<eqyiel>
is it possible to disable validation of the manual on unstable? I thought `services.nixosManual.enable = false;` would do it, but some recent change is causing it to fail with `manual-combined.xml fails to validate`
silver has quit [Read error: Connection reset by peer]
<clever>
eqyiel: it sys the nixos-manpages and home-manager depend on it
<eqyiel>
oh, derp
<eqyiel>
it's probably due to home-manager then
<eqyiel>
clever: thanks
<clever>
yeah, you need to tell home-manager to also not build nixos man pages
thc202 has quit [Ping timeout: 260 seconds]
UNIcodeX__ has joined #nixos
isHavvyGhosting is now known as isHavvy
benley has quit [Quit: WeeChat 1.9.1]
markus1189 has quit [Ping timeout: 256 seconds]
UNIcodeX_ has quit [Ping timeout: 256 seconds]
TonyTheLion has quit [Quit: leaving]
markus1189 has joined #nixos
maingo has quit [Quit: Connection closed for inactivity]
<disasm>
gchristensen: still around?
romildo has joined #nixos
<gchristensen>
yep
<disasm>
gchristensen: you had pinged me earlier :) Looks like we're both on at the same time now.
<gchristensen>
w00t
muzzy has joined #nixos
romildo has quit [Client Quit]
<jackdk>
On nixpkgs unstable, I cannot run `nix-shell -p "haskell.packages.ghc822.ghcWithPackages(p: [p.blank-canvas])"` because it's pulling base-compat-batteries-0.10 which pulls base-compat-0.10 but also text-show-3.7 which pulls base-compat-0.9.3. I note that `hackage-packages.nix` has "base-compat" and "base-compat_0_10_1"; should there be similar for
<{^_^}>
→ ef30e659 by @dtzWill: gllvm: 2018-04-28 -> 1.2.0
<{^_^}>
→ 54b90e5f by @dtzWill: Merge pull request #39900 from dtzWill/update/gllvm-1.2.0
<elvishjerricco>
jackdk: That probably means that someone did `foo = super.foo.override { bar = self.bar_x_y_z; };` in one of the `configuration-xyz.nix` files. I consider this *extremely* bad form, since it leads to inconsistencies in the package set like you're seeing here. It's better to just change `bar` in the package set rather than changing it in one specific dependent.
jperras has joined #nixos
<jackdk>
that someone was me. I'll dig up the PR, and if you can help me get this right I'll put in another one
<elvishjerricco>
jackdk: So the base package set is based on a Stackage LTS, which has base-compat-0.9.3, but no base-compat-batteries (meaning those have been added by us). It's usually better to leave the Stackage-derived packages as untouched as possible, so I'd probably prefer it if we simply pulled in older versions of base-compat-batteries and its dependents that are compatible with base-compat-0.9.3.
freeman42x]NixOS has quit [Ping timeout: 240 seconds]
markus1189 has quit [Ping timeout: 240 seconds]
mbrgm has quit [Ping timeout: 256 seconds]
<jackdk>
okay, so how is hackage-packages.nix generated and which version of a package does it decide to cabal2nixise?
mbrgm has joined #nixos
<clever>
jackdk: check the git history on it, and read the most recent commit msg
hakujin has quit [Ping timeout: 255 seconds]
<jackdk>
clever: clever. thanks
<jackdk>
ah, base-compat-batteries never had an 0.9 release. that's probably why this is gnarly
<{^_^}>
[nixpkgs] @romildo opened pull request #39902 → enlightenment: add dependency on mesa_noglu → https://git.io/vpgW1
muzzy has quit [Ping timeout: 260 seconds]
aarvar has joined #nixos
<jackdk>
so if we have no stackage-lts-compatible base-compat-batteries (even lts-11.7 is still on base-compat-0.9.3), what's the best way forward?
aarvar has left #nixos [#nixos]
LysergicDreams has quit [Quit: Lost terminal]
<elvishjerricco>
jackdk: Hm. That's disappointing. Is there a previous version of blank-canvas that doesn't depend on base-compat-batteries?
LysergicDreams has joined #nixos
<jackdk>
yeah the next-to-recent might work since it only depends on base-compat >= 0.8 && < 0.10. how do I ask for a specific version of a package when I do a ghcWithPackages call?
<elvishjerricco>
jackdk: Well for one thing, we should make the default `blank-canvas` be that version instead of current version. But until that's done you can do `ghcWithPackages (p: [(p.callHackage "blank-canvas" "x.y.z" {})])
<elvishjerricco>
jackdk: Would be good to submit a PR to configuration-hackage.yaml to make that change the default
<jackdk>
yeap, happy to do that. It looks like a line in the extra-packages dict?
blonkhart has quit [Ping timeout: 260 seconds]
blonkhart has joined #nixos
<spear2>
where does sshd_config go? i don't find it in /etc/ssh/sshd_config?
<spear2>
it's kept in the store?
daveo has quit [Ping timeout: 264 seconds]
jrolfs_ has joined #nixos
hakujin has joined #nixos
<jackdk>
elvishjerricco: I've added a line to configuration-hackage2nix.yaml but I'm having trouble making update-nixpkgs.sh run. It's complaining of an unknown package monad-par-extras, and I've checked out all-cabal-hashes as a subdir
<elvishjerricco>
jackdk: You don't need to run update-nixpkgs.sh yourself. Just check in the change to configuration-hackage2nix.yaml and it'll be there next time peti updates the package set.
<jackdk>
but how else can I test that I haven't broken the build?
hakujin has quit [Ping timeout: 256 seconds]
jrolfs_ has quit [Ping timeout: 260 seconds]
<jackdk>
elvishjerricco: also, should I revert the override on the dep of base-compat-batteries, which is setting base-compat = base-compat_0_10_1?
Fare has quit [Ping timeout: 264 seconds]
<elvishjerricco>
jackdk: I guess we should keep the base-compat-batteries change, since it doesn't build otherwise.
<elvishjerricco>
I just never test my changes to configuration-hackage2nix.yaml, since it's pretty basic :P But I guess you're right; if you figure out how to test the change, let me know
<jackdk>
thanks for your help elvishjerricco. I'll have to leave it there for now
muzzy_ has joined #nixos
iqubic` has joined #nixos
iqubic` has left #nixos [#nixos]
sir_guy_carleton has quit [Quit: Leaving]
iqubic has quit [Ping timeout: 255 seconds]
iqubic has joined #nixos
fresheye1 has quit [Ping timeout: 268 seconds]
hakujin has joined #nixos
<zybell>
gchristensen:mosh runs on UDP and is compat to ssh. But better is:no large files in the first place. Someone at debian(?not sure)uses a very small PXE image that essentially only includes curl and the boot is continued over http,where you dont have to change an image but single files.
<jackdk>
is this running at a different level to ipxe?
muzzy has joined #nixos
<gchristensen>
I essentially have no choice but to make a large pxe image.
hakujin has quit [Ping timeout: 248 seconds]
muzzy__ has joined #nixos
drakonis has joined #nixos
<muzzy_>
Hey can someone explain, in simple terms, what exactly Nixops is?
hakujin has joined #nixos
<bhipple[m]>
NixOps is basically terraform for NixOS + cloud. It allows you to write Nix expressions that declaratively specify a set of machines, cloud resources, what should be on those machines, etc.
<jackdk>
gchristensen have you looked at ipxe? you can chainload it from your vendor pxe rom if you don't have the ability to flash the rom, and it can download kernel/initrd over http
<gchristensen>
yes, it is actually ipxe
<muzzy_>
okay. Does it install nixos to those machines or do the machines have to exist as Nixos? Sorry I just need to get this clear in my head
<Myrl-saki>
Is there a musl stdenv?
<bhipple[m]>
muzzy_: The machines don't need to exist; it can create them from thin air in EC2 using the API, then install your NixOS on it, etc. Same for managing attached storage, AWS accounts, etc.
benley has joined #nixos
<justanotheruser>
Does anyone have a better guide to deploying nixos AMIs than this? https://nixos.wiki/wiki/Install_NixOS_on_Amazon_EC2 ? It references outdated environment and a command, variables, "ec2-add-keypair", which googling with quotes doesn't show any useful documentation
<bhipple[m]>
muzzy_: The tutorial in the NixOps manual is actually quite good, and can be done in <30 minutes. It involves spinning up a small set of machines in a cluster with a nix expression
hakujin has quit [Ping timeout: 264 seconds]
<muzzy_>
alright, So I followed the nixops guide (with libvirtd) and deployed a webserver VM, interesting, it showed up in virt-manager. Is this akin to simply making a VM from a nixos ISO?
ivanivan has joined #nixos
Ridout has quit [Quit: Lost terminal]
Fare has joined #nixos
jrolfs has quit [Ping timeout: 264 seconds]
jrolfs has joined #nixos
jrolfs has quit [Ping timeout: 240 seconds]
blankhart has quit [Ping timeout: 240 seconds]
orivej has quit [Ping timeout: 240 seconds]
blonkhart has quit [Quit: WeeChat 1.9.1]
jperras has quit [Quit: WeeChat 2.1]
hakujin has joined #nixos
Fare has quit [Ping timeout: 264 seconds]
<spear2>
error: Nix daemon out of memory?
srl295 has quit [Quit: Connection closed for inactivity]
jrolfs_ has joined #nixos
<spear2>
forgot to swapon, do i need to re-generate hardware config?
hakujin has quit [Ping timeout: 240 seconds]
<spear2>
guess not it worked now =)
odi has joined #nixos
asuryawanshi has joined #nixos
asuryawanshi has quit [Remote host closed the connection]
jrolfs_ has quit [Ping timeout: 256 seconds]
<iqubic>
spear2: if you turn on swap, that might fix the memory issues.
<spear2>
yes that was it =)
<iqubic>
Until Nix 2.0 can do builds with less memory than it currently uses, I'm sticking with whatever the last 1.x release was.
<jcrben>
wondering if that's normal. also wondering if I'm going to run out of space before it's done
ivanivan has joined #nixos
drakonis has quit [Read error: Connection reset by peer]
Fare has joined #nixos
<ivanivan>
when I add a new variable to environment.variables in my nixos config, they don't show up in existing or new shells for my current user, but do show up for new logins such as `sudo -i`
<ivanivan>
can anyone enlighten me as to how to pick up new variables without rebooting?
<ivanivan>
i guess i need to log out and back into the x session?
<bhipple[m]>
ghostyy: you can add an overlay with `foo = <older version>`, or you can git checkout the nixpkgs commit that built your nixos, change the pkg, and rebuild from it if you want something more temporary
odi_ has quit [Read error: Connection reset by peer]
hakujin has quit [Ping timeout: 264 seconds]
krey has joined #nixos
<kuri0>
i edited patches.nix to include some of my patches but when i build the vm it uses the kernel from the cache server
<kuri0>
and doesn't build my custom patched kernel
Fare has quit [Ping timeout: 264 seconds]
ericsagnes has quit [Ping timeout: 256 seconds]
<dmj`>
peti around?
wangoe has quit [Ping timeout: 240 seconds]
WilliButz has quit [Ping timeout: 240 seconds]
zemm has quit [Ping timeout: 240 seconds]
zemm has joined #nixos
WilliButz has joined #nixos
wangoe has joined #nixos
aarvar has joined #nixos
asuryawanshi has joined #nixos
<muzzy_>
Do I use Nix-build to edit software before it compiles?
Rusty1_ has quit [Quit: Konversation terminated!]
<muzzy_>
xmr-stak isn't working for me. Apparently I need to use a different AMD library... so I'll need to supply that file before it compiles. How do I do this in nixos?
<sphalerite>
kuri0: you'll probably want to put your patches in the boot.kernelPatches option.
jedai42 has joined #nixos
jedai has quit [Ping timeout: 276 seconds]
griff_ has joined #nixos
hamishmack has joined #nixos
jD91mZM2 has joined #nixos
Myrl-saki has joined #nixos
lukec has quit [Ping timeout: 255 seconds]
<kuri0>
sphalerite, what if i do want to edit patches.nix ?
<jD91mZM2>
Hi everyone! I got NixOS installed yesterday, and I'm currently trying to set it up (apparently I had a lot more things installed than I remembered). My biggest issue so far has been the VPN. I got OpenVPN set up finally, but I have not yet figured out how to make a killswitch. UFW doesn't seem to be available, and Nix's default firewall doesn't seem to support limiting the outgoing connections. What would I
<sphalerite>
kuri0: you shouldn't need to edit patches.nix. You can just put your patches in your configuration.
<jD91mZM2>
use to do this?
<kuri0>
sphalerite, i want to edit it i'm making a fork of nixpkgs with some tweaks
<sphalerite>
jD91mZM2: yes, the default firewall configuration is fairly limited. You can add your own stuff using networking.firewall.extraCommands
<jD91mZM2>
Aha! Guess I don't have a choice but to learn iptables :P
<sphalerite>
kuri0: patches.nix only contains a set of available patches, they aren't enabled byefault
<sphalerite>
kuri0: you need to put them in boot.kernelPackages, use an override on the kernel package, or modify the set of patches passed to it in pkgs/top-level/all-packages.nix
<jD91mZM2>
How would I compile a custom program with runtime dependencies? Should I make a package?
<trikl[m]>
jD91mZM2: I use nftables for it
<sphalerite>
jD91mZM2: I have a friend who has a setup like that as well, I've asked them and can let you know when I hear back if you hang about here :)
<trikl[m]>
Hehe. The only relevant bit is the output chain
winem_ has joined #nixos
<sphalerite>
jD91mZM2: if it's for development, you probably want to use nix-shell because otherwise it will rebuild the whole thing on every small change
<sphalerite>
jD91mZM2: for actually using it in your configuration or deploying it to other machines with nix, yes, package it
<jD91mZM2>
I tried to run nix-shell -p libpulseaudio, and it works when compiling, but not when actually running
<jD91mZM2>
"libpulse.so.0: cannot open shared object file: No such file or directory"
<sphalerite>
It may be trying to load it dynamically at runtime rather than linking against it during the loading
<sphalerite>
in which case you'll need to use LD_LIBRARY_PATH
<jD91mZM2>
(Or package it?)
<sphalerite>
if it is dlopening it it will also need the wrapper with LD_LIBRARY_PATH in a package AFAIK
lukec has joined #nixos
<jD91mZM2>
Ah, bummer
<sphalerite>
you can check if the dlopen thing is the case using readelf -d <executable>
<jD91mZM2>
btw what's the difference between nix-shell -p and nix run?
robstr has joined #nixos
xy2_ has joined #nixos
<sphalerite>
if it has a NEEDED entry for libpulse you shouldn't be having this issue
<sphalerite>
does it have a RUNPATH entry with /nix/store/<hash>-libpulseaudio-x.y.z/lib in it?
<jD91mZM2>
It has a bunch of other paths, but not libpulse for what I can ee
<eisbaer>
hi, i'm trying to forward port 22 from a vm on my host to the internet, but i can't get it to work with "networking.nat". when i check the iptables rules they seem to be okay.
<sphalerite>
I have my hopes for getting that working before 18.09
<jD91mZM2>
Btw when instaling binutils I'm getting priority issues for `ld`. Which one should I give the most priority? gcc or binutils?
<sphalerite>
Not install it :/
<jD91mZM2>
Why not?
<sphalerite>
basically the way the tools are packaged is not actually suited for installing into user environments, so it doesn't work the way you might expect most of the time
<sphalerite>
oh actually I do have binutils installed to have strings and readelf available
<sphalerite>
but yeah I don't recommend having gcc installed, because it just doesn't work very well standalone (rather than in nix-shell)
jacob_ has joined #nixos
hakujin has joined #nixos
<jD91mZM2>
Aha
<sphalerite>
Anyway, as for what the command does — the RUNPATH entry in an executable's dynamic section gives the loader additional paths in which to look for the libraries, much like LD_LIBRARY_PATH but embedded in the executable. That patchelf command modifies it to add the path to libpulse
<sphalerite>
a wrapper like LD_LIBRARY_PATH="$LD_LIBRARY_PATH:/nix/store/ilh6z1mr335dkr22jvzpx61x0vlwd4wc-libpulseaudio-11.1/lib" ./target/debug/xidlehook is probably more convenient though, and should work just as well
<jD91mZM2>
Your command didn't work for me on the release binary
<jD91mZM2>
It updated the binary, but it still complains when running
<sphalerite>
your path to libpulseaudio is probably different
<jD91mZM2>
Yeah I updated it
<sphalerite>
oh, huh.
<jD91mZM2>
I mean I probably just messed something up
<jD91mZM2>
Lemme try LD_LIBRARY_PATH
<jD91mZM2>
Oh wait
<sphalerite>
make sure the path you pass does actually contain libpulse.so.0 :)
<jD91mZM2>
I linked it to -dev
<jD91mZM2>
derp
<sphalerite>
:)
hakujin has quit [Ping timeout: 240 seconds]
<andreabedini>
hello, nix beginners here (on macos) with a question
<andreabedini>
my NIX_PATH is /nix/var/nix/profiles/per-user/root/channels
<jD91mZM2>
IT WORKS! YAY!
<andreabedini>
rather than /nix/var/nix/profiles/per-user/andrea/channels as I'd expect
<guesie>
hi! I'm having trouble installing nixos.. Actually I had the same issue before, but I don't remember how to solve it: I created a boot and a root partition, both ext4, however I'm getting: 'Warning.. GPT partition contains no BIOS boot partition... Embedding is not possible'
<jD91mZM2>
You need a BIOS partition, like it says
<jD91mZM2>
256MiB should be enough, fat32, with the legacy bootable flag
<sphalerite>
jD91mZM2: because the build environment is quite tightly coupled to bash, it sources stdenv's setup.sh which defined a bunch of functions
<sphalerite>
jD91mZM2: so in a nix-shell you can do `genericBuild` to just try building the package right where you are.
Tucky has joined #nixos
<sphalerite>
jD91mZM2: you can of course do nix-shell --run zsh (or whatever your favourite shell is), just that the functions won't be defined. That's acceptable for many use cases but not all
<guesie>
jD91mZM2: alright, actually I had fat before.. with the same error, but I did not set the legacy boot flag (I'll try to do that)
<jD91mZM2>
In fdisk you can press "x, A, <number>, r"
<sphalerite>
guesie: are you sure BIOS boot is what you want?
<[Leary]>
So when packaging a binary that looks for runtime dependencies in the PATH, it gets wrapped in a script that builds it a PATH variable holding the executables it will look for. But what if you're packaging a shell script instead? It seems a bit silly to wrap it with another. Is there a good way to generate the script at build time so that it directly calls ${program}/bin/program instead of looking for program in the PATH?
<makefu>
guesie: you need an extra "grub partition" for gpt. it does not fit into the mbr anymore
periklis has joined #nixos
<sphalerite>
guesie: you may want to set GRUB to use EFI boot. Unfortunately I need to go now though so I can't help you work that out
<Myrl-saki>
Why can't we use `config = { };` in nixops?
<guesie>
sphalerite: Actually I'm not sure if I need 'BIOS boot', I'm not too much into these things. But thanks for the hint, maybe someone else can help me out here
<jD91mZM2>
You do need some sort of boot partition, but if your system supports EFI then you of course want to use that
<guesie>
jD91mZM2: ok, I suppose it supports EFI. I created the EFI partition with fdsik.. whcih file system should I chose?
<jD91mZM2>
Choose the partition type "EFI System"
<Myrl-saki>
error: Module `:anon-1:anon-1' has an unsupported attribute `environment'. This is caused by assignments to the top-level attributes `config' or `options'.
<guesie>
jD91mZM2: did so
<jD91mZM2>
Type 1 I believe
<jD91mZM2>
Ok well then you're set
<guesie>
jD91mZM2: I guess I'd have to create a file system via mkfs.., no?
<jD91mZM2>
Nope, I don't think so at least
<jD91mZM2>
Otherwise just try fat32
<guesie>
jD91mZM2: ok, well see ;)
<jD91mZM2>
How come the NixOS repos have gotten an updated Discord, yet I'm on nixos-unstable and haven't gotten the update?
<jD91mZM2>
I've tried nix-channel --update and nix-env --upgrade
<Myrl-saki>
Uhh
muzzy__ has quit [Ping timeout: 256 seconds]
muzzy_ has quit [Ping timeout: 256 seconds]
muzzy has quit [Ping timeout: 256 seconds]
<LnL>
channels only update when a set of tests pass and all the builds are finished
<stumble>
Hang on - not clear to me whether you want rebuilds to use unstable, but if you do then you need to prefix the command LnL gave you with sudo.
<jD91mZM2>
Honestly it's not entirely clear to me either. But I ran it with sudo and now both root and user are using unstable
<jD91mZM2>
"warning: name collision in input Nix expressions" when running an upgrade
<stumble>
Right - I had this problem yesterday.
<LnL>
I personally don't like using unstable because it can take a while for it to receive security updates, and with nix there's no reason to limit yourself to a single channel
<stumble>
It's alerting you to the fact that the system channel and the user channel have the same name.
<jD91mZM2>
aha
<stumble>
It's not necessarily a problem that they have the same name.
<jD91mZM2>
What do you recommend if I just want to use the latest user software but don't care about system version?
<jD91mZM2>
Should I use stable on root and unstable on user?
<stumble>
So you definitely (!) don't have a major problem - the system is on rolling release, which is what you want, and your user profile is on the same channel although it won't auto-update - you have to update user packages manually (or automate it yourself of course).
<stumble>
NixOS doesn't distinguish much between system packages and other packages. The point of the sudo stuff is that the root channel installs things available to all users.
<stumble>
If you're the only user then the way you have it set up is fine.
<jD91mZM2>
What do you mean update manually? nix-env --upgrade won't work?
<LnL>
that should look for updates across all channels
<stumble>
Yes that will work. :-) I meant that you need to do that.
<jD91mZM2>
If I'm using the same channel for root and user, is there a point to having the channel added to user?
<LnL>
but you'll have to update the root and user channels separately
<LnL>
no
<LnL>
unless you don't want it to be global for some reason
<stumble>
BRB
<stumble>
(I agree with LnL though.)
<jD91mZM2>
So I want `nix-channel --list` to be empty for the user but nixos-unstable for root?
thc202 has joined #nixos
tertle||eltret has quit [Quit: Connection closed for inactivity]
Acou_Bass has quit [Quit: byeeeeeeeeeeeeeee]
<LnL>
I have a nixos and nixos-unstable root channel on my system and use unstable for stuff I want to be more recent
<robstr>
Hey, I have my shell.nix file containig `(import ./nix/release.nix {} ).app.env`, now I want to spawn a postgresql database too, before i did something like `stdenv.mkDerivation {` with a shell hook. Can someone help puzzling them together ? :)
lonokhov has joined #nixos
<Myrl-saki>
How do I know the list of packages that Hydra builds?
sie has quit [Ping timeout: 256 seconds]
<{^_^}>
[nix] @edolstra pushed to master « LocalStore::addValidPath(): Throw error when the CA assertion is wrong »: https://git.io/vpgSV
<Myrl-saki>
Why is release-17.09 still being evaluated?
<vaibhavsagar>
Myrl-saki: it's still getting updates
<vaibhavsagar>
they haven't dropped support for it yet
<Myrl-saki>
vaibhavsagar: Why is 17.09 unique?
<vaibhavsagar>
Myrl-saki: it's not, the same thing happened with 17.03 after 17.09 came out
<Myrl-saki>
With nix 2, how to downgrade from 18.03 to 17.09?
<vaibhavsagar>
Myrl-saki: make sure you have nix.package = nixStable2
<vaibhavsagar>
in your configuration.nix, then replace the nixpkgs channel and do a nixos-rebuild switch
<vaibhavsagar>
why do you want to do this though?
<Myrl-saki>
vaibhavsagar: i686 rustc is broken, and the X server now indirectly depends on rustc.
<Myrl-saki>
By broken, I mean not being compiled by the hydra.
<vaibhavsagar>
hmm, that's unfortunate
<Myrl-saki>
vaibhavsagar: Tracking it was a pain.
<vaibhavsagar>
have you raised an issue?
<Myrl-saki>
vaibhavsagar: I think it's a "soft" requirement though.
<vaibhavsagar>
I don't know what that means in the context of Nix
<vaibhavsagar>
there's some flag that can exclude it from buildInputs?
<Myrl-saki>
vaibhavsagar: Let me check nixpkgs for a second before saying anything wrong. :P
<vaibhavsagar>
can you build your own rustc-i686?
<Myrl-saki>
vaibhavsagar: That's what I'm doing right now.
<Myrl-saki>
vaibhavsagar: My main problem is that rustc is a *huge* project, and might use up the RAM.
<Myrl-saki>
vaibhavsagar: It should be possible to use the binaries by using an overlay though.
<vaibhavsagar>
maybe nixpkgs-mozilla?
<Myrl-saki>
vaibhavsagar: Yeah, that's what I'm thinking. That opens up another can of worms though.
<vaibhavsagar>
hmm, you're right
<Myrl-saki>
vaibhavsagar: Give me a few, I'll just use nix-instantiate and tree it.
<vaibhavsagar>
you can also do `nix why-depends` IIRC
<Myrl-saki>
vaibhavsagar: why-depends doesn't work for buildtime dependencies yet.
<vaibhavsagar>
see
<vaibhavsagar>
*I see
humanoyd has joined #nixos
<Myrl-saki>
Okay.
<Myrl-saki>
I'm actually more confused now.
<Myrl-saki>
Anyways.
<Myrl-saki>
This is what I mean is a soft requirement.
<{^_^}>
[nixpkgs] @nlewo opened pull request #39914 → gcc49: backport retpoline support → https://git.io/vpg73
robstr has quit [Quit: WeeChat 1.9.1]
Myrl-saki has joined #nixos
<Myrl-saki>
Uhh
<Myrl-saki>
Did anyone have a reply?
<Myrl-saki>
PHone died.
<srhb>
Myrl-saki: No, still looking. Haven't tried that before.
<srhb>
Myrl-saki: But am I right to think you want to inject an apply that filters out the icons from systemPackages?
<Myrl-saki>
srhb: Yep!
mkaito has quit [Ping timeout: 240 seconds]
troydm has quit [Quit: What is Hope? That all of your wishes and all of your dreams come true? To turn back time because things were not supposed to happen like that (C) Rau Le Creuset]
hakujin has joined #nixos
humanoyd has quit [Ping timeout: 255 seconds]
<tilpner>
Myrl-saki - If you don't intend to use nixos-icons, you can overlay a dummy derivation onto it. It will still be installed, but won't require imagemagick anymore
<Myrl-saki>
tilpner: Yeah, that also seems to work.
<{^_^}>
[nixpkgs] @zimbatm pushed to release-18.03 « fetchs3: allow to name the derivation output (#39823) »: https://git.io/vpgdg
<srhb>
Myrl-saki: How come?
<Myrl-saki>
srhb: I mean. What is `imports`, actually? @_@
<srhb>
Myrl-saki: The nixos imports module bit
<srhb>
Like
<Myrl-saki>
I always thought imports was just importing files.
<srhb>
Modules
<srhb>
That is, things of shape { ... }: { config = ...; options = ...; imports = ...; } or alternatively a bare set of those instead of a function
<srhb>
Or alternatively none of the top level sets, but just the contents of config
<Myrl-saki>
srhb: What should I use nixpkgs.config for?
<srhb>
It's not nixpkgs.config
<srhb>
I'm confused, what are you asking?
<Myrl-saki>
I mean, what is nixpkgs.config for then?
<srhb>
Configuring nixpkgs
<srhb>
(Not NixOS)
<srhb>
Well, technically the nixpkgs that NixOS uses.
hiroshi has quit [Ping timeout: 256 seconds]
simukis has joined #nixos
<srhb>
The reason you're thinking files instead of modules is that usually you have imports like ./path-to-file
<srhb>
But that file, then, in turn, contains a module
<srhb>
But there's nothing stopping you from just defining it inline.
<Myrl-saki>
srhb: Question. I believe files is literally a file, and it's the module evaluator that converts the file, with, say, `import`?
<Myrl-saki>
srhb: So, `imports = [ file ];` is the same as `imports = [ (import file) ];`
humanoyd has quit [Quit: WeeChat 2.1]
<Myrl-saki>
I got sidetracked a bit. Lmao.
<Myrl-saki>
srhb: Basically, I'm saying that there's a processing stage that checks the type of each element of imports, and if it's a file, then it uses `import` on it.
<Myrl-saki>
srhb: Is my guess correct?
<srhb>
It sounds correct, a bare path should be a store path.
<srhb>
I didn't check the module source though :)
<Myrl-saki>
Either way, thanks. I got someone into NixOS, but this was a *major* roadblock for him.
raynold has joined #nixos
<srhb>
Honestly I wish there wasn't so much magical "if this type then that else if this type then ..."
<srhb>
What's wrong with having the user normalize everything into the canonical module form?
<kiloreux>
I have a library that I usually load with LD_PRELOAD and I am looking to use it with a nix package. Does nix has anything similar to LD_PRELOAD ?
<Taneb>
srhb: yeah
<srhb>
Taneb: I did, but promptly forgot to act on it, sorry. I've poked it now.
<Taneb>
srhb: thank you :)
<Taneb>
(this is my first time making an issue to something like nixpkgs and I'm a bit anxious about the whole thing)
<Myrl-saki>
srhb: Why is it called options?
<srhb>
Taneb: Please don't be, if there's something missing quality wise we will make sure to tell you what we need. Or if we forget, you can always poke us here :)
<Taneb>
:)
<__monty__>
Taneb: You can get your feet wet working on the ranger project, we welcome any and all contributions ; )
<srhb>
Myrl-saki: options are the declarations of what attributes there are, their types, postprocessing etc. config is the values (the ones you're used to settings in configuration.nix)
<Myrl-saki>
srhb: Right.
<Taneb>
__monty__: not quite what I meant by "like nixpkgs" :P
<Myrl-saki>
srhb: Thanks.
<__monty__>
Taneb: Just evangelizing : )
<Myrl-saki>
srhb: Why can't configuration.nix use `options` and `config`?
<srhb>
Myrl-saki: Who says it can't!
<Myrl-saki>
srhb: Nixpkgs. :P
<Myrl-saki>
srhb: I get a module error.
<srhb>
Oh, really, I thought it was just a regular module.
<srhb>
Sure you didn't do something wrong?
<Myrl-saki>
That's a posibility.
<Myrl-saki>
srhb: But this is what I get.
<Myrl-saki>
error: Module `:anon-1' has an unsupported attribute `boot'. This is caused by assignments to the top-level attributes `config' or `options'.
<Myrl-saki>
srhb: That's why you saw me using `nixpkgs.config` :P
<srhb>
Myrl-saki: I think you're doing something else wrong. I just rewrote mine to canonical form, and it's fine.
<Myrl-saki>
srhb: Can you show me?
<srhb>
Myrl-saki: canonical form being { ... }: { imports = ...; options = ...; config = ...; }
seanparsons has quit [Quit: ZNC 1.6.5 - http://znc.in]
<srhb>
Myrl-saki: Which prat specifically? I can't show you my whole config sadly.
<Myrl-saki>
srhb: I have `boot` and other whatnots.
<srhb>
Unless you're overriding their declarations
<Myrl-saki>
Well
<Myrl-saki>
This is weird.
<Myrl-saki>
I did that, with it under config, and yeah, it works.
<srhb>
The form you're used to seeing is just a flattened version that has only the config attributes
<srhb>
It's implicit that they belong to config in that case.
<Myrl-saki>
Thanks, learning a lot.
seanparsons has joined #nixos
<Myrl-saki>
srhb: So, I could technically do `{ ... }: { imports = ...; options = ...; ...}`?
<Myrl-saki>
srhb: Where everything else is a config?
<srhb>
Yes. Though you may run into some surprising issues where you can't merge _in-module_
<srhb>
Like, merges happen between modules, not in the same module
<Myrl-saki>
srhb: What do you mean by that?
periklis has quit [Ping timeout: 240 seconds]
<Myrl-saki>
srhb: Oh wait, I think I know what you mean.
<srhb>
For instance, you cannot set environment.systemPackages twice in the same module, but you _can_ assign it twice if you do it in two separate modules, and they will merge.
<Myrl-saki>
srhb: So, `options` won't apply to the current module?
<Myrl-saki>
Ohhh
<srhb>
Yes it will, but it might not merge the way you think it will.
<jD91mZM2>
Related to merging, is there any way to make it not merge? For example if I want to completely override my DNS stuff without keeping the default?
andreabedini has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<srhb>
jD91mZM2: lib.mkForce
aramiscd has quit [Quit: WeeChat 2.0.1]
<srhb>
Myrl-saki: Essentially, the same rule applies. You cannot define a thing twice in the same module.
andreabedini has joined #nixos
<srhb>
(And if it doesn't merge anyway, you can't define it twice at all.)
<Myrl-saki>
srhb: Thanks, I'll give the source code a read.
<srhb>
Myrl-saki: Take this, it's dangerous to go alone. *hands pickaxe*
<jD91mZM2>
mkForce didn't seem to make any difference?
<srhb>
jD91mZM2: Oh, uh, what are you doing again?
<jD91mZM2>
Trying to update networking.nameservers without keeping the default
<jD91mZM2>
pkgs.lib.mkForce ["1.1.1.1" "1.0.0.1"]
<Myrl-saki>
srhb: I've given it a read, even with my Haskell knowledge, I'm still lost. :P
<srhb>
jD91mZM2: The default is [] though?
<jD91mZM2>
Oh. So I just need to disable DHCP somewhere else. My bad.
<srhb>
jD91mZM2: So no need to do any forcing.
<srhb>
jD91mZM2: You can look at its value with nixos-option networking.nameservers
<{^_^}>
→ 46aa5200 by @ckauhaus: vulnix: 1.6.1 -> 1.6.3
<{^_^}>
→ b30c6d79 by @7c6f434c: Merge pull request #39912 from ckauhaus/vulnix-1.6.3
<jD91mZM2>
Yeah, I thought I had done that. Apparently not.
<srhb>
:)
zybell_ has joined #nixos
waleee has joined #nixos
<srhb>
Myrl-saki: It's not the most easily read part of the repo. :)
<srhb>
Myrl-saki: If you have a specific question I can try to dig up the relevant bit.
andreabedini has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Myrl-saki>
srhb: Thanks!
Ariakenom has quit [Ping timeout: 268 seconds]
spear2 has joined #nixos
spear2 has quit [Quit: Leaving]
spear2 has joined #nixos
asuryawanshi has quit [Remote host closed the connection]
asuryawanshi has joined #nixos
asuryawanshi has quit [Remote host closed the connection]
asuryawanshi has joined #nixos
spear2 has quit [Client Quit]
<jD91mZM2>
Does anybody know how to disable the nameserver that's detected by DHCP?
spear2 has joined #nixos
<srhb>
jD91mZM2: I suppose that depends which network stack you're using.
<jD91mZM2>
I mean in /etc/resolv.conf
<srhb>
Disabling it would usually mean "prevent it from writing to /etc/resolv.conf" which depends on the network stack.
<jD91mZM2>
Oh
<jD91mZM2>
I don't know exactly what a network stack is to be honest :P
<srhb>
How did you configure your networking? :)
<jD91mZM2>
I'm using NetworkManager
<srhb>
Wired, wireless, networkmanager, ... are all keywords here
<spear2>
can anyone suggest an audio player besides deadbeef? the interfaces is buggy (missing buttons, volume)
<srhb>
jD91mZM2: Hmm, not sure how to prevent it from doing that.
<jD91mZM2>
So NetworkManager is the one adding my default gateway?
<srhb>
Probably dhclient through networkmanager, but not sure what that has to do with DNS.
freeman42x]NixOS has joined #nixos
<jD91mZM2>
I mean adding my default gateway to /etc/resolv.conf
<srhb>
Do you mean an entry like "nameserver (default gateway)" ?
<goibhniu>
spear2: FWIW I use vlc, but tomahawk is nice too.
<jD91mZM2>
Yep. Seems like there is networking.networkmanager.dhcp. Maybe I want to try emptying that
Ariakenom has joined #nixos
<jD91mZM2>
Never mind, emptying that didn't work because it has to be one of specified options
<stumble>
I set my nameserver explicitly, and that overrides networkmanager.
<stumble>
I mean, I use networking.nameservers = [ "..." "..."];
<jD91mZM2>
Yeah, me too. But /etc/resolv.conf still has the other entries below
<stumble>
Heck!
<stumble>
That's bad.
<srhb>
jD91mZM2: So, this is not configurable currently
<srhb>
jD91mZM2: Usually you would want to set dns=None (or something like that) in NetworkManager.conf
<stumble>
I actually override both with openvpn options, and then only my vpn nameservers appear in /etc/resolv.conf.
<stumble>
But I didn't think that was necessary. :-(
<jD91mZM2>
Darn it. Would be great if I could somehow use the chattr hack
<srhb>
jD91mZM2: But right now, the only options are dnsmasq, systemd-resolved, unbound and default.
<srhb>
You could propose extending the module.
<jD91mZM2>
Like a pull request? Exciting!
<srhb>
jD91mZM2: Look at nixpkgs/nixos/modules/services/networking/networkmanager.nix under configFile
<srhb>
jD91mZM2: And under dns
<srhb>
jD91mZM2: That should give you an idea of how it works currently.
mkaito_ has joined #nixos
<sphalerite>
jD91mZM2: for networkmanager, you can set it in the connection properties. In nmtui there's "Ignore automatically obtained DNS parameters"
<sphalerite>
not sure about other UIs
Myrl-saki has quit [Remote host closed the connection]
<srhb>
sphalerite: Imperative config?!
<srhb>
:-)
<sphalerite>
srhb: welcome to networkmanager :)
Myrl-saki has joined #nixos
<srhb>
When are we getting a good, declarative setup that handles passwords properly too :(
yegortimoshenko has joined #nixos
<sphalerite>
personally I'm not awfully fussed about it being stateful for wifi connections, since connecting to a new network isn't something I'd want to rebuild my system for every time
spear2 has quit [Remote host closed the connection]
seanparsons has quit [Read error: Connection reset by peer]
knupfer has quit [Quit: knupfer]
<srhb>
Nah, I see the point. I'd accept a decent imperative ux even.
<sphalerite>
nm isn't perfect but it's not awful either :p
warbo has joined #nixos
<xnaveira[m]>
When I configure my windowManager to be i3 and no desktopManager everything works fine and my machine logs in directly into i3. If I enable gnome3 in desktopManager it logs into gnome3, no matter what i wrote in default. How can I have gnome3 installed but login default to i3?
<sphalerite>
xnaveira[m]: set desktopManager.default to "none", I think
<xnaveira[m]>
that's what i did Linus but it logins to gnome3 anyway
<xnaveira[m]>
sphalerite
zeronineseven has joined #nixos
<sphalerite>
with desktopManager.default = "none" and windowManager.default = "i3"?
<xnaveira[m]>
well then i get i3
<xnaveira[m]>
but no gnome3 installed
<xnaveira[m]>
as soon as i enable gnome3 in desktopManager i can only login to gnome3 regardless defaults etc
<{^_^}>
[nixpkgs] @Synthetica9 opened pull request #39917 → gcc7: Removed java as a target language → https://git.io/vpghr
<xnaveira[m]>
yes, not only that but i can't choose sesssion
<symphorien>
If you use lightdm, it is stateful and remembers what you chose last time you logged in
<xnaveira[m]>
let me recheck though
<symphorien>
So log out, choose none+i3, log in and it should default to i3 from now on
<xnaveira[m]>
yeah but i can't choose at the login screen
seanparsons has joined #nixos
<xnaveira[m]>
no mouse
<sphalerit>
which display manager are you using? If it's slim, the default, you can switch sessions using F1/F
<xnaveira[m]>
and if i enable gdm then i can only enter username
<sphalerit>
2
freeman42x]NixOS has quit [Ping timeout: 240 seconds]
<zeronineseven>
Hi! Newbie here. Could somebody please explain how can I disable default systemd mount (e.g sys-kernel-debug) via configuration.nix?
silver has joined #nixos
<sphalerite>
zeronineseven: systemd.mounts."sys-kernel-debug".enable = false; I think
<xnaveira[m]>
what you mean Linus Ctrl-alt-F2?
<sphalerite>
xnaveira[m]: no, just pressing the F1 and F2 keys on the keyboard
<sphalerite>
that should cycle through the session options in slim.
<xnaveira[m]>
that does nothing
jD91mZM2 has quit [Quit: WeeChat 2.0]
<xnaveira[m]>
ahhh
<xnaveira[m]>
i see
<xnaveira[m]>
sory
<sphalerite>
Assuming you are using slim that is. What does the login screen look like?
<xnaveira[m]>
right, i'll test that
<xnaveira[m]>
wel ii was using it
<xnaveira[m]>
i guess you mean the one with nixos logo
<xnaveira[m]>
but then i switched to gdm
<sphalerite>
yep that's the one
<xnaveira[m]>
but i didnt know about the f1/f2 stuff
<xnaveira[m]>
let me try that
<xnaveira[m]>
thx
<sphalerite>
yeah it's not obvious…
jD91mZM2 has joined #nixos
<jD91mZM2>
LightDM doesn't seem to read my .profile? Nor does it unlock my GNOME keyring.
<zeronineseven>
sphalerite: "nixos-rebuild switch" yields "The option value `systemd.mounts' in `/etc/nixos/configuration.nix' is not of type `list of submodules'."
<sphalerite>
zeronineseven: oh pff systemd.mounts is different from all the other systemd.* options. Try systemd.mounts = [{where = "/sys/kernel/debug"; enable = false;}];
jrolfs has joined #nixos
pie__ has quit [Ping timeout: 260 seconds]
<xnaveira[m]>
using k=lightdm or f1/f2 in slim solve the problem, thank you guys
<jD91mZM2>
No, I mean ~/.profile. Pretty much every single other system sources it for environment variables when you're logging in... But LightDM on NixOS does not seem to do that.
<jD91mZM2>
I did, however, find that there's an option for unlocking the GNOME keyring. I'm going to try that. I'll be right back
jD91mZM2 has quit [Quit: WeeChat 2.0]
blankhart has joined #nixos
logzet has joined #nixos
fendor has quit [Remote host closed the connection]
hakujin has quit [Ping timeout: 240 seconds]
jD91mZM2 has joined #nixos
<jD91mZM2>
Okay, security.pam.services.lightdm.enableGnomeKeyring works
andreabedini has joined #nixos
armbed has quit [Ping timeout: 260 seconds]
lord| has quit [Ping timeout: 264 seconds]
MercurialAlchemi has quit [Ping timeout: 240 seconds]
alex`` has quit [Quit: WeeChat 2.1]
<__monty__>
jD91mZM2: That's not a general rule, many shells source it but not all.
kiloreux has quit [Ping timeout: 260 seconds]
alex`` has joined #nixos
<jD91mZM2>
It's not a general rule, but I've never really seen a system which doesn't until now
<zybell_>
mixos is different. *All* sorts of different ;-)
srhb has quit [Quit: Quit]
<jD91mZM2>
Would it be sane to use nixos.vimPlugins over vim-plug?
<zybell_>
*NixOS
<sphalerite>
jD91mZM2: yes!
<jD91mZM2>
What do I do if a plugin is missing?
<zybell_>
install
<waleee>
jD91mZM2: the downside is that lazy loading configuration is more verbose
<sphalerite>
jD91mZM2: reasons you might want to: updating in lockstep with the rest of your user or system environment, so you get the rollbacks and stuff as well. Reasons you might not want to: plugin availability, update frequency
<sphalerite>
jD91mZM2: but you can add your own plugins, just a sec
<jD91mZM2>
Oh, I can just make a PR to vim-plugin-names?
<jD91mZM2>
I mean to be honest I don't actually want to use the precompiled binary but rather compile it from source
<sphalerite>
that's the spirit! :D
nD5Xjz has joined #nixos
alexteves has quit [Ping timeout: 240 seconds]
smichel has quit [Ping timeout: 260 seconds]
smichel_ is now known as smichel
<jD91mZM2>
I'm making a shell.nix currently. I don't want it to auto-fetch my stuff, but rather just put me in the right environment. But there's an assertion on buildRustPackage that requires a cargoSha256 entry. What do I do?
pikajude has joined #nixos
<sphalerite>
jD91mZM2: it looks like the languageclient stuff is already set up in nixpkgs to build from source so you should be good to go
hakujin has quit [Ping timeout: 248 seconds]
<jD91mZM2>
oo ok
<jD91mZM2>
Should my neovim plugins really be set system wide though?
<jD91mZM2>
Don't mind me I'm just nervous. I'll go ahead and switch.
<sphalerite>
jD91mZM2: I think for building rust stuff with nix, carnix is all the rage. Last time I tried it it wasn't great fun though, I think it depends mostly on how simple the dependencies are. If the whole dependency closure is release versions of (pure-rust||widely-used) crates, carnix should work well AFAIK
wchresta_ has joined #nixos
<sphalerite>
jD91mZM2: you don't need to do it system-wide, you can have it installed in your user profile as well
deepfire has joined #nixos
<jD91mZM2>
Oh?
<sphalerite>
yeah, like for my neovim.nix as linked above you can do `nix-env -f neovim.nix --arg pkgs 'import <nixpkgs> {}' -i`
srhb has quit [Read error: Connection reset by peer]
<{^_^}>
→ afea46bb by @peti: ghc-8.4.1: drop obsolete version, we have 8.4.2 now
<{^_^}>
→ b33d1d64 by @peti: hackage-packages.nix: automatic Haskell package set update
<{^_^}>
→ 92256a94 by @peti: haskell-tar: drop obsolete override for ghc-8.4.x
<sphalerite>
jD91mZM2: you can use it in your system configuration by putting (import ./neovim.nix {inherit pkgs;}) in your systemPackages
freusque has joined #nixos
<jD91mZM2>
Oh so your neovim configuration creates a neovim wrapper package that includes a config?
yegortimoshenko has quit [Ping timeout: 260 seconds]
jtojnar has quit [Ping timeout: 268 seconds]
<jD91mZM2>
I think I might just stay with vim-plug then... Seems easier :P
ericsagnes has joined #nixos
freusque has quit [Client Quit]
wchresta_ has joined #nixos
<jD91mZM2>
Related to that - running a newly compiled program fails. Just "bash: bin/languageclient: No such file or directory"
<jD91mZM2>
Running it with strace gives me execve("bin/languageclient", ["bin/languageclient"], 0x7ffd3e67c5b0 /* 169 vars */) = -1 ENOENT (No such file or directory)
freusque has joined #nixos
cmcdragonkai has joined #nixos
andreabe_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<cmcdragonkai>
Does anybody know how to integrate "switch users" into XMonad? Similar to how in KDE Plasma you can switch users and it will go back to SDDM.
<sphalerite>
in fact, dm-tool from lightdm works with sddm too.
<sphalerite>
I think this may not work with slim, but it should definitely work with lightdm, gdm, and sddm
<cmcdragonkai>
sphalerite: Does that lock the screen again?
<cmcdragonkai>
I'm using XMonad and sddm.
<cmcdragonkai>
If I run that, what happens to my current session, and how would I login back to my saved sesison?
NinjaTrappeur has quit [Quit: WeeChat 2.1]
MercurialAlchemi has joined #nixos
<sphalerite>
no, it doesn't lock the screen. You switch back to your existing one with the appropriate keystroke for switching to the tty (usually ctrl-alt-f7), or by logging in as the already-logged-in user
NinjaTrappeur has joined #nixos
freusque has joined #nixos
jtojnar has joined #nixos
yegortimoshenko has joined #nixos
zeronineseven has quit [Quit: Konversation terminated!]
hakujin has joined #nixos
<cmcdragonkai>
Another problem, just installed NixOS, and rebooted. After rebooting `ip link` doesn't show the wireless interface anymore.
<cmcdragonkai>
What could be the reason for this?
<goibhniu>
maybe it needs firmware?
<cmcdragonkai>
how do I know what firmware it needs?
<cmcdragonkai>
it worked when I booted from USB
freusque has quit [Client Quit]
andreabedini has joined #nixos
<goibhniu>
you can search for the chip name to check
<{^_^}>
[nixpkgs] @globin pushed commit from R. RyanTM to master « opendht: 1.7.0 -> 1.7.1 »: https://git.io/vp2qt
<cmcdragonkai>
I had `enableAllFirmware = true` before, but this resulted in some upstream URL error getting broadcom drivers or something.
<nschoe>
Hi everyone, how can I check if there is a hydra built binary for a given package (kicad in my case). Because when I try to install kicad, it compiles, but fails. I'd like to check if hydra has the same error. Is this possible?
<srhb>
nschoe: That's usually a sign that hydra _does_ have the same error.
<nschoe>
srhb, I thought so, but the end error message is https://pastebin.com/EL2zCDaD but I don't think this is acautlly related to disk space, so I wante to double check ^^
hakujin has quit [Ping timeout: 260 seconds]
freusque has joined #nixos
<nschoe>
srhb, this gives me "error: path '/nix/store/x8agj7jfmhl766miacqri2iw0ibxlxw1-kicad-4.0.7' is not valid".
<srhb>
So Hydra does not have it in cache.
freusque has quit [Client Quit]
<srhb>
Or well, cache.nixos.org does not
toby1851 has quit [Remote host closed the connection]
ckauhaus has quit [Quit: Leaving.]
fendor has quit [Ping timeout: 265 seconds]
jensens has quit [Remote host closed the connection]
daveo has joined #nixos
freusque has joined #nixos
<rizary>
anyone use sass compiler? i got `Errno::EACCES: Permission denied @ rb_sysopen - /nix/store/b2id4wdwlsxakz6kk5dr8jwxzx10l8zc-app/static/index.css
<rizary>
`
freusque has quit [Client Quit]
<rizary>
when try to build derivation
freeman42x]NixOS has joined #nixos
peacememories has joined #nixos
<rizary>
i put it in the buildPhase like this: `sass --update -C --force --sourcemap=none --stop-on-error -t compressed $src/static/stylesheet/main.sass:$src/static/index.css`
ajp has quit [Quit: No Ping reply in 180 seconds.]
zybell_ has quit [Read error: Connection reset by peer]
<nschoe>
okay. But if it doesn't have it in cache, it means it could not build it, right?
<sphalerite>
MaCXyLo: it probably won't let you delete the active one, I think for that you'll need to do rm /nix/var/nix/profiles/test
<jD91mZM2>
Does nix have a syntax for inheriting all the things?
<makefu>
system profile should reside in '/nix/var/nix/profiles/system-profiles/test'
<sphalerite>
MaCXyLo: that only removes the generations, you'll need to do nixos-rebuild boot again after that to regenerate the boot menu without the extra profile
<jD91mZM2>
If I have a package which requires stdenv, pkgs, whatever, is there a way to just call it with (import ./package.nix {inherit all the things;})?
<sphalerite>
oh yeah,add system-profiles into those paths, thanks makefu :D
<nschoe>
srhb, OMG
<nschoe>
srhb, oh wait this is on unstable
<sphalerite>
jD91mZM2: sort of. In the case where it's stdenv and other stuff that's at nixpkgs's top level, you'd use callPackage
<srhb>
nschoe: But I'm not. So disregard everything I said until we can figure it out :P
<nschoe>
srhb, what do you mean?
<srhb>
nschoe: I'm not on unstable, and I do not see kicad in cache.
<sphalerite>
jD91mZM2: as for a builtin that returns the current scope as a set, I don't know of one.
<srhb>
nschoe: So, at some point there appears to have been a permanent error on the hydra builds (output limit exceeded) so the build was disabled intely on Hydra by setting hydraPlatforms to empty list (as opposed to say, linux, darwin, ...)
<nschoe>
srhb, okay but where is this "hydraPlatforms = []" set?
<jD91mZM2>
sphalerite: Nice. Where can I import this callPackage?
<sphalerite>
jD91mZM2: it's in pkgs
<nschoe>
I don't see if in kicad's default.nix.
<MaCXyLo>
sphalerite: thanks! that works.
<srhb>
nschoe: In the meta section of pkgs/applications/science/electronics/kicad/default.nix
<{^_^}>
[nixpkgs] @globin pushed commit from R. RyanTM to master « di: 4.45 -> 4.46 »: https://git.io/vp2Z3
<nschoe>
(but I did not know what that meant anyway).
<nschoe>
So someone apparently gave up? ^^
<sphalerite>
jD91mZM2: oh yeah, callPackage takes an additional argument which is overrides. So typically you'll want to pass the empty set after the path as well
iqubic has quit [Remote host closed the connection]
<nschoe>
srhb, I don't suppose I can somehow ask hydra to try and build noe pakcage for me, though? ^^
<{^_^}>
→ abc001d4 by @edolstra: Merge pull request #39904 from r-ryantm/auto-update/aws-sdk-cpp
<{^_^}>
→ aafe5fb2 by R. RyanTM: aws-sdk-cpp: 1.4.33 -> 1.4.40
NinjaTrappeur has joined #nixos
blankhart has joined #nixos
<MaCXyLo>
i'm using mate as desktop environment on nixos. several virtual folders doesn't work; e.g. computer, trash and network. any idea howto fix this?
<steveeJ>
Mic92_: do we need to provide libraries for rustup within the rustup expression?
<nick_l>
clever: nothing there of interest.
wchresta_ has joined #nixos
kiloreux has quit [Ping timeout: 264 seconds]
<{^_^}>
[nixops] @boxyoman opened pull request #941 → Fixed a bug creating route53 record sets → https://git.io/vp2ui
hakujin has joined #nixos
katyucha has quit [Ping timeout: 264 seconds]
guesi has joined #nixos
jrolfs has joined #nixos
hakujin has quit [Ping timeout: 268 seconds]
<jD91mZM2>
In the "maintainers" field of nixpkgs meta, is it the packager's name or the developer's name?
<guesi>
hi! I installed nixos with a luks setup basically following the instructions from (https://gist.github.com/martijnvermaat/76f2e24d0239470dd71050358b4d5134). The installation was successfull. However when booting I'm getting 'No boot device found.' mhh, does anyone have a suggestion for me?
Fare has quit [Ping timeout: 248 seconds]
Tucky has quit [Remote host closed the connection]
<logzet>
jD91mZM2: packagers name usually
<logzet>
symphorien: It seems that the half of my store is dead
<logzet>
^^
<symphorien>
well I use nix-shell a lot, and all what it uses remais dead. depending on how you use nix, this could be plausible
jrolfs has quit [Ping timeout: 256 seconds]
Ariakenom has joined #nixos
kiloreux has joined #nixos
<logzet>
I'm also using nix-shell a lot, I don't remember running nix-collect-garbage on this system and I experimented with modifying xkeyboardconfig some time ago, that triggers a LOT of rebuilds
<logzet>
So this seems completely plausible
<{^_^}>
[nixpkgs] @rbvermaa merged pull request #39654 → GCE: pull the ssh host keys from the metadata service as expected by NixOps. → https://git.io/vp4aw
<{^_^}>
→ 3b8c7424 by @AmineChikhaoui: pull the ssh host keys from the metadata service as expected by NixOps.
<{^_^}>
→ f514a689 by @AmineChikhaoui: barf if pulling the ssh host keys fails
<{^_^}>
→ 4a86f8c9 by @AmineChikhaoui: remove the entire temporary directory
<{^_^}>
[nixpkgs] @rbvermaa pushed 3 commits to release-18.03: https://git.io/vp2gh
<{^_^}>
→ 3a48839b by @AmineChikhaoui: remove the entire temporary directory
<{^_^}>
→ 01db5220 by @AmineChikhaoui: pull the ssh host keys from the metadata service as expected by NixOps. Issues: #38623 https://github.com/NixOS/nixops/i…
<{^_^}>
→ 3a2ad3e4 by @AmineChikhaoui: barf if pulling the ssh host keys fails
wchresta_ has quit [Ping timeout: 268 seconds]
muzzy has joined #nixos
muzzy_ has joined #nixos
muzzy__ has joined #nixos
hakujin has joined #nixos
moriarty has joined #nixos
<Taneb>
Hmm, does there exist any linter or styler for .nix files?
<moriarty>
is nixos good for cryptocurrencies?
drakonis has joined #nixos
<simpson>
moriarty: I think that NixOS is a neutral force when it comes to acting on the cryptocurrency market, but I'm not an expert.
<simpson>
Taneb: My vim has linting for Nix but I'm not sure how I set it up.
<etu>
moriarty: I've seen derivations for several kinds of coins. I don't know how well maintained any of them are. But I think they should work.
toby1851 has quit [Remote host closed the connection]
hakujin has quit [Ping timeout: 264 seconds]
<etu>
jtojnar: I'm trying to package the gsconnect gnome extension. This almost works: https://ptpb.pw/nBsP/nix But I get errors like this: "Gio.DBusError: Error calling StartServiceByName for org.gnome.Shell.Extensions.GSConnect: GDBus.Error:org.freedesktop.DBus.Error.Spawn.ChildExited: Process org.gnome.Shell.Extensions.GSConnect exited with status 127"
<{^_^}>
[nixpkgs] @peterhoeg opened pull request #39932 → qmlbox2d and gcompris: do not fix Qt version to 5.9 → https://git.io/vp2at
<hodapp>
I want my derivations on the blockchain >:O
<etu>
jtojnar: It seems to be because "gjs doesn't exist" and I don't know why.
<moriarty>
simpson, thanks for the heads up
<moriarty>
etu, cheers :)
__Sander__ has quit [Quit: Konversation terminated!]
<jD91mZM2>
Is there any tools for printing where a package is located that can be used from bash?
<jD91mZM2>
I have a startup script which starts a bunch of applications, and deja-dup-monitor is not in $PATH
<tilpner>
jD91mZM2 - nix-build --no-out-link '<nixpkgs>' -A hello
<jtojnar>
etu: do you have any other errors in journal?
<Lisanna>
by default a remote builder will have to get all the build inputs from you (the host on which the build was started), since the remote host might not have access to the same binary caches that your host does. But that option will have the remote builder attempt to use its own binary caches when fetching build inputs
<{^_^}>
[nixpkgs] @xeji pushed commit from @r-ryantm to master « electrum-ltc: 3.1.2.1 -> 3.1.3.1 (#39440) »: https://git.io/vp2iE
Izorkin has joined #nixos
lassulus has quit [Ping timeout: 240 seconds]
lassulus has joined #nixos
Izorkin_ has quit [Ping timeout: 260 seconds]
<clever>
Myrl-saki: there is an option in nix to have the remote end use a binary cache when possible
<Lisanna>
clever I showed them
Fare has quit [Ping timeout: 248 seconds]
<clever>
ah, i see
<Myrl-saki>
That aside
<Myrl-saki>
How do I multi-user
<Myrl-saki>
Rather
<Myrl-saki>
How do I setup a multi-user.
<Myrl-saki>
Just simply sourcing nix.sh gives me broken things.
<Myrl-saki>
/nix/var/nix/profiles/per-user/$USER/profile doesn't exist, for example.
<clever>
Myrl-saki: i tend to mutate single into multi, i start by chowning /nix/store to root:root, and i tend to drop nix.sh entirely and set things up myself
<stammon>
In my nixos config I import several other files using a imports = [....] statement. I would like to pass down a variable / argument to the imported files. Can I somehow do that?
<clever>
Myrl-saki: then i run nix-daemon as root, and fix the errors as i see them
wchresta_ has joined #nixos
<Myrl-saki>
clever: Makes sense.
<clever>
Myrl-saki: the default and user profile should be in $PATH, something in $NIX_PATH, the user should own his own profile dir
<Myrl-saki>
clever: So, nix.sh shouldn't be used for multi-user?
<Myrl-saki>
Or can it still be used?
<clever>
not sure
<Myrl-saki>
clever: What I did was nix.sh, then path things up.
<Myrl-saki>
patch*
jedai has joined #nixos
anderslundstedt has quit [Quit: leaving]
<sphalerite>
stammon: you can declare options in them, like with all the modules included with nixos
<clever>
Myrl-saki: this symlink may also have to be manually created
<clever>
Myrl-saki: i think nix tries to auto-create it with the users name, the one in the user may also have to be updated to not be default anymore
jedai42 has quit [Ping timeout: 240 seconds]
<sphalerite>
heh, nixos-install does not care if there is an existing OS in the root filesystem
blankhart has quit [Ping timeout: 248 seconds]
<Myrl-saki>
clever: Stupid question, should I, uh, `chown -R root:root /nix/`?
kier has quit [Ping timeout: 240 seconds]
<stammon>
sphalerite: Why haven't I thought of this, thx
jedai has quit [Ping timeout: 276 seconds]
<clever>
Myrl-saki: yeah
<Myrl-saki>
clever: Also, in my NixOS machine, /nix/store is root:nixbld?
<clever>
oh yeah, you also need a nixbld group with write to the store, and build users in the group
seanparsons has quit [Read error: Connection reset by peer]
<clever>
do not add yourself to the group, its literally suicide
<clever>
nix will kill the chosen user, so any nix build command is asking nix-daemon to kill your entire session :P
<Myrl-saki>
> literally suicide
<Myrl-saki>
I was gonna say word abuse of "literally", but that's a pretty cool interpretation. :P
kier has joined #nixos
Tehnix has quit [Remote host closed the connection]
Tehnix has joined #nixos
<Myrl-saki>
> tmw I accidentally DoS'ed someone with nixos-rebuild
<clever>
ive seen users in here before trying to diagnose why X dies whenever they try to build something
mteufel is now known as mt
wchresta_ has quit [Ping timeout: 260 seconds]
katyucha has joined #nixos
asuryawanshi has joined #nixos
<Myrl-saki>
This is a stupidly obvious question
<Myrl-saki>
Rather, this is a stupid question with an obvious answer
Izorkin_ has joined #nixos
<Myrl-saki>
But when my laptop and desktop transfer nix stores through wifi, my pings spike up to triple digit.
cryp2nomicon has quit [Ping timeout: 240 seconds]
hakujin has joined #nixos
<Myrl-saki>
But the not-so-obvious part is that this seems to only happen with nix transfers. I don't think I've observed this with rsync.
<Myrl-saki>
Nor scp.
<Myrl-saki>
Got any idea?
<tilpner>
Measure transfer speeds, check if there's a difference between rsync and nix copy
<clever>
scp does things one file at a time, and waits for the remote end to sync
blankhart has joined #nixos
Izorkin has quit [Ping timeout: 240 seconds]
lejonet has quit [Ping timeout: 240 seconds]
<Myrl-saki>
clever: How about rsync?
seanparsons has joined #nixos
<clever>
not sure
lejonet has joined #nixos
<Myrl-saki>
tilpner: I'll check when I can.
<Myrl-saki>
tilpner: Also, I think it's just bias on my part, since `nix copy` seems to copy faster than rsync.
<Myrl-saki>
how it does that, i haff no idea at all.
<clever>
Myrl-saki: oh, and i have also noticed similar lag spikes, my laptop is a build slave and is on 802.11g, and synery lags noticably when its transfering things
<sphalerite>
Myrl-saki: rsync does one file at a time too
<Myrl-saki>
sphalerite: So basically, it's because nix copy is highly parallel?
<sphalerite>
I don't think it is
jensens has quit [Ping timeout: 260 seconds]
<sphalerite>
I think it might copy up to 4 things simultaneously, but even then I'm not sure if it does it over multiple TCP connections
<clever>
i think it does 1 thing at a time over 1 ssh, but it doesnt wait for confirmation that things are working
<sphalerite>
but yeah that would probably explain it. Opening multiple TCP connections for one logical transfer is bad behaviour really
<clever>
ive seen only a single `nix-store --write` on the remote end
<clever>
and ive seen it claim to be copying 4 things, then fail because the 1st was unsigned
<cransom>
i haven't done a bake off, but http/2 might just be better at saturating the stream(s) than rsync for this case.
<clever>
it shoved 4 NAR's worth of data into the socket before getting a reply back
<Myrl-saki>
unable to open SSH connection to 'ssh://####': cannot connect to '####'; trying other available machines...
<cransom>
because rsync is just a wrapped ssh and you are subject to the buffers it's using
<Myrl-saki>
But it works if I use sudo.
<Myrl-saki>
Not if I use nix-build
knupfer has joined #nixos
<Myrl-saki>
Err
<clever>
Myrl-saki: ssh keys?
<Myrl-saki>
clever: Is it a matter of nixbld users not having access to the ssh keys?
<sphalerite>
Myrl-saki: no, this would be root performing the SSHing
<clever>
remote builds are not initiated from the nixbld group
<sphalerite>
Myrl-saki: this can be a matter of private keys as well as the remote end's host pubkey
<Myrl-saki>
sphalerite: clever: It works if I use `sudo nix-build`, but not `nix-build` alone.
anderslundstedt has joined #nixos
<benley>
anyone remember how to deal with `nixos-rebuilt boot` failing because the /boot filesystem is 100% full?
<benley>
I guess I should have made it bigger than 250mb.
<sphalerite>
benley: remove old generations, manually remove kernels and initrds from /boot, rerun nixos-rebuild boot
<Myrl-saki>
Why do I have to manually enable it though?
<elvishjerricco>
ixxie: What do you mean "adding ... to path?" Environment variables are just part of the environment, and the path is just a particular environment variable.
<Myrl-saki>
sphalerite: Works now! :D
<ixxie>
elvishjerricco: right xP of course.... its been a while since I had to mess with those things
katyucha has joined #nixos
<ixxie>
elvishjerricco: when I set an environment variable it persists right?
<Ralith>
is there any way to pass `nix build` an expression on the command line?
<sphalerite>
Ralith: pass it as a single parameter enclosed in parens
cryp2nomicon has joined #nixos
<elvishjerricco>
ixxie: You mean like when you do `FOO=bar` in a shell? It'll remain in memory for the rest of that shell but it won't be in other shells
<ixxie>
elvishjerricco: I wanna set my DIGITAL_OCEAN_AUTH_TOKEN so it should persist; on the otherhand putting it in any file seems unsecure
<Ralith>
sphalerite: thanks!
<elvishjerricco>
ixxie: Ah. Yea that's a harder problem. It's gotta be saved somewhere; there's no way to persist it without writing it to something persistent like a hard disk
Fare has joined #nixos
<ixxie>
yeah but I am wondering what is the most secure way to do it
<ixxie>
normally a password is hashed and salted somehow when stored to disk AFAIK
Myrl-saki has quit [Remote host closed the connection]
Myrl-saki has joined #nixos
katyucha has quit [Ping timeout: 260 seconds]
romildo has joined #nixos
katyucha has joined #nixos
markus1199 has quit [Ping timeout: 264 seconds]
markus1189 has joined #nixos
<romildo>
I am packaging https://github.com/madmaxms/theme-obsidian-2 (a Gnome theme, based upon Adwaita-Maia dark skin). The project name ends with a number. How should I name it in nixpkgs?
<elvishjerricco>
ixxie: Well customer passwords are stored that way for somewhat different reasons. Specifically, you don't *want* access to your users' passwords. But you *need* access to your server credentials, so they have to be on disk. Only real protection you can give them is to encrypt them and prompt you for a password every time they're loaded.
asuryawanshi has quit [Remote host closed the connection]
asuryawanshi has joined #nixos
<ixxie>
elvishjerricco: right
<ixxie>
elvishjerricco: honestly I am at a loss at how I am meant to be using NixOps with Digital Ocean...
szicari has joined #nixos
acarrico has joined #nixos
lonokhov has quit [Quit: WeeChat 2.0]
<elvishjerricco>
ixxie: At least with AWS, I just have my secret key stored in plaintext on my file system in `~/.aws/credentials`, and keep my whole disk encrypted. Then nixops just uses the AWS API, which looks in `~/.aws/credentials` for the key. Not sure what the parallel would be fore DO
cryp2nom1con has joined #nixos
cryp2nom1con has quit [Client Quit]
<nliadm>
is it possible to tell if a derivation came from an overlay?
<nliadm>
I'd like to be able to easily build everything in my overlay
cryp2nomicon has quit [Quit: leaving]
cryp2nomicon has joined #nixos
boothead has joined #nixos
Myrl-saki has quit [Remote host closed the connection]
<boothead>
Hi folks, I'm having trouble getting my system to boot with / and /nix mount points both pointing to lvm volumes.
Myrl-saki has joined #nixos
<boothead>
I get a message at boot time indicating a timeout of some sort
<tilpner>
nliadm - You can self: super: let all = { foo = ...; }; in all // { allFromOverlay = all; }
<tilpner>
nliadm - Then you can build attrValues allFromOverlay
<elvishjerricco>
ixxie: It'd be nice if NixOps would encrypt it's state file. Then you could just set your secrets with nixops set-args and they'd be secure
<boothead>
The problem was after I earlier ran nixos-install after I blew away /nix
<clever>
boothead: do you know how to edit the cmdline in grub with e?
ryantrinkle has quit [Ping timeout: 248 seconds]
wchresta_ has joined #nixos
<boothead>
clever: yep, i've done that before. what should I add?
<nliadm>
oh, I guess that works. thanks tilpner
<clever>
boothead: boot.shell_on_fail
<clever>
boothead: then you can get a shell when it fails to mount the root
drakonis has quit [Read error: Connection reset by peer]
<boothead>
clever: nice - I've got that far - now I have i) and f) as options
<boothead>
(as well as r and * that I had before)
<clever>
boothead: you can now open a shell, and check what lvdisplay, pvdisplay, and vgdisplay say
<tilpner>
nliadm - self: super: let allFromOverlay = { inherit allFromOverlay; foo = ...; }; in allFromOverlay works too
xcmw has quit [Ping timeout: 240 seconds]
myrl has joined #nixos
Myrl-saki has quit [Quit: WeeChat 2.0]
<boothead>
clever: those commands aren't available..
myrl is now known as Myrl-saki
<clever>
boothead: what does tabtab show as being available?
daveo has quit [Ping timeout: 240 seconds]
<boothead>
fdisk -l is telling me that /dev/dm-0 doesn't contain a valid partition table... dm-0 is what the symlink in /dev/mapper for my root partition points to..
<clever>
boothead: generally, the dm-0 device shouldnt be partitioned
<clever>
boothead: thats basically partitioning your partitions
<ixxie>
elvishjerricco: hmm just wondering what to do since I don't feel qualified to make a security PR to NixOps xD
<boothead>
clever: not many commands available, looks like a busybox or something...
<clever>
boothead: lvm should be available
<boothead>
ah yes - just spotted that :-)
<clever>
boothead: you may need to use "lvm pvdisplay"
<boothead>
lvdisplay shows me what I expect from the graphical installer set up
<boothead>
as does pvdisplay
<clever>
boothead: does it say the access is read/write?
<{^_^}>
[nixpkgs] @edef1c opened pull request #39938 → switch-to-configuration: use Net::DBus to retrieve the list of units → https://git.io/vp25y
<clever>
boothead: and is the status available?
<boothead>
clever: yup
<boothead>
yup to both
<clever>
then you should be able to just mount each LV to the right sub-dir of /mnt-root/
<clever>
and then exit out of that shell to resume the boot
<clever>
then we can review your configuration.nix
<{^_^}>
[nixpkgs] @romildo opened pull request #39939 → theme-obsidian2: init at 2.5 → https://git.io/vp2dq
wchresta_ has quit [Ping timeout: 248 seconds]
<boothead>
ah - have to unmount /mnt-root first I guess?
<boothead>
mount: mounting /dev/localvol/root on /mnt-root failed: Device or resource busy
<boothead>
never mind - seems to be mounted already - but I get: couldn't mount as ext3 dues to feature incompatibility when I try to mount /dev/localvol/store /mnt-root/nix
<clever>
boothead: ahh, so your iso formatted /nix with new features, and the kernel you installed cant mount it
<boothead>
it's ext4 - after reporting the problem mount seems to think that it is mounted
<boothead>
ctrl-d will continue the boot process right clever ?
<clever>
boothead: try exit and see if it can continue to boot
freeman42x[nix] has joined #nixos
<boothead>
ok it continued to boot, but now it's doing the same thing that made me try the reinstall in the first place.. I have messages like rcu_sched detected stalls on CPUs/tasks
<boothead>
blocking rcu_structure:
<clever>
boothead: due to the deterministic nature of nix, it is likely to recreate the same problem if installed from the same nixpkgs
xcmw has joined #nixos
<boothead>
hmmm, good point :-)
<boothead>
yay determinism (except this time)
<drakonis>
yay
<clever>
boothead: id try an older generation of a channel, but its all gone now
<boothead>
I did install from a newer installer iso - I assume that would have made nix channels different?
<{^_^}>
[nixpkgs] @matthewbauer pushed 3 commits to master: https://git.io/vp2Fs
<{^_^}>
→ 3216cf6c by @matthewbauer: trash-cli: cleanup
<{^_^}>
→ 143978a4 by @matthewbauer: treewide: remove platform assertions
<{^_^}>
→ b45ef79b by @matthewbauer: Merge pull request #39907 from matthewbauer/work
<boothead>
i have tried both 18.03 and whatever the last 17.xx version I was on was
<boothead>
and I refomatted the partition I mount at /nix/store in between all this too
<boothead>
anything else I can try for more debugging?
<boothead>
I would have assumed that this was something to do with kernel packages etc, and I'm pretty sure that's been upgraded clever?
<clever>
google the error and see what others have done
daveo has joined #nixos
reinzelmann has quit [Quit: Leaving]
<boothead>
clever I've found a few mentions but all point towards an intermittent hard to find bug. Anything I could have missed in the re-install process that may have left me on old versions?
<clever>
boothead: nixos already has those by default
<clever>
boothead: you wouldnt even be able to mount /nix if dm_mod was missing
<clever>
boothead: and you should be able to check lsmod after you get an initrd shell to confirm
jrolfs has quit [Ping timeout: 240 seconds]
<boothead>
Hmm I had kernelPackages = pkgs.linuxPackages_latest; which I've just commented... I've seen a few bugs reported about that - clutching at straws now :-)
wchresta_ has quit [Ping timeout: 240 seconds]
<clever>
boothead: i do find it strange that the problem only manifests after you proceed to stage-2
<clever>
it seems stable during stage-1
<boothead>
well that didn't work - got a kernel panic straight away that time!
<clever>
you can also add boot.shell_on_fail to boot.kernelParams
<clever>
then it will be in the grub config by default
TonyTheL1on has joined #nixos
rihards has quit [Quit: rihards]
srl295 has joined #nixos
<boothead>
clever: isn't stage-1 where i'm having to drop into a shell?
TonyTheLion has quit [Ping timeout: 268 seconds]
<[Leary]>
BTW boothead you should try running a memtest before you dive deep into debugging. I've seen bad ram manifest as rcu_sched stalls and failed boots.
<clever>
boothead: yeah
<boothead>
[Leary]: hmm, the 18.03 iso and my windows partition boot up fine..
humanoyd_ has quit [Quit: WeeChat 2.1]
TonyTheL1on has quit [Ping timeout: 260 seconds]
<ghostyy>
hi, i have a question, i have a machine with two disks, each LUKS encrypted with a keyfile, which correspond to two devices in a ZFS mirror configuration, the keyfiles are stored inside a filesystem image which is itself encrypted with LUKS (just a passphrase, no keyfile)
<ghostyy>
the goal being that i can decrypt both of them by inputting just one password
<ghostyy>
my question is, how do i set this up in nixos?
<ghostyy>
the image (and disks) also all have detached headers
TonyTheLion has joined #nixos
<wilornel>
Hi #nixos! I was wondering if anyone here was able to install matplotlib for python?
<drakonis>
hmm
<drakonis>
i need to do tex
jD91mZM2 has quit [Quit: WeeChat 2.0]
<ghostyy>
the only part im really have trouble with is mounting the imagefile after i decrypt it
jD91mZM2 has joined #nixos
jD91mZM2 has quit [Client Quit]
<wilornel>
I did `nix-env -iA nixos.freetype; nix-env -iA nixos.libpng; nix-env -iA nixos.python; nix-env -iA nixos.python36Packages.pip; nix-env -iA nixos.pkgconfig; pip install matplotlib;` and it still wont find freetype and libpng
<srhb>
wilornel: This seems to work for me: nix-shell python3 -p 'python3.withPackages (ps: with ps; [ matplotlib ])'
<drakonis>
has anyone done anything involving tex on nixos
jD91mZM2 has joined #nixos
<ghostyy>
i also wouldnt be opposed to simply writing a shell script to do all of this and having nixos execute it in the initramfs but im not completely certain hyow to do that either, ive got an extra initrd but im not sure that commands in there can ask for input or whatever
<ghostyy>
does anyone have any experience with that? setting up weird crypto configs in nixos, or executing custom scripts in the initramfs?
<steveeJ>
Mic92_: can you try to build a crate that depends on the "built" crate? I can't get it to compile in my regular shell, even if I install zlib in my user environment. I need to use a 'nix shell -p zlib' to make rustc see the zlib
TonyTheLion has quit [Ping timeout: 256 seconds]
jD91mZM2 has joined #nixos
<jD91mZM2>
How would I configure my monitors in nix? I'm looking at services.xserver.monitorSection, but it's only for the first one.
nckx has quit [Client Quit]
nckx has joined #nixos
TonyTheLion has joined #nixos
<clever>
jD91mZM2: i just configure it with xfce after i login
<jD91mZM2>
I used to have some kind of script like that too, but LightDM doesn't respect that
<wilornel>
would be nice if nix could remember how long some packages took to build and not necessarily remove them when you run something like nix-collect-garbage ..
<drakonis>
naw
<drakonis>
there's package pinning i think?
jayq has joined #nixos
TonyTheL1on has joined #nixos
<steveeJ>
jD91mZM2: I use autorandr
<Mic92_>
steveeJ: well nix-shell -p is not a enought to link something against zlib
<Mic92_>
you need a default.nix
TonyTheLion has quit [Ping timeout: 268 seconds]
<steveeJ>
Mic92_: I wouldn't know that, all I know is that it works in this case
<Mic92_>
otherwise $NIX_CFLAGS_COMPILE is not set
<Mic92_>
zlib is not a dependency of rustc
<steveeJ>
Mic92_: sure it's not, but it's a dependency of one of the crates
<Mic92_>
then you need to add it yourself
winem_ has joined #nixos
<Mic92_>
we cannot add all dependencies just because some crate needs it
<steveeJ>
Mic92_: I'm not asking for that, no worries. I'm wondering what could be done to make this comfortable
<steveeJ>
when using nix-shell, my regular shell is not available of course
<jD91mZM2>
I think I might be looking for services.xserver.xrandrHeads, but I have to go (again!) and can't test it right now :|
nckx has quit [Quit: Updating my GNU GuixSD server — gnu.org/s/guix]
andrewmiller1 has joined #nixos
TonyTheL1on has joined #nixos
nckx has joined #nixos
<jayq>
hey friends. vps borked & disks are under a new UUID. I can mount the system from the install disk and correct the UUID's in hardware-configuration.nix, but what's the proper way to issue a `nixos-rebuild switch` from a mounted nix installation (if possible?).
<steveeJ>
Mic92_: direnv sounds interesting, I'll give it a try, thanks!
<radvendii>
clever: gotcha. and just to clarify this is mysql users that needn't necessarily be associated with any linux users
<ixxie>
hmm its stuck waiting on a lock
doyougnu has joined #nixos
TonyTheL1on has joined #nixos
<radvendii>
clever: is there a way to use initialScript that doesn't store the password in plaintext in the configuration.nix file?
<radvendii>
that's a bad idea, right?
<clever>
radvendii: not really
<clever>
radvendii: you could also use a custom systemd service that reads passwords from outside the store, or generates one on the fly with pwgen and saves it to a known path
kwork_ has quit [Ping timeout: 260 seconds]
TonyTheLion has quit [Ping timeout: 240 seconds]
<radvendii>
Mm. How risky is this? If I have this on a server, will anyone be able to read the password if they know where to look?
<ixxie>
is there a way to clear the cache?
<radvendii>
There was something in some documentation that said that it's world-readable, but that's only if they have access to the computer, right?
<clever>
radvendii: anybody with shell access can read all of /nix/store/
<radvendii>
but they have to be able to login first, yeah?
<clever>
radvendii: yeah
<radvendii>
okay, then that's fine
<clever>
radvendii: or find an exploit in a program that allows executing commands
<{^_^}>
→ 2cbaa163 by @obadz: mono: create mono4 & mono5 aliases and fix gtk-sharp/pdfmod builds
<{^_^}>
→ b9d19d48 by @xeji: Merge pull request #39886 from obadz/mono
<jayq>
clever: can't find `nix-enter` (server is running 17.09 still). Searched in some code repos & web. Where can I read more about this?
<steveeJ>
Mic92_: yeah, maybe reread my message it's quite accurate
<ixxie>
did something in Nix 2.0 change the way nix-channel prints? it seems to be very silent or stuck
<ixxie>
or slow
<ixxie>
well
<ixxie>
I did just delete the cache xD
<clever>
jayq: nixos-enter i think is the name
TonyTheL1on has joined #nixos
<Mic92_>
steveeJ: you can overwrite in ~/.direnvrc the use_nix bash function as shown in the wiki, so you can establish whatever mapping you need
<jayq>
clever: yeah, thought it might have been, but same outcome. chroot didn't help much (I assume nix[os]-enter does that but more?)
<Mic92_>
`direnv stdlib` shows the current definition
<clever>
jayq: nixos-enter will chroot and setup env vars for you
<Mic92_>
you will still need a local .envrc so.
TonyTheLion has quit [Ping timeout: 240 seconds]
<radvendii>
clever: can I just use the rootPassword field and then login as root?
<steveeJ>
Mic92_: ah, right, I will look into it!
<steveeJ>
is there a way to write '' within a '' ... '' scope without escaping it? I'm looking for a replacement for ''
<ghostyy>
argh... when i try to decrypt my encrypted image file with my header file i get "Attaching loopback device failed (loop device with autoclear flag is required)."
<srhb>
steveeJ: Without escaping it? No.
<ghostyy>
even though im already doing the trick where you cat the header to /dev/ram0 and read it from there
<steveeJ>
srhb: no trick like using ''' instead of '' for the scope? :-D
<wilornel>
srhb: I'm not sure how to translate that to a shell.nix ... wherever I would have inserted for example postgresql, I put int (python35.withPackages ( ps: with ps; [matplotlib] ))
<clever>
steveeJ: '' foo \'\' bar''
<steveeJ>
using ' inside the string is not uncommon, and escaping it is tedious
<clever>
steveeJ: actually, that doesnt seem to work
<clever>
steveeJ: but '' foo ${"''"} bar'' does
<steveeJ>
it also makes the snippets unportable
<ghostyy>
maybe i need to cat the image file to ramfs too?
<{^_^}>
[nixpkgs] @dtzWill opened pull request #39945 → git: disable failing test w/musl for now → https://git.io/vpaTY
<clever>
steveeJ: but maybe you just want builtins.readFile
TonyTheL1on has quit [Ping timeout: 248 seconds]
<srhb>
steveeJ: I don't understand what you meant. ''' is exactly how you escape '' :)
<steveeJ>
srhb: instead of writing var = '' ... some string \'' ''; maybe we could use var = ''' some string '' ''';\
<srhb>
steveeJ: Cute idea.
<wilornel>
the python version is different when I do it your way (3.6.5) and my way (3.6.4)
ckauhaus has quit [Quit: Leaving.]
TonyTheLion has joined #nixos
<srhb>
steveeJ: And when we need ''' inside it we can just do '''' and only the n-length(') finishes the string.
<steveeJ>
srhb: yeah, it's not a new idea
<srhb>
I was being a bit tongue-in-cheek, but okay :-P
<srhb>
I suppose it's a solution.
shabius has quit [Quit: Leaving]
<steveeJ>
srhb: I gracefully ignored that :P
<srhb>
Very gracefully!
<steveeJ>
clever: I don't even know how you came up with surrounding it with an expression marker :-D
<clever>
steveeJ: "doublequote" strings dont treat '' specially
<steveeJ>
In retrospect I understand that, but it's not an obvious solution to the problem, as it makes it even less portable
<clever>
steveeJ: using builtins.readFile is the simplest portable option, but then you have no substitution
<steveeJ>
I'm using this for shell configuration, and it does make sense to keep these snippets portable to share them
<steveeJ>
clever: that is a good idea, and simple
shabius has joined #nixos
<steveeJ>
clever: thanks, as always :-)
<clever>
though you can also mix in ''stuff ; source ${./file1.sh} ; stuff''
<clever>
then file1.sh is just read as-is, and the rest around the source cmd is parsed by nix
hakujin has quit [Ping timeout: 260 seconds]
<steveeJ>
clever: will ${./file1.sh} evaluate to the path or the content?
TonyTheL1on has joined #nixos
<clever>
a path to a copy of it in /nix/store/
colemickens has joined #nixos
<clever>
the path is based on the hash of the contents
<steveeJ>
fancy, will it ever be GC'ed?
<wilornel>
I don't understand the idea behind `python3` argument in `nix-shell `nix-shell python3 -p 'python3.withPackages (ps: with ps; [ matplotlib ])'`
<clever>
steveeJ: nix knows that the resulting derivation depends on that path and wont allow it to be deleted first
jrolfs has quit [Ping timeout: 260 seconds]
TonyTheLion has quit [Ping timeout: 268 seconds]
<clever>
steveeJ: so the script that the string generates has to be deleted first
<steveeJ>
clever: so there's a GC root pointing back to the dependent script?
<srhb>
wilornel: Did I write that? If so, it was a mistake, sorry
<clever>
steveeJ: it becomes a runtime dependency, so it doesnt need a root
<srhb>
wilornel: it should just be -p python3.withPackages ...
<steveeJ>
oh, cause the derivation is also in the nix store, understood
<srhb>
instead of just the url as a string, that is.
<mightybyte>
I was fine with that, but it required a hash. I did my normal method of setting the hash to 00000... and seeing what it said it should be, but it didn't tell me what it should be.
<srhb>
mightybyte: nix-prefetch-url --unpack can help with that. And I think there's an issue to add the usual real-hash-error
rosa has quit [Quit: rosa]
<srhb>
Unless fetchTarball does more sanitizing... Hmm.
<mightybyte>
Ahh, my bad...I had forgotten to call import on the result of fetchTarball
<mightybyte>
Apparently the error message just happened to match up to what I would expect if it had that behavior.
<jayq>
Trying to install `nixos-enter`. The manual hints to two ways: (a) `nix-build '<nixpkgs/nixos>' -A config.system.build.nixos-enter`, and (b) `nix-env -iE "_: with import <nixpkgs/nixos> { configuration = {}; }; with config.system.build; [ nixos-enter ]`. They both fail when using 'nixos-enter'; however, replacing 'nixos-enter' with 'nixos-generate-config' works fine for both. I've searching with `nix-env -qaP` with no results. It's
<jayq>
clearly in the source code, but I don't know how to derive the selection path from that alone. What should I try next? Any docs I should read over specific to this?
<bbarker>
In Nix 2.0+, is there a reason nix-env -if foo.nix doesn't link in binaries to the PATH somewhere (or maybe it is just that my environment is broken)?
<{^_^}>
[nixpkgs] @cleverca22 opened pull request #39953 → runInLinuxVM: fix ext4 and crc32c-intel interactions → https://git.io/vpaCo
Maxdamantus has quit [Ping timeout: 264 seconds]
<mightybyte>
Does the fetchFromGitHub downloader work with private repos? It doesn't look like it is for me.
Maxdamantus has joined #nixos
<mightybyte>
But I could swear I've used it for private repos in the past.
erasmas has quit [Quit: leaving]
deepfire has quit [Ping timeout: 264 seconds]
deepfire has joined #nixos
<{^_^}>
[nixpkgs] @cleverca22 opened pull request #39954 → runInLinuxVM: fix ext4 and crc32c-intel interactions → https://git.io/vpaCF
Nosti has joined #nixos
Nosti has left #nixos ["Leaving"]
<clever>
jayq: which version of nixpkgs are you trying that on?
<semantimancer>
Anyone here maybe willing to help me with a /boot problem?
dbe_ has quit [Ping timeout: 260 seconds]
jrolfs has joined #nixos
<semantimancer>
Only been using nixOS a few days. Whenever I boot, I have to run nixos-rebuild switch from root or else my config doesn't seem to take effect. Also means I can't seem to boot into X.
rsa has quit [Ping timeout: 245 seconds]
jrolfs has quit [Ping timeout: 256 seconds]
jrolfs has joined #nixos
<clever>
semantimancer: your /boot must be mounted when you run nixos-rebuild, and you need to add it to hardware-configuration.nix
<srl295>
`/etc/nixos/` seems to be a good place for a git repo to track history, any downsides to this?
<semantimancer>
Thanks, clever. Would you happen to know a good example of what it should look like in hardware-configuration.nix, for a UEFI system? Most of what I'm seeing online uses grub.
<clever>
semantimancer: in general, you can just run nixos-generate-config, and it will recreate hardware-configuration.nix, based on what is currently mounted
<clever>
semantimancer: thats why it has a comment saying to avoid editing the file
<semantimancer>
Okay. So maybe I just didn't have it mounted when I ran that the first time.
<semantimancer>
Yeah, just saw that.
<clever>
yeah, thats a fairly common mistake, they run generate-config without /boot mounted, then mount /boot to fix other errors, but the config doesnt know of it
hakujin has quit [Ping timeout: 260 seconds]
<srl295>
`environment.systemPackages = with pkgs; [ wget emacs-nox ];` << the hyphen in `emacs-nox` doesn't seem to work here, what's the right syntax?
<{^_^}>
[nixpkgs] @lionello opened pull request #39957 → nix-prefetch-git: fix handling of submodules with spaces → https://git.io/vpa8D
<semantimancer>
I'll give that a try, clever. Appreciate it.
semantimancer has quit [Quit: Lost terminal]
<clever>
srl295: the attribute name appears to be emacs25-nox
Maxdaman1us has joined #nixos
Maxdamantus has quit [Ping timeout: 264 seconds]
<Myrl-saki>
clever: So... uh... I just resorted to building i686 packages instead. Just as insane but I guess way more practical?
<srl295>
clever: thought i tried that, guess not
<clever>
Myrl-saki: which i686 packages?
<Myrl-saki>
clever: For starters, webkitgtk. :P
<Myrl-saki>
clever: Next I'll retry rustc.
<Myrl-saki>
I had to do some patches for webkitgtk to work, such as disabling GLES2
<Myrl-saki>
There were some header conflicts, and god knows why.
<srl295>
i'm having trouble getting docker to startup (even manually).. systemctl says 'no service docker.service' anyone know ?
jackdk has joined #nixos
spear2 has joined #nixos
<clever>
srl295: you need to set virtualisation.docker.enable = true; in configuration.nix
<srl295>
I have `virtualisation.docker.enable = true` and added the package `docker` . Seems I can startup using `dockerd` but not `systemctl`
<clever>
srl295: and youve done nixos-rebuild switch?
<srl295>
yes and also sublaunched vm
<clever>
sublaunched vm ?
<srl295>
i thought i set virtualisation.docker.enable but i set virtualiZation.docker.enable