <nixpkgs> {}; in pkgs.python2"' does not?
<qyliss^work>
nix-shell -p python2 is like nix-shell -E "let pkgs = import <nixpkgs> {}; pkgs.stdenv.mkDerivation { name = "nix-shell"; buildInputs = [ pkgs.python2 ]; }"
<LnL>
sb0: the first is a shell with python available while the second is a shell to build python itself
<obadz>
patchelf help pls? I'm using patchelf --set-rpath write the expression for a binary and it's clearly working on some dlls but not on others (that are in the same directory as the ones that are working!). Any pointers on how to debug?
dian has quit [Quit: Page closed]
init_6 has joined #nixos
Tucky has quit [Quit: WeeChat 2.2]
<{^_^}>
[nixpkgs] @dgarzon opened pull request #55691 → nodePackages_10_x.@angular/cli: init at 7.3.1 → https://git.io/fh7vP
<symphorien>
Maybe you need to patchelf the libs as well
<symphorien>
Sorry that was meant to obadz
ThatDocsLady has joined #nixos
<nekolyanich>
Hi all! Failed to build haskellPackages.brick-skylighting (missing dependencies). How to override haskellPackage's dependency version?
silver has joined #nixos
ThatDocsLady_ has joined #nixos
init_6 has quit [Ping timeout: 246 seconds]
rycwo has joined #nixos
<{^_^}>
[nixpkgs] @tobim opened pull request #55698 → compiler-rt-7.0.1: Fix build on darwin → https://git.io/fh7fQ
ThatDocsLady has quit [Ping timeout: 258 seconds]
<__monty__>
nekolyanich: You mean the dependencies brick-skylighting requires? `doJailbreak` is probably the answer.
init_6 has joined #nixos
<nekolyanich>
__monty__: Nice, i'll try doJailbreak. I'm try to build haskellPackages.matterhorn. It requires brick-skylighting.
<__monty__>
nekolyanich: And brick-skylighting doesn't satisfy the version bounds, or a dep of brick-skylighting doesn't? In the first case jailbreak matterhorn, in the latter brick-skylighting.
<petrkr>
sphalerite, it can be only my instance, but maybe someone can test that too
<gregoire>
I have a problem with my config I can't figure out why It doesn't work while It previously did. My `configuration.nix` contains `nixpkgs.config = import ../nixpkgs/config.nix;` where config.nix contains `{allowUnfree = true;}`. When doing `nixos-option nixpkgs.config` I do get my `allowUnfree = true;`. However, when running `sudo nixos-rebuild switch` I get `Package foo has an unfree license refusing to evaluate`. When setting `NIXPKGS_ALLOW_UNFREE=1`,
<gregoire>
it does work. I've spent the last 2 hours trying solving this without any luck. Any idea?
rycwo has quit [Remote host closed the connection]
kleisli has joined #nixos
<{^_^}>
[nixpkgs] @mdorman opened pull request #55700 → gtk+-3.24.3: Fix for missing symbols → https://git.io/fh7Jt
<hyper_ch2>
Mic92: howdy
<ottidmes>
gregoire: in all liklihood you are pointing to the wrong NixOS configuration.nix
matijja has joined #nixos
johnnyfive has quit [Quit: I quit now.]
<gregoire>
Even when using `-I nixos-config=/path/to/configuration.nix` it doesn't work. Also, If I was pointing on the wrong configuration (which I'm sure I don't), I wouldn't see `allowUnfree = true;` when running `nixos-option nixpkgs.config` right?
<gregoire>
But thanks for the suggestion
kleisli has quit [Ping timeout: 258 seconds]
<ottidmes>
gregoire: well it could, if you are using e.g. NIXOS_CONFIG, then that has priority over <nixos-config>
<gregoire>
$NIXOS_CONFIG is not set in my config
<ottidmes>
gregoire: we can test it: nix eval --json '(with import <nixpkgs/nixos> { }; config.nixpkgs.config)' (should give false) nix eval --json '(with import <nixpkgs/nixos> { configuration = /path/to/configuration.nix; }; config.nixpkgs.config)' (should give true)
<petrkr>
sphalerite, Now I tested latest kernel (4.19.20 vs 4.19.21) and has same problem as 4.14.98 vs 4.14.99
<petrkr>
20 (98) works 21 (99) does not
<ottidmes>
gregoire: run with sudo to get the same env as sudo nixos-rebuild is ran in
<gregoire>
Both lines returns nothing in my terminal
<gregoire>
But it doesn't fail
<ottidmes>
wut, I'd expect it to either give back the nixpkgs config as JSON or at least crash
<obadz>
symphorien: thanks, that seems to have worked.
<nekolyanich>
__monty__: Thanks for help, but brick-skylighting not builds with doJailbreak. Looks for me, libraries are changed their api.
<ottidmes>
what does that mean? in what way does it refuse?
<gregoire>
Oups sorry, It refuses as my package has an unfree license. Please note that I'm using home-manager to install this package: I also have a symlink to my config.nix in ~/.config/nixpkgs
<ottidmes>
ah, that changes things, I don't use home-manager and it probably needs its own way of configuring it, no idea, hopefully someone else can help who uses home-manager
<__monty__>
nekolyanich: Hmm, that's a bigger problem. You'll have to override its deps with callHackage or something.
<__monty__>
nekolyanich: That way you can specify their versions.
<gregoire>
ottidmes: I already did everything that home-manager expects. Also, as using NIXPKGS_ALLOW_UNFREE=1 solves the issue, I don't believe hte problem is from home-manager
<gregoire>
Let me try moving the package directly to my configuration.nix :)
<arianvp>
anybody here do Agda development on Nix
<arianvp>
how do I set up agda2-mode with extra libraries from nixpkgs?
nD5Xjz has quit [Ping timeout: 268 seconds]
nikivi has quit [Quit: ZNC is awesome]
<nekolyanich>
__monty__: I'm not confident in haskell packages infrostructure, so i'll try this later.
<gregoire>
ottidmes: I've now moved the problematic package to my configuration.nix (postman) and not using home-manager. I still have the same problem
<nekolyanich>
__monty__: thanks for help
<arianvp>
tilpner: that's a bit too global for my taste. i dont really install things into my nix profile
abcrawf has quit [Remote host closed the connection]
<arianvp>
wonder if someone has some kind of magic shell.nix lying around
nikivi has joined #nixos
<gregoire>
ottidmes: Ok, very weird thing, only postman has problems with my config, I have other unfree packages which works fine
<__monty__>
nekolyanich: Here's an example where I did this for the dhall package: http://ix.io/1AX8
<gregoire>
Could it be a problem with the licence of postman itself?
abcrawf has joined #nixos
<gregoire>
Ohoho, starting to narrow down the problem
<ottidmes>
gregoire: if you have a local checkout try and changing the license to licences.unfree, and see what happens I guess, otherwise maybe try overrideAttrs to change it
<tilpner>
arianvp: Oh, and it doesn't need to be that profile. You could probably buildEnv in shell.nix and set an env var to the output. Then you would use that env var from your emacs config instead of a user-global profile
<tilpner>
arianvp: But I'm just guessing, I don't use emacs or agda
<{^_^}>
[nixpkgs] @artempyanykh opened pull request #55703 → haskell.packages.ghc822/844: Fix configuration and build of GHC822/844 → https://git.io/fh7J5
<gregoire_>
ottidmes++
<{^_^}>
ottidmes's karma got increased to 13
<gregoire_>
It worked :)
knupfer has joined #nixos
Tucky has joined #nixos
iyzsong has quit [Read error: Connection reset by peer]
<ottidmes>
I have asked before and since I have yet to find an answer maybe I should create an issue, but why does --arg and --argstr not work with `nix eval '(...)'` despite being listed in its help message?
knupfer has quit [Remote host closed the connection]
knupfer has joined #nixos
<{^_^}>
[nixpkgs] @Vskilet opened pull request #55706 → nixos/roundcube: more options → https://git.io/fh7UD
<Vskilet>
My pull request is awful I think ^^
agander has quit [Ping timeout: 240 seconds]
<jD91mZM2>
Uhhh... umm... ok so if I created a recipe for package A and package B, where B is an optional dependency, would it be okay to worst case scenario send a `broken = true` package B if I finally got it to compile but not work at runtime?
<ottidmes>
jD91mZM2: AFAIK that will not be accepted, I think you will have to submit it without dependency B if that were to happen. Instead you could ask for help to make it not broken here in the IRC or maybe by creating a Nixpkgs issue for it, requesting help.
<ottidmes>
Twey: strace to find out?
gregoire_ has quit [Quit: Leaving]
orivej has quit [Remote host closed the connection]
<Twey>
brk(NULL) = 0x25b1000
<jD91mZM2>
ottidmes: Right, thanks. There doesn't seem to be much info on this program so I don't think expect anybody to help, but I'll send details here in a sec
<{^_^}>
ArcticaProject/nx-libs#652 (by realsimix, 1 year ago, closed): Abort of nxagent-3.5.99.13 on CentOS 7
kreisys has joined #nixos
kreisys has quit [Client Quit]
o1lo01ol1o has joined #nixos
<{^_^}>
[nixpkgs] @7c6f434c pushed 2 commits to release-18.09: https://git.io/fh7kd
reinhardt has quit [Quit: Leaving]
endformationage has joined #nixos
<{^_^}>
[nixpkgs] @smaret opened pull request #55711 → Aplpy: init at 2.0.rc2 → https://git.io/fh7kj
samrose has joined #nixos
schjetne has joined #nixos
matijja has quit [Ping timeout: 250 seconds]
yokyolu has joined #nixos
<yokyolu>
nix-shell gives this warning: dumping very large path (>256 MiB); this may run out of memory
<yokyolu>
my guess is that it copies stuff to tmpfs
<yokyolu>
lots of stuff
<tilpner>
That's fine, you're just importing a large directory into the store
<yokyolu>
how do i make it stop?
<tilpner>
Don't import large files/directories
<srhb>
yokyolu: By not importing a large path, basically.
schjetne has quit [Ping timeout: 240 seconds]
<yokyolu>
how do i know what's imported?
<tilpner>
Well, import is the wrong term
<clever>
yokyolu: did you do src = ./.; ?
rpg has joined #nixos
ikitat has joined #nixos
<yokyolu>
i did `if lib.inNixShell then null...`
<tilpner>
And this occurs when running nix-shell?
<yokyolu>
yes
<tilpner>
Can you paste the rest of your expression?
<samrose>
anyone run across an issue where /<nix store>/bin/ld cannto find -libzmq although I have installed nixpkgs zeromq4 ?
<srhb>
samrose: "installed" in what sense?
<clever>
,libraries samrose
<{^_^}>
samrose: Don't install libraries through nix-env or systemPackages, use nix-shell instead. See https://nixos.wiki/wiki/FAQ/Libraries for details.
<srhb>
,libraries samrose
<{^_^}>
samrose: Don't install libraries through nix-env or systemPackages, use nix-shell instead. See https://nixos.wiki/wiki/FAQ/Libraries for details.
<srhb>
:|
<samrose>
srhb like in configuration.nix
alex``` has quit [Ping timeout: 258 seconds]
<srhb>
samrose: You're not supposed to do that, and it won't work, as detailed in the above link :)
<yokyolu>
it's generated with callCabal2nix and src is set with overrideAttrs
<srhb>
yokyolu: is src a large directory? :)
drakonis has quit [Quit: WeeChat 2.3]
<srhb>
yokyolu: you can figure this out with nix-shell -vvv and see which store paths are realized after the warning, but it'll be a long log (>2 /tmp/yourlog)
<samrose>
srhb: ok understood. Do you have advice for building out rust in package or module where you need to utilize rust nightly, and it appears that mozilla rust overlay is meant for nix-shell ?
<srhb>
samrose: You can use the mozilla rust overlay to build with nix-build just fine
<yokyolu>
srhb: thanks! it is src after all! but why?
<dckc>
umask... quite probably. good catch.
<srhb>
yokyolu: Why what exactly?
agander has joined #nixos
hyper_ch2 has quit [Quit: Page closed]
<samrose>
srhb: have you ever run across examples of people using rust overlay in hydra builds for their custom OS or custom packages?
<dckc>
same permission denied failure with umask 002 :-/
* dckc
heads to meeting...
nikivi has quit [Quit: ZNC is awesome]
<srhb>
samrose: Yes, I've done that.
<samrose>
any examples you might share? I would be happy to document success on nixos.wiki
nikivi has joined #nixos
<srhb>
samrose: I can't share it exactly, but I can give you the idea of what I did. i'm not sure this is the most optimal way to do it with recent rust versions though, it might be simple enough to just override things locally in nixpkgs without using the mozilla overlay
<{^_^}>
[nixpkgs] @mpisanko opened pull request #55714 → upgrade leiningen to latest (2.9.0) → https://git.io/fh7I1
Tucky has quit [Quit: WeeChat 2.2]
<samrose>
srhb: I have a feeling this is probably simpler than I am making it. Any clues you might offer will likely help and thanks!
<srhb>
samrose: Then, I used makeRustPlatform with rustc = rustChannel.rust; cargo = rustChannel.cargo;
<srhb>
samrose: From there, rustPlatform has an attribute buildRustPackage with the versions described by the manifest file.
<srhb>
samrose: That's it basically.
<yokyolu>
srhb: overrideAttrs (drv: { src = if lib.inNixShell then null else ./. ; }) didn't do what i thought it would
<yokyolu>
srhb: the thing that's being overriden is made by callCabal2nix
<yokyolu>
if i just do it manually it works
kyren has quit [Remote host closed the connection]
<samrose>
it seems a bit sketchy to me to launch a service with nix-shell, what is the consensus on this out there?
<samrose>
(probably just my ignorance)
<gchristensen>
it certainly wasn't intended for that purpose, and I don't think many people do that
<srhb>
samrose: I think it's sketchy.
<srhb>
samrose: (I've seen it before too)
<samrose>
ok, made sense to me, as you lose determinism when things are built and deployed via shell
<srhb>
samrose: I usually see it used as a way to get around the different kinds of sandboxing, and I think that's ugh.
<srhb>
As in, deliberate indeterminism.
<srhb>
non*
<samrose>
got you ok thank you
schjetne has joined #nixos
kreisys has joined #nixos
<srhb>
yokyolu: I can't think of a simple explanation to that offhand, probably need to see a reproducible example.
<infinisil>
It does make sense if you need it for development for your project
<infinisil>
E.g. a webserver for a web project
<cransom>
is there way to prevent hydra from running a second set of jobs for a jobset if the first set hasn't completed yet?
<infinee>
infinisil: nice example
<infinee>
webserver, mq, etc
<cransom>
hrm. maybe a previous build as an input?
<srhb>
cransom: Aside from having an actual dependency? No, I don't think so.
<srhb>
cransom: That'll do it.
<yokyolu>
how to pass an argument to a derivation called by callPackage?
<srhb>
yokyolu: callPackage thing { arg = foo };
<yokyolu>
hm
schjetne has quit [Ping timeout: 268 seconds]
<yokyolu>
i just got enlightened a bit
<srhb>
yokyolu: The purpose of callPackage is to fill in all the arguments with the defaults of its scope. But you can override that by filling in that attribute set.
<infinee>
Is there no way to specify version of a package in my shell.nix? the examples I've seen only use pkgs like ruby which have version specific named pkgs. but the pkg I'm using doesn't have any version info appended to the name of pkg
lewo has quit [Read error: Connection reset by peer]
lewo has joined #nixos
Ariakenom_ has joined #nixos
<srhb>
infinee: Need more information. Is the version you want in nixpkgs or not?
dbmikus has joined #nixos
<yokyolu>
srch: i needed to pass it through overrideCabal i think
<srhb>
yokyolu: It depends. You might actually want override.
Ariakenom has quit [Ping timeout: 250 seconds]
<infinee>
srhb: when I do `nix-env -qa delve` I see 2 versions listed
xkapastel has joined #nixos
<yokyolu>
so with `src = if inNixShell then null else ./.;` can i pin the shell verison of derivation to gcroots? or do i need do write a different nix file for that?
<srhb>
yokyolu: I'm not really sure what pinning to a gcroot means in this sense.
jasongro` has joined #nixos
<srhb>
yokyolu: What are you trying to achieve exactly?
rcshm has joined #nixos
<srhb>
yokyolu: I *think* what you're trying to avoid is the reimport slowness, in which case you need a fixed output derivation if you really do have that much source that needs to go into the store.
jasongrossman has quit [Ping timeout: 246 seconds]
winem_ has quit [Ping timeout: 272 seconds]
alex`` has joined #nixos
periklis has joined #nixos
<srhb>
yokyolu: Is that right?
<yokyolu>
srhb: yes! this! thank you
<srhb>
yokyolu: Right. Once you've made a fixed-output derivation of your src, it means you've got a hash attached to the final path. If nix determines that the path with *that hash* already exists, it won't have to reimport. This does mean that *you're* now responsible for bumping that hash, or new changes will not take effect in your nix-shell. Just trying to outline the consequences here.
<srhb>
This can be very confusing behaviour.
johanot has quit [Quit: WeeChat 2.2]
<yokyolu>
srhb: ok, so as i get it every time i do whatever in ./. hashes change and it puts ./. to store again and again and again ....
<srhb>
Right.
<srhb>
And it can't determine whether the hash changed _before_ the import if it's not a fixed output drv
e-user has quit [Remote host closed the connection]
<yokyolu>
srhb: so i just make a shell-specific.nix with `src = null;`
<srhb>
Possibly.
<srhb>
I guess my main question is: Do you really really really have 256MiB of sources that you need to import?
<yokyolu>
srhb: or override regular.nix in shell.nix
init_6 has quit []
<srhb>
I mean, that's a lot!
<srhb>
You may benefit from filtering, splitting up the package, ...
<yokyolu>
srhb: it is. i havent figured how to make nix ignore binaries and other junk
<yokyolu>
is there somethnig like .gitignore for nix?
<Taneb>
yokyolu: look up filterSource in the nix manual
<yokyolu>
Taneb: thanks
<yokyolu>
will do
Ariakenom_ has quit [Quit: Leaving]
alex`` has quit [Ping timeout: 258 seconds]
<clever>
yokyolu: you can also use builtins.filterSource or lib.cleanSource, to limit what gets copied to /nix/store as it imports
<catern>
hey #nixos, so LyX links against Hunspell. I'm running LyX but it doesn't seem to be picking up Hunspell dictionaries. How do I get LyX to pick up the Hunspell dictionaries?
<catern>
I could override the hunspell argument to LyX with hunspellWithDicts, but that would cause it to be rebuilt, right?
<{^_^}>
[nixpkgs] @flokli pushed commit from @WhittlesJr to master « nzbget: Fix script for copying default config file template (#51235) »: https://git.io/fh7Lz
<srhb>
catern: Either accept the rebuild or restructure the LyX as a LyXWithDicts thing that is a thin wrapper over a dict-agnostic hunspell-enabled LyX, but changes the hunspell being used to be the one with your dicts
<srhb>
catern: You still fundamentally get the rebuild, but now it's a small rebuild.
jasongro` has quit [Quit: ERC (IRC client for Emacs 26.1)]
<catern>
actually, I don't think overriding to hunspellWithDicts will help at all
<catern>
since that just wraps the command-line hunspell tool - it won't help with the library
<catern>
(in fact overriding hunspell = hunspellWithDicts would probably break)
<srhb>
catern: If the runtime hunspell will actually accept DICPATH (what a name...) maybe you can just wrap LyX similarly?
<srhb>
I'm not sure how LyX uses it internally
<srhb>
If it's just calling out to either some code or an executable that obeys that env var, it seems trivial.
<catern>
Apparently LyX has a configuration option to override it, so my user is just using that now
<catern>
I wish my users would actually ask on #nixos on Freenode when I tell them to, rather than forcing me to proxy their questions to here :)
<{^_^}>
hunspell/hunspell#413 (by Jookia, 2 years ago, closed): Suggest applications support the DICPATH variable
<srhb>
catern: That would make it very simple, yes.
<catern>
That issue suggests that DICPATH is just a convention, not something libhunspell knows about, and it's up to each application to respect it (well, at least that's how it was, anyway - the issue suggests that Hunspell has gained some helpers for making that easy)
<srhb>
catern: hunspell.cxx references DICPATH, but that may be just the command line and not the library.
jasongrossman has joined #nixos
jasongrossman has quit [Remote host closed the connection]
<dckc>
I think the umask issue goes beyond my umask... root's umask or the build user's umask or something is also an issue:
<dckc>
> ls -ld /nix/store/5hdmx9yk7gr71b98j4vh9271k0zg5jis-nix-2.2.1
<dckc>
drwx------ 9 root root 4096 Dec 31 1969 /nix/store/5hdmx9yk7gr71b98j4vh9271k0zg5jis-nix-2.2.1
<{^_^}>
undefined variable 'ls' at (string):218:1
Jetien_ has quit [Ping timeout: 258 seconds]
schjetne has quit [Ping timeout: 246 seconds]
<ottidmes>
gchristensen: since you were the one recommending diffoscope to me, any idea how to force it to use hexdump diff rather than text diff? the text diff is useless to me, I basically get -x\n+x, i.e. they seem the same but are reported different
<infinisil>
ottidmes: You can pipe it into xxd first to get a hex view
<infinisil>
(I haven't tried it, but that's what I'd do)
<ottidmes>
hmm, weird bug, nix eval does something with stderr, that nix-instantiate does not, which causes bugs when autocompleting, when I do 2> /dev/null it works like nix-instantiate, when I dump the stderr into two files and compare them with a hex few they are both empty
<ottidmes>
infinisil: yeah thanks, thats indeed what I ended up doing (hexdump -C in my case)
Soo_Slow has quit [Remote host closed the connection]
<dckc>
ugh... any suggestions on how to deal with this umask stuff? I tried starting a root shell, doing `umask 022` and then running the installer, but it says "Please do not run this script with root privileges."
<{^_^}>
[nixpkgs] @Taneb opened pull request #55721 → haskellPackages.equivalence: dontCheck in GHC 8.6 → https://git.io/fh7qG
neminis has joined #nixos
<srhb>
Oh good, it's a gtk bug and a ghc bug all in one.
<neminis>
hello, do you know how to use gtest on nixos? i get linking errors even with all variables set correctly
<samueldr>
srhb: mixed gdk_pixbuf ?
<samueldr>
are you on 18.09?
<samueldr>
if you strace before your panic; do you see gdk_pixbuf 2.38 trying to load the png module for gdk_pixbuf 2.36?
<srhb>
samueldr: gtk_cell_accessible_parent_get_row_header_cells is not exported despite being in the gir files, so gi-gtk linkage breaks. ghc has a bug (that's fixed in 8.8.1) that causes it to panic with linker not initalized when -jN for N != 1, instead of emitting the linker error for some reason
<srhb>
At least I think that's what's going on.
<samueldr>
ah, if it isn't gdk_pixbuf then I have nothing to contribute, sorry :)
<srhb>
:-)
<srhb>
I'm just venting, I don't want to rebuild gtk, but what can you do.. :-P
<srhb>
samueldr: Thanks, I'm likely to run into that.
johnnyfive has joined #nixos
ub has quit [Quit: Leaving]
<LnL>
neminis: what kind of variables?
* srhb
sort of wants to grep pkgs.*.src for "gtk_cell_accessible_parent_get_(row|column)_header_cells" now...
<neminis>
LnL, well i have gtest installed system-wide, i use nix-shell to inherit gtest and then my builder script knows ho
<dckc>
in install-multi-user, in _sudo(), how do I add `umask 022` to sudo "$@"? I tried sudo sh -c "umask 022; $@" but that's not quite right somehow. Would somebody please go back to 1970 and change shell quoting to be sane?
<neminis>
how to set the environment variables for make
<LnL>
you're building something outside of nix?
kreisys has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
greymalkin has joined #nixos
<neminis>
yes it's not yet a full derivation, i am trying to build the executable of my project using the builder i set up in the .nix file for that project
arjen-jonathan has quit [Ping timeout: 240 seconds]
knupfer has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @catern opened pull request #55722 → pythonPackages.pkgconfig: use same setupHook as pkgconfig → https://git.io/fh7qN
shibboleth has joined #nixos
<dckc>
@simpson, any chance you have a clue for me on multi-user nix install with umask 077 in pam config?
<simpson>
dckc: Not really. Multi-user Nix on non-NixOS has been pain and suffering every time; I wouldn't bother.
<simpson>
Like, given your current trajectory, maybe containers are a better path than trying to bend CentOS or SUSE into behaving.
ixxie has joined #nixos
<averell>
what is considered a feature under the feature freeze?
dermetfan has joined #nixos
<WilliamHamilton[>
hi, I get an error when issuing the command `nix-prefetch-github Z3Prover z3`
rycwo has joined #nixos
<WilliamHamilton[>
I'd like to know if you can reproduce it, and why it happens
<dckc>
ok... any particular sort of containers, simpson?
<ottidmes>
WilliamHamilton[: no bug here
<srhb>
WilliamHamilton[: I get a hash mismatch.
<ottidmes>
WilliamHamilton[: I get back a seemingly valid JSON response
<WilliamHamilton[>
srhb: I get a hash mismatch too! But let me post the error for the others
<ottidmes>
srhb: you are on unstable?
<srhb>
WilliamHamilton[: I suspect it's a newer nix version doing that?
<srhb>
ottidmes: Yep.
<ottidmes>
WilliamHamilton[: yeah got to be that, the new error mismatch
<{^_^}>
seppeljordan/nix-prefetch-github#4 (by bbigras, 2 weeks ago, open): Error: Internal Error: Calculate hash value for sources
<dckc>
in fact, my nix goal is to address a problem we're having with containers: we run into different issues in test and prod, and copying the image takes forever. (or something. I wasn't doing that part of the work)
sigmundv has quit [Ping timeout: 244 seconds]
<WilliamHamilton[>
ok, thanks, so either I downgrade or do it manually, right? I'll try manually
<bbigras>
I reported that issue if someone needs me to test anything.
<dckc>
we had R-Studio server running in a container, but adding some R packages is really unwieldy, so we're looking at installing outside containers. The R-Studio server part is sorta done, and I got R-with-packages done with nix, and I was hoping to put them together. oh well.
<srhb>
WilliamHamilton[: But now we have "hash mismatch in fixed-[...]" etc. :)
<srhb>
Should be simple to try a regex for both.
<srhb>
bbigras: WilliamHamilton[ ^ if any of you have got the python chops :)
<WilliamHamilton[>
srhb: just for my knowledge, how would I get the right sha without prefetch?
<bbigras>
you could use the git (not github) one maybe
<srhb>
WilliamHamilton[: It's in your error message.
<srhb>
WilliamHamilton[: Got: ...
<WilliamHamilton[>
srhb: btw, about the z3 haskell-bindings issue, tell me if you want me to prepare a pr without the offending functions for haskell-z3
<bbigras>
I use `nix-prefetch-git` meanwhile
<WilliamHamilton[>
srhb: thanks!
<srhb>
WilliamHamilton[: But using fetchFromGitHub and providing a wrong hash works just fine.
<srhb>
WilliamHamilton[: Oh, right.. I don't actually use z3, I just dove into the debugging.. :-P If you want to, go for it!
<srhb>
WilliamHamilton[: Nice, do poke if you need testing :)
<WilliamHamilton[>
thanks ottidmes !
kreisys has joined #nixos
reinhardt has joined #nixos
<WilliamHamilton[>
srhb: sure :)
ryantrinkle has quit [Ping timeout: 246 seconds]
<jD91mZM2>
Ok so I am making a nx-libs recipe, and compiling that repository creates both the nxagent and nxproxy binaries. Now, there already is a recipe for only nxproxy which doesn't compile nxagent. I don't think I can disable nxproxy compilation from nxagent, however. What do I do? 1) Delete the nxproxy binary afterwards, even if it still has to be compiled. 2) Deprecate the existing nxproxy recipe and just use
<jD91mZM2>
nx-libs. 3) Just leave both
rycwo has quit [Remote host closed the connection]
<mdash>
jD91mZM2: probably #2, perhaps putting nxagent and nxproxy in different outputs if that makes sense
<averell>
that would be ideal
<averell>
i'd love to see an x2goserver module
yokyolu has quit [Remote host closed the connection]
Soo_Slow has quit [Remote host closed the connection]
<pbb>
hi, I am setting up a raspberry pi with a hifiberry amp (a little board you put on top and it works as an amplifier). with linuxPackages_rpi a kernel module is missing: snd_soc_bcm2708_i2s
knupfer has joined #nixos
<pbb>
how can I get this kernel module
<pbb>
it seems to be included as a module by default in the rpi kernel
<pbb>
(but on nixos it's not there)
<jD91mZM2>
mdash: Multiple outputs but still one package? That sounds cool
<jD91mZM2>
There's also some library stuff and I'm not sure what belongs to what
phizzz has joined #nixos
nschoe has quit [Ping timeout: 268 seconds]
ryantrinkle has joined #nixos
jasongrossman has joined #nixos
<pbb>
hm, the module seems to have gone in some kernel version and is not needed anymore.
<pbb>
then I'm really wondering why I can't see the sound card
<phizzz>
when i run a nix-shell based on a shell.nix, it sometimes appears to rebuild. isn't the environment only supposed to be built once?
<catern>
hey #nixos, I'm prototyping a new service infrastructure, can anyone suggest a relatively complex service packaged in Nixpkgs? e.g., a webapp with a dependency on a database service backend and a load balancer/proxy/something in front?
<srhb>
catern: Hydra is pretty close
<catern>
interesting point
<cransom>
if you collect garbage or update a channel, it will build new.
<srhb>
It's.. Probably not an ideal example with the weird postgres setup. But you get the idea from reading it, I think :)
<phizzz>
on average once a day, when i enter the nix-shell, it does: these paths will be fetched.... copying path '/nix/store/blah-stdenv-linux' from 'https://cache.nixos.org'...
<pbb>
catern: pixelfed.org
<catern>
Hydra requires root though right? or root equivalent?
<catern>
Because it messes with the nix-daemon in trusted-user ways?
<catern>
pbb: is this in nixpkgs?
<phizzz>
cransom: was that comment for me?
<srhb>
catern: I don't think so, but now that you ask I'm suddenly not sure.
<pbb>
catern: nope, it's a php application and it's a pain to package
<srhb>
catern: I think the only root-access happens during setup.
<catern>
I suppose it's not actually an issue - it might be cool to be able to run Hydra in a container with its own dedicated nix-daemon
<srhb>
catern: (Setting up the db users etc.)
<tilpner>
catern: Maybe matrix-synapse, if you don't want to do anything with nix-daemons
<srhb>
catern: gitlab is also an example
<pbb>
oh gitlab is such a pain
<pbb>
with all its different services
<catern>
OK, to be clear I don't actually want things that are *painful* :) just non-trivial...
<srhb>
:-P
<catern>
(maybe I will do gitlab though, if it does have a bunch of services with deps between them...)
<srhb>
catern: It does.
knupfer has quit [Ping timeout: 268 seconds]
<catern>
(what would be really nice is if I could identify some webapp or application that I actually care about, so I'm more motivated to set it up :) )
<pbb>
phabricator is also similar, there is something in nixpkgs but it's so old you can forget that
<catern>
(hydra kinda falls in that category...)
WhittlesJr has joined #nixos
<catern>
pbb: to be clear I don't want to package something new - I want something that is already packaged and working in Nixpkgs - because I'm going to port it to my new thing
<pbb>
ok, I see
<srhb>
catern: If this is to provide an example, I find that anything that is a "simple service that listens locally and needs an nginx proxy" tend to be nice, clear examples of how it works.
<jD91mZM2>
I think it should be made a little more obvious in the PR template that nix.useSandbox is true by default. Sure threw me off, and I didn't mark them as such when I could have :)
<srhb>
jD91mZM2: It'd need a lot of disclaimers unfortunately. (only on recent Nix, only on Linux, ...)
<srhb>
But yes!
phreedom has quit [Ping timeout: 256 seconds]
phreedom_ has joined #nixos
<neminis>
in which pkg is "ar" ? how can I find it?
<ar>
,localte bin/ar
<ar>
,locate in/ar
<{^_^}>
Couldn't find in any packages
<ar>
can't type today
<ar>
,locate bin/ar
<{^_^}>
Found in packages: swift, arduino, androidndk, arduino_core, binutils-unwrapped
<ar>
neminis: i am currently located in warsaw hackerspace, Wolność 2A, 01-018 Warszawa, Poland ;)
<neminis>
lol
<tilpner>
ar++
<srhb>
ar: If there ever was a moment to shine... :-)
<neminis>
thx for the help
jD91mZM2 has quit [Ping timeout: 245 seconds]
<steveeJ>
samueldr: that's a really weird issue with that brother pinter driver ignoring the page size. how deep have you investigated this?
<samueldr>
not really
<samueldr>
99% of my use is through cloud printing because printers are evil
<samueldr>
(and is 99.999% of the time left unplugged from the power)
<samueldr>
steveeJ: it might as well be a dumb thing easy to fix for someone that knows their way around linux printing, or even just looks :)
agander has quit [Ping timeout: 250 seconds]
ikitat has joined #nixos
<steveeJ>
samueldr: do you think it's Nix specific?
<samueldr>
I don't know, haven't got a non-nixos machine to test
xkapastel has quit [Quit: Connection closed for inactivity]
<samueldr>
(and once my printing needs were over I relegated this to the bottom of the tasks list)
<steveeJ>
those are two understandable points :)
<samueldr>
but my guts tell me it shouldn't happen on platforms for which the driver was written for; otherwise it'd have been spotted already?
<samueldr>
my two best assumptions are that either the package assumes it can write system files somehow, or there is *something* missing from the packaging that makes it work, but not entirely
<steveeJ>
further than writing the nix expression for the driver I don't know anything about cups. are the job parameters passed via the command line or stdin to the filter/dcriver?
<steveeJ>
I think we'd see an error in your former case. the latter is tough
stites has quit [Quit: ZNC 1.6.6+deb1ubuntu0.1 - http://znc.in]
matijja has quit [Ping timeout: 246 seconds]
stites has joined #nixos
JosW has joined #nixos
<{^_^}>
[nixpkgs] @vincrusher opened pull request #55725 → Fix alpha-embedded Target on Hydra → https://git.io/fh7OG
zupo has joined #nixos
rauno has joined #nixos
bpa has joined #nixos
<pbb>
has anyone used device tree overlays with nixos on the raspberry pi?
<pbb>
dtdebug=1 in the boot configuration does not give me a single line in dmesg
kreisys has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<judson>
I'm using the vimUtils packages, and it seems like autoload directories aren't getting loaded. I need to dig in more, but does anyone recognize that issue?
<bbarker>
has anyone used NixOS on Azure? sorry for what may be a naive question; i saw some github issues around it, and am not currently an Azure user, but may be one very soon ...
rpg has joined #nixos
<samrose>
running into the error src/libexpr/primops.cc:56: void nix::EvalState::realiseContext(const PathSet&): Assertion `store->isStorePath(ctx)' failed. on nixos-install although the iso builds without error on hydra
acarrico has quit [Ping timeout: 246 seconds]
ddellacosta has joined #nixos
xkapastel has joined #nixos
<{^_^}>
[nixpkgs] @7c6f434c merged pull request #55138 → random cleanups and a tiny fix → https://git.io/fhSih
endformationage has quit [Ping timeout: 244 seconds]
<bbigras>
I don't understand yet which is the best way to package Rust applications. For my stuff at work I build using `nix-shell` and then ship with docker with `pkgs.dockerTools.buildLayeredImage`
<samrose>
ottidmes bbigras this is *exactly* what I am struggling with now too
<samrose>
earlier today in this channel it was confirmed to me that nix-shell is not the way to go
<ottidmes>
bbigras: I will try that one thanks, it will probably be as slow as my current test case du-dust
<samrose>
I got a clue earlier today about how to possibly build up the version of rust and cargo that would live behind `rustPlatform`
<ottidmes>
samrose: nixos-unstable + rustPlatform.buildRustPackage + RUSTC_BOOTSTRAP = 1; seems to be the best you can do, if you want to contribute it to Nixpkgs as well at some point
<samrose>
ottidmes: yes if I actually figure it out, I will definitely try to contribute it
<ottidmes>
samrose: that hackernews link I gave you last time did not pan out, it worked for me after some tweaking
alex`` has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @Ma27 pushed commit from Averell Dalton to release-18.09 « nextcloud-client: fix crash »: https://git.io/fh7sT
jabranham has joined #nixos
jabranham has joined #nixos
jabranham has quit [Changing host]
<yokyolu>
nix-copy-closure fails with `cannot connect to ...`, -vvv says that `bash:nix-store: command not found` after `sending: nix-store --serve --write`
ikitat has quit [Remote host closed the connection]
ikitat has joined #nixos
alex`` has joined #nixos
lawlesseel has quit [Remote host closed the connection]
<srhb>
yokyolu: Sounds like nix-store is not on PATH on the remote end?
<clever>
yokyolu: also, that error message is far better then it used to be, it used to treat "command not found" as a series of 32bit ints, and then complain with the 1st int, lol
<monotux>
just wanted to say how I've missed using nixos, setting up things (and using version management with it) in configuration.nix is so handy!
<yokyolu>
hm
rycwo has joined #nixos
<bsima>
monotux: +1 - I use nix on macos, for generating docker images, for all my configs (home-manager)...
<bbigras>
Is there a filter command or something that I can use with `src =` to include only some files from the current folder?
<yokyolu>
ssh ... echo $PATH
<srhb>
bbigras: filterSource
<yokyolu>
.../nix/var/nix/profiles/default/bin...
ambro718 has joined #nixos
<bbigras>
srhb: thanks!
<stites>
anyone have tips for setting up nixos on zfs with a mirrored root pool? I tried to use the shortcut of https://github.com/a-schaefers/themelios but upon boot I got "mounting zroot/NIX/nix on /mnt-root/nix failed""
<srhb>
yokyolu: (different shell rc files are usually loaded depending on the terminal etc)
<stites>
srhb: yes, I think so
<srhb>
stites: There's no difference from non-mirrored really, except for adding the mirror :)
<yokyolu>
srhb: that explains it
Ariakenom has joined #nixos
<pareidolia>
Is there a way to add services to a slice in NixOS?
<srhb>
stites: (None of that affects the way the datasets are handled regarding mounts etc.)
<stites>
srhb: so I'm currently getting a failed stage 1. Should I drop into the interactive shell, remount, and try to regenerate hardware-configuration.nix?
<{^_^}>
[nixpkgs] @dotlambda pushed commit from @pmiddend to master « pythonPackages.typing-extensions: 3.6.6 -> 3.7.2 (#55719) »: https://git.io/fh7sV
_kwstas has quit [Remote host closed the connection]
Glider_IRC_ has quit [Ping timeout: 246 seconds]
<monotux>
bsima, I currently only use it for a VPS, but will probably return to using it on my laptop at some point. I have a FreeBSD machine with a lot of jails that I manage by hand (and tooling, of course) but it feels so risky as it would require a lot of work to replicate it, while using a NixOS machine with a single textfile (plus all application data,
<monotux>
of course) feels so much more...correct!
_kwstas has joined #nixos
_kwstas has quit [Remote host closed the connection]
vk3wtf has joined #nixos
zupo_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<stites>
hmm... $stage2Init is set to some location in the nix store, but when I pivot I get a directory not found (going to try addressing that)
rcshm has quit [Remote host closed the connection]
<clever>
stites: my only thought right now is that its maybe slow for one of the vdev's to come online, and stage2 is giving up
_kwstas has joined #nixos
<clever>
stites: you could also try using /nix/var/nix/profiles/system/init as your stage2 init
_kwstas has quit [Remote host closed the connection]
<stites>
but in the stage-1-init.sh is set to /init, though which is a special type of file
<{^_^}>
[nixpkgs] @oxij opened pull request #55732 → all-packages.nix: move defaults to package files continues → https://git.io/fh7so
<stites>
damn. I ran a `mkdir -p $stage2Init && pivot_root /mnt-root $stage2Init` but got a " device is busy"
<gchristensen>
anyone know where I can find the official documentation on shebang?
<pareidolia>
symphorien: Does the slice configuration support resource limiting?
aleph- has joined #nixos
<symphorien>
systemd.slices.<name>.sliceConfig is passed as is to systemd as the [Slice] section of the unit, and systemd.slices.<name>.unitConfig similarly as the [Unit] setion
<symphorien>
all other options are just "convenient shortcut" to using those two
_kwstas has joined #nixos
zupo has joined #nixos
_kwstas has quit [Remote host closed the connection]
<symphorien>
so everything systemd supports with unit files is possible with the nixos module system
zupo has quit [Client Quit]
<pareidolia>
symphorien: Ah, those are "freeform" maps that won't give me the "option not defined" error
<stites>
clever: I think I'm just going to do a fresh reinstall. I was mostly using the "a-schaefers/themelios" script as a quick provision, but I think I need to do a more detailed install
<symphorien>
pareidolia: yes you lose "typecheking"
<symphorien>
also merging can be problematic
<pareidolia>
I know hehehe
<pareidolia>
I was already praying and looking up
<pareidolia>
Ah, I totally didn't notice that the options search has pagination xD
<pareidolia>
I see it now
<symphorien>
at some point, I investigated if it was possible to have systemd check the syntax of our unit files at compilation
<clever>
stites: you could also upgrade it from a single vdev to a mirror, using zpool attach
<symphorien>
but it turns out systemd-analysze verify needs a running systemd to work
markus1189 has quit [Ping timeout: 245 seconds]
<symphorien>
so it is impossible to use it in the sandbox
_kwstas has joined #nixos
markus1189 has joined #nixos
_kwstas has quit [Remote host closed the connection]
npmccallum has joined #nixos
<stites>
@clever yeah I think that'll be my go-to. I'll just zap the other disk.
<stites>
I'm hoping to preserve a seperate freebsd datastore -- would you happen to know if the linux-zfs will still be able to read the freebsd-zfs disks?
_kwstas has joined #nixos
_kwstas has quit [Remote host closed the connection]
<clever>
stites: i think it will work both ways, as long as you dont touch `zpool upgrade` or `zpool enable`
shibboleth has joined #nixos
erasmas has quit [Quit: leaving]
<pareidolia>
symphorien: Systemd NetClass directive is apparently removed. Great.
<stites>
clever: thanks for the pointers! for the `justdoit` script, when are you running that script?
<stites>
it looks like you are just adding it to the systemPackages
agander has quit [Ping timeout: 250 seconds]
<clever>
stites: that module is added to the installer image, so the binary is in $PATH when you boot the installer
<clever>
stites: the main thing in that directory, is a set of nix scripts to generate some kexec utils
<{^_^}>
[nixpkgs] @dotlambda pushed commit from @pmiddend to master « pythonPackages.androguard: 3.3.3 -> 3.3.4 (#55728) »: https://git.io/fh7s5
<clever>
stites: so you just upload the tarball to any linux machine, unpack it to /, run /kexec_nixos, and boom, its running nixos entirely from ram
acarrico has joined #nixos
<clever>
ssh back in, justdoit, and boom, its now nixos, all previous data is toast (youll want to not do that, i believe)
<symphorien>
pareidolia: can you check with nixos-option for example that this is not just an artifact of merging ?
_kwstas has joined #nixos
_kwstas has quit [Remote host closed the connection]
ixxie has quit [Quit: Lost terminal]
_kwstas has joined #nixos
<pareidolia>
symphorien: Systemd says Support for option NetClass= has been removed and it is ignored
<stites>
I think it's a little late for me to use this, now, since I've already nuked the two SSDs I use for zroot
<symphorien>
ah I thought you complained that the option would silently remove this attribute
<pareidolia>
I'm trying to limit bandwidth of a couple services together
<clever>
stites: you could also make a modified ISO with justdoit.nix in the imports, or just do a "normal" install to a usb stick, with justdoit.nix in the imports section
<pareidolia>
Every time I find an approach, it is deprecated for some reason
<symphorien>
pareidolia: if they share a gid, you can do it with iptables
<symphorien>
(or an uid)
yokyolu has quit [Ping timeout: 245 seconds]
<pareidolia>
symphorien: I could add both users to a common group, yes. Can you point me to a description of this function
<pareidolia>
Because every time I get stuff about bloody "tc" or "trickle"
<symphorien>
man iptables-extension, search for --gid-owner
<symphorien>
this is for selecting packages
<symphorien>
*packets
<pareidolia>
Ah, tc is an interface to iptables. Heh.
<pareidolia>
"No manual entry for iptables-extension"
<pareidolia>
ah, should be "extensions"
<symphorien>
and search for --limit for rate limiting
<symphorien>
hum this is in packets per unit of time
knupfer has quit [Remote host closed the connection]
vk3wtf has quit [Ping timeout: 252 seconds]
<pareidolia>
symphorien: Ok, so the group doesn't have to be the primary group of the users?
<stites>
@clever I just kicked off a `justdoit`!
<symphorien>
no idea
<clever>
:D
<symphorien>
actually I have never used this
<pareidolia>
We'll see
arjen-jonathan has joined #nixos
dermetfan has quit [Ping timeout: 240 seconds]
<stites>
it looks like it just nukes sda, which is fine for me
<stites>
so my fingers are crossed
<stites>
ah, it's stalling on importing zroot again
<gchristensen>
I had a bug where nixos couldn't import the root on monday, but I think that has been fixed
<gchristensen>
also it isn't monday
<clever>
gchristensen: what about the timeout thing you added, to make it wait if half the pool devices are missing?
<gchristensen>
oh that wasn't me
<stites>
hmmmmm... I could be satisfied just loading nixos on a non-zfs root and using zfs to load a pool post-boot.
<stites>
but that isn't really documented anywhere
<stites>
is that because it's trivial to set up?
<clever>
stites: can you record a video of it booting and hanging?
zupo has joined #nixos
<stites>
sure...
<Ariakenom>
I was using the live CD on usb to install nixos. Does anyone know if the wireless set up is supposed to work like when booting from usb? I didn't get it to work.
<clever>
Ariakenom: you need to set wireless.enable = true; in your configuration.nix file, and then nixos-rebuild switch
<lejonet>
Hmm, if I want a interface link to be set up, but no IP assigned to it, would I just create a networking.interfaces.<name> with an empty ipv4.addresses or something to achieve that?
<Ariakenom>
clever: I did set that before the install
ambro718 has quit [Ping timeout: 240 seconds]
<noonien>
Ariakenom: i prefer enabling networkmanager, instead of doing wireless.enable
<noonien>
i don't really like using wpa_supplicant
<clever>
Ariakenom: then you just need to run wpa_passphrase, and feed the resulting config into /etc/wpa_supplicant.conf, or abandon supplicant and use NM
<pareidolia>
symphorien: I've got it working when I target the transmission user, but not yet the group...
<vaibhavsagar>
Ariakenom: this has bitten me every time I've tried to install NixOS on a new computer
<symphorien>
then target the user; is it a problem ?
<symphorien>
btw, doesn't transmission support rate limiting ?
<pareidolia>
Yes, but I want to lump it together with some other stuff :)
<symphorien>
ah I see
<Ariakenom>
I suppose the live CD was using networkmanager but the tutorial recommended wpa_supplicant?
<lejonet>
Ariakenom: I would also give you a hint to look at iwd, if you don't like networkmanager and think wpa_supplicant is too much work
ninjatrappeur[m] has joined #nixos
<pareidolia>
symphorien: Ah, as I suspected
<pareidolia>
Only the primary group is examined for the criterium
<pareidolia>
Oh well, no problem defining a rule for earch user
kalbasit has joined #nixos
<pareidolia>
They all go through the same bottleneck
<symphorien>
you can define a chain for everything which must be rate limited
<pareidolia>
Yep
eadwu has joined #nixos
<Ariakenom>
Thanks for your help.
<Ariakenom>
I have more info for when I'll give it another go
<clever>
`zpool detach pool device` can be used to downgrade that mirror to a single-disk pool, but this is a new/different problem, and its not booting from the justdoit based install
xkapastel has quit [Quit: Connection closed for inactivity]
npmccallum has joined #nixos
drakonis has quit [Quit: WeeChat 2.3]
<stites>
So I think this is pretty good. justdoit didn't zap the old sdb -- I think I would just have to boot into the livecd again and zero out /dev/sdb, then run justdoit again and it should work.
<stites>
right?
npmccallum has quit [Client Quit]
<clever>
stites: you likely dont even have to re-run justdoit, or even zero out sdb, just tell the bios to boot sda instead
<clever>
stites: just pay attention at grub, and note the date/time of the nixos its showing, that should reveal which disk it is
<clever>
or the name of the pool its importing, once it starts to boot
<phizzz>
i appreciate your help. i am having difficulty finding documentation for a lot of the nixos stuff
<instantepiphany>
Hi guys, I am trying to upgrade my router (a PC that runs nixos with a couple network cards) from 18.03 to 18.09. I have updated the `nixos` channel as root to 18.09, but when I run `nixos-rebuild switch` I get: https://pastebin.com/Zqjtq5T2
simukis has quit [Quit: simukis]
<instantepiphany>
"error 9 while decompressing xz file" is the thing that jumps out at me - not sure what this indicates. Some kind of chicken and egg problem with the build env?
<gchristensen>
phizzz: reading source code will definitelyhelp
<clever>
instantepiphany: if you run it again, what happens?
zupo_ has joined #nixos
<instantepiphany>
I'll try again - I tried 3 times in a row last night and had the same issue, but this morning I ran garbage-collect -d and verified everything in the store in case there was some issue there.
<phizzz>
gchristensen: can someone change the topic to "read the source first, then post your questions..." :P
<gchristensen>
phizzz: we aren't that rude!
<gchristensen>
we're glad to help :) even if it means _we_ read the source
ikitat has quit [Ping timeout: 244 seconds]
zupo has quit [Ping timeout: 246 seconds]
<phizzz>
just kidding, you all have been very helpful. i'd just rather find the info
* phizzz
looks for the source
<instantepiphany>
clever: it is downloading more paths now - if it has the same error again should I use --keep-going to isolate which path is having the error?
<gchristensen>
phizzz: search.nix.gsc.io will be helpful :)
<gchristensen>
ok, gotta run. good luck, phizzz!
<clever>
instantepiphany: yeah, that can help
<instantepiphany>
clever: https://pastebin.com/1mVadv2y - so my understanding is the build succeeded, but in switching to the new derivation something, somewhere, failed. What now?
<instantepiphany>
Probably due to my LANGUAGE and LC_ALL not being set.
<clever>
instantepiphany: and if that fails, `nix-store --verify --check-contents`
alex`` has quit [Ping timeout: 245 seconds]
<instantepiphany>
clever: happy to do that, but please correct me if I misunderstand - the only error in my latest paste is that perl complains that it can't find supported & installed locale settings - could that simply be I have misconfigured my locale somehow?
ikitat has joined #nixos
vk3wtf has quit [Quit: WeeChat 2.3]
<clever>
instantepiphany: `Compilation failed in require` looks like a worse issue, that stopped switch from switching
<instantepiphany>
clever: Verifying nix store had 30 or so paths show this type of message: `path '/nix/store/n6x9cnkk6b4q4wg3b74v6z178d3ah81r-file-5.34' was modified! expected hash 'sha256:1jar6j26nzy3yqqzh3jqy3b1vw9dh72rv8y7wbrm3sjyr1y3934r', got 'sha256:038zcbaiynysrgbn23fdmjwad38516dj2fjc1ysmhzxyiq7dq5cb'`. Possibly bad RAM or HD?
vk3wtf has joined #nixos
<clever>
instantepiphany: sounds like an improper shutdown might have corrupted the disk
<clever>
instantepiphany: run `nix-store --repair-path` on each of those paths
<instantepiphany>
Yeah there were a few improper shutdowns last night as the nvidia derivation was causing a system crash somehow, I disabled the driver and xserver as I don't use this machine graphically. I would have thought paths should *not* be added to the nix store until they have had their hash verified.
<clever>
instantepiphany: the problem is that some filesystems like ext4 can loose the contents of a file at an improper shutdown
<clever>
so after adding the path to the store, its contents just change
rpg has quit [Ping timeout: 257 seconds]
<instantepiphany>
clever: Oh, like a queued write to disk - I should have tried REISUB :)
ikitat has quit [Ping timeout: 272 seconds]
<clever>
instantepiphany: yeah
<clever>
personally, ive switched fully to zfs, and ive not had any issues like that on it
acarrico has quit [Ping timeout: 250 seconds]
schjetne has joined #nixos
<instantepiphany>
zfs is on my "would like to have time to try that" list :)