leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
Jonathan__ has joined #nixos
Jonathan__ has quit [Client Quit]
stepcut has joined #nixos
stepcut has quit [Remote host closed the connection]
wolfman154 has joined #nixos
<ajs124>
Guest4910: what do you mean by "hash mismatches"?
wolfman154 has quit [Client Quit]
wolfman154 has joined #nixos
wolfman154 has quit [Client Quit]
<Guest4910>
ajs124 - as an example (there are many that seem to occur) "hash mismatch importing path '$path'; expected hash '$sha...1', got '$sha...2' " -- obviously anything with a "$" varies depending on the path in question -- It's a graphical console, so I can't copy out the message
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fjtK1
<{^_^}>
[nixpkgs] @Ma27 pushed commit from @NickHu to release-19.03 « zsh-prezto: 2017-12-03 -> 2019-03-18 »: https://git.io/fjtKM
Guest4910 has quit [Ping timeout: 255 seconds]
jeremy has joined #nixos
jeremy is now known as Guest90579
stepcut has joined #nixos
acarrico has joined #nixos
acarrico has left #nixos [#nixos]
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
lizard_queen has joined #nixos
lizard_queen has quit [Client Quit]
<Guest90579>
ajs124 - as an example (there are many that seem to occur) "hash mismatch importing path '$path'; expected hash '$sha...1', got '$sha...2' " -- obviously anything with a "$" varies depending on the path in question -- It's a graphical console, so I can't copy out the message
<aanderse>
anyone else notice something really fishy about ogl apps in 19.03+?
<aanderse>
a bunch of them don't compile anymore because bad headers
i1nfusion has quit [Remote host closed the connection]
i1nfusion has joined #nixos
bhipple has joined #nixos
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
Erasmus has joined #nixos
mmlb977450853 has joined #nixos
bhipple has quit [Read error: Connection reset by peer]
mmlb97745085 has quit [Ping timeout: 246 seconds]
kreisys has joined #nixos
aanderse_ has joined #nixos
aanderse has quit [Ping timeout: 264 seconds]
Anton-Latukha has quit [Quit: Leaving.]
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
alex``` has quit [Ping timeout: 245 seconds]
aanderse_ has quit [Ping timeout: 246 seconds]
aanderse has joined #nixos
jasongrossman has joined #nixos
schneid3306 has joined #nixos
npmccallum has joined #nixos
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
dbmikus has quit [Ping timeout: 255 seconds]
orivej has joined #nixos
silver has quit [Read error: Connection reset by peer]
Supersonic112 has joined #nixos
Supersonic has quit [Disconnected by services]
Supersonic112 is now known as Supersonic
schneid3306 has quit [Quit: ZZZzzz…]
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
jluttine has joined #nixos
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
freeman42x has quit [Ping timeout: 252 seconds]
sigmundv_ has quit [Ping timeout: 252 seconds]
lovesegfault has quit [Quit: WeeChat 2.4]
lionello has joined #nixos
<lionello>
Morning all.
<lionello>
I'm struggling with building a minimal docker image with dockerTools.buildImage. One of my derivations pulls in openssl, and then gcc et al end up in the docker container, adding +200MiB.
buffet_ has joined #nixos
buffet has quit [Ping timeout: 250 seconds]
<lionello>
I think there's something I can do avoid that using propagatedBuildInputs but can't find a good example of it
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
kreisys has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
kreisys has joined #nixos
_Geeko_ has left #nixos [#nixos]
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
infinee has joined #nixos
elvishjerricco has joined #nixos
elvishjerricco has quit [Changing host]
o1lo01ol1o has quit [Remote host closed the connection]
leat1 has quit [Remote host closed the connection]
jasongro` has joined #nixos
leat1 has joined #nixos
drakonis has joined #nixos
jasongrossman has quit [Ping timeout: 255 seconds]
jasongro` has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
hlolli_ has joined #nixos
leat1 has quit [Remote host closed the connection]
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
o1lo01ol1o has joined #nixos
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
Lears has quit [Ping timeout: 250 seconds]
o1lo01ol1o has quit [Ping timeout: 250 seconds]
hlolli_ has quit [Remote host closed the connection]
hlolli_ has joined #nixos
ninjin has quit [Ping timeout: 256 seconds]
ninjin has joined #nixos
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
<tA>
howdy, im trying to use the sdl haskell library, and when i run nix-shell --pure i get "no avaliable video device" and "XDG_RUNTIME_DIR not set", my nix files are here: https://git.lain.church/tA/tAre any ideas?
<tA>
i think i need to add SDL to the build inputs, but when I do that it says theres no thing, so im deffo doing something wrong, double checked with nix search
<mabel>
I just saw the "nixos" attribute defined in all-packages.nix... it seems like a really interesting way to build a nixos environment for a container
leat1 has quit [Remote host closed the connection]
<mabel>
nixops doesn't have a Docker backend, and Nix's docker infrastructure is so lightweight that it doesn't actually set up anything like /etc
<mabel>
so I'm going to try an experiment: use pkgs.nixos (or this profile-docker thing) to define a simple container base, then extend that with modules for each container I want to create
<mabel>
then use dockerTools.buildLayeredImage to construct the tarballs
<srhb>
mabel: I'm sure sure what you expect to gain from the nixos parts.
<jackdk>
I don't know the docker infra super-well, but you can dockerTools.buildImage atop alpine, say
<mabel>
srhb: I've been fighting for a day to make a nix expression for Hydra inside docker containers
<mabel>
it's very simple to set services.hydra.enable = true; in NixOS
<mabel>
if I base on alpine, I can't use buildLayeredImage, additionally
<mabel>
additionally, I need to set special options for nix to make it suitable for Hydra's use... afaict I can't use options.nix to have it generate the right config
<mabel>
so then I have to figure out where to binplace /etc/nix.conf, how to structure that, etc... all the drudgery nixos usually frees me from
orivej has quit [Ping timeout: 255 seconds]
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
<srhb>
mabel: I imagine that's difficult yeah. I mean, even getting systemd to run in docker used to be tricky
<srhb>
mabel: And NixOS, in essence, is just nix + nixpkgs + systemd
random_yanek has joined #nixos
<mabel>
well, nix + nixos + nixpkgs + systemd. the modules in nixos make life much easier
<srhb>
mabel: But in that case, it might be easier to just use the module system
<srhb>
mabel: I guess the dream situation is to decouple each module from the implementation (ie. systemd) but we're not really there yet, though some people have done some things in this relation.
<mabel>
now, I know that home-manager uses systemd --user units. if only I could start those units as --user, I wouldn't need the container to be privileged at all
<mabel>
(I mention home-manager only because that's the one example I know of where a nix derivation compiles down into totally usermode systemd stuff)
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
rauno has joined #nixos
drakonis has quit [Remote host closed the connection]
enteee has joined #nixos
<{^_^}>
[nixpkgs] @xrelkd opened pull request #59199 → cargo-expand: init at 0.4.10 → https://git.io/fjtPP
domogled has joined #nixos
<mabel>
srhb: it totally worked btw! I had to exec it privileged but that example succeeded for me
<srhb>
mabel: Cool! :)
<enteee>
hi, I would like to change the run-script build by build-vm (different -vga option, etc.). How can I achieve this in my nixos configuration?
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
<enteee>
I think the module used by nixos-rebuild build-vm is nixos/modules/virtualisation/qemu-vm.nix which declares options such as virtualisation.qemu which seems to be doing what i want, but how do I set those so that nixos-rebuild build-vm picks them up?
slack1256 has quit [Remote host closed the connection]
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
<clever>
enteee: just add them to configuration.nix, like any other option
<mabel>
base image is ~750MB though, boot time is several seconds and I need privileged execution to run systemd. it was a good experiment to try, still
<enteee>
clever: thx, but for example adding virtualisation.memorySize = 500; does not seem to work
<enteee>
clever: so I am not quite sure how to add them. Or how would you do this?
crmlt has joined #nixos
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
arjen-jonathan has joined #nixos
<lionello>
I'm making some progress. Now using allowedRequisites=[] and it complains gcc and others ends up in the container
Tucky has joined #nixos
wfranzini has joined #nixos
reinhardt has joined #nixos
tippenei1 has joined #nixos
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
tippenei1 is now known as tippenein
tippenein has left #nixos [#nixos]
zupo has joined #nixos
wfranzini has quit [Remote host closed the connection]
m0rphism has joined #nixos
<clever>
enteee: how exactly does it fail?
wfranzini has joined #nixos
endformationage has quit [Ping timeout: 250 seconds]
<enteee>
clever: The option `virtualisation.memorySize' defined in `/etc/nixos/configuration.nix' does not exist.
<clever>
enteee: does the build still finish?
<enteee>
clever: umm, so nixos-rebuild test exists with code 1. so i guess no
<clever>
enteee: what about nixos-rebuild build-vm --fast ?
<Aleksejs>
So, I solved my yesterday's problem with pip... by throwing away nixos and installing raspbian :( I couldn't find any working solution how to install specific python packages on nixos
leat1 has quit [Remote host closed the connection]
realrokka has quit [Remote host closed the connection]
crmlt has quit [Read error: Connection reset by peer]
realrokka_ has joined #nixos
<srhb>
Aleksejs: It's pretty well-documented in the nixpkgs manual, did you read that?
<Aleksejs>
srhb: tried that
crmlt has joined #nixos
<srhb>
Aleksejs: What's the problem?
leat1 has quit [Remote host closed the connection]
<Aleksejs>
It couldn't findnspecific packages
leat1 has joined #nixos
<srhb>
Maybe that package hasn't been made yet?
stites has quit [Ping timeout: 245 seconds]
Unode has quit [Ping timeout: 245 seconds]
gentauro has quit [Ping timeout: 245 seconds]
makefu has quit [Ping timeout: 245 seconds]
aramiscd has quit [Ping timeout: 245 seconds]
<Aleksejs>
Well, pip install rpi_281x on raspbian worked well
<srhb>
Aleksejs: If you insist on using pip, raspbian is probably a better fit than NixOS anyway. :)
stites has joined #nixos
delroth has quit [Ping timeout: 245 seconds]
gentauro has joined #nixos
<srhb>
Aleksejs: Nothing wrong with that.
aramiscd has joined #nixos
<srhb>
Nix works best when you go all-in -- that means packaging everything in Nix.
makefu has joined #nixos
Unode has joined #nixos
<srhb>
(meaning: No pip)
Supersonic has quit [Ping timeout: 245 seconds]
xantoz has quit [Ping timeout: 245 seconds]
delroth has joined #nixos
Supersonic has joined #nixos
<Aleksejs>
I tried all solutions that I could find, I've spent about 6 hours and couldn't find a way to install that package. If you know the way to install it then I'll give it a try
<srhb>
Aleksejs: I don't have time to do a walkthrough now, but if you pastebin your attempts, maybe someone will chime in and explain what went wrong. :)
<makefu>
Aleksejs: packing for NixOS is very different from other distributions. in general you do not mutate the global state, but create a build recipe which exactly defines how the complete package will be built. in your case for scdl you would need to create such definition yourself unfortunately as it is not yet packaged. in your case the package does some very, say interesting steps when running. namely it tries
<makefu>
build it with nix-build -E 'with import <nixpkgs> {};pkgs.python3.pkgs.callPackage ./default.nix {}'
lionello has quit [Ping timeout: 246 seconds]
sam_ has quit [Remote host closed the connection]
jackdk has quit [Ping timeout: 252 seconds]
domogled has quit [Quit: domogled]
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
<makefu>
the derivation compiles the source code for your platform (when run on said platform)
<Aleksejs>
Thanks! I will try it later today
<makefu>
cool!
__monty__ has joined #nixos
goibhniu1 has joined #nixos
<makefu>
when you test the library you need to include it in your environment by (pkgs.python3.pkgs.callPackage ./default.nix {}) inside your python.withPackages { } stuff
<makefu>
otherwise your standard interpreter will not find it
crmlt has quit [Ping timeout: 255 seconds]
frem has quit [Quit: Connection closed for inactivity]
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
lovesegfault has joined #nixos
gal_bolle has joined #nixos
drager has joined #nixos
xantoz has joined #nixos
<drager>
Hi, how can ssh-add my id_rsa automatically on login? I.e propmt for password
<typetetris>
Calling a shell as a login shell stopped working for me. After `bash -l` PATH is not set correctly and doesn't contain /run/current-system/sw/bin for example. This makes tmux unusable. It seems __NIXOS_SET_ENVIRONMENT_DONE=1 is already contained in the environment of the new `bash -l` process, preventing /etc/profile to source `/nix/store/bjpffpfkx4rxzgypf43ryqxqgcw974aa-set-environment`. It seemed to work differently
<typetetris>
before, because login shells stopped working for me with an upgrade done yesterday. Is this considered a bug? Against which part/program should I report it?
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
hyper_ch2 has joined #nixos
<NickHu>
drager: Check out https://www.funtoo.org/Keychain - it's packaged in nix and also home-manager if you are using that
jasongrossman has quit [Quit: ERC (IRC client for Emacs 26.1)]
<drager>
NickHu: Thanks!
jasongrossman has joined #nixos
<typetetris>
Okay, just tripped over my own feet.
<typetetris>
Followed the example for setting MANPATH, INFOPATH in nixpkgs manual, which also absolutely sets PATH.
Guest66068 has quit [Read error: Connection reset by peer]
<NickHu>
I'm encountering the strangest build error - https://paste.xinu.at/1Nrz/ - but the archive extracts fine when I do it myself! 06be9865c6f418d851ff4494e12406568353b891ffe1f596b34693c387af26c7 /nix/store/0fpf4zfxzkz5yxcdp4wbyy1z52yzj1kw-acl-2.2.53.tar.gz
<NickHu>
Has anyone seen anything like this?
Makaveli7 has joined #nixos
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
sinner has joined #nixos
sinner is now known as Guest73949
winem_ has joined #nixos
Piece_Maker has joined #nixos
callahad has quit [Quit: Ping timeout (120 seconds)]
Acou_Bass has quit [Quit: byeeeeeeeeeeeeeee]
Jackneill has quit [Ping timeout: 245 seconds]
Tucky has quit [Ping timeout: 245 seconds]
gspia has quit [Ping timeout: 245 seconds]
EoF has quit [Ping timeout: 245 seconds]
xantoz has quit [Ping timeout: 245 seconds]
gspia_ has joined #nixos
callahad has joined #nixos
Piece_Maker is now known as Acou_Bass
Jackneill has joined #nixos
EoF has joined #nixos
drager has quit [Quit: WeeChat 2.2]
Tucky has joined #nixos
xantoz has joined #nixos
reinhardt has quit [Quit: Leaving]
<NickHu>
Maybe the chroot that it's building in is somehow running out of space? I've tried multiple times and it gives the same error
knupfer has joined #nixos
sam_ has joined #nixos
leat1 has quit [Remote host closed the connection]
knupfer has quit [Client Quit]
knupfer has joined #nixos
leat1 has joined #nixos
<sphalerite>
NickHu: what about /mnt/tmp? :)
<NickHu>
sphalerite: Doesn't exist
<NickHu>
Should it?
<sphalerite>
hm
<NickHu>
I'm building with nix-build --store /mnt ...
<sphalerite>
I'm not sure
<NickHu>
I have my hard drive mounted to /tmp, so 30GB free
i1nfusion has quit [Remote host closed the connection]
i1nfusion has joined #nixos
<NickHu>
I was thinking about perhaps just unpacking the archive manually, and remounting the store rw (gasp) and pointing the src in the derivation there - this wouldn't change hashes and stuff would it?
<__monty__>
Systemd question: I don't mind having polkit ask for a password. I do mind having to type it multiple times for one systemctl command. Can I have pkttyagent remember the password for a single systemctl invocation?
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
i1nfusion has quit [Remote host closed the connection]
Anton-Latukha has joined #nixos
leat1 has quit [Remote host closed the connection]
<Taneb>
Is there any sort of wrapper function in nixpkgs which takes a derivation creating a single file and makes a derivation which puts that file into a directory?
i1nfusion has joined #nixos
<Taneb>
I guess it's not too hard to do with runCommand...
leat1 has joined #nixos
ubert has joined #nixos
wfranzini1 has joined #nixos
<symphorien>
there is writeScriptBin for /bin
wfranzini has quit [Ping timeout: 245 seconds]
wfranzini1 is now known as wfranzini
<Taneb>
This is more appropriate for something like /share
<averell>
don't know, but maybe nix-shell --pure comes close enough?
<cmacrae>
Hey peeps o/ Working my way through Nix Pills to better understand things so I don't have to come in here so often asking these sorts of questions! One thing I'm trying to tackle at the moment; is there a way of conditionally including an attr in a set? I've written a pseudo implementation here to better express what I'm trying to achieve: https://git.io/fjtMB
<cmacrae>
Plan is to use the resulting set with builtins.toJSON, which I'm already doing... I just want to be able to conditionally craft one of the attrs based on the system hostname
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
<immae>
cmacrae: would something like workSpecificAttr = if config.networking.hostName == "work-laptop" then ({ some = "work"; ...}) else null; be enough for you?
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
domogled has joined #nixos
<ashkitten>
jomik: what editor
<ashkitten>
have you specified the path to rls
<ashkitten>
does it show errors
<jomik>
Yes, neovim currently. Using CoC-nvim with a custom language server. It executes "${pkgs.rls}/bin/rls" and it does give syntax errors and such.
<jomik>
I do get some errors from rustc.
<jomik>
through rls I suspect - I can see that it is running in `CocOpenLog`
<ashkitten>
does your language client have completion support
<ashkitten>
oh wait, it says it's compiled into rls
<jomik>
That's what I assume it is :P
<jomik>
It should be at least, yah.
<ashkitten>
then i have no idea
<ashkitten>
ask #rust or whatever
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
<jomik>
Hmm, good idea
jomik has quit [Quit: WeeChat 2.4]
<chris___>
JOIN #haskell
jomik has joined #nixos
<ashkitten>
welcome back, lol
domogled has quit [Quit: domogled]
<jomik>
I wrote /quit by mistake xD
<jomik>
#rust is not on freenode.
<ashkitten>
it's on moznet
<ashkitten>
irc.mozilla
<cmacrae>
johanot: Well, it works from a testing perspective. So, thank you! However, when I try and evaluate the hostName, nix is upset that the value is a function and not a set. I've updated the gist: https://git.io/fjtMB - any ideas?
<cmacrae>
johanot: Hmmm, just tried with parens around varying parts (whole statement, just optionalAttrs, just optionalAttrs and its boolean eval, etc.), all to the same failure
<siraben>
tilpner: how did you get the "riot" command to work? in your .bashrc?
<johanot>
cmacrae: oh.. didn't see the error msg at all actually, just assumed.. Well.. The error is "with lib.attrsets".. You just need "with lib;"
Jetien_ has quit [Ping timeout: 268 seconds]
<tilpner>
siraben: I put that wrapper into my user packages
<tilpner>
siraben: So into users.users.tilpner.packages
<siraben>
tilpner: ah, so what does it look like there?
<cmacrae>
johanot: Ah I don't think it's that, sadly. It works using 'with lib.attrsets' when the bool eval for 'optionalAttrs' is just 'true'. You can see I put a comment in there of what works :) Just tried using just 'with lib' and it didn't work
<cmacrae>
(Really appreciate your help on this btw, sorry to pester!)
chris___ has quit [Ping timeout: 256 seconds]
<johanot>
cmacrae: what is "nixos" in your NIX_PATH?
iyzsong has joined #nixos
arjen-jonathan has quit [Ping timeout: 268 seconds]
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
<cmacrae>
johanot: Ah... I don't have a 'nixos' entry in my NIX_PATH. I was going off examples for how to be able to evaluate 'config.networking.hostName'... so I suppose I approached that wrong
<cmacrae>
What would I need to import to be able to evaluate that?
<siraben>
tilpner: cool. So that's how you keep it declarative, wow.
ma27 has quit [Quit: WeeChat 2.4]
<siraben>
I'm glad that since I switched to NixOS I have never had my system in an unusable or unpredictable state
<siraben>
The heavy space usage is the only minor annoyance I suppose, but how else would it be possible to have a declarative system
<tilpner>
(I do recommend the overlay for pretty-ness, but do that whenever you're comfortable with it)
jomik has joined #nixos
<siraben>
tilpner: just got it working, thanks.
<tilpner>
\o/
<tilpner>
jomik: You pinged out before "vorner | jomik, I use nvim+LanguageClient-neovim+ncm2+rls. I just type „variable_name.“ and the completion selection simply appears."
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
<jomik>
tilpner: Oh, thanks :D My computer has started randomly freezing. Worrying that it may be a NixOS 19.03 issue :o
<jomik>
Can't get past it without rebooting.
<cmacrae>
Thanks for the comment johanot! <3 I think this is a better approach
cfricke has quit [Ping timeout: 268 seconds]
<cmacrae>
Just got to figure out how I can call builtins.toJSON on the resulting import
razzy has joined #nixos
<johanot>
cmacrae: see my second comment on the gist :-)
sam_ has joined #nixos
<ajs124>
Is there a way to set kernel.features for linuxPackages_custom or linuxManualConfig?
<ajs124>
I'm asking because otherwise opengl.driSupport32Bit doesn't work
orivej has joined #nixos
<razzy>
hi, i am retrying nixos on RPI B. totally newbie, trying to make wifi work.
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
civodul has joined #nixos
<razzy>
is there only vi editor?
<__monty__>
Nano/pico might be included be default too.
<johanot>
razzy: on the install medias, "nano" is usually present
<__monty__>
Oh, and there's always ed of course : >
<johanot>
__monty__: right :)
Enzime__ is now known as Enzime
<Enzime>
does anyone know what the difference between `users.users` and `users.extraUsers` is?
<Enzime>
and which I should use in my /etc/nixos/configuration.nix?
<srhb>
Enzime: THe difference cis that users.extraUsers is deprecated, iirc.
<Enzime>
has anyone here set up remote building?
<srhb>
Enzime: (But users.extraUsers still exists as an alias for users.users)
<etu>
Enzime: users.extraUsers is an alias of users.users
<srhb>
Enzime: Lots of people.
<cmacrae>
Thanks so much johanot! Sadly, when I try this, it seems that (when exposing parameters, like so '{ hostName }:') I get 'value is a function while a set was expected'
<Enzime>
can I use the nixbld* users?
<Enzime>
or do I need to make a custom one
<srhb>
Enzime: I prefer not to use the nixbld users.
<razzy>
hmm, is there a way to copy from help?
<Enzime>
srhb: any reason in particular?
<srhb>
Enzime: Mostly because I can't easily understand the security implications of allowing them to do things remotely at all
silver has joined #nixos
<srhb>
Especially in a trusted fashion
<srhb>
I feel like it's breaking part of the isolation at the very least.
<srhb>
Enzime: Besides, it feels arbitrary to allow some(all?) of them to do this.
<srhb>
Rather not :)
<johanot>
cmacrae: In order for me to help debug that, I guess I need to see a new gist :-)
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
<cmacrae>
Of course, thanks! I'll update the gist
<enteee>
clever: thanks for your help already :)
<LnL>
don't use nixbld users for other things, nix will kill all it's processes before/after builds
<Enzime>
is there anything I should do when I make this special account for remote building to lock it down or anything?
adamt has joined #nixos
jomik has quit [Quit: WeeChat 2.4]
Jetien_ has joined #nixos
<cmacrae>
There ya go johanot - I've updated it to show what I'm trying :)
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
<symphorien>
Enzime: processes owned by nixbld* users will be killed by the nix-daemon at the end of builds
<symphorien>
so you can't really use them
winem_ has joined #nixos
<cmacrae>
Thank you johanot! Will take a stab at this again with your comments <3
<gchristensen>
oh dear don't use nixbld Enzime :D
<Enzime>
it appears I've caused a ruckus
lionello has joined #nixos
<gchristensen>
:)
<ninjin>
nh2[m]: Thank you for the offer, got Hetzner working without a hitch now – not using Nixops though since that did not work for me.
npmccallum has quit [Ping timeout: 246 seconds]
DigitalKiwi has quit [Read error: Connection reset by peer]
DigitalKiwi has joined #nixos
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
DigitalKiwi has joined #nixos
DigitalKiwi has quit [Changing host]
pointfree has quit [Ping timeout: 240 seconds]
<ninjin>
Essentially, wrote Bourne script to drive the install – I am too senile to remember things otherwise – and adapted the “Install from other OS” instructions from the manual to work inside the Hetzner rescue system as root.
hexagoxel has quit [Quit: ZNC 1.6.5+deb1+deb9u1 - http://znc.in]
<{^_^}>
[nixpkgs] @andir pushed 2 commits to staging-18.09: https://git.io/fjtSL
<ninjin>
Key things I learned: 1.) Partition, then RAID for `/boot` et al. 2.) Encrypted swap with new keys for each boot are most easily done using a swapfile. I think that was about it really, the rest I just learned by putting together things from the manual, the Arch wiki, and random searches.
Myhlamaeus has quit [Ping timeout: 264 seconds]
erictapen has joined #nixos
orivej has quit [Ping timeout: 255 seconds]
wolfman154 has joined #nixos
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
fusion809 has joined #nixos
<{^_^}>
[nixpkgs] @basvandijk merged pull request #58255 → Add Prometheus 2 service in parallel with 1.x version (continuation) → https://git.io/fjUvu
Soo_Slow has quit [Remote host closed the connection]
wolfman154 has quit [Client Quit]
wfranzini1 has joined #nixos
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
wfranzini has quit [Read error: Connection reset by peer]
wfranzini1 is now known as wfranzini
crmlt has joined #nixos
hexagoxel has joined #nixos
lionello has quit [Quit: leaving]
linarcx has quit [Quit: Quit]
linarcx has joined #nixos
leat1 has quit [Remote host closed the connection]
wfranzini has quit [Ping timeout: 246 seconds]
leat1 has joined #nixos
arjen-jonathan has joined #nixos
sam_ has quit [Remote host closed the connection]
sam_ has joined #nixos
wolfman154 has joined #nixos
orivej has joined #nixos
wolfman154 has quit [Client Quit]
knupfer has joined #nixos
linarcx has quit [Remote host closed the connection]
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
logzet has joined #nixos
cfricke has joined #nixos
crmlt has quit [Ping timeout: 245 seconds]
linarcx has joined #nixos
hyper_ch2 has quit [Quit: Page closed]
<linarcx>
Guys. Anyone here try to install nixos an guixsd on same partition and dual boot between them?
<linarcx>
*and
<gchristensen>
fascinating
<linarcx>
Why?
<ajs124>
linarcx, you can most certainly do that with zfs (if guixsd supports zfs, which it might not, because CDDL). When I first started using NixOS, I installed it next to Arch Linux on the same partition, in a different dataset.
<gchristensen>
having two distros competing to manage /etc and /var is interesting
<gchristensen>
I know it can be done, but man, fascinating :)
<linarcx>
Ha ha ha:)
wfranzini has joined #nixos
<linarcx>
ajs124: I don't know about zfs on guixsd. But yes, nixos nicely detect all other os and distributions without any problem. Guixsd only detect itself :)
werner291 has joined #nixos
<etu>
ajs124: btrfs! ;)
leat1 has quit [Remote host closed the connection]
<musicmatze>
I have no rights to close issues... but maybe one of you (maybe YOU) could just mark them all as closed because they're all stalled?
stepcut has quit [Remote host closed the connection]
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
stepcut has joined #nixos
<andi->
Anyone using wayland and firefox on 19.03 that is able to confirm if XWayland is still required for Firefox or not? I added support a while back but haven't checked lately (not yet on wayland)
<MiloIgnis[m]>
I think it was as of a week or two ago. (I tried running with and without the GDK_BACKEND=wayland - is that the right way to do it?)
waleee has joined #nixos
razzy has left #nixos ["ERC (IRC client for Emacs 26.1)"]
<andi->
mhmm
stepcut has quit [Remote host closed the connection]
<andi->
the GDK_BACKEND thing was enabled by the change I did but I somehow guessed it wouldn't be required to run if there wasn't an XWayland running already
<andi->
the reason GDK_BACKEND=wayland is still a good idea is when your terminal (supposedly running on XWayland) wants to start the browser. If that wouldn't be set then firefox would fallback to the X11 rendering.
sam_ has quit [Remote host closed the connection]
<andi->
MiloIgnis[m]: if you use the package `firefox-wayland` instead of firefox that variable should be set automatically. It really is the same package just with a wrapper that enforces wayland in the above described scenario.
ddellacosta has joined #nixos
<MiloIgnis[m]>
Ah! I hadn't found that - I'll try it tonight after work
<andi->
please give feedback :) I am trying to get the wayland support where it should be.. I realized that all I use on my computer is really Firefox+terminal and both support wayland
sam_ has joined #nixos
<MiloIgnis[m]>
I had been running it from Sakura, but I confirmed beforehand that XWayland wasn't running
<MiloIgnis[m]>
<andi- "please give feedback :) I am try"> Same, as per daily usage. Sway's pretty great too
leat1 has quit [Remote host closed the connection]
<MiloIgnis[m]>
I appreciate your efforts!
leat1 has joined #nixos
stepcut has joined #nixos
jasongrossman has quit [Ping timeout: 255 seconds]
stepcut has quit [Remote host closed the connection]
sam_ has quit [Remote host closed the connection]
knupfer has quit [Ping timeout: 245 seconds]
kreisys has joined #nixos
jasongrossman has joined #nixos
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
stepcut has joined #nixos
<werner291>
Hmm... I'm probably missing something obvious, but I'm trying to set up nginx on my NixOS box. I can download the index.html page with wget when I'm logged into the server, but I can't seem to reach it from the outside world. It might be a firewall issue, but somehow openss doesn't seem to have any trouble running on an alternative port.
sam_ has joined #nixos
<werner291>
*openssl
<{^_^}>
[nixpkgs] @lsix opened pull request #59212 → python3Packages.django_2_2: init at 2.2 → https://git.io/fjtHw
hlolli has joined #nixos
<werner291>
Oh... NixOS has a default-enabled firewall.
<hlolli>
Am I doing something wrong with `leaveDotGit = true;` when I stat the .git directory, it doesn't exists?
<{^_^}>
[nixpkgs] @matthewbauer pushed commit from @nomeata to master « macdylibbundler: init (#59173) »: https://git.io/fjtHN
mmlb977450853 has joined #nixos
wfranzini has quit [Read error: Connection reset by peer]
crmlt has joined #nixos
wfranzini has joined #nixos
wfranzini has quit [Read error: Connection reset by peer]
<hlolli>
Is volth ever here on irc? I'm actually trying to bump graalvm, and the mx build tool is driving me to madness. They pull from github in every build, and the cache system is based on python's flock. It seems almost impossible to get this working with nix. I could patch the hell out of the mx.py file, but it would make it almost unuseable for other than building graalvm on nix.
mmlb9774508530 has joined #nixos
<simpson>
hlolli: Patching crappy software to force it to obey Nix is very Nixy.
linarcx has quit [Remote host closed the connection]
<hlolli>
would it make sense to have an alternative mx just for building. The mx command basiaclly runs my.py, I could create mx_patched.py just for installing graal, and the old mx would work just fine, installing into home directories...
mmlb977450853 has quit [Ping timeout: 245 seconds]
<hlolli>
Or copy the mx.py to mx_patched.py and patch that with patches.
mmlb97745085305 has joined #nixos
jasongro` has joined #nixos
mmlb9774508530 has quit [Ping timeout: 245 seconds]
leat1 has quit [Remote host closed the connection]
<simpson>
hlolli: Yeah, sounds like a temp dir of some sort will be needed. It's unfortunate that they did not immediately understand *why* Nix wants this.
<hlolli>
The only way to do this, would be to make this lockfile from flock(1) https://github.com/graalvm/mx/blob/master/mx.py#L6455-L6463 but then I'd need to append "flock lock" to the fetchgit command, and the .git directory would need to be present to test the revision match.
hexagoxel has joined #nixos
<hlolli>
alternativelt, just skip this whole thing and point to the directory directly without any checks.
<simpson>
Oh jeez, yeah, upstream mx should fix that to put the flock in $TMPDIR or some other standard place. This assumption of writing to the build tree is not correct in other environments either.
dsx has joined #nixos
ddellacosta has quit [Ping timeout: 268 seconds]
Soo_Slow has joined #nixos
o1lo01ol1o has joined #nixos
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
drakonis has joined #nixos
jomik has joined #nixos
<{^_^}>
[nixpkgs] @vcunat pushed 2 commits to release-19.03: https://git.io/fjtQi
iyzsong has quit [Ping timeout: 252 seconds]
ddellacosta has joined #nixos
dhkl has joined #nixos
leat1 has quit [Remote host closed the connection]
<hlolli>
Feel free to comment on this ticket, or anyone, I'm not particularly articulate when arguing for good practices. Life is too short for chasing python build tool "experts" :-) I think an internal_patched_mx.py makes most sense to get this running.
leat1 has joined #nixos
<Myrl-saki>
Okay, I'm kind of stupid. What's the proper way to mount a POSIX file system with a certain UID? bindfs?
<manveru>
simpson: btw, i got the tahoe-lafs HEAD and gridsync 0.4.1 built the other day, i think it might be worth putting those into nixpkgs
<{^_^}>
[nixpkgs] @FRidh pushed 334 commits to staging-next: https://git.io/fjtQQ
<manveru>
before someone else has to brave the python madness :)
nschoe has joined #nixos
<Myrl-saki>
I wonder if it's also possible to do this with namespaces, or is that just nuking a fly?
<simpson>
manveru: Sure, I'll review the PR.
<manveru>
cool, thanks
Makaveli7 has quit [Read error: Connection reset by peer]
<manveru>
simpson: do you know how to make tox tests work?
<simpson>
manveru: Nope. I think that we should just run the test suite directly for the given Python, though, rather than trying to wedge in tox.
<manveru>
they're pretty nasty... shadow the python path env var, and try to download the deps in a virtualenv and stuff :(
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
orivej has quit [Ping timeout: 255 seconds]
xkapastel has joined #nixos
<OmnipotentEntity>
I'm having a bit of trouble with pip2nix, it's giving an python import failure that indicates that cmdoptions does not exist within pip. Was wondering if anyones run face first into a similar problem where the pip package seems to be incomplete.
<simpson>
omnipotententity: How many dependencies are there? Is handwriting a Nix expression feasible?
<OmnipotentEntity>
I'm not certain, it might be? But even if I do workaround the problem manually, I'm worried that this error is telling me something I ought to know about my installation.
<OmnipotentEntity>
There's only 4 lines within the requirements.txt, so I could probably manually write a nix expression
kleisli_ has joined #nixos
kleisli has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @matthewbauer opened pull request #59216 → binutils-wrapper: Try to avoid adding unnecessary -L flags → https://git.io/fjt50
leat1 has quit [Remote host closed the connection]
jomik has joined #nixos
leat1 has joined #nixos
werner291 has quit [Ping timeout: 256 seconds]
<yorick>
so my placeholder literally ends up in the output file
<yorick>
as /1rz4g4znpzjwh1xymhjpm42vipw92pr73vdgl6xs1hycac8kf2n9
winem_ has joined #nixos
<yorick>
steps to reproduce: pkgs.writeTextFile { name = "test"; text = "${placeholder "out"}";}
<NemesisD>
does anyone have suggestions on interactively debugging nix derivations? i'm trying to write one and i'm getting a file does not exist error even though the file definitely does, so I think it may be an environment difference between my native shell and the build shell
<NemesisD>
i'd like to just drop into the shell that's running my build script during the build phase if possible
leat1 has quit [Remote host closed the connection]
<__monty__>
NemesisD: nix-shell yourexpression.nix is how that's usually done, then run the build phases manually.
leat1 has joined #nixos
<jomik>
tilpner: IT WORKS!
<jomik>
Had to set `let $RUST_SRC_PATH = '${pkgs.rustPlatform.rustcSrc}'` in my vimrc :)
<NemesisD>
__monty__: the only thing is that it is just leaving me in the cwd, do i like create a dir in /tmp and work from there or something?
<__monty__>
The paths used should be in the nix store afaiui?
<__monty__>
Like, running through it manually should result in the same things happening as when you nix-build it.
leat1 has quit [Remote host closed the connection]
<NemesisD>
__monty__: so if i do `nix-shell -A conda-multiplatform --pure default.nix` and then paste my exact installPhase, it succeeds, if i run `nix-build -A conda-multiplatform --no-out-link` it fails partway
leat1 has joined #nixos
cfricke has quit [Quit: WeeChat 2.4]
<__monty__>
Maybe it's not the installPhase that goes awry? This is where I have to tap out tbh.
<NemesisD>
the error is even stranger, `/nix/store/fcjrmvdfnmwz7q3sb94xhy6081a9l6i7-Miniconda3-4.3.31-Linux-x86_64.sh: line 353: /nix/store/lq7g1dz66cncckp4gysxvx9plygwvrj3-conda/pkgs/python-3.6.3-h6c0c0dc_5/bin/python: not found`but if i run `file -L /nix/store/lq7g1dz66cncckp4gysxvx9plygwvrj3-conda/pkgs/python-3.6.3-h6c0c0dc_5/bin/python` the file exists, is an executable
<Akii>
yorick I can surely get around that with installing git first but what's the proper way to express this dependency?
devalot has joined #nixos
<NemesisD>
__monty__: fair enough. the package has an unpackPhase and buildPhase of `true`
<Akii>
yorick and then, where do I put the SSH keypair? Is a system wide configuration or what
wangoe has joined #nixos
<devalot>
Is there a way to override a NixOS module option type? There's a service with a bool option that when set to false generates an invalid configuration file. I need it to be a string.
<__monty__>
NemesisD: Did you hardcode that path in some way maybe? (I'm basically stalling until someone who knows better steps in.)
<NemesisD>
__monty__: my src is a big shell script installer for miniconda. my installPhase is `set -x && chmod +x $src && $src -p $out -b -f`
<infinisil>
devalot: Not really no
<NemesisD>
-b for batch mode so it won't ask questions, -f for force to override the output as i debug, -p configures the installation prefix
adamt has quit [Ping timeout: 250 seconds]
dbmikus has quit [Ping timeout: 258 seconds]
<Akii>
oh wait, I can just put them in roots home folder
<Akii>
hm
<Akii>
maybe?
<wangoe>
Hi, does anyone know a good way to change a declarative containers port without having to undeploy and delete it first? Whenever i use declarative nixos containers and redeploy them via nixops, the container@*.service fails with "Failed to register machine: xxx already exist"
<devalot>
infinisil: That's a bummer.
crmlt has joined #nixos
<infinisil>
devalot: What specifically are you talking about?
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
<devalot>
GitLab smtp starttls setting. It does a toString on a bool when generating some Ruby code. If you use false then you get an empty string which is a syntax error.
<infinisil>
Ah yeah damn
ubert has quit [Quit: Leaving]
polman has joined #nixos
<infinisil>
devalot: Easiest way to fix it is probably disabling the module, copying it, importing the copy, and fixing the type in the copy
<devalot>
I was worried I'd have to do that. Thanks.
mmlb97745085305 has quit [Ping timeout: 245 seconds]
<infinisil>
devalot: Can you submit a PR to fix the problem as well?
<infinisil>
> lib.boolToString false
<{^_^}>
"false"
<infinisil>
Not sure if this is valid ruby, but this function generates "false" and "true"
<devalot>
That would be the expected behavior.
<devalot>
I'll send in a PR.
johanot has quit [Quit: WeeChat 2.4]
drakonis has quit [Quit: WeeChat 2.3]
wfranzini has quit [Remote host closed the connection]
<Akii>
yorick thanks, works like a charm :3
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
wangoe has quit [Quit: WeeChat 2.4]
drakonis has joined #nixos
<NemesisD>
i'm following a guide on debugging derivations and it says to run nix-shell and run all the phases in order, for some reason, e.g. `buildPhase` seems to run the default build phase and `$buildPhase`is set to the one i've customized in my derivation, why is that?
<Myrl-saki>
Does anyone understand how blkid work? It's better than iterating through /dev/disk/by-uuid/* for nixos-generate-config.
<Myrl-saki>
I'm reading the source code, and it feels like I'm trying to read a C++ program written in C.
TronDD-w has joined #nixos
drakonis_ has quit [Ping timeout: 250 seconds]
<NemesisD>
also for some reason multiline phases get packed into 1 line, so my $installPhase is flattened into 1 line with no &&s or anything inbetween the commands
<enteee>
I am trying to configure nixos-rebuild build-vm in my nixos configuration. I would like to change the default for the -vga argument, adding virtualisation.qemu.options = [ "-vga virtio" ]; to my configuration.nix makes nixos-rebuild build-vm --fast work, but every other nixos-rebuild operation breaks. Does somebody know why this does not work?
leat1 has quit [Remote host closed the connection]
<srhb>
enteee: Presumably because the qemu module is only conditionally imported? The error message will probably make it easier for us to debug.
leat1 has quit [Remote host closed the connection]
<azazel>
anyone knows why glusterfs is so old in nixos, maybe there's a specific reason
leat1 has joined #nixos
shibboleth has joined #nixos
<srhb>
azazel: The last update mentions that the newer versions weren't tested extensively, so the maintainer opted for the strictly tested version.
<enteee>
srhb: The option `virtualisation.cores' defined in `/etc/nixos/configuration.nix' does not exist.
<azazel>
srhb: thanks
<srhb>
enteee: Yep, that's the issue.
<enteee>
srhb: where the line is equal to: virtualisation.cores = 4
<enteee>
srhb: so how can I then define options only if the corresponding module is imported?
tinga has quit [Ping timeout: 245 seconds]
<enteee>
srhb: i am pretty new to nix, or am I missing something. Is there an other approach to easily bootstrap a actual usable (with more than 512 mib ram etc..) vm from my nixos config easily?
<{^_^}>
[nixpkgs] @xeji pushed commit from @r-ryantm to master « python37Packages.pymediainfo: 3.0 -> 3.1 (#57436) »: https://git.io/fjtdH
<srhb>
enteee: I think the easiest solution is likely to forgo nixos-rebuild for building your vms and writing a custom expression that does the same thing, but with your option set as well.
nschoe has joined #nixos
<enteee>
srhb: hmm, but wat is the idea behind nixos-rebuild build-vm then if not what i am trying to do?
<srhb>
enteee: I *do* think you can set the memory, can't you?
<{^_^}>
[nixpkgs] @xeji pushed commit from @r-ryantm to master « polyml: 5.7.1 -> 5.8 (#58753) »: https://git.io/fjtdd
<srhb>
Maybe not, it appears gated behind the same import mechanism. Hm.
<enteee>
srhb: sort of, i can do nixos-rebuild build-vm --fast ,with the virtaualization.* options ins my configuration.nix. but every non --fast operation fails
<srhb>
enteee: An easier approach may be to have a separate config that imports your _own_ config along with the qemu stuff, and then point nixos-rebuild to that config whenever you need to build the vm.
<srhb>
enteee: (That is, /etc/nixos/configuration-vm.nix which imports /etc/nixos/configuration.nix)
<srhb>
Then you can nixos-rebuild -I nixos-config=/etc/nixos/configuration-vm.nix
<srhb>
enteee: Uh, actually, I'm surprised --fast is what makes the difference, now that I think about it.
<enteee>
srhb: ok I see..
<srhb>
I would expect _all_ vm actions to work with the virtualisation options, and all non-vm to fail with them..
<{^_^}>
[nixpkgs] @xeji pushed commit from @r-ryantm to staging « nspr: 4.20 -> 4.21 (#57335) »: https://git.io/fjtFU
dbmikus has quit [Ping timeout: 245 seconds]
dbmikus_ has joined #nixos
npmccallum has joined #nixos
winem_ has quit [Ping timeout: 250 seconds]
<enteee>
srhb: hmm probably not the best place to learn nix with ;) anyways I really like the build-vm functionality..
<srhb>
enteee: It is cool, indeed, but it does need some love
<srhb>
I think the entire thing could be salvaged by gating the functionality behind an enable option and conditionally importing the enable = true from nixos-rebuild instead of the import triggering the whole mechanism
<srhb>
That would allow you to always refer to the options, but...
<srhb>
It's a significant rewrite of some core functionality :)
nschoe has quit [Read error: Connection reset by peer]
<NemesisD>
is the nix-build environment somehow isolated from reaching into the nix store? if i run `file -L /nix/store/xgwr5gli31v0p28byxlmnmss9hwmi93l-conda/pkgs/python-3.6.3-h6c0c0dc_5/bin/python` the file is located and the symlink is dereferenced, same as in a nix-shell, but in nix-build the file doesn't exist?
<srhb>
enteee: Maybe you'd like to open an issue describing the problem? I might also have missed some much cleaner solution that someone else will think of.
logzet has quit [Ping timeout: 240 seconds]
<srhb>
NemesisD: Under the sandbox chroot, you can only access actual buildInputs and friends to prevent accidental stateful leakage.
<srhb>
NemesisD: (In short: Yes.)
knupfer has quit [Ping timeout: 250 seconds]
logzet has joined #nixos
<enteee>
srhb: i also tried importing nixos/modules/virtualisation/qemu-vm.nix in my configuration.nix but that destoys my whole system upon rebuild :)
<NemesisD>
srhb: i think i'm doing something wrong then. that path is extracted as part of the install of my derivation
<srhb>
enteee: Yeah, that's what I meant by the import triggering the functionality
<enteee>
srhb: I will open an issue.
erasmas has quit [Ping timeout: 245 seconds]
<srhb>
enteee: Cool, thank you :)
<Myrl-saki>
Yeesh. I tried to check how blkid works, and holy craapp, that's a lot of things going on.
<srhb>
NemesisD: Can you elaborate?
<Myrl-saki>
I was thinking of reimplementing some stuff in Bash or Perl, but dammnnn, that's practically impossible.
<Myrl-saki>
I'm actually quite surprised at how many things exist to make this work out.
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
erasmas has joined #nixos
<enteee>
srhb: or can I maybe achive this using an overlay?
<srhb>
enteee: In the same sense that you can achieve this with a custom nix expression, yes.
<NemesisD>
srhb: sure. so i'm trying to write a derivation for conda (i realize nixpkgs has one but it isn't set up how i need it). i have a `src` that's their shell script installer which has an embedded tarball in it. i give a prefix of `$out` for the installation. it seems like it uses the prefix also as a temporary space for setting up the package and i think that's where it runs into trouble
<srhb>
enteee: But I think it's more work than what I suggested.
<srhb>
NemesisD: Can you share the expression?
<NemesisD>
srhb: i confirmed that setting sandbox to false succeeds. i'm wondering if it would be wiser to set my install prefix to the build's PWD and then in the install phase move that directory to $out
<srhb>
NemesisD: Oh, it's trying to alter the permissions on the src drv itself?
<Myrl-saki>
Anyways, that aside. Would anyone find a PR that uses libuuid for findStableDevPath useful?
<NemesisD>
srhb: that's probably not a good idea is it?
domogled has joined #nixos
<srhb>
NemesisD: It's a separate store path, hence it's read-only
<srhb>
NemesisD: cp $src ./.
<srhb>
Good to go.
<srhb>
NemesisD: (Imagine if any derivation could modify any other store path willy-nilly)
leat1 has quit [Remote host closed the connection]
<NemesisD>
srhb: would that ./. need to be interpolated or just as a literal in the install script?
leat1 has joined #nixos
<srhb>
NemesisD: It will be the build directory, whatever that is, unless you've done something to change it.
<srhb>
NemesisD: So a better idea is probably: cp $src ./install, chmod +x ./install. bash ./install ...
<NemesisD>
srhb: ok, i've done that but i'm still getting the "missing" files issue. would it make sense for me to do this in buildPhase instead of installPhase?
<NemesisD>
srhb: is there a buildInput i could provide?
leat1 has quit [Remote host closed the connection]
<NemesisD>
srhb: also i think it may be unpacking the interpreter from the archive itself rather than downloading it
leat1 has joined #nixos
<srhb>
Yes, you're right on that last point (otherwise the download would also be prohibited)
<srhb>
NemesisD: If it can be made to use the nixpkgs python, sure.
Jetien_ has quit [Ping timeout: 255 seconds]
domogled has quit [Ping timeout: 246 seconds]
<srhb>
NemesisD: I imagine that's specific to the installer, if anything.
thePirateKing has joined #nixos
<Myrl-saki>
How small can /boot get?
<srhb>
Myrl-saki: I had a 16M one earlier today.
<NemesisD>
srhb: conda is like a python package manager that lets you have per-project sandboxes that specify their exact interpreter and all python deps (not my choice, i'm at the mercy of my data science team). i'm basing my work off this but this derivation gives me a FHSEnv rather than a conda executable: https://github.com/NixOS/nixpkgs/blob/master/pkgs/tools/package-management/conda/default.nix
slack1256 has joined #nixos
<srhb>
NemesisD: Sorry, I don't know the tool, but I expect the reason for the FHSEnv is exactly why we're running into these issues now, and probably will be as we fix up things continuously.
cmacrae has quit [Ping timeout: 246 seconds]
<Myrl-saki>
Hm.
<Myrl-saki>
My /boot/initrd is a 200+ MB.
<srhb>
Myrl-saki: I didn't have anything interesting embedded in my initrd.
<srhb>
If you start adding, say, a rescue system or something, it grows quickly.
<NemesisD>
srhb: i was under the assumption that it was because conda is primarily a development tool, so its sensible to give you an interactive shell to drop into that has conda available and nothing else
<Myrl-saki>
More like I'm starting to get worried that I've been doing things wrong.
Neo--- has joined #nixos
<srhb>
NemesisD: I wouldn't know, sorry.
<srhb>
Myrl-saki: Could be? :P
<srhb>
NemesisD: Maybe you can ask jluttine who's listed as a maintainer :)
<srhb>
Myrl-saki: (Meaning, with --install-bootloader)
<{^_^}>
[pass] @zimbatm pushed to master « README: fix link »: https://git.io/fjtFi
<srhb>
Myrl-saki: Which works just fine with nixos-rebuild
<Myrl-saki>
Yeah. nixos-rebuild boot should work, I believe?
<NemesisD>
srhb: i should also mention that the install succeeds if i turn sandboxing off, so i'm not sure if this lib64 thing is actually the problem
<srhb>
Myrl-saki: Depending on your system, you may need --install-bootloader -- I forget exactly what efi things are gated behind that. if it's only firmware, it might be unnecessary, but I'd rather not brick your system for you :-)
<azazel>
NemesisD: what's the reason are you using conda and not pip?
<srhb>
NemesisD: I am a bit puzzled by that frankly. I wonder if it somehow finds a system python in that way. Anyway, i checked and the missing python3 error that I got is definitely due to the loader in /lib64 not existing.
<Myrl-saki>
srhb: Problem is that I've been trying to fit a LiveCD with less than 2 GB. Seems like it's quite impossible now, with the system closure being 2 GB immediately.
<Myrl-saki>
I can probably compress it down to 1 GB with squashfs, but the initrd would be large regardless..
linarcx has joined #nixos
<srhb>
Myrl-saki: Well, yeah, if the store is included, sure..
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
<Myrl-saki>
(It's a rescue drive, so all drivers included and all that jazz.)
<thePirateKing>
conda allows for sandboxing
<thePirateKing>
which makes sysadmins happy
<NemesisD>
azazel: my data science team uses conda to specify their environments (python version + packages). i have the task of carrying that out into production and having new environments be specifiable at runtime. so right now when they push a payload i install/update the conda env and then run their scripts after activating the given conda env
<srhb>
exarkun: It sounds like you want to get rid of the mock and unzip arguments from the function and instead use buildInputs = with pkgs; ...
<NemesisD>
azazel: so in that case i couldn't use nix to target an exact python version and/or set of python packages because they're not known at container build time
Xal has joined #nixos
<srhb>
exarkun: The function is a canonical nixos module, which has an argset like { pkgs, config, options }: ... -- not a callPackage argset.
<srhb>
(Erm, it's a nixos module function, it's _not_ canonical)
<srhb>
enteee: Great, thank you! :)
<TronDD-w>
How do I get a python module installed with pip and have something like Ansible see it? I tried the nix-shell venv method and the venv python interpreter can see the module but ansible still doesn't.
<srhb>
TronDD-w: Is your goal to have ansible see a module or strictly to use pip to do it?
d3vnu11 has joined #nixos
d3vnu11 has quit [Remote host closed the connection]
<TronDD-w>
My goal is to get ansible to see the module
d3vnu11 has joined #nixos
d3vnu11 has quit [Remote host closed the connection]
<srhb>
TronDD-w: Then I think you will have a better and nixier time forgoing pip and overriding ansible via nix itself.
<srhb>
TronDD-w: Probably by adding things to its propagatedBuildInputs
<NemesisD>
azazel: hope that makes sense. so if i were to try to not use conda, i could force my data scientists to pick a specific python version and ship requirements.txt files instead of conda envs and use pip in production to install them, but i'd get no isolation between environments and there'd be a mismatch between their dev and production
<exarkun>
srhb: ah .. ha. Thanks. Can you help me understand why the result is that error about environment.systemPackages?
<srhb>
exarkun: Erm, what error?
<srhb>
Oh, I see it
<Myrl-saki>
Okay, I'm doubly confused now, since seems like I've actually been installing bootloader the whole time. Lmao
<azazel>
NemesisD: you can get isolation doing "python -m venv adir"
<Myrl-saki>
Oh wait no. I had GRUB set to nodev.
<TronDD-w>
srhb: Alright, let me try running down that path
<azazel>
NemesisD: when using python3
<srhb>
TronDD-w: I think overridePythonAttrs will be helpful
Soo_Slow has quit [Ping timeout: 268 seconds]
<exarkun>
srhb: oh hm and the error remains the same if I change the signature to { pkgs, ... }
<azazel>
NemesisD: virtualenv is the most straightforward way of obtaining isolation in python
<exarkun>
srhb: If you could, what tools would you employ to understand the failure better? Can I inspect what value is there?
<exarkun>
oohh..
leat1 has quit [Remote host closed the connection]
<NemesisD>
azazel: interesting. okay so let me walk through this. i'd set up my nix environment with a specific, recent python 3 version. i'd ask my datascience team to give me a requirements.txt for pip instead of a conda env file (i'm hoping conda can produce those). when that file changes, i'd somehow run pip under virtualenv to install to that sandbox's virtual env dir and then run their python script with python3 -m venv theirenv
<NemesisD>
theirscript.py?
<srhb>
exarkun: I stare really hard at the code.
<srhb>
:/
<exarkun>
yes :/ :/ :/
leat1 has joined #nixos
<srhb>
exarkun: But your next error, if any, should hopefully be more helpful..
<noob_>
is there any way to understand how to use the "substituters" of a nix.conf (this is a path to a web cached binary?) to download the linked binary, and then, use putty to place this on a virtual machine running linux with nix on it, instead of getting it from the web chach?
<noob_>
if a fresh linux is placed on a virtual machine, without internet access, then there should be some way to place a downloaded version of the cache onto the virtual machine, and tell the nex-env to search for them there, by editing the nix.conf file?
<srhb>
noob_: Are you asking how to mirror an entire cache?
<Myrl-saki>
Since seems like my old installation is actually correct.
<srhb>
noob_: That store path is unique for that package and all its dependencies combined.
<srhb>
noob_: So, once we've evaluated the path, we can go ask a substituter: Do you have this path?
knupfer has joined #nixos
<exarkun>
So, how is it that there are packages that fail to build in nixpkgs master@HEAD?
<thblt>
I guess I just need to create a derivation for the layout...
<srhb>
noob_: That's more or less all there is to it. So, in order to determine the paths you want, just evaluate every expression that you care about.
<thePirateKing>
I am sure youve been there though lol
shibboleth has quit [Quit: shibboleth]
<thblt>
But how do I install this file? That's the part where I feel stupid
<infinisil>
thblt: Read the section?
<noob_>
srhb: if the "nix-env -f githubRepoURL.tar.gz -iA command" is called without the nix.conf file having the public key and substituter url, then it tries to build the binary, but if the nix.conf exists and is correct, it just links the binary. so how is this supposed to work offline?
<noob_>
is there a way to get the binaries it retrives and somehow have the "nix-env ..." command above use these instead of trying to fetch them from online?
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
<srhb>
noob_: It's not just the binaries, it's also the source of every dependency.
<noob_>
oh...
<srhb>
The short answer is: No, there's no way to do this _completely_ offline short of traversing the dependency and realizing every source as well..
<srhb>
s/dependency/dependency tree
<noob_>
isnt the source whats in the github repo?
<noob_>
so that would just need to be stored locally also
<srhb>
noob_: First, am I assuming you're referring to reflex-platform or something like that?
<noob_>
idk what that is
<infinisil>
thblt: Oh you mean actually install the files in xkb. Yeah by default that's not possible, xkb's fault I'm pretty sure. But there is #47764 that does it in a hacky way. That PR is unfortunately not ready
<noob_>
its just a part of the instructions. which are, place that in a nix.conf file in the right place, and run the "nix-env ... " command above
<srhb>
noob_: Note that it's referring to the alacritty sources on github -- in a different repo
<srhb>
noob_: Meaning, a project _may_ have only local sources, but it's likely to have non-local ones too.
<noob_>
oh
<srhb>
noob_: Transitively at least, it gets quite deep, depending at build time on, say, the sources for gcc or libc or curl or openssl or ...
<ee1943>
I'm trying to build a program that depends on the LLVM and clang libraries. My derivation.nix has the following line buildInputs = [ llvm_8 clang_8 ]. Unfortunately clang_8 fails when it builds one of its headers and so I can't get a build working.
<noob_>
so it wouldnt be as easy as just downloading one github repo and the corresponding binary cache, because it could have more source
<srhb>
noob_: We can probe any _specific_ attribute for its build time deps, but we can't easily say: Given a cache, give me all the possible packages. Even if you had that big a disk, it's usually infeasible.
<ee1943>
Is there any way to override the clang_8 cmake targets to avoid building that target? I don't need that specific executable it's trying to build
<Myrl-saki>
BOOT_DEV="/dev/disk/by-uuid/$(blkid -o value -s UUID "${DISK_IMAGE}1")"
<srhb>
noob_: I mean, if you had access to download a binary cache (eg. list every path that exists in it, then get it...) that would be sort of feasible, but it's a little backwards, and again, the binary caches are usually huge.
<noob_>
srhb: so is there no way to store something which can be transferred onto a clean virtual machine running linux?
<srhb>
noob_: Yeah, nix-copy-closure as listed above, but that doesn't really seem to link with your questions above..
<srhb>
noob_: Rather than speaking in hypotheticals, perhaps we should deal with your actual issue at hand?
<srhb>
If it's something like: "How do I transfer the specific build time deps of this one (or these 10) package(s)" we have great tools to do just that.
<noob_>
srhb: ok, it might help to explain the reason for wanting to do this. it might not be that the internet is not available, but it might just be that the user takes their cache down... so that something which builds today might not build next year, and its just that instead of storing the whole virtual machine with the build on, it would be good to be able to store something much smaller, like, just the binary cach and source code
<srhb>
noob_: That should never happen. If a cache is not available, we can just build from source
<srhb>
noob_: That's how the caches came into existence after all.
<noob_>
have tried, and the build fails
<srhb>
noob_: Then a _source_ has disappeared from the internet
<vdemeester>
gchristensen: ^^ is this the correct way (to update containerd on 18.09 with security fixes)
<fendor>
git blame is slow
<noob__>
im confused, part of what your saying is "this is not possible" and part is "we have tools to do this"
trh_ has joined #nixos
goibhniu2 has joined #nixos
ixxie has joined #nixos
noob_ has quit [Ping timeout: 256 seconds]
<srhb>
noob_: OK, it's not impossible, but it's certainly not common practice to publish binary caches that rely on source artifacts that are not available on the internet for everyone to use.
goibhniu1 has quit [Ping timeout: 245 seconds]
<srhb>
noob__: And binary caches are usually generated in exactly the same way you'd build a package from scratch if there's no binary cache available. It's the exact same process.
<noob__>
its an active project and its constantly changing and is likely to break or be unstanble at various times
<trh_>
hey folks! i'm attempting to write a default.nix file for a development environment for a PureScript library. I wrote the expression for the extra package I need right in the file and then used it as an overlay to nixpkgs. However, I'm getting an error about not being able to convert a function to JSON and I'm not able to find much more information about it.
<noob__>
thats the reason for wanting to make sure that whatever these build instructions produce today can be reproduced at another time
<noob__>
so, to remove all of the internet cache dependencies
<srhb>
noob__: Right. For the exact command you pasted, this is doable
fusion809 has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @vcunat pushed to release-19.03 « Merge #59003: pidgin: fix gst plugin path for audio/video »: https://git.io/fjtbX
fusion809 has joined #nixos
<trh_>
if I uncomment the line in `pkgs` that's `nixpkgs.mkdocs` then it works just fine
<trh_>
so it may be an issue with how I've put the expression for `mkdocs` in there
<srhb>
noob__: The process is to evaluate the build time dependencies of the expression and copy it to some other machine.
<noob__>
srhb: would it be worth opening a /query to avoid cluttering the chat with these instructions or might it be helpful for others? not sure if this is a really simple request or not
<srhb>
Sure, we can do that.
domogled has quit [Client Quit]
<noob__>
thanks
<{^_^}>
[nixpkgs] @matthewbauer opened pull request #59225 → cc-wrapper: make machine configuration configurable → https://git.io/fjtbM
<infinisil>
fendor: Chances are they are actually broken though, in which case you'd have to fix them. Or use another nixpkgs version in which it's not broken
leat1 has quit [Remote host closed the connection]
<_Geeko_>
clever i did not know well which things i need to put in the nix config
<infinisil>
Oh I thought you talked about a nixpkgs haskell package
<azazel>
nh2[m]: Thanks! But to better understand, the version of glusterfs is still 4.0 because later releases had issues on NixOS ? It's so difficult to test new releases?
<fendor>
infinisil, i was, wrong error message, tried to build it from source now
<nh2[m]>
azazel: oh no, I just switched to Ceph, so I haven't been working on GlusterFS in NixOS at all. I just wanted to get that last PR in to not throw away previous work. The question the previous commenter asked was why I don't *directly* go to a newer version
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
<azazel>
nh2[m]: ah, so you switched to ceph, so that's the reason why the package wasn't updated... and you find it to be more resource hungry than gluster?
<infinisil>
fendor: Ah yeah, you can try building it by changing `import (import ./fetch-nixpkgs.nix) {}` to `import (import ./fetch-nixpkgs.nix) { config.allowBroken = true; }`
<nh2[m]>
azazel: no, I'm happy with the resource usage of Ceph
<fendor>
infinisil, ok thanks!
<nh2[m]>
azazel: in general I found it a lot less buggy than gluster (which is why we switched)
pie___ has joined #nixos
pie__ has quit [Remote host closed the connection]
<azazel>
nh2[m]: it's incredible that they got to version 4 (or 6 for what matters) with a service without configuration files
<{^_^}>
[nixpkgs] @xeji pushed commit from @r-ryantm to master « i2pd: 2.23.0 -> 2.24.0 (#58696) »: https://git.io/fjtNM
crmlt has joined #nixos
<jonreeve[m]>
I guess, put another way, how can I check my home.nix file for errors? I have a feeling there's an error in there somewhere, but nix isn't telling me where, and I can't get a trace
<jonreeve[m]>
The error is `error: value is an integer while a set was expected, at /nix/store/7qvd11gb3nvgqdvgs1apv7hjzp5a4laz-nixos-19.09pre174426.acbdaa569f4/nixos/lib/attrsets.nix:234:39`
<jonreeve[m]>
But I don't see anything meaningful at that file location.
<colemickens>
--show-trace didn't help jonreeve[m] ?
<colemickens>
it usually dumps on me, but I think it usually helps me narrow it done. Otherwise I aggressively comment out parts of the suspected offender and uncomment until I find the error.
<jonreeve[m]>
colemickens: Nope. `--show-trace` doesn't seem to have any effect. It just outputs the same thing as without it.
Qubasa has joined #nixos
<colemickens>
aminechikhaoui: ah, the GH issues are much newer though, strange...
<colemickens>
aminechikhaoui: have you tried issuing hte sysctl?
<aminechikhaoui>
yeah but it seems that config doesn't even exist, so I guess the grubby stuff will just fail in the same way
<aminechikhaoui>
unless it does some magic to modify the kernel
<colemickens>
"that config doesn't exist" ?
ixxie has joined #nixos
<colemickens>
oh I see, grubby is some sort of grub config tool in centos maybe?
jabranham has joined #nixos
<colemickens>
Yes, that's exactly what the grubby stuff does, AFAICT.
<aminechikhaoui>
probably, I meant that e.g `sysctl -a | grep user_namespace.enable` doesn't return anything
leat1 has quit [Remote host closed the connection]
<colemickens>
but I'm not sure, and not a centos user!
leat1 has joined #nixos
<fendor>
how does nix decide which packages to use? just the ones that are at haskellPackages? depending on the compiler, of course
<colemickens>
I think you need to enable "user_namespaces" using the kernel option.
<infinisil>
fendor: Nix just uses what you specifyp
<colemickens>
then at runtime, you can enable "unprivileged_userns_clone" on top of that?
<infinisil>
fendor: It's a declarative language after all
<fendor>
infinisil, but it doesnt seem to use the same build plan as cabal?
<infinisil>
fendor: I need more context to be able to give any meaningful help
<fendor>
so, actually, i probably have to declare every dependecy by hand, since cabal2nix does not fix the versions?
<infinisil>
fendor: You need to call cabal2nix's result with haskellPackages.callPackage
<fendor>
i thought, it would be a good to learn stuff
<fendor>
infinisil, yes, but then it uses old packages, sometimes too new packages etc..
<infinisil>
fendor: But hie-nix isn't a haskell project
<colemickens>
aminechikhaoui: actually, the comment you linked (runc) to even says something similar, though it has a different grubby option name
<fendor>
infinisil, yeah, right, the real project i am trying to build is haskell-ide-engine
<infinisil>
fendor: a cabal2nix version of hie was once in nixpkgs, but soon dropped due to the high maintenance of it
<infinisil>
fendor: Because hie uses lots of custom dependencies, and in nixpkgs we only have one version of everything, so a lot of overrides were needed to get it to work
<aminechikhaoui>
colemickens yeah I guess I'll try some of those suggestions and see if something works
<infinisil>
fendor: If you want to learn about cabal2nix, I recommend any other project than hie :)
<aminechikhaoui>
bummer that there is no single official documentation on how to do that in Centos
<colemickens>
yeah, I'd expect there to be a Wiki page for something like that, especially given that features usage in containers/k8s/sandboxing/etc. :(
<{^_^}>
[nixpkgs] @exarkun opened pull request #59231 → fix build on aarch64 → https://git.io/fjtNN
<infinisil>
(as you can see, that was actually the motivation for domenkozar to make hie-nix :))
CharlesHD has joined #nixos
<CharlesHD>
Hi there
<fendor>
infinisil, yeah and I wanted to prefer cabal2nix to stack2nix, but apparently, this doesnt exaclty work.
<CharlesHD>
When I try to upgrade nixos to 19.09 from 18.09 I loose my wifi.
<infinisil>
fendor: It would still be possible, but would take some work to get the overrides right
crmlt has quit [Ping timeout: 255 seconds]
dsiypl4__ has quit [Read error: Connection reset by peer]
<infinisil>
stack2nix has the advantage that you don't need to fiddle with overrides, but the tradeoff is that you don't have nixpkgs' caches
trh_ has joined #nixos
drager has joined #nixos
dsiypl4__ has joined #nixos
<infinisil>
(and can't share stuff with nixpkgs in general)
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
<fendor>
well, then ill give that up
crmlt has joined #nixos
<fendor>
thanks!
<CharlesHD>
I'm not sure if it is nm not recognizing my card, or drivers that do not install, etc. Any hints on how I should proceed ?
<drager>
I'm having some trouble with lags in videos in fullscreen in firefox. I suppose this has todo with my graphics drivers. I tried setting xserver.videoDrivers = [ "nvidia" ] but then the displayManager wouldnt start...
<infinisil>
CharlesHD: How do you specify wifi in your configuration.nix?
random_yanek has quit [Ping timeout: 250 seconds]
ryantrinkle has joined #nixos
<drager>
I also tried hardware.nvidiaOptimusDisable = true etc
<ryantrinkle>
i'm having trouble with locale inside a buildFHSUserEnv; has anyone gotten en_US.UTF-8 fully working in there?
<ryantrinkle>
running `locale` fails
<__monty__>
drager: Is hardware acceleration enabled in about:config?
Ralith_ has quit [Ping timeout: 245 seconds]
<CharlesHD>
infinisil: I declare my device in networking.wlanInterfaces { "wlan-station0" = { device = "wlo1" ;};
<wolfman154>
infinisil: 14:05 wolfman154: Do I need to set my custom keyboard layout in home-manager instead of configuration.nix in orders for it to work?
<wolfman154>
I’m confused?
orivej has joined #nixos
<infinisil>
Haven't I helped you with that weeks ago already? I thought you figured it out back then
random_yanek has joined #nixos
random_yanek has quit [Max SendQ exceeded]
<wolfman154>
infinisil: I got it to work in the global configuration.nix after removing gdm and gnome3, but after I set up xmonad in home-manager my custom keyboard layout stopped booting
random_yanek has joined #nixos
<wolfman154>
infinisil: do you mind me sending you the paste bin links to my configuration files?
<enteee>
I am struggling to get displaylink, i3 with 2 monitors working. The monitors turn on but the picture I see is a non refreshing stripped version of the displayManager background. I tried already changing resolution / refresh rate but with no luck. Does somebody have experiece with this hacky topic?
palo has quit [Quit: WeeChat 2.2]
goibhniu2 has quit [Ping timeout: 246 seconds]
jabranham has quit [Ping timeout: 255 seconds]
goibhniu1 has joined #nixos
mthst has joined #nixos
slack1256 has quit [Remote host closed the connection]
Guest73949 has quit [Ping timeout: 245 seconds]
leat1 has quit [Remote host closed the connection]
<enteee>
globin: no, but i tried using compton but it didn't change anything
<globin>
enteee: I sometimes have that when I have compton running and have to kill it to make my setup use the correct displays
palo has joined #nixos
<enteee>
globin: hmm did you find out why compton breaks your setup?
<globin>
nope, haven't had the time yet
<enteee>
ok
Myhlamaeus has quit [Ping timeout: 250 seconds]
<kyren>
hey, I'm not totally sure this is a nixos issue, but I'm trying to get nextcloud running inside of sway, it works fine, but it doesn't remember the account / password that I give it
<kyren>
I believe this is related to bug #38266 maybe?
<{^_^}>
https://github.com/NixOS/nixpkgs/issues/38266 (by alexeymuranov, 1 year ago, open): owncloud-client and nextcloud-client prompt for password despite that the keyring is unlocked
winem__ has joined #nixos
winem_ has quit [Ping timeout: 264 seconds]
kleisli_ has quit [Ping timeout: 246 seconds]
<kyren>
I think it should be working, but clearly I don't understand enough about how keyrings on linux work, I'm trying to use gnome-keyring as my keyring manager so that nextcloud will remember its account, any help would be appreciated
wolfman154 has joined #nixos
<kyren>
I've created a login keychain in seahorse, and it is now unlocked on login (I changed security.pam.services.login.enableGnomeKeyring to true to make this work)
leat1 has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @timokau pushed commit from @r-ryantm to master « ocserv: 0.12.2 -> 0.12.3 (#58730) »: https://git.io/fjtx8
<azazel>
aminechikhaoui: have you tried with sysctl kernel.unprivileged_userns_clone=1 ?
leat1 has joined #nixos
Neo--- has quit [Ping timeout: 258 seconds]
<kyren>
I've also added gnome3.gnome-keyring to services.dbus.packages, and frustratingly nextcloud just acts as though there is no keyring at all
<azazel>
kyren: I just have my encryption key in GPG and Nextcloud asks me for its password when it starts
<azazel>
I don't think I have any gnome package (or kde) package installed
<kyren>
I'm not married to gnome-keyring, I'll take whatever works I guess with libsecret, how do you set that up exactly?
johanot has quit [Quit: WeeChat 2.4]
<azazel>
kyren: it's a commandline tool
znc-user has joined #nixos
<kyren>
azazel: yeah I'm vaguley familiar with gpg I just don't use it... you just, have a gpg keyring at all and it works? should I follow some basic guide on using gpg for the first time?
<azazel>
kyren: there are tons of tutorials online about how to create a gnupg key
<azazel>
you probably should re-login after having enabled that
<esmerelda>
And saw it was open source
Guest85 has joined #nixos
<esmerelda>
And so I went to look for it and it's been pulled from github
<esmerelda>
Arrggh
<azazel>
maybe it's on gitlab?
<wolfman154>
infinisil: it’s telling me undefined variable lib, I was was wondering if you could paste bin the syntax, I’m not sure why it’s not working?
Guest85 has quit [Client Quit]
<esmerelda>
Unlikely, I think Azazel. The person still has 150 other projects on github
Guest85 has joined #nixos
__monty__ has quit [Quit: leaving]
Guest85 is now known as jurpeedurp
knupfer has quit [Ping timeout: 240 seconds]
<esmerelda>
The interesting thing to me though was that they mentioned it being extremely tricky to set up sandboxed nix builds within Docker
<manveru>
but i had to abandon this approach and is also the reason i'm not working on scylla anymore
<esmerelda>
Oh bummer!
wolfman154 has joined #nixos
<esmerelda>
Does Linux not do nested unprivileged cgroups?
<azazel>
esmerelda: you can try gitlab's runner and its bash executor
<manveru>
my original implementation used remote builds from the docker container to ec2, but that was against company policy, so i was left without any viable workers :(
<wolfman154>
infinisil: I’m still new to nixos, that’s why I’m asking so many questions
<etu>
,locate epstopdf
<{^_^}>
Found in packages: tetex, vimPlugins.vimtex, texlive.combined.scheme-medium
ixxie has quit [Ping timeout: 255 seconds]
<esmerelda>
Ahh makes sense. Hm, currently I'm trying to set up everything on GKE, which I assume lets me use privileged containers
<esmerelda>
Alternatively, though inadvisably, I could spin containers in the build pod in real-time, and simply use the container itself for sandboxing
thePirateKing has quit [Quit: Leaving]
nuh^ has quit [Ping timeout: 246 seconds]
nuh^ has joined #nixos
nuh^ has quit [Max SendQ exceeded]
nuh^ has joined #nixos
nuh^ has quit [Max SendQ exceeded]
nuh^ has joined #nixos
nuh^ has quit [Max SendQ exceeded]
<wolfman154>
Try adding `systemd.user.services.setxkbmap.Install.WantedBy = lib.mkForce []` in your home-manager configuration
nuh^ has joined #nixos
nuh^ has quit [Max SendQ exceeded]
nuh^ has joined #nixos
nuh^ has quit [Max SendQ exceeded]
sinner has joined #nixos
OmnipotentEntity has quit [Remote host closed the connection]
leat1 has quit [Remote host closed the connection]
nuh^ has joined #nixos
<wolfman154>
It’s telling me undefined variable lib, why is that?
<esmerelda>
Although I guess at that point, I could also just ditch GKE and maintain my own k8s, then run build on the VM root
drakonis has joined #nixos
rcshm has quit []
kumikumi has quit [Ping timeout: 250 seconds]
<kyren>
azazel: hey, I'm sorry for asking stupid questions here I'm very unfamiliar with gpg.. do I need to generate a key or just enable gpg-agent or what do I need to do here, I'm trying to look at gpg guides but I don't know what specifically I need to do to let libsecret store passwords using gpg-agent
<esmerelda>
wolfman154: do you have lib in the top { ... }?
winem__ has quit [Ping timeout: 250 seconds]
CharlesHD has joined #nixos
<CharlesHD>
re
Makaveli7 has quit [Quit: WeeChat 2.4]
<CharlesHD>
Is there some i3 users around here ?
<azazel>
kyren: I just know that I've enabled the agent and I have a a key generated. Now when the nextcloud client starts a dialog pops up asking for the password of that key...
<CharlesHD>
I have the following issue : i3.configFile option is bypassed in nixos
Guest79538 has quit [Ping timeout: 246 seconds]
<azazel>
CharlesHD: what do u mean with `bypassed`?
<CharlesHD>
i've set it up to ~/.config/i3/config but when i launch i3 --moreversion it tells me that it loads a different configuration file from the /nix/store
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<wolfman154>
esmerelda: in home-manager I just have { config, pkgs, ... }: Up top
esmerelda has quit [Ping timeout: 240 seconds]
<wolfman154>
esmerelda: so I need to add lib here?
<azazel>
oh... Mine is correct, I'm checking my config
<CharlesHD>
this configuration file indeed correspond to an old version of the configuration file I have, but the new additions I did to ~/.config/i3/config are not transmitted to the /nix/store version
<CharlesHD>
seems weird that nix stores the i3 configuration file.
fendor has joined #nixos
kumikumi has joined #nixos
T_S_ has joined #nixos
<kyren>
azazel: okay, I've tried this and no luck, I don't know how libsecret works internally and I'm not finding any good hints in the nextcloud logs either, oh well thank you anyway
<MiloIgnis[m]>
<andi- "MiloIgnis: if you use the packag"> Tried it out! It's a little odd - it seems to start up using Wayland as the backend, but it still causes XWayland to spawn. Unlike with regular firefox though, if I then kill XWayland, firefox-wayland continues to work fine. Also, the text of the firefox UI is tiny and different, though I imagine that's not your fault (I've got a high dpi screen)
<wolfman154>
Is it just a place holder?
<ee1943>
How long does it typically take to get a first post approved for Discourse?
<ee1943>
Got stuck in the spam filter :(
<azazel>
CharlesHD: mine is reading /home/azazel/.config/i3/config and I don't have anything in config more than the enable
<kyren>
azazel: do you happen to use a login manager? If so I bet that has something to do with it
leat1 has quit [Remote host closed the connection]
<wolfman154>
15:14 wolfman154: { config, pkgs, ... }: What does the ... mean?
<azazel>
kyren: yes, I'm using lightdm
<wolfman154>
15:15 wolfman154: Is it just a place holder?
leat1 has joined #nixos
<kyren>
azazel: okay, I suspect that's it, I just need to find a login manager that works with sway then
<CharlesHD>
azazel: tried to remove the configFile from my configuration.nix but it still read in the nix/store :-/
<T_S_>
does anyone use the Agda and or AgdaStdlib packages on nixos? I am trying to get a hello world program running but it complains about locating the standard library.
<manveru>
wolfman154: it means you can pass more arguments in the set, but they're ignored
<manveru>
wolfman154: try `({a, ...}: a) {a = 1; b = 2;}` without the ... :)
dnlkrgr has quit [Ping timeout: 255 seconds]
<infinisil>
wolfman154: Yeah just add lib to the list there
<wolfman154>
manveru: would { config, pkgs, lib, ...} Work, I need to add lib
<manveru>
yeah
<wolfman154>
manveru++
<{^_^}>
manveru's karma got increased to 10
<wolfman154>
infinisil++
<{^_^}>
infinisil's karma got increased to 81
jabranham has joined #nixos
<wolfman154>
Thanks, a little frustrated today :(
<wolfman154>
I’m hanging in there though
<wolfman154>
:)
<azazel>
kyren: try loading the agent with "gpgconf --kill gpg-agent && gpgconf --launch gpg-agent" and star the nextcloud-client from that terminal
<kyren>
sigh, now the question is: is there a login manager that works with sway :D
<azazel>
kyren: ah, no, I can answer that
<kyren>
azazel: oh, well darn!
kleisli has joined #nixos
<kyren>
azazel: wait just to make sure, you can answer that question and the answer is in the negative: there is no login manager that works with sway, correct?
<azazel>
kyren: it's one of the reason I'm still using i3, and the other being that I have yet to understand if X11 apps that make use of the tray area do work in sway
<azazel>
kyren: the last time I searched, the answer was negative, but I can be wrong
<kyren>
azazel: I believe they *don't* and only ones that use some... newer tray protocol independent of X11 do, but they work worse
<kyren>
azazel: I currently have a wlroots patch in my configuration to make steam clicking work
<kyren>
*clicking things in steam popups
<andi->
MiloIgnis[m]: yeah I see that.. It just seems like the default font there is way different. Not sure how much love that part of the UI receives. Do you know how it looks like on e.g. fedora?
CharlesHD has quit [Ping timeout: 256 seconds]
leat1 has quit [Remote host closed the connection]
<MiloIgnis[m]>
andi-: I don't! Not the biggest deal, I know firefox on wayland is still a bit fresh. Notifications come up as blank full windows under sway too - not sure if that's a firefox problem or a sway problem
<MiloIgnis[m]>
Any idea about why XWayland is being started needlessly?
<andi->
The matrix quoting is really a bit annoying..
fendor has quit [Ping timeout: 245 seconds]
Thra11 has quit [Ping timeout: 250 seconds]
Akii has joined #nixos
slack1256 has joined #nixos
<jurpeedurp>
Hi everyone, my build for clang_8 on mac os is failing with fatal error: 'xpc/xpc.h' file not found. Any idea what could be going wrong?
<MiloIgnis[m]>
andi-: My bad, I didn't realize!
<andi->
No biggie :)
<jurpeedurp>
I wasn't able to find a relevant package in the repo, so I don't think its a missing dependency.
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
ryantrinkle has quit [Ping timeout: 245 seconds]
<wolfman154>
infinisil: I added that setting too but the keyboard layout still has not changed, maybe I made a mistake somewhere? Have you looked at the 4 links in the #home-manager channel?
Ralith_ has joined #nixos
Glider_IRC_ has joined #nixos
znc-user has quit [Remote host closed the connection]
kreisys has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Glider_IRC__ has quit [Ping timeout: 245 seconds]
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
arjen-jonathan has quit [Ping timeout: 255 seconds]
<infinisil>
wolfman154: I have not, but home-manager#543 is related to what you have, basically you want to turn off the setxkbmap service, because that one overrides your systems keyboard config
<jurpeedurp>
I'm running on mac slack1256, would that command still work or is there a different one I should be using?
<slack1256>
-i install by the name of the package "clang", but -A says it look on tree data structure that is nixpkgs for the node "clang_8"
rjmill has quit [Ping timeout: 245 seconds]
tmaekawa has joined #nixos
alexozer has quit [Quit: WeeChat 2.4]
reallyme1orable has quit [Quit: Lost terminal]
alexozer has joined #nixos
<wolfman154>
slack1256: when I type firefox in terminator in xmonad, it splits half the screen for firefox and the other half for terminator, how do I set it so applications automatically open into full screen?
<jurpeedurp>
Oh, that makes sense! Is there also a way to disable one of the binaries it fails to build?
<slack1256>
jurpeedurp: It should work just fine on mac or nixos or any distro with nix
<trh_>
does anyone know an example of a `default.nix` file that contains an overlay within the file (all within 1 file, not several)?
tmaekawa has quit [Client Quit]
<jurpeedurp>
I get the following error by running that command: "attribute 'nixos' in selection path 'nixos.clang_8nix-env' not found"
<slack1256>
wolfman154: that is more a pure xmonad question than a xmonad on nixos question. There are lot of documentation material on the web to set up full-screen, mod-space should iterate between layouts
<jurpeedurp>
do you know why that may be? :/
<wolfman154>
slack1256: do you know a link that you can paste bin?
<jurpeedurp>
ah, switching it to nixpkgs it seems to have worked!
<slack1256>
`nix-env -iA nixos.clang_8"` works on my machine
Jonathan__ has joined #nixos
<slack1256>
jurpeedurp: oh maybe that it
swerlk has quit [Ping timeout: 245 seconds]
jabranham has quit [Remote host closed the connection]
<slack1256>
wolfman154: I don't have one at hand, but google your way. Also #haskell and #xmonad are more specific channels for this
leat1 has quit [Remote host closed the connection]
<jurpeedurp>
thanks so much slack1256! Especially for taking the time to explain rather than just giving me an answer. I really appreciate it.
sigmundv_ has joined #nixos
leat1 has joined #nixos
<slack1256>
jurpeedurp: no probs :-)
<slack1256>
nixos is kind of weird, but lots of stuff have an explanation behind
akapav_ has quit [Ping timeout: 252 seconds]
kreisys has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
akapav has joined #nixos
alexozer has quit [Quit: WeeChat 2.4]
linarcx has quit [Quit: WeeChat 2.4]
wolfman154 has quit [Remote host closed the connection]
swerlk has joined #nixos
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
jurpeedurp has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
leat1 has quit [Remote host closed the connection]
znc-user has joined #nixos
leat1 has joined #nixos
trh_ has quit [Ping timeout: 245 seconds]
<nh2[m]>
azazel: yes, no config files and doing everything with interactive commands is a bad idea. Did you see my pains I went through to bootstrap it automatically because of that? https://github.com/nh2/nixops-gluster-example
fendor has joined #nixos
fendor has quit [Client Quit]
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
dbmikus__ has joined #nixos
dbmikus_ has quit [Ping timeout: 245 seconds]
Guest28728 has quit [Remote host closed the connection]
<azazel>
nh2[m]: yes, I had a look at that! Actually, I'm testing gluster out for using it in kubernetes deployment, but I'll test ceph as well... It's nice that i doesn't need block devices, but the management is terrible. I've compiled heketi (github.com/heketi/heketi) which is sort of a "gluster manager", but I think I'll try ceph before heketi...
_Geeko_ has joined #nixos
<_Geeko_>
hi guys any idea's what i can do to try to get rid of stutter while playing csgo?
<azazel>
nh2[m]: It would be nice if you had a ceph configuration example as well ;-)
<_Geeko_>
seems only on nixos i got issues with that atm
<azazel>
csgo?
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<_Geeko_>
counter strike global offensive
<azazel>
I'm not really in to games since my C64 era
<nh2[m]>
azazel: it doesn't include any nixops bits, only the NixOS service, but that might already be useful to you
jasongrossman has joined #nixos
<nh2[m]>
we certainly want to upstream the stuff at some point soon and perhaps I can make also another nixops example deployment with it
leat1 has quit [Remote host closed the connection]
leat1 has joined #nixos
leat1 has quit [Client Quit]
Soo_Slow has quit [Remote host closed the connection]
Soo_Slow has joined #nixos
<_Geeko_>
any of you as like a nix config example for dual boot with linux distro?
<azazel>
nh2[m]: nice, thanks! It will take some time to dig it out in it's current (in flux) state mmmm... so you are using the 13.x version but the PR has been more than half a year in the working...
<Jonathan__>
Hi! I am brand new to Nix and NixOS. I am thrilled with the idea of Nix and am planning to switch to it soon from Arch.
<azazel>
Jonathan_: welcome
<Jonathan__>
I could use some help getting wifi working though.
<Jonathan__>
I installed using a wired connection...but...wifi is nice...
<nh2[m]>
azazel: you mean the PR linked at the very bottom of my issue? I'm using the version I put into that branch, in production for almost 2 years now
<azazel>
_Geeko_: to boot another os, I usually switch the boot loader to GRUB and then you can have a try with the os-prober (maybe you are lucky and it sees the other linux) or you can manually use https://nixos.org/nixos/options.html#boot.loader.grub.extraentries to add an entry that points to the other's linux grub.cfg or directly to a kernel entry from /boot/grub/grub.cfg on the other system
<Jonathan___>
So, probably a dumb question. But is the proper way to change the config to edit `/etc/nixos/configuration.nix` and then run that reload command?
<_Geeko_>
thats how i do it as well :D
<gchristensen>
yep
<azazel>
mh2[m]: ah, good to know!
<_Geeko_>
azazel just need this command? boot.loader.grub.useOSProber = true;
<Lisanna>
anyone have a minimal nixos config to make the installed emacs or vim or whatever actually useful for editing git commits
<gchristensen>
what does that mean?
<Lisanna>
line length help
<Lisanna>
*maybe* auto text reflow to correspond with those line lengths but I'm not being picky
<azazel>
_Geeko_: yes, but It could not work because both the linux installation install the bootloader on the same block device... beware! If you're going to install another linux now, in another partition, you should copy the final grub.cfg on an usb-stick so that you have it and you may change the other's linux grub config to boot nixos instead...
<azazel>
_Geeko_: yes, but It could not work because both the linux installation install the bootloader on the same block device... beware! If you're going to install another linux now, in another partition, you should copy the final grub.cfg on an usb-stick so that you have it and you may change the other's linux grub config to boot nixos instead...
<azazel>
eh, nixos boot loader has overwritten the other, I suppose
dhkl has joined #nixos
jluttine has joined #nixos
<aswanson>
does anyone have a working RStudio/kera+tensorflow setup? I don't know how to get the python packages into rstudio's environment the NixOS way
<azazel>
_Geeko_: if you know in which partition the other linux's /boot/grub/grub.cfg resides, you can manually add an entry to nixos' grub
<_Geeko_>
its on /dev/sda3 my other linux
Soo_Slow has quit [Remote host closed the connection]
<Jonathan___>
Eh, I rebooted my machine after changing the config. And the boot loading is asking whether to boot to `NixOS (Generation 1)` or gen 2`.
<gchristensen>
great!
<_Geeko_>
thats good ye
<_Geeko_>
means generation 2 is your last build
<gchristensen>
if you screw something up real bad, reboot and pick the second newest one and you'll be sorted :)
<Jonathan___>
That is so awesome...!
<Jonathan___>
Why is Gen 2 on the bottom though?
<Jonathan___>
if it is default?
<_Geeko_>
its by default selected
<_Geeko_>
anyway
<Jonathan___>
Oh, cool.
<_Geeko_>
your last build
<Jonathan___>
How many generations does is keep?
<gchristensen>
all of the ones you haven't deleted ;)
<gchristensen>
some of my systems have 1,000 generations, some just a few
<Jonathan___>
(:
<Jonathan___>
lol
<_Geeko_>
gchristensen
<_Geeko_>
can you help me out a bit with dual boot?
uranther has quit [Quit: Connection closed for inactivity]
<Jonathan___>
Yay, wifi is working, you guys are awesome.
mbrgm_ has joined #nixos
<azazel>
_Geeko_: the key thing to set is the "search parameter", that probably in your case has to be set to "--hint hd0,msdos3" (if your hd has a classic partition types)
<gchristensen>
Jonathan___: great to hear, congratulations!
Erasmus has left #nixos ["END-AUMF"]
mbrgm has quit [Ping timeout: 252 seconds]
mbrgm_ is now known as mbrgm
dhkl has quit [Ping timeout: 255 seconds]
<azazel>
_Geeko_: the fs label has to be set too, see the note at the end of the examples
<hlolli>
I'm trying to use clang, but I get this error. I'm not a c++ dev so not sure what this means. gcc-7.4.0/include/c++/7.4.0/stdlib.h:38:12: error: no member named 'abort' in namespace 'std' using std::abort;
<_Geeko_>
azazel first need to figure out that cfg location
<azazel>
_Geeko_: you said earlier that it's on sda3