romanofskiWork has quit [Ping timeout: 244 seconds]
romanofskiWork has joined #nixos
pie_ has quit [Remote host closed the connection]
pie_ has joined #nixos
stepcut has quit [Remote host closed the connection]
<sophiag>
anyone here use a yubikey? i just got the 5, but chrome69 isn't recognizing it and apparently firefox62 isn't supported :/
camsbury has joined #nixos
<sophiag>
wondering if i should try libu2f
cross has joined #nixos
<gchristensen>
go to about:config, search for u2f, toggle to true, works great
<gchristensen>
(ff)
stepcut has joined #nixos
<ottidmes>
bbarker: this seems to work for me: nix-shell -p ocaml-ng.ocamlPackages_4_05.ocaml
camsbury has quit [Ping timeout: 260 seconds]
<sophiag>
gchristensen: thanks. although now gitlab is hanging when trying to recognize it. not the same issue as with chrome
<ottidmes>
I started using the TOPT feature of Bitwarden (like Google Authenticator), and use Google Authenticator to access my Bitwarden itself, so I no longer need a yubikey for that. What other uses does a yubikey has, that might win me over to buy one myself?
<sophiag>
what finally got me to do it was travelling and remembering i still had some stuff using sms
<gchristensen>
u2f is nice because it isn't phishable
<sophiag>
yeah, this was basically the push for me to do the "right thing"
<sophiag>
and the yubikey5 uses fido2 which can be passwordless, although seems like nothing supports it yet
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fpqZ6
<ottidmes>
gchristensen: as compared to Google Authenticator?
<gchristensen>
right
dbmikus has joined #nixos
maximiliantagher has quit [Remote host closed the connection]
<bbarker>
ottidmes, thanks!
maximiliantagher has joined #nixos
<ottidmes>
I guess I will just buy one then, they are not cheap, but not extremely expensive either, not an expense I should spare for something as important as getting access to all my other paswords ;)
<ottidmes>
gchristensen: hah! "Google has recently revealed that since the company adopted U2F security keys for all of its employees last year", that alone tells you that Google Authenticator itself is not enough :P
maximiliantagher has quit [Ping timeout: 244 seconds]
<gchristensen>
seems you can use u2f to log in with nixos, too, neat.
<ottidmes>
any reason not to go for: YubiKey 5 NFC, that one seems the most useful, I do not even own a USB C port, and the mini version seems less useful than one on your keychain (since I will carry that on me almost always and e.g. not my laptop per se)
<sophiag>
ottidmes: that's the one i got
<ottidmes>
sophiag: thats good to know :)
<ottidmes>
gchristensen: ooh nice, I did do that with LUKS and a key shaped USB stick, so I am in favor of the idea :)
<sophiag>
and remember to set hardware.u2f.enable = true :p
lassulus_ has joined #nixos
romanofskiWork has quit [Ping timeout: 240 seconds]
lassulus has quit [Ping timeout: 246 seconds]
lassulus_ is now known as lassulus
<gchristensen>
security.pam.enableU2F not sure how it works exactly
maximiliantagher has joined #nixos
pita has joined #nixos
maximiliantagher has quit [Ping timeout: 268 seconds]
romanofskiWork has joined #nixos
jhillyerd has joined #nixos
freeman42]NixOS has joined #nixos
freeman42x]NixOS has quit [Ping timeout: 246 seconds]
detran has joined #nixos
c0bw3b_ has quit [Remote host closed the connection]
<gchristensen>
this reminds me about how insufficiently our PAM modules are
<gchristensen>
insufficiently configurable*
<gchristensen>
imo a u2f token isn't "suffient" auth, but our config makes it so
sigmundv__ has quit [Ping timeout: 240 seconds]
rprije has quit [Ping timeout: 252 seconds]
Supersonic has quit [Ping timeout: 260 seconds]
sigmundv has joined #nixos
nckx has quit [Quit: Updating my GNU GuixSD server — gnu.org/s/guix]
nckx has joined #nixos
Supersonic has joined #nixos
dbmikus has quit [Ping timeout: 276 seconds]
orivej has joined #nixos
slack1256 has quit [Remote host closed the connection]
stepcut has quit [Read error: Connection reset by peer]
stepcut_ has joined #nixos
<kyren>
I've never personally done that actually, there are lots of other ways as well
<jhillyerd>
hmm, but that seems to assume there is a package available for it somewhere. I guess I could check older channels
<kyren>
you can use nix-env with the '-f' flag to point at a specific nixos version and install an attribute from that, if you want it installed imperatively
<kyren>
yeah that's what you have to do, figure out when it was in nixpkgs as that version, then install it from that version of nixpkgs
<jhillyerd>
Will try. nixos may win me over yet... my first weekend.
<kyren>
I don't use nix-env imperatively, so if you just want it available to do development with, you could use nix-shell or direnv with use_nix like I normally do
<jhillyerd>
So far I've been able to put everything in my configuration.nix; but have been experimenting with nix-shell some.
stepcut_ has quit [Remote host closed the connection]
freeman42x]NixOS has quit [Ping timeout: 244 seconds]
<kyren>
I'm not sure why it's not in the path but there is an elm repl in the store and it's v0.18, you probably know the details better since you have it working with 0.19 already
<jhillyerd>
seems like that may be working
<kyren>
I'm glad you're liking nixos, I only recently really converted myself, it took me a good 2 or 3 tries I think
<kyren>
I think the v0.18 package might be weird, in the v0.19 one `elm repl` seems to work but not in v0.18, I dunno what's up exactly, but if you just need it for development porpoises then there's lots of tricks you can do once you start understanding the nix language
rpg has joined #nixos
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<jhillyerd>
there were massive changes between 0.19 and 0.18, I think in 18 most things were elm-make instead of elm make etc
<jhillyerd>
it would make more sense to allow people to install either easily, but I think there were a lot of changes to the way nixos packages elm too; so I think they just wanted to start fresh
<jhillyerd>
(based on what I read in the commit log)
<kyren>
ahaaaaa, that's why I can't find elm-repl in current nixpkgs, s/elm/elm-repl/ in the command above and it works
<kyren>
or, maybe it works, I dunno as long as one of us understand it :P
Dedalo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
drakonis has joined #nixos
drakonis1 has joined #nixos
drakonis_ has quit [Read error: Connection reset by peer]
sophiag has left #nixos ["ERC (IRC client for Emacs 26.1)"]
dramforever has joined #nixos
<dongcarl>
Hi all, I'm writing a service file for the bitcoin daemon. Wondering what's the most idiomatic way of making one option influence the default of other options?
<dongcarl>
For example, the listening port of the daemon changes when it runs in `testnet` mode
Rusty1 has quit [Quit: Konversation terminated!]
<dongcarl>
also wondering which directory it should go in in `services`, I'm thinking... `misc`?
<srhb>
colemickens: Thanks. Looks like mine's fine..
<srhb>
But that might just be because they aren't up for renewal yet.
<colemickens>
I wonder if it's regressed for new users?
<colemickens>
or that too
peterzky has joined #nixos
<srhb>
Guess I can try setting up a fresh subdomain..
jtojnar2 has quit [Ping timeout: 252 seconds]
jakub has joined #nixos
<jakub>
Hello, i am a total beginner and can't seem to find help in the manuals: I have just installed nix on debian, created a new profile for some tinkering, installed packages in that profile and now i would like to run a nix-shell with that profile (I haven't created any derivation yet), i guess this is not an intended use case, but how would i go about it if it is possible.
<srhb>
jakub: That's not generally how you use nix-shell, indeed
<srhb>
jakub: Normally, profiles are symlinked in place (eg in ~/.nix-profile)
<srhb>
jakub: While shells are used by making "packages" that are collections of whatever you need in a particular ephemenral environment
peterzky has quit [Quit: peterzky]
<jakub>
srhb: i thought the profile looked very much like such a package
<{^_^}>
[nixpkgs] @Mic92 merged pull request #50176 → docker-compose: rename from docker_compose, python3 by default → https://git.io/fptk6
<srhb>
jakub: In many ways it does, in the sense that the profile is usually one big buildEnv (a "package" created by symlinking together a bunch of "packages")
freeman42x]NixOS has joined #nixos
<jakub>
srhb: so to make use of nix to create a virtual environment I need to switch to defining derivations straight from the beginning, is that it?
<srhb>
jakub: I would recommend you do that, yes. It's much more lightweight and intuitive that making shells somehow enter profiles :)
rprije has joined #nixos
<jakub>
srhb: alright, thank you, i hoped i could postpone digging into the syntax etc for a bit before i get the hang of the architecture etc, but nevermind :)
<srhb>
jakub: fwiw it's really lightweight and you can totally just cargo cult a small shell.nix to get started with. What's the project type?
<srhb>
jakub: Perhaps I can point you in the direction of a starting point.
vidbina has joined #nixos
<jakub>
srhb: in the end I am looking to build a virtual machine disk image and I noticed there are some utilities in nix to handle containers/namespaces, I have my prototype in plain bash, which I would like to make more robust by converting it to a nix package (if that makes any sense)
<srhb>
jakub: Totally.
<srhb>
I'm not super familiar with the subject space though, so I'm not sure what to point you at.
<jakub>
srhb: I might come back with specific questions when I hit a dead end, until then, thank you for setting me on the right track :) much appreciated
<srhb>
jakub: Sure thing. Have fun :)
sigmundv__ has joined #nixos
<ekleog>
jakub: if you can afford having your VM disk image run nixos inside, you may be interested in `nixos-rebuild build-vm` and friends
mayhewluke has quit [Ping timeout: 272 seconds]
<ekleog>
though I wouldn't necessarily advise this as a starting point for nix :°
<WilliamHamilton[>
hi, I'm trying to open a nix-shell with some python packages (I don't usually use python, but I want to try a project that uses it). Here's what I've come up with: https://pastebin.com/c9VLMQ82
<WilliamHamilton[>
when I try to open `nix-shell -p that-thing` it says: `error: undefined variable 'arsenal-2' at (string):1:183`
<WilliamHamilton[>
so, why `arsenal-2` instead of `arsenal-2.1`? What's the meaning of that error?
<WilliamHamilton[>
oh wait maybe I know
<WilliamHamilton[>
yes, solved it, never mind
mayhewluke has quit [Ping timeout: 240 seconds]
mayhewluke has joined #nixos
<glasserc>
When I do "nix-env -iA nixos.jq.man", I don't see a man page for it in ~/.nix-profile, but when I do "nix-build '<nixos>' -A jq.all", I see it in result-man. Is there a way to install man pages for just one package?
maximiliantagher has quit [Ping timeout: 244 seconds]
freeman42x]NixOS has quit [Ping timeout: 245 seconds]
<{^_^}>
[nixpkgs] @hlolli opened pull request #50270 → use newer clj2nix which passes pkgs as an argument to deps.nix → https://git.io/fpqbj
<glasserc>
Another question, is there a way to get nix-env -i to tell me what symlinks it created in the user environment?
<gchristensen>
the only thing it does is create a new profile in /etc/profiles/per-user/youruser, and updates /nix/var/nix/profiles/per-user/grahamc/profile to point to the new one
jD91mZM2 has joined #nixos
detran has quit [Ping timeout: 244 seconds]
kp__ has joined #nixos
phreedom has quit [Remote host closed the connection]
<qyliss>
Can I have a NixOS option be recursively merged? eg if one module sets option = { foo = [ 1 ]; }; and another sets option = { foo = [ 2 ]; };, the value of the option becomes { foo = [ 1 2 ]; } ?
<steveeJ>
qyliss: I think it's automatically like that in the module system
<qyliss>
It is at the top level
<qyliss>
eg if I set option to { foo = []; } and { bar = []; }, those get merged
<qyliss>
But it doesn’t seem to be recursive
<qyliss>
I’m guessing I need to do something with lib.types in the option declaration
camsbury has joined #nixos
camsbury has quit [Remote host closed the connection]
camsbury has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<gchristensen>
I think you want a submodule for that
<infinisil>
qyliss: Just specify the type accordingly, the merging should take care of that
<infinisil>
So the type should be `attrsOf (listOf int)` in this case
<glasserc>
gchristensen: However, I only see one new symlink in the user environment, but nix-env -i reported that it "created 597 symlinks in user environment"
<glasserc>
Oh, wait, maybe I understand
<gchristensen>
some symlinks are all inside of the directory /nix/var/nix/profiles/per-user/youruser/profile points to
<glasserc>
That number is total, not new
<glasserc>
It's the number of symlinks in the new environment, rather than the number of new symlinks in the environment
<gchristensen>
yae
johann__ has quit [Quit: Leaving.]
<glasserc>
OK, that makes sense. Unfortunately it doesn't really help me with what I want, which is getting the man page when I nix-env -i a package
johann__ has joined #nixos
oida has quit [Remote host closed the connection]
maximiliantagher has joined #nixos
johann__ has quit [Client Quit]
johann__ has joined #nixos
<slabity>
For `packageOverrides`, can you override a package that comes from an overlay? Nix is telling me it can't find the package I'm specifying.
maximiliantagher has quit [Ping timeout: 245 seconds]
<slabity>
In this case I'm using the mozilla overlay and trying `rust = super.latest.rustChannels.nightly.rust.override {...};`
erictapen has joined #nixos
<ottidmes>
for a package I want to copy over a directory, reconstructing the parent directory structure and symlink instead of copy the executables, I can do it with first doing a find for the executables, and looping over them and doing the mkdir and ln -s, explicitly, and then just use cp for the rest, but is there a better way than this manual loop?
johann__ has quit [Client Quit]
johann__ has joined #nixos
<qyliss>
Submodules are doing what I wanted :)
<infinisil>
qyliss: Nice :)
<qyliss>
(The context here is that I want to be able to write s6-rc services in Nix)
jabranham has joined #nixos
<infinisil>
Ohhh I saw your toots about it, very cool
jperras has quit [Read error: Connection reset by peer]
stepcut has joined #nixos
<ottidmes>
with nix-index I am getting: error: attribute 'hackage-db_2_0_1' missing, at /wheel/fork/nixpkgs/pkgs/development/haskell-modules/configuration-ghc-8.2.x.nix:92:22
jperras has joined #nixos
<slabity>
I am getting a lot of 'Error fetching the file listing for store path '...' failed'
<{^_^}>
[nixpkgs] @peti pushed 3 commits to haskell-updates: https://git.io/fpmfW
<infinisil>
I think that can happen if you have certain overlays
<{^_^}>
[nixpkgs] @peti pushed 3 commits to haskell-updates: https://git.io/fpmfB
<romildo>
I am packaging the Deepin Desktop Environment and I would like some advice on how to deal with hard coded paths in the source.
<slabity>
romildo: Changing source code would best be done with a patch
<slabity>
Unless you want to work upstream and expose it as a build option
<romildo>
For instance, in the startdde package I have found (and did not fix yet) the following paths: http://dpaste.com/0EMACHR
hyper_ch2_ has quit [Quit: Page closed]
asymmetric_ has joined #nixos
<romildo>
Some of the paths are provided by required dependence files. Those I can fix by renaming the path to the exact one installed by the dependence.
<slabity>
Might not be the best idea, but you could always create an FHS derivation and symlink those paths to the actual paths
<romildo>
Some of the paths are provided by *optional* dependence packages. I do not want to make the dependence obligatory. How to fix them?
<clever>
steveeJ: next step then is to read the nixos-install script to see what its trying to run
mmercier has joined #nixos
<arianvp>
Mic92: you around?
mmercier has quit [Client Quit]
<romildo>
slabity, is it reasonable to use /run/current-system/sw as a prefix for most of those hard coded paths?
<ottidmes>
romildo: no, you should use the nix store paths
<ottidmes>
romildo: you cannot assume someone wants all dependencies in their system environment
o1lo01ol1o has joined #nixos
sc_infu is now known as inf
<ottidmes>
romildo: so just use e.g. "${pkgs.jq}/bin/jq"
<simpson>
Indeed, you could conversely make the safe assumption that nobody wants to have to have non-Nix-provided dependencies.
<romildo>
For binaries can I rely on the PATH environment variable for the system finding its location?
<slabity>
romildo: Not unless the program calling those binaries will resolve the paths
erictapen has joined #nixos
<ottidmes>
romildo: more importantly, if you assume /run/current-system, you assume NixOS, while Nix with Nixpkgs can be used outside NixOS
<slabity>
If it's looking for an absolute path like /usr/bin/whatever then it's safe to say it needs that exact path
<slabity>
If it's looking for a binary name like and being executed by some sort of shell that resolved PATH, then you can rely on it.
Itkovian_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<romildo>
I am anticipating cyclic dependencies, because there is some package A that uses a (optional) binary from package B, but B depends on A to be compiled. Any hints for this situation?
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/fpmLO
<simpson>
Make A take the dep optionally, build A, make B hard-require A, build B, build A with B.
endformationage has joined #nixos
<NickHu>
Is there a path like $NIX_CC/nix-support/dynamic-linker which takes you to the 32bit multilib linker you should be using for 32bit executables on an x86_64 system?
johann__ has joined #nixos
<romildo>
Is it better to edit those hard coded paths using a patch file, or using commands like sed and substituteInPlace explicitly in the nix expression??
<steveeJ>
clever: I gave up on that install and rebooted into a different rescue image on the VPS. the first was "sysrescue" based on gentoo. on debian that issue is gone, but I ran into a few others :D
erasmas has joined #nixos
dbmikus_ has joined #nixos
Dedalo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Mic92>
arianvp: yes
<Mic92>
Ideally nixos-enter would be bootstrable from nixpkgs so one can use non-nixos rescue systems.
johann__ has quit [Quit: Leaving.]
<Mic92>
Actually it is, but it is not so obivicious or documented.
dbmikus has quit [Ping timeout: 272 seconds]
<ottidmes>
I have cp --no-clobber, which works, but "... are the same files" for each existing file, is this normal? If I just do 2>/dev/null I might miss important errors
<steveeJ>
Mic92: non-nixos or non-nix? I was able to install nixos from the debian system. still have to confirm it's bootable though ;)
<steveeJ>
an option would be to use kexec in the rescue system
drakonis has quit [Read error: Connection reset by peer]
Dedalo has joined #nixos
mmercier has joined #nixos
<NickHu>
Also where can I find the 32bit libGL.so.1 on a x86_64 system?
averell has joined #nixos
<clever>
NickHu: hardware.opengl.driSupport32Bit = true; and it will put it into the LD_LIBRRARY_PATH for you
<NickHu>
clever: It doesn't seem to; at the path /run/opengl-driver-32/lib, I only have libGLES...*
<NickHu>
and libGLX_nvidia.so
<clever>
NickHu: that might be nvidia to blame, i saw a recent PR where they dropped 32bit support
jschievink has joined #nixos
kai_w has joined #nixos
<NickHu>
I found one in /nix/store/r9b4bhgn4745nb08m3yalalz0swwj6j7-nvidia-libs/lib
<clever>
NickHu: run nix-store --query --roots on that path
<NickHu>
Is there in general a way to get to 32bit libs? I'm trying to patchelf a binary package and it's 32bit
romildo has quit [Quit: Leaving]
<jschievink>
If I run nix-build twice in a row without changing anything and the package hash still changes, how would I debug that?
<jschievink>
(that's not supposed to happen, right?)
<clever>
NickHu: pkgsi686Linux contains all of nixpkgs, but in 32bit
<NickHu>
Ah great
<clever>
jschievink: if your using src = ./.; then you just added a result symlink to the source, and now the source is different
<samueldr>
jschievink: nix-build will put a `result` symlink in the CWD, which if it is the source, will change the source
<clever>
jschievink: src = lib.cleanSource ./.; will clean that up
<jschievink>
ooh thanks, that makes sense
<clever>
cleanSource also gets rid of things like .o files, which can break builds
<clever>
nix wipes all timestamps, then make gets confused and thinks the .o is up to date, and doesnt build anything
<NickHu>
clever: Do you know if there is a path like $NIX_CC/nix-support/dynamic-linker which takes you to the 32bit multilib linker?
__Sander__ has quit [Quit: Konversation terminated!]
<clever>
NickHu: if you use pkgsi686Linux.stdenv.mkDerivation, then $NIX_CC will point to the 32bit gcc/glibc
<clever>
and then you dont have to deal with multi-lib at all, its just a pure 32bit env
Itkovian has joined #nixos
maximiliantagher has quit [Remote host closed the connection]
mmercier has quit [Quit: mmercier]
<NickHu>
ah ok, I suppose that has some advantages
<clever>
pkgsi686Linux.callPackage would give you 32bit for all of your inputs as well
<clever>
NickHu: oh, and look at something like the old skype package, its 32bit as well
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
<lejonet>
infinisil: haha, that should work xD "Your services are not required anymore"
<Mic92>
infinisil: There are projects like openwrt or android (the project, not apps) that will unlikely make into nixpkgs. Also some projects that might have development requirements for their tests, that we don't need/want in nixpkgs, but a required for developers
hedning has quit [Quit: hedning]
<betaboon>
clever: thanks. i was just stupidly looking at the wrong file ...
<Mic92>
arianvp: oh gosh. We need a test for that.
jvassev_ has joined #nixos
<clever>
Mic92: if left alone, i would be more likely to turn all of WRT into a single nix expression that builds the entire thing purely :P
<arianvp>
Mic92: yes :D
<arianvp>
I can add a regression test, but I have no idea how to fix it :P
<Mic92>
arianvp: You can add LD_DEBUG to its service to see what it loads.
<lejonet>
infinisil: awh, doing langFortran = false; doesn't seem to stop it from trying to pull in gfortran :( so maybe your suggestion is the only real way :P
<lejonet>
(tho the nixpkgs default.nix seem to say that langFortan always is false)
jvassev_ has quit [Ping timeout: 256 seconds]
maximiliantagher has joined #nixos
<ottidmes>
is there a nix/shell variable that contains the hash used in $out, I could of course extract it from $out, but I'd rather not. I want to use it to store the build version to make sure a systemd script is only run once for every new build
Ariakenom has joined #nixos
<{^_^}>
[nixpkgs] @nlewo merged pull request #49414 → dockerTools.buildImageWithNixDb: simplifications and switch to closure info → https://git.io/fx7rR
<arianvp>
Dezgeg: okay so it does seem to load the libraries
<arianvp>
oyu're right
mayhewluke has quit [Ping timeout: 276 seconds]
<qyliss>
Mic92: that’s awesome! I was thinking about proposing something like that the other day, because I have a bunch of shell.nix fils lying around for various FOSS projects.
maximiliantagher has quit [Ping timeout: 268 seconds]
zolk3ri has quit [Remote host closed the connection]
Thra11 has joined #nixos
deni_ is now known as deni
<arianvp>
Ughh
mayhewluke has joined #nixos
<arianvp>
Mic92: Okay so good news is nss_modules loading isn't broken
<samueldr>
gchristensen: some checkups to do: is it thinking it's printing A4 even though you select Letter in your software, and the printer knows it has Letter in? (personal experience)
erictapen has quit [Ping timeout: 244 seconds]
<{^_^}>
[nixpkgs] @FRidh pushed to staging-next « Revert "Python: reduce amount of warnings when building with `buildPython*`" »: https://git.io/fpmZH
<{^_^}>
[nixpkgs] @FRidh pushed 95 commits to staging-next: https://git.io/fpmnT
<Mic92>
qyliss: yes, having them in one place allow to keep those recipies up-to-date.
<ottidmes>
I want to write: sed -i '$s|$| grep -Ev '\''(Agent.Rpc.SourceEvent Verbose|Microsoft.Cascade.Rpc.(Request|Response)Message)'\''|' in a script written as a '' ... '' string, how to escape them?
<clever>
,escape''
<{^_^}>
'' two single quotes: ''' bash curly bois: ''${} newline: ''\n tab: ''\t any character x: ''\x ''
<{^_^}>
[nixpkgs] @FRidh pushed commit from @costrouc to master « pythonPackages.ovito: init at 3.0.0 »: https://git.io/fpm8z
<mog>
does anyone use virtualbox on nixos? i thought i installed everything but when i run it VirtualBox: cannot run `/nix/store/zp0xizchw3i15xynwyx51phv18idk9l7-virtualbox-5.2.14/libexec/virtualbox/VirtualBox': No such file or directory
<clever>
mog: virtualisation.virtualbox.host.enable = true; is the only thing you need in your config, nothing has to be installed
<mog>
i have that set
<clever>
mog: what does `ls -lh /nix/store/zp0xizchw3i15xynwyx51phv18idk9l7-virtualbox-5.2.14/libexec/virtualbox/VirtualBox` report?
<mog>
if i ls /nix/store/zp0xizchw3i15xynwyx51phv18idk9l7-virtualbox-5.2.14/libexec/virtualbox/
erictapen has quit [Ping timeout: 268 seconds]
<mog>
there is a lot of stuff there including VBoxHeadless which does run just no VirtualBox
<clever>
mog: i'm guessing thats a broken build of vbox
<mog>
wait
<mog>
im an idiot
<mog>
i also have virtualisation.virtualbox.host.headless = true; elsewhere in my config
<mog>
i bet its overriding
<clever>
that would do it
<clever>
it removed all gui components
<mog>
yup i share configs across boxes
<clever>
virtualisation.virtualbox.host.headless = lib.mkForce false; lets you override it further, on just one box
<Ralith>
ottidmes: the DNS entries are getting installed correctly, that's not the problem
<mog>
tht fixed it
jschievink has quit [Quit: WeeChat 2.3]
stepcut has joined #nixos
elgoosy has joined #nixos
averell- has joined #nixos
averell has quit [Ping timeout: 252 seconds]
averell- is now known as averell
stepcut has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @vcunat opened pull request #50280 → xorg: init xf86-video-vboxvideo at 1.0.0 → https://git.io/fpmBN
<{^_^}>
[nixpkgs] @dasJ opened pull request #50281 → nixos/luksroot: Check whether the device already exists → https://git.io/fpmRq
kp__ has quit [Ping timeout: 252 seconds]
JosW has joined #nixos
drakonis has joined #nixos
drakonis has quit [Read error: Connection reset by peer]
drakonis has joined #nixos
drakonis has quit [Read error: Connection reset by peer]
drakonis1 has joined #nixos
drakonis_ has quit [Ping timeout: 264 seconds]
locallycompact has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @timokau pushed commit from @r-ryantm to master « python36Packages.netcdf4: 1.4.1 -> 1.4.2 (#49988) »: https://git.io/fpmgu
pointfourone has joined #nixos
<LtZechs>
Hi my name is Daniel and I'm trying to install Borg Backup with Nix. I'm running into an issue when I run (nix-env -iA borgbackup). I'm receiving the error (error: attribute 'borgbackup' in selection path 'borgbackup' not found). I also tried running (nix-env --install borgbackup) and I recieved an error as well (installing 'borgbackup-1.1.7' error: unexpected end-of-file). I'm running OS X Mojave (10.14.1 (18B75)). Any help would be
<cransom>
obfusk: if you don't quote "$x", bash is going to eat your xmll since it has the raw < > in it.
aleph- has joined #nixos
<obfusk>
cransom: but bash doesn't expand < or > inside a $variable. had I used ''echo ${x} > $out'', then yes (but that would have resulted in obvious errors)
<obfusk>
also: I can't reproduce the behaviour if I run it by hand in nix-shell (see the gist)
<averell>
how could i rollback a channel fetch assuming i cleared the locally built profiles already?
phreedom has joined #nixos
stepcut has joined #nixos
<{^_^}>
[nixpkgs] @marsam opened pull request #50285 → python3Packages.async_generator: enable in python3.5 → https://git.io/fpmr9
<infinisil>
LtZechs: I think I've seen similar errors when my network had problems
Ariakenom has quit [Quit: Leaving]
<infinisil>
Try searching for this error in the issue tracker
maximiliantagher has quit [Ping timeout: 260 seconds]
<infinisil>
LtZechs: The issue has certainly nothing to do with borgbackup
<cransom>
obfusk: i'm not sure where to narrow it down, but i think it's being eaten by shell expansion.
<infinisil>
LtZechs: Nix could be part of the problem, but most likely it's just your network, because it works for me no problem
<infinisil>
LtZechs: This chat here is #nixos, no more PM's pls
<LtZechs>
infinisil: my apologies. Thanks for all the help.
LtZechs has quit [Quit: Page closed]
<infinisil>
Oh already gone, I just wanted to ask for the full error and direct them to the nix issue tracker..
<mkaito>
people don't know how to irc :P
sigtrm has quit [Ping timeout: 276 seconds]
<schmittlauch[m]>
I've got a question on the FHS env: If programs running inside such a sandbox want to trigger a (xdg-)open of a file using its default program, does this default program have to be present within that env as well?
<schmittlauch[m]>
Practical example: the nixpkgs expression for "zotero" packages it to run inside such an FHS env. I encounter that double-clicking on pdf files in that program doesn't open them.
sigtrm has joined #nixos
dycan has joined #nixos
drakonis1 has quit [Quit: WeeChat 2.3]
maximiliantagher has joined #nixos
drakonis1 has joined #nixos
romildo has joined #nixos
<schmittlauch[m]>
It pretty much looks like the default PDF viewer needing to be present within that env. Adding okular to the env makes the opening work
<romildo>
I want to test some nixos module using a vm, but when I start the vm with result/bin/run-nixos-vm there is kernel panic. It used to work before. Any clues?
maximiliantagher has quit [Ping timeout: 244 seconds]
maximiliantagher has joined #nixos
vk3wtf has joined #nixos
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<jabranham>
Does anyone have an example of the proper ports to open in the firewall to let steam in-house streaming work?
<clever>
jabranham: i think 25565 and 27036 was steam
<clever>
i forgot to comment them when opening things up
<jabranham>
clever: thanks. I can't try it right now but will use it tonight.
<clever>
in my case, the "host" was a windows box, and the client was nixos, with the above config
pie_ has quit [Quit: Leaving]
<jabranham>
ah. I've got nixos as the host.
<clever>
i was using that feature to run a "windows only" game on nixos
kenogo has quit [Quit: WeeChat 1.9.1]
rprije has joined #nixos
<jabranham>
clever: they've also got that new thing that lets linux run all/most "windows only" games. I think it's called proton? Just in case you hadn't heard of it
<clever>
yep, but it doesnt work with .net based things
<__monty__>
And mono isn't sufficient?
<clever>
space engineers for example doesnt work
<clever>
__monty__: good idea, i'll need to install it again under steam and see what mono does with it
maximiliantagher has quit [Remote host closed the connection]
<obfusk>
cransom: thanks for trying to help. but in that case 1) why doesn't it produce an error (I'd expect it to)? 2) why doesn't it eat the other tags? 3) why can't I reproduce it by running genericBuild by hand in a nix-shell?
<dycan>
tilpner: thank you. I'll take a look if the problem is in there.
<tilpner>
"Cloud Networking" is probably allowed to cause all sorts of problems, including Forbidden
<tilpner>
So I'd probably just wait until Google says to try again
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
<dycan>
tilpner: I see. Thank you for pointing it out.
juhulian has joined #nixos
<c0bw3b_>
aanderse: not your changes but cmake should be in nativeBuildInputs (libcec pr)
c19tch has joined #nixos
<cransom>
obfusk: thesea re good questions, i don't have an answer. my thought was that because that the build steps/phases are created as bash functions, thats why they are blown away. i don't have the missing link where to go from the xml to the output though. it's somewhere.
<Ashy>
hmm, i'm really struggling to get pip packages working with python
dycan has quit [Remote host closed the connection]
<Ashy>
what's the best way to write or generate nix derivations for a pip package that isn't packaged up yet?
<simpson>
I normally patch nixpkgs, but I expect to send all of my packages upstream.
<Ashy>
i'd be happy to do that, these are public pip packages that i didnt write
maximiliantagher has quit [Remote host closed the connection]
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
c0bw3b_ has quit [Remote host closed the connection]
<Ashy>
simpson: to work in your own fork of nixpkgs on nixos what do you need to do?
<Ashy>
add a new channel?
<simpson>
Ashy: I have a git checkout of nixpkgs, and I use the $NIX_PATH environment variable to control whether I'm using it instead of system channels.
<clever>
Ashy: channels are optional, if you just want to test the build, you can just run `nix-build -A hello` in the root of the clone
stepcut has quit [Remote host closed the connection]
stepcut has joined #nixos
<Ashy>
what i'm really after is a way to land in a nix-shell with the extra pip packages available
<Ashy>
and then just: nix-shell -> python "import requests" ...
<clever>
Ashy: ah, then you want either `nix-shell -I nixpkgs=$HOME/apps/nixpkgs` in the dir with your shell.nix, or change line 1 to `with import /home/clever/apps/nixpkgs {};`
<clever>
mudri: does the string en-latin9 exist anywhere in your nixos config?
fragamus has joined #nixos
Zer000 has joined #nixos
<mudri>
clever: i18n.consoleKeyMap = "en-latin9"
<clever>
mudri: i'm guessing that that keymap isnt in the build anymore, so its not being found
<Zer000>
I updated a server I have running gitlab from 18.03 to 18.09 (by adding a new nix channel and then doing a rebuild) and my gitlab database has the right schema (all the tables are there) but it is empty!
<Zer000>
bye bye project metadata i guess
<clever>
mudri: let me check closer...
<Zer000>
It's weird because I have hydra running on the same server and it is fine. There must be a migration script in gitlab that wipes data.
<clever>
Zer000: did you change the stateVersion? (or is it missing)
<clever>
Zer000: i had a problem with my hydra a few months ago during a similar upgrade, the postgresql datadir changed from /var/lib/something to /var/lib/something/1.2.3
<clever>
and then the new dir was empty, so the db was gone
<clever>
double-check what your postgresql is using (ps aux should help), and if the data is in a related path
<Zer000>
clever, but my db is not gone. the problem is specific to gitlab. Both hydra and gitlab were on the same server and only gitlab has data loss
<Zer000>
hydra is all there
<clever>
Zer000: if you check in /etc/systemd/system/ for the gitlab service file, does it name the postgres db to use?
<clever>
-r--r--r-- 1 root root 1449 Dec 31 1969 result/share/keymaps/i386/colemak/en-latin9.map.gz
<clever>
mudri: i can confirm that this file is present on the rev you named, and missing on master
<clever>
mudri: its in the `nix-build -A kbd` derivation
<clever>
mudri: and a bisect points me to a commit titled: kbd: Rename some keymaps.
<clever>
nh2: ran it twice and it passed both times
<ottidmes>
I am trying to silence a program, after digging in the open source bit I have determined that its the closed source bit that is logging verbose, even though its command line argument says otherwise. I tried redirecting to /dev/null, but I just checked /proc/<pid>/fd/0 and it points to some socket rather than /dev/null. I have exhausted all my potential leads to getting it silent, except for a very dirty trick
<ottidmes>
of using gdb to change the output of the running process to /dev/null. Anybody got an alternative?
<{^_^}>
[nixpkgs] @samueldr pushed 2 commits to release-18.09: https://git.io/fpmMK
<clever>
ottidmes: ok, try doing `strace -ff -o logfiles -s 1000 <that prog>` and then search it for any `connect(0,` strings or similar
mmercier has quit [Quit: mmercier]
<nh2>
srhb: hmm dammit. Will try to make another example
<clever>
nh2: what is the state of your sandbox flag?
hyperfekt has quit [Ping timeout: 256 seconds]
<ottidmes>
clever: -o can be an absolute path?
<nh2>
clever:
<nh2>
# grep sandbox /etc/nix/nix.conf
<nh2>
sandbox = true
<nh2>
extra-sandbox-paths =
<clever>
ottidmes: i think so, but the `-ff` will also append to that string
romanofskiWork has joined #nixos
<ottidmes>
clever: seems to work with abs path :)
<nh2>
clever: I have the feeling that it must be the sandbox, nothing else makes sense to me, but it really looks like it's on. But maybe it isn't on despite my setting or something?
<clever>
nh2: if you take the raw `.drv` from it, and `nix-copy-closure` to another machine, and then `nix-build --check /nix/store/foo.drv`, can you make it work on one, and break on another?