<freeman42x[m]>
I don't understand installation procedure for taffybar under NixOS: https://github.com/taffybar/taffybar or rather, I can't find it anywhere in there
markus1189 has joined #nixos
<keithy[m]>
im getting path '/nix/store/j08r79pwmpj6cx3czm4b9i5jsw66ln9x-dhall-json-1.6.3' does not exist and cannot be created errors a lot should I sudo?
<elvishjerricco>
keithy[m]: Still, garbage collection never hurts
<xcmw>
What do errors like this mean? When I use sudo it works but I feel like I should not need to. error: opening lock file '/nix/store/287bi0aiz96rj5s77zvfkknfwb13a2dx-MisoTest.lock': Permission denied
<keithy[m]>
ok
<keithy[m]>
and rebuild again?
jbrock has quit [Quit: jbrock]
<elvishjerricco>
xcmw: That means you installed Nix with a different user than yourself and didn't make it a multi-user install
jbrock has joined #nixos
<xcmw>
I only have one user
<elvishjerricco>
xcmw: Then you probably installed it as root
<xcmw>
I feel like I didn't use sudo before and that this a something new
<xcmw>
Could be wrong though
<elvishjerricco>
xcmw: Which you should only do if you're making a multi-user install
<elvishjerricco>
xcmw: You may have run just one command long after installing with root or something
<keithy[m]>
the pinebookpro stability might be an issue
<keithy[m]>
its frozen on my twice
<keithy[m]>
* its frozen on me twice, attempting with only nix-env -i running no browsers (i3 manjaro)
pie_[bnc] has quit [Changing host]
pie_[bnc] has joined #nixos
turona has quit [Ping timeout: 265 seconds]
<{^_^}>
[nixpkgs] @offlinehacker merged pull request #86143 → Add `jq` into path for `nix-prefetch-docker` → https://git.io/JfqSr
<{^_^}>
[nixpkgs] @offlinehacker pushed 2 commits to master: https://git.io/Jfqhj
turona has joined #nixos
proofofkeags has joined #nixos
<{^_^}>
[nixpkgs] @worldofpeace opened pull request #86163 → Gnome and Pantheon: install nixos wallpapers → https://git.io/Jfqjq
jkachmar has quit [Quit: Connection closed for inactivity]
proofofkeags has quit [Ping timeout: 260 seconds]
jbrock has quit [Quit: jbrock]
jbrock has joined #nixos
<aleph->
Well seems I'll be patching the prometheus module. Need to add openstack support to the scrape config.
jbrock has quit [Client Quit]
jbrock has joined #nixos
waleee-cl has quit [Quit: Connection closed for inactivity]
<gchristensen>
is there a way to do this: config = lib.mkMerge (mapAttrsToList (name: cfg: { ... })); in a NixOS module? when I do this, even with an empty function in the mapAttrsToList, I get an infinite recursion
<elvishjerricco>
gchristensen: You can't write modules that are dynamic in the top level options they generate. That makes it possible for them to generate options that should have been included in the `mapAttrsToList` function; hence the infinite recursion
<gchristensen>
right. ack
<elvishjerricco>
top level options they provide definitions for* is more accurate wording
<gchristensen>
annoying, but ack :)
AluisioASG has quit [Quit: No Ping reply in 180 seconds.]
sarcasticadmin has quit [Ping timeout: 260 seconds]
<jasongrossman>
Hi!
<hyper_ch>
lately I'm having a weird problem with my Lenovo notebook: When I shut it down, it seems to work fine. At the end screen gets turned of but the power button led and others remain lit on... also the cpu (?) fan starts up and get noticeably loud and it just won't stop until I press the power button.
jasongro` has joined #nixos
jbrock has quit [Quit: jbrock]
slack1256 has quit [Ping timeout: 260 seconds]
jbrock has joined #nixos
jbrock has quit [Remote host closed the connection]
jbrock has joined #nixos
AluisioASG has quit [Read error: Connection reset by peer]
<hyper_ch>
pjt_014: how? it reaches power off state, screen goes out, there doesn't seem to be any noticeable action going on
<pjt_014>
It could still be in a state where dmesg can operate. Having something like dmesg -w > log.txt run while 'shutting down' would tell you if that's the case
<pjt_014>
because I can't tell exactly what state it's in by that description only
<zeta_0>
clever: i don't have either of those 2 options set so i guess i can set that picom option, does it have to be set in configuration.nix or can i set the option home.nix instead?
jbrock has quit [Quit: jbrock]
jbrock has joined #nixos
<clever>
zeta_0: then you need to figure out which file is setting compton
jbrock has quit [Client Quit]
jbrock has joined #nixos
chagra has joined #nixos
<zeta_0>
clever: it turns out i had the compton option set in home.nix, so i renamed it to picom, i don't remember why i set that up in the 1st place, so i'm going to need to look that picom stuff up
<{^_^}>
[nixpkgs] @jonringer pushed commit from @jerith666 to master « jid: 0.7.2 -> 0.7.6 »: https://git.io/Jfmfn
mir100 has quit [Ping timeout: 250 seconds]
chagra_ has quit [Ping timeout: 260 seconds]
<zeta_0>
cool, that change removed the trace warning
Jonathan21 has joined #nixos
endformationage has quit [Quit: WeeChat 2.6]
mir100 has joined #nixos
<Jonathan21>
Hey, brilliant Nix people. I am having a crash with Sway in the unstable branch, so I submitted an issue with Sway, but they say it is a NixOS issue.
<Jonathan21>
They made this statement, but is this actually true?! """On NixOS you'll have to recompile as soon as you upgrade one of the dependencies, since your libs go out of sync otherwise."""
<clever>
Jonathan21: anything in /nix/store must follow that rule
<clever>
Jonathan21: opengl behaves weirdly on nixos, and ignores the previous rules
<simpson>
The rest of what they said makes sense; they recommend a certain overlay.
<clever>
Jonathan21: you must get sway from the same version of nixpkgs as nixos, or things break
<Jonathan21>
Oh, because multiple versions of OpenGL does not work.
<simpson>
That looks like a funkily-linked libdrm. Wonder how that happened?
<clever>
simpson: the GL.h it was compiled from is out of sync with the libGL in /run/opengl-driver/
<clever>
because 2 nixpkgs versions got mixed up
<clever>
opengl is the only place where nix impurely mixes things up like that
peelz has joined #nixos
<simpson>
clever: That makes zero sense to me. libGL's ABI is fixed, and moreover the symbols that are messed up are from libdrm, one of the libraries underneath the GL stack. But whatever fixes it, I guess.
<Jonathan21>
Okay.
<clever>
simpson: but libGL may be pulling in an old libdrm via RPATH, which isnt in sync with the libdrm that sway was built against
<clever>
simpson: so it spreads like a virus, and all of GL's deps could be the "wrong" version for sway
<simpson>
clever: Sure. This is an abstraction violation for sway, but who knows whether it's their fault.
<clever>
id say its due to mixing nixpkgs revs, and it wouldnt happen if sway and nixos came from the same rev
<simpson>
clever: It *is* possible to just go ham and ioctl() the DRM node without libdrm, and further to open up an X socket without libxcb. All of these libraries, on Linux, are comfort and cushioning.
<simpson>
Ah, sure. Mixing multiple nixpkgs doesn't go well.
jbrock has quit [Quit: jbrock]
jbrock has joined #nixos
aminechikhaoui has quit [Quit: Ping timeout (120 seconds)]
jbrock has quit [Client Quit]
jbrock has joined #nixos
<zeta_0>
emacs is not recognizing the environment variables in the shellHook i set in shell.nix for my lorri setup, so i'm just going to put all the environment variables in the `home.sessionVariables` option in home.nix, i hope that does not cause problems down the road
kahiru has quit [Ping timeout: 250 seconds]
kahiru_ has joined #nixos
<zeta_0>
anyways, what does --pure mean in nix-shell --pure ? i've never used the --pure option before, so i'm wondering what are some cases where it needs to be used?
<clever>
zeta_0: it means env vars wont leak in from the shell above nix-shell
Jonathan21 has quit [Remote host closed the connection]
aminechikhaoui has joined #nixos
<zeta_0>
clever: thanks, that makes sense
<JameySharp[m]>
simpson: open an X socket without libxcb? then what did I write it for? 😅
<JameySharp[m]>
that was a funny comment to walk in on 😆
<{^_^}>
[nixpkgs] @lblasc opened pull request #86168 → sof-firmware: init at 1.4.2 → https://git.io/JfmJs
<simpson>
JameySharp[m]: I've been working on an XML parser in Monte so that I can connect to X from Monte without XCB. Yet another one of my many poor life decisions.
<colemickens>
Las, you can use an overlay for that, but it will trigger a rebuild of a lot of things.
hax404 has quit [Ping timeout: 272 seconds]
lord| has joined #nixos
hax404 has joined #nixos
proofofkeags has joined #nixos
<la-s>
I wasn't sure if that would work so I just copied the tree and changed mesa/default.nix, but it makes me recompile a bunct of things that are seemingly unrelated like libav
zupo_ has quit [Read error: Connection reset by peer]
<colemickens>
yeah, mesa is at the bottom of a very large inverted tree
<colemickens>
If you just need to build an additional driver there's a trick you can use.
<colemickens>
Otherwise I don't know of a good option in Nix.
<deuloma>
What is the way to pass a variable into nix-shell? I have a default.nix which uses rust unstable if a variable is set, for tests, I would like to override it to run test in both scenario
<deuloma>
so, run nix-shell as play but also run nix-shell with one variable overriten
<colemickens>
The AWS tab on the download page doesn't look right.
jgeerds has quit [Ping timeout: 265 seconds]
jbrock has quit [Quit: jbrock]
jbrock has joined #nixos
jbrock has quit [Client Quit]
jbrock has joined #nixos
hax404 has quit [Ping timeout: 240 seconds]
hax404 has joined #nixos
smatting_ has joined #nixos
<nschoe>
Hi everyone, for those who were here last night (for me), just know that... the GHC cross compilation termiinated during the night... and it was successful!
svends has quit [Ping timeout: 240 seconds]
civodul has joined #nixos
<nschoe>
Hum, hi again. Now I'm trying to copy the closure I've just cross-compiled on my rPi, but I can't do it. No matterif I use 'nix copy' (with or without --no-check-sigs, with 'ssh-ng://', with 'ssh://') or 'nix-copy-closure', I always get error: cannot add path '/nix/store/2gmj7d2ij7gyqk2mbs2czmii9pc8g81c-linux-headers-4.19.16' because it lacks a valid signature
<nschoe>
I've tried copy fro mlaptop to rPi with --to and from the rPi wih --from but same error.
<srk>
\o/
<la-s>
well you can try copying it manually
<nschoe>
la-s: well I don't know all dependencies.
<la-s>
nschoe: nix-store -qR
<nschoe>
la-s: thanks
<nschoe>
But that doesn't feel very nix-like ^^
chagra has quit [Ping timeout: 260 seconds]
<nschoe>
Also with nix copy isn't some thing "else' performed rather that just a copy?
<la-s>
well, it works lol; I've even used it before
<la-s>
I think it's pretty much just a literal copy
<srk>
nschoe: are you copying as root, right?
<srk>
on the rPi
<seku>
noob here, been playing around a bit with gchristensen's idea of deleting / on every boot. It worked, but of course it deleted my users as well, so i can no longer log in xD. Parsing his configuration file, i saw that he stores them in secrets... somewhere. Is there something like a tutorial for that? :)
knupfer1 has joined #nixos
<kahiru_>
I just defined my users in configuration.nix and put hashed passwords in there
<nschoe>
srk: no I was not.
<nschoe>
srk: but I've just tried now and it seems to work.
<seku>
kahiru_, thanks ill try that! just for my understanding, /etc/ gets wiped too, so configuration.nix as well? is it copied/rebuilt in the iniramfs or something like that?
<nschoe>
srk: anyway to make my normal user 'trusted'? should I?
<seku>
thanks, DigitalKiwi!
alexherbo23 has joined #nixos
<kahiru_>
seku: /etc gets indeed wiped. I put the configuration to a persistent mountpoint and then added something to configuration.nix to create symlinks in /etc/nixos pointing to the persistent location
<srk>
nschoe: I'm not sure if that would help, it depends on the signatures of the build products iirc
<xfix>
about tmpfs, you want to keep /etc/nixos somewhere safe probably if you want to be able to use nixos-rebuild
<xfix>
let me check my own configuration, because i did something similar in a virtual machine
<seku>
interesting, it sounds a bit more involved than i tought :)
<seku>
maybe i should just make /etc/nixos a persistent dataset as well?
<nschoe>
srk: yes but I can always use --no-check-sigs, no?
alp has quit [Ping timeout: 265 seconds]
<nschoe>
srk: because the issue here is that it "forces" me to log in as roo, which is not good. Or log in as root from teh rPi (with sudo -i) and use --from.
<srk>
nschoe: right! or setup proper signatures but that's probably overkill :)
orivej has joined #nixos
<kahiru_>
xfix: ah, so I went a bit longer way around it
<xfix>
this means on every boot, /etc/nixos becomes a symlink to /data/etc/nixos
<xfix>
of course, this depends on what the exact persistent storage path is
<nschoe>
srk: :/
alp has joined #nixos
<nschoe>
srk: okay, will keep copy from the Rpi (logged as normal user, elevated to roo twith sudo -i). Because signing derivations afte reach rebuild (I'm in development mode) will be a pain.
<nschoe>
Thanks for the help:)
<srk>
yw!
<seku>
xfix, thanks, that's just what i need :D
<xfix>
in my case, i made partitions for /nix, /data, /home and /boot
<nschoe>
And *it works*. Cross compiled to rPi 3, even for Haskell programs is not broken.
<nschoe>
I have cross-compiled a *haskell* program on the laptop, targetted for an rPi3, copied to closure and executed the thing on the rPi.
<nschoe>
(Sorry just needed to share the joy)
<manveru>
nschoe: congratulations :)
orivej has quit [Ping timeout: 260 seconds]
<Valodim>
woo :)
<nschoe>
Thanks for the support and help :)
<ikwildrpepper>
nschoe: awesome, congrats
<immae>
xfix: do you use tmpfs on root filesystem? I think someone was interested in that a few days ago, if you’re available to share some code
<xfix>
well, i experimented with that in a virtual machine
<nschoe>
(and if you guys wans to do this, make damn sure to have enough space on /tmp and limit --cores 2 -j 2 unless you have 16GB+ of RAM, because GHC is a greedy bastard to compile :D)
<xfix>
i don't do that on an actual computer... yet
<immae>
sure
<immae>
aanderse: ^
ris has quit [Ping timeout: 258 seconds]
<etu>
I've been using tmpfs for root for months and months
<xfix>
it's honestly rather straightforward, you just mount tmpfs under /mnt during installation
<immae>
(He might have found a solution inbetween though)
<etu>
and recently switched over my /home to tmpfs as well
<xfix>
nixos-generate-config will recognize that
<xfix>
of course, put a real mount for /nix, /boot and whatever you want really
<immae>
It’s not for me, someone who asked a few days ago about it :)
<xfix>
no need to back up /nix or /boot, but you probably want to back up mountpoints with actual state
<immae>
xfix: you mount tmpfs under /mnt and then it will be written as is in fstab is that it?
<xfix>
you are using NixOS, /etc/fstab is an implementation detail :)
<xfix>
but yes
<immae>
right, it’s easier than what I though indeed :)
<xfix>
what you probably are thinking of is hardware-config.nix
<xfix>
hardware-configuration.nix*
<immae>
yes
<xfix>
and nixos-generate-config can handle tmpfs
<immae>
yes, my question was more "you can mount a tmpfs as root early in the boot process" but I get your idea
<xfix>
yes
<immae>
Thanks xfix / etu
<xfix>
it makes sense to allow having /nix on a separate partition, so of course the boot process is prepared for that
<immae>
yes sure
hax404 has quit [Ping timeout: 240 seconds]
<wucke13>
Can one packageOverride be used in another packages override?
<xfix>
in fact, having /nix on a separate partition isn't very contrived, for instance you may want to do that to disable snapshoting for /nix, and so on
<etu>
And one probably want a bunch of bind-mounts and symlinks specified in the config to some kind of persistent storage
<xfix>
technically only .local/share/fish/fish_history needs to be persistent, but i suppose this saves on time running completion generation script every time you reboot the machine
<colemickens>
etu: has someone tried to submit a persistence module like that upstream?
<etu>
colemickens: The ones on top are upstream things, such as symlinks in etc and bind-mounts. The thing on bottom is using a module for home-manager that isn't upstream.
<colemickens>
yeah, I went and found the module from talyz or whatever. it'd be cool to see that in HM
deuloma has quit [Remote host closed the connection]
<lemsip>
how would I go about making a package from a shell.nix? I just want an executable that runs a single command.
ilikeheaps has quit [Quit: Konversation terminated!]
<hyperfekt>
xfix: automatically? where does it do that? i thought you'd have to trigger that? generating the completions from man pages is exactly what NixOS does for you
<xfix>
fish shell creates ~/.local/share/fish/generated_completions directory if missing
stree has quit [Read error: Connection reset by peer]
bvdw1 has joined #nixos
stree has joined #nixos
corpix_ has joined #nixos
Scriptkiddi has quit [Ping timeout: 246 seconds]
ajs124 has quit [Ping timeout: 260 seconds]
das_j has quit [Ping timeout: 260 seconds]
mrpi has quit [Ping timeout: 260 seconds]
<{^_^}>
[nixpkgs] @Profpatsch opened pull request #86194 → pandoc: remove GHC from the runtime closure (2.4GB -> 180MB) → https://git.io/JfmYj
corpix has quit [Ping timeout: 240 seconds]
rembo10_ has joined #nixos
Bunogi has quit [Ping timeout: 246 seconds]
Scriptkiddi has joined #nixos
wolfshappen has quit [Ping timeout: 256 seconds]
lurkless has quit [Ping timeout: 256 seconds]
bvdw has quit [Ping timeout: 256 seconds]
reactormonk has quit [Ping timeout: 256 seconds]
bvdw1 is now known as bvdw
gxt has quit [Ping timeout: 240 seconds]
unacceptable has quit [Ping timeout: 256 seconds]
WilliButz has quit [Ping timeout: 264 seconds]
rembo10 has quit [Ping timeout: 264 seconds]
aszlig has quit [Ping timeout: 260 seconds]
Bunogi has joined #nixos
unacceptable has joined #nixos
lurkless has joined #nixos
reactormonk has joined #nixos
aszlig has joined #nixos
WilliButz has joined #nixos
gxt has joined #nixos
<hyperfekt>
xfix: any chance you have python on $PATH? it seems like the source you were looking at doesn't have its python path patched during the build
mrpi has joined #nixos
das_j has joined #nixos
ajs124 has joined #nixos
<xfix>
i do
wolfshappen has joined #nixos
<hyperfekt>
that's it then, thanks. i'll get that fixed
<xfix>
cool :)
<xfix>
Python is an optional dependency for fish, so fish pretty much ignores it being missing
<xfix>
and the shell never informs about it being missed, so it could be easily missed
<xfix>
about it not existing*
<xfix>
a particular Python version doesn't matter, fish shell checks python3, python2 and python in order
jbrock has quit [Ping timeout: 260 seconds]
jbrock has joined #nixos
<adisbladis>
Hm
CMCDragonkai1 has joined #nixos
<adisbladis>
We should patch those shebangs
CcxWrk_ has joined #nixos
CcxWrk_ is now known as CcxWrk
CcxWrk has quit [Killed (tolkien.freenode.net (Nickname regained by services))]
fpob_ has joined #nixos
<adisbladis>
Ah, it's not even a shebang
fpob has quit [Ping timeout: 256 seconds]
<adisbladis>
That's pretty dumb
<{^_^}>
[nixos-homepage] @garbas pushed to landing-page « a bit better main description, but i'm lost with words »: https://git.io/JfmOW
<adisbladis>
xfix: hyperfekt: We could patch the fish function __fish_anypython
<xfix>
sure
<hyperfekt>
adisbladis: haha if that was used consistently we could only do that, yes
<adisbladis>
hyperfekt: It would fix fish_config at least
opticnerve has joined #nixos
<xfix>
this probably could be fixed to use __fish_anypython, i think it's a bug that generating man pages doesn't do that
<hyperfekt>
adisbladis: the fish_update_completions command already works, that's not the issue
<adisbladis>
What's your issue then? :)
<xfix>
i will submit a patch soon
<xfix>
for fish shell specifically to use __fish_anypython here
<hyperfekt>
wait nvm i'm still on an old version.
<xfix>
yeah, the git master uses __fish_anypython in fish_update_completions
<xfix>
so yeah, this could return a constant value in NixOS
<xfix>
echo /nix/store/whatever
jbrock has quit [Quit: jbrock]
jbrock has joined #nixos
<hyperfekt>
okay, so we patch fish to use __fish_anypython everywhere upstream, we patch __fish_anypython to actually return a path in the nix derivation, and we patch the fish interactive init to create the completion directory in the fish module
<xfix>
i think fish should use __fish_anypython everywhere already
<xfix>
other than python completions, but here it's correct to not use __fish_anypython
<adisbladis>
sphalerite: Yeah! tmpfs _all_ the things!
<hyperfekt>
rheesa would be proud
<adisbladis>
Running Docker on tmpfs was annoying
alp has joined #nixos
<hyperfekt>
haha
<etu>
aanderse: tmpfs is really fast though ;D
<aanderse>
woo! i'm glad to hear
<hyperfekt>
xfix: now i see why i don't have that directory created even with python on path, i have my own __fish_config_interactive that cuts out like 80% of the stuff
<aanderse>
etu: looks like /root needs to be persistent?
<etu>
adisbladis: I still do that, but I only have one container that I build very rarely. We'll when I need to build more with docker.
<etu>
aanderse: I have that bind mounted
<etu>
aanderse: /persistent/home/root /root auto bind,noauto,x-systemd.automount 0 2
<aanderse>
etu: yeah i noticed that. i assume there is a reason it need to be persistent then?
<etu>
aanderse: No real need, but I kinda like to have the shell history intact
<{^_^}>
[nixos-homepage] @garbas merged pull request #414 → Use fd to rebuild untracked files → https://git.io/JftTp
<{^_^}>
[nixos-homepage] @garbas pushed 2 commits to master: https://git.io/Jfm3S
<aanderse>
etu: ahh, ok
<hyper_ch>
and also .ssh/known_hosts is nice to have :)
<aanderse>
i usually just sudo... maybe i'll skip the persistent /root
<etu>
aanderse: Sure :)
never_released_ has joined #nixos
<aanderse>
this is a home desktop, not a server
<adisbladis>
As a data point: I don't have persistent /root anywhere
<adisbladis>
I don't care about root shell history
never_released has quit [Ping timeout: 256 seconds]
<etu>
hyper_ch: Well, I don't ssh from root user. And .ssh is reasonable to have persistent for ssh keys so known_hosts is in there as well
<adisbladis>
etu: Luckily podman doesn't have the same shortcomings as Docker in the file system drivers
<etu>
adisbladis: Ond most systems I could probably drop that :D
<hyper_ch>
:)
<adisbladis>
etu: Feels a bit ¯\_(ツ)_/¯
<adisbladis>
etu: Interesting idea though, also use home-manager for root :P
<adisbladis>
And get the same persistent symlinking
<etu>
That would be nice
<etu>
I should probably do that
<hyperfekt>
xfix: adisbladis: i made the upstream PR
<{^_^}>
#86195 (by adisbladis, 36 minutes ago, open): fish: Patch __fish_anypython function to return build input python3
<seku>
things are moving fast :O
jakobrs has joined #nixos
<jakobrs>
Does NixOS + NixOps work well on the Pi 4?
pjan has joined #nixos
<hyper_ch>
etu: what is the effect of that bind mount: /persistent/home/root /root auto bind,noauto,x-systemd.automount 0 2 ?
maddo has joined #nixos
<hyper_ch>
sphalerite: why have /etc/nixos on tmpfs?
<{^_^}>
[nixops] @adisbladis opened pull request #1327 → Restore private ipv4 support → https://git.io/JfmsU
chagra has joined #nixos
<etu>
hyper_ch: It will create a systemd automount unit for /root that has a placeholder filesystem "mounted" on /root that will trigger a the mount of /persistent/home/root on /root whenever something accesses /root
<hyper_ch>
and why not just mount it like normal?
<etu>
Because speed :p
<hyper_ch>
:)
<adisbladis>
Aint nobody got time for mount
<aanderse>
lol
<etu>
hyper_ch: I don't think it matters tbh when you have it all on a zfs volume, but it's something I've used to do in the past because spinning rust with ext4 may take time to mount etc. And then I didn't want mounts blocking the boot process since they weren't needed for the boot really.
itorres has left #nixos ["Kicked by @appservice-irc:matrix.org : issued !quit command"]
<linarcx>
Hey guys, i'm using nix shell. Added libedit library. It's working perfect. But i wana to access it's manual page too. So `man history` won't show anything to me. How can i access man pages of installed package inside nix shell?
<nschoe>
noonien: they are doen when you instantiate nix-shells yes. But they should be cleared on boot.
<sphalerite>
hyperfekt: just joking, taking the concept beyond sane extents
<sphalerite>
er
<sphalerite>
hyper_ch: ^
<hyper_ch>
sphalerite: ok :)
<hyper_ch>
I thought you might have some nix wizzardry up your sleeve for doing so
linarcx has quit [Ping timeout: 265 seconds]
turlando has joined #nixos
<turlando>
Hello
<hyper_ch>
hmmm, the Dell PERC H200 that I ordered unRAIDed through ebay works fine... I have now for each server one of those. I just wonder if I should order maybe 1 or 2 more as backups
<hyperfekt>
that reminds me of when i tried to write a toSQL function for nix
<turlando>
I'm new to Nix and NixOS and I'm planning to migrate my server (now running FreeBSD) to Nix. I'm reading some documentation and performing some tests in a VM. Here https://nixos.wiki/wiki/NixOS_on_ZFS#Pool_Layout_Considerations it suggests to create a local/nix volume and a safe volume. Should it be the same if I create just a nix volume not residing inside local, right?
<hyperfekt>
aanderse: i wanted to define an sqlite database in nix because a program stored its settings in there. that function would convert a submodule specifying a database and its contents to SQL statements generating such a database
<aanderse>
ah, ok
<edcragg>
hi, i'm having difficulty working out how to install a plugin for nixops, for which there is almost no documentation. i'm trying to install https://github.com/input-output-hk/nixops-packet, is someone able point me in the right direction?
rogue_koder has joined #nixos
rogue_koder_ has quit [Read error: No route to host]
<turlando>
sphalerite: thanks for your reply. So it should be safe to put the nix store in an unencrypted volume too. What about compression though? It's going to contain mostly compressed archives?
<{^_^}>
[nixpkgs] @arianvp opened pull request #86208 → linux: do not depend on systemd indirectly → https://git.io/JfmWv
cjpbirkbeck has joined #nixos
<seku>
turlando: noob here too, but iirc zfs checks the first few chunks of a file to see if it is compressible. if it is,compression is applied. I usually just leave compression on on the pool level, thus nix store as well.
<turlando>
seku you're right, zfs compares the first 1/8 of data and if the compressed version is smaller than the uncompressed version then it keeps the compressed one, otherwise discards compression. On freebsd I'm used to disable compression for the distfile and packages directory since they contains only compressed archives, thus I was asking. I agree it's harmless and a minor optimization
sarcasticadmin has joined #nixos
<xfix>
ZFS compression is worth it even if all files are already compressed
<turlando>
thanks for the suggestions
<turlando>
What about the safe volume? It reflects a normal *nix tree, right?
lord| has quit [Read error: Connection reset by peer]
<xfix>
fyuuri: have you enabled gpg-agent: `programs.gnupg.agent.enable = true`?
<fyuuri>
Also, since upgrading to 20.03 tmux does not load ~/tmux.conf anymore. Does anybody else have the same problem?
<fyuuri>
@xfix: Not that I know of! THank you very much for the hint. Did not seem to be necessary before 20.03.
<fgaz>
how do I make a package available for fstab mounts?
<fyuuri>
Ill try and come back :)
<fgaz>
I'm trying to set up s3fs
aveltras has joined #nixos
<fgaz>
adding it to systemPackages doesn't seem to work
<xfix>
fyuuri: it's in release notes
<xfix>
- GnuPG is now built without support for a graphical passphrase entry by default. Please enable the gpg-agent user service via the NixOS option programs.gnupg.agent.enable. Note that upstream recommends using gpg-agent and will spawn a gpg-agent on the first invocation of GnuPG anyway.
<fyuuri>
xifix: I must be blind :(. I read through it and did not notice
<turlando>
xfix I could be wrong but I think that the recordsize is the biggest block ZFS will read, not the smallest
<turlando>
read/write
<xfix>
oh right
<xfix>
okay, better example, if you have 160KB file with record size set to 128KB, with compression you will have 160KB file, not 256KB file
<turlando>
makes sense now :)
<xfix>
good on you on verifying whatever you read on the internet :)
<fgaz>
maybe I should use systemd.mounts instead of fileSystems?
iyzsong has quit [Ping timeout: 244 seconds]
o1lo01ol1o has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @afrepues opened pull request #86213 → emacs-modes: build MELPA package sets without null packages → https://git.io/Jfmlg
<fyuuri>
@xfix : thank you so much! It works like a charm :)
<rooke>
Running into an odd error with lorri + direnv. The set up was working until it suddenly stopped. When I enter the directory lorri complains that it can't find the `EVALUATION_ROOT`. Building the enviroment by hand with `lorri shell` still completes and other directories are working. Reboot didn't fix the problem, any ideas what may be causing it?
<fyuuri>
Does anyone have an example nixos compatible fish configuartion using the fisher plug-in manager?
fabianhjr has joined #nixos
<fgaz>
aha, fixed it by using `${pkgs.s3fs}/bin/s3fs#buket` instead of `s3fs#bucket`
<rooke>
er of course right as I ask I manage to fix it... not sure exactly why it caused the issue or why it was working before but changing the `.envrc` to be `eval "$(lorri direnv)"` instead of `eval '$(lorri direnv)'` fixed it
<{^_^}>
[nixpkgs] @elohmeier opened pull request #86214 → obs-studio: 25.0.3 -> 25.0.8, use addOpenGLRunpath → https://git.io/Jfmli
<fyuuri>
Files in ~/.config/fish/conf.d/ are loaded too late in the loading process
<noonien>
is it common tofor blue man to not display device names?
eoli3n_ has joined #nixos
<noonien>
for blueman*
CMCDragonkai1 has joined #nixos
cosimone has quit [Quit: Quit.]
cosimone has joined #nixos
eoli3n_ has quit [Client Quit]
<nschoe>
Is there some context somewhere for funny nix hashes? 'cause I just saw "-L/nix/store/eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee-gdbm-1.18.1/lib" pass on my build.... :-)
<fyuuri>
Does anyone have a pointer for me regarding fish and nixos? Can I somehow make it only load the config in ~/.config/ ?
jbrock has quit [Quit: jbrock]
jbrock has joined #nixos
aveltras has quit [Read error: Connection timed out]
aveltras has joined #nixos
<{^_^}>
[nixpkgs] @Mic92 pushed commit from @mmilata to release-20.03 « libmicrodns: 0.1.0 -> 0.1.2 »: https://git.io/Jfm8O
<gustavderdrache>
if you see a whole bunch of e's, it's likely that it was a removeReferencesTo call in order to keep the closure size down
jbrock has joined #nixos
o1lo01ol1o has joined #nixos
<nschoe>
gustavderdrache: oh :/ You just crushed my hopes ^^
<nschoe>
gustavderdrache: just kidding. Thanks :)
<gustavderdrache>
haha, it's all good
hoijui has quit [Ping timeout: 260 seconds]
<gustavderdrache>
my shoes aren't that dirty :P
<gustavderdrache>
in that specific context, it looks like there's a reference to gdbm that someone wanted to remove in a script
<nschoe>
gustavderdrache: stupid question: why would anyone want to remove a reference?
<gustavderdrache>
so here's an example
<gustavderdrache>
i'm packaging $FOO for nixpkgs
<gustavderdrache>
and $FOO is an autoconf-powered program
<gustavderdrache>
and so one of the things it helpfully does is keep track of all of its configure flags
<gustavderdrache>
so you can see that $FOO was compiled with, say, gdbm support
<gustavderdrache>
but most library packages are split: there's a dev (headers) and a lib (.so files) version, so to speak
<gustavderdrache>
so what this means is that just to support foo --help, nix is forced to assume that $FOO needs both the gdbm library files and its headers
proofofkeags has joined #nixos
<gustavderdrache>
(this is what "closure" means - it's like a function closure)
<nschoe>
I see.
<etu>
gustavderdrache: Have you seen the updated php packaging? :)
<gustavderdrache>
etu: i did! i'm still really amused that i accidentally inspired that
<etu>
gustavderdrache: :)
<gustavderdrache>
i can't wait to give it a proper spin
<gustavderdrache>
it might replace a bunch of awkward PHP packaging i've had to do for work
<etu>
gustavderdrache: We have some more improvements coming
<etu>
gustavderdrache: Well, you had some really cool things going on there, we just needed to work a bit more on it :)
<gustavderdrache>
nschoe: long story short, it's how you tell nix "no, really, i don't need this"
<gustavderdrache>
etu: eyes emoji
<gustavderdrache>
etu: if you need an extra pair of hands, i'm more than happy to accidentally break things due to my newbieness
<worldofpeace>
@arianvp:matrix.org: I'm guessing by desktop icons you just mean icon themes. When you install an icon theme they should be at either /run/current-system/sw/share/icons, or under nix-profile with nix-env, so it should work instant because the enviroment doesn't need to change to detect them (xdg data dirs)
endocrimes has quit [Quit: running from the computers]
endocrimes has joined #nixos
rien has quit [Read error: Connection reset by peer]
andi- has joined #nixos
endocrimes has quit [Client Quit]
endocrimes has joined #nixos
<the_pumpkin_man[>
Hello, is there a way ton configure something system wide even if it isn't implemented in NixOS ?
<the_pumpkin_man[>
Because I'm trying to use udevil to mount device, but I need to edit the configuration for that
justanotheruser has joined #nixos
<the_pumpkin_man[>
And tried to search in nixos-option, there's a programs.udevil.enable option but nothing more (hell, why not a configFile or extraConfig ?)
<asbachb>
Maybe a strange question: But does nixos default firewall block outgoing ssh connections?
Thra11 has quit [Ping timeout: 260 seconds]
<gchristensen>
no
<Guest215>
I don't think so
<gchristensen>
Guest215: you've gone from bbigras to Guest215 :)
asbachb has quit [Remote host closed the connection]
philr_ has quit [Ping timeout: 260 seconds]
<Guest215>
haha yeah. I'm using matrix and I had "Bruno" in most rooms but I kept getting notifications destined to others Brunos. So I changed my main name or whatever to bbigras and I use "Bruno" in like 1 room with a friend.
asbachb has joined #nixos
<asbachb>
Mh strange. I'm unable to access my server via ssh or connect to github.
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
<Kyndig>
I tried manually creating a profile link to ...3s60widcb-user-environment, which is what the first user's profile pointed to, but that resulted in a nix-env error about nix/share
<Kyndig>
it seems all the documentation stops short of telling you how to actually install the imported store items.
<nschoe>
,locate ar
<numkem>
say I'd like to provide my own packages that I can put in configuration.nix but outside of nixpkgs itself. Are overlays the way to go?
<{^_^}>
[nixpkgs] @flokli pushed 4 commits to release-20.03: https://git.io/Jfmgz
<{^_^}>
[nixpkgs] @symphorien opened pull request #86231 → nix-du: fix with strictDeps → https://git.io/Jfmgo
<numkem>
the idea would be to make it available to multiple machines
o1lo01ol1o has quit [Ping timeout: 264 seconds]
<ar>
nschoe: warsaw, poland; this time not at the hackerspace
<{^_^}>
[nixpkgs] @zimbatm pushed commit from @evils to master « foldingathome: 7.5.1 -> 7.6.9 (#86171) »: https://git.io/Jfm2s
jkarni has joined #nixos
<nschoe>
ar: ^^ sorry :D
<nschoe>
cole-h: okay it's just that i'm trying to package a library and it complains that 'ar: not found'.
trubi has joined #nixos
<aanderse>
so what's the command to rescue an install from nixos live usb?
Henson has joined #nixos
morgrimm has joined #nixos
<cole-h>
nixos-enter I think you're looking for?
<cole-h>
aanderse: ^
<morgrimm>
Hey all - I'm trying to customize ndrxp's flake repo for my configs, and I'm getting an error when trying to use a symlink as a path in it.
<cole-h>
bqv: ^ Maybe you can help
<morgrimm>
It's missing a `lastModified` attribute - I assume Nix needs it to hash it
<Henson>
hi folks. What's the best way to handle inter-software dependency in NixOS? Say the derivation of one piece of software has variables that are set using NixOS options, and a second piece of software takes the first software's derivation as an argument in order to call an executable in the first software. In the NixOS configuration file I'll need to make the second package refer to the first...
<{^_^}>
[nixpkgs] @worldofpeace pushed 10 commits to master: https://git.io/Jfm2F
<Henson>
somehow. Do I make an NixOS option for the second software to explicitly specify the first? Do I have the first software insert a hidden option in the system configuration indicating its derivation that the second piece of software picks up on without the user ever having to specify it?
<glittershark>
https://nixos.org/nix/manual/#chap-distributed-builds says "you need to ensure that the PATH of non-interactive login shells contains Nix." for using remote builders - but doesn't give any info on *how* to do that
dooygoy has quit [Ping timeout: 264 seconds]
<Henson>
or is there a better way I can accomplish the goal?
<glittershark>
I'm getting exactly the described error message
<hyper_ch>
Henson: what are you trying to achieve specifically?
jgeerds has joined #nixos
<glittershark>
my .zshrc on that machine sources ~/.nix-profile/etc/profile.d/nix.sh, and if I actually log in to a shell on that machine `which nix-store` succeeds
<glittershark>
but nix ping-store says nix-store is not a command
dooygoy has joined #nixos
<glittershark>
is nix ping-store not sourcing my .zshrc? if so what *does* it source?
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/JfmaJ
<nschoe>
Quick question: I have added 'binutils-unwrapped' in my 'buildInputs'. I entered a nix-shell and ran 'unpackPhase' and 'configurePhase'.
<nschoe>
Looks like 'ar' is not in the path.
<nschoe>
I can see that 'ar' is in /nix/store/....binutils-2.31.1/bin, but 'which ar' doesn't show it.
<Henson>
hyper_ch: I have multiple pieces of software that depend on each other because they call each other's executables. At the derivation level I simply need to pass the derivations of parent software into children software and they can find the executables. But when incorporating this into a NixOS configuration, I want to keep the software modules separate from each other and wrap the different...
<nschoe>
I thought adding a package in 'buildInputs' would add its bin/ in the $PATH
rogue_koder has quit [Ping timeout: 246 seconds]
<cole-h>
nschoe: buildInputs is for libraries. You probably want it in nativeBuildInputs.
<cole-h>
(and you probably don't want -unwrapped, just normal binutils will do fine)
<nschoe>
cole-h: okay, trying this.
<nschoe>
thanks
<Henson>
hyper_ch: derivation variables in NixOS options. But I need to figure out the correct way at the NixOS configuration level of making the different software aware of each other. I can't pass derivations for the software around to each other because they're now controlled using NixOS options and not by explicitly defining the software using callPackage with arguments. I'm trying to keep the...
<Henson>
hyper_ch: interfaces to configuration the software separate and clean, while being able to handle the inter-package dependencies.
rogue_koder has joined #nixos
<hyper_ch>
no idea
<{^_^}>
[nixpkgs] @Ericson2314 merged pull request #86080 → meson: Clean up build and infra → https://git.io/JftpE
<{^_^}>
[nixpkgs] @Ericson2314 pushed 4 commits to staging: https://git.io/Jfmac
<{^_^}>
[nixpkgs] @ThibautMarty opened pull request #86236 → treewide: fix modules options types where the default is null → https://git.io/Jfma8
<nschoe>
cole-h: I've used 'binutils' instead of the -unwrapped and put it in 'nativeBuildInput' : stil not 'ar' in the PATH, and this time, the derivation path is '/nix/store/....binutils-wrapper-2.31.1/' and no 'ar' in the bin/
alp has quit [Ping timeout: 272 seconds]
Darkmatter66 has quit [Ping timeout: 240 seconds]
<cole-h>
AFAIK, binutils should be available by default. How exactly are you running this?
<aanderse>
cole-h any ideas if nixos-enter says it isn't a nixos installation?
<turlando>
Using ZFS on NixOS is it mandatory to use mountpoint=legacy for / and /nix?
Darkmatter66 has joined #nixos
<cole-h>
aanderse: Sorry, I was just recalling the binary from memory -- I don't actually use NixOS yet x)
<{^_^}>
[nixpkgs] @prusnak opened pull request #86237 → Update Electron 6,7,8 → https://git.io/JfmaS
<glittershark>
obviously, because it's a non-interactive shell
<nschoe>
cole-h: ah thank.
<nschoe>
So presumably I have to patch something to replace ar with the ar from binutils. Now are you sure it's the 'binutils' and not -wrapped needed?
asbachb has quit [Ping timeout: 240 seconds]
<nschoe>
I can't seem to find 'ar' (even prefixed) in 'binutils'/bin
o1lo01ol_ has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
jbrock has quit [Quit: jbrock]
jbrock has joined #nixos
jbrock has quit [Remote host closed the connection]
jbrock has joined #nixos
gustavderdrache has quit [Ping timeout: 265 seconds]
<cole-h>
In that case, yeah, use -unwrapped :P
alp has joined #nixos
<glittershark>
I needed to source the nix-profile in zshenv not zshrc >.<
stigo has quit [Remote host closed the connection]
stigo has joined #nixos
<sphalerite>
ar: nschoe can't find you!
<sphalerite>
(scnr)
fabianhjr has joined #nixos
fabianhjr has quit [Max SendQ exceeded]
dooygoy has quit [Quit: WeeChat 2.8]
<nschoe>
sphalerite: it's not a lack of trying, though!
FRidh has quit [Ping timeout: 244 seconds]
fabianhjr has joined #nixos
fabianhjr has quit [Max SendQ exceeded]
fabianhjr has joined #nixos
fabianhjr has quit [Max SendQ exceeded]
mallox has quit [Quit: WeeChat 2.8]
<sphalerite>
nschoe: re having things use the right ar — often they'll look at the AR environment variable, so maybe start by setting that and see if it magically fixes everything?
fabianhjr has joined #nixos
fabianhjr has quit [Max SendQ exceeded]
mallox has joined #nixos
<{^_^}>
[nixpkgs] @glittershark opened pull request #86241 → openexr: Switch to cmake-based build → https://git.io/Jfmws
knupfer has joined #nixos
knupfer1 has quit [Quit: knupfer1]
<nschoe>
sphalerite: well I've beeng issuing about sixty billion "grep" commands to try and find this "AR" thing to be able to --substitute but so far.. none
<nschoe>
I am wondering about my sanity at this point ^^
<nschoe>
sphalerite: but it makes sense to try and export AR and manually call buildPhase indeed. Thanks.
numkem has quit [Remote host closed the connection]
numkem has joined #nixos
jasom has quit [Ping timeout: 246 seconds]
knupfer has joined #nixos
eoli3n__ has joined #nixos
tpanum- has joined #nixos
<energizer>
nix-copy-closure prints "copying 9023 paths" and then every minute or so it says "copying path /nix/store/... to ssh://foo". does that mean it's gonna take ~9023 minutes to finish?
<risson>
Are you sure you have a good enough bandwidth?
<risson>
Or maybe you just started with some heavy paths
<risson>
I'm getting cannot copy /nix/store/0cgij5flg2azj1byy3wyn47qbxyv1ppz-grub-2.04/share/grub/unicode.pf2 to /boot when switching configuration inside a chroot
tpanum has quit [Ping timeout: 260 seconds]
proofofkeags has joined #nixos
<risson>
Yes /boot is mounted, but apparently I cannot write to it from inside the chroot
iqubic` has quit [Remote host closed the connection]
cruxeternus has joined #nixos
himmAllRight has joined #nixos
jbrock has quit [Client Quit]
jbrock has joined #nixos
<fresheyeball>
I have a thing I want to build like this
<fresheyeball>
nix-build -A foo.bar.baz
<fresheyeball>
which worked fine
user_0x58 has quit [Ping timeout: 244 seconds]
<fresheyeball>
but now foo.bar.baz is a lambda that requires a set with field "color"
iqubic has quit [Ping timeout: 240 seconds]
<fresheyeball>
I tried nix-build -A foo.bar.baz --argstr color "red"
asbachb has joined #nixos
<fresheyeball>
but it seems to pass the argument, not to the lambda in question, but into the default.nix file
<fresheyeball>
how can I get around this?
<asbachb>
When I enable `networking.networkmanager.enable = true;` I'm able to connect to my wifi. Firefox is working too. But I'm unable to do ssh for example.
<asbachb>
Any ideas how to investigate?
<fresheyeball>
asbachb: do you have a firewall?
iqubic has joined #nixos
<asbachb>
fresheyeball: Only default configuration. Already tried to disable the firewall completely.
<fresheyeball>
what is the error you are getting?
<hexa->
asbachb: try ssh -v
niso_ has joined #nixos
<kraem>
are you trying to connect by ip or a domain name? (not sure but i think web browsers usually use their own dns resolver, which could explain firefox working)
<edrex>
Hi all, I getting my user env set up on an old 32 bit thinkpad, and wondering if there are any big i686 binary caches out there. I guess the build verifiability of nix doesn't help if there's no trusted party specifying the hash keys ya?
<asbachb>
hexa-: I don't think it's a unitymedia problem. When I disable `NetworkManager` and switch back to `networking.wireless.networks` it works without problems.
never_released_ has quit [Quit: ZNC 1.7.5 - https://znc.in]
niso_ has quit [Ping timeout: 246 seconds]
niso has joined #nixos
fabianhjr has joined #nixos
koleesch has joined #nixos
domogled has quit [Quit: domogled]
<asbachb>
Flussh all iptables. Still does not work.
never_released has joined #nixos
nschoe has joined #nixos
never_released has quit [Remote host closed the connection]
jbrock has quit [Quit: jbrock]
jbrock has joined #nixos
proofofk_ has joined #nixos
jbrock has quit [Client Quit]
pamplemousse has joined #nixos
asbachb has quit [Ping timeout: 240 seconds]
cr4y1 has quit [Remote host closed the connection]
jbrock has joined #nixos
cr4y1 has joined #nixos
asbachb has joined #nixos
<asbachb>
hexa-: Got it to work. Seems that the NetworkManager profile was kind of broken. Deleted it and redone the connection. Seems so work now. Thank you for your time.
<{^_^}>
[nixpkgs] @adisbladis opened pull request #86249 → podman: Wrap packages required to run containers → https://git.io/Jfm6h
<nschoe>
clever: ah.Well regarding our previous conversation from earlier, I decided to go ahead and try to package library bcm2835 anyway, Even if I don't use it in the end (I'm still aiming at usign one of the other two Haskell I2C lib).
<nschoe>
Anyway so I'm writing a .nix file to package bcm2835. And it was failing in the 'buildPhase' b/c 'ar not found'.
never_released has quit [Remote host closed the connection]
<nschoe>
clever: so I have been trying for about 2 hours now to find a way to make 'ar' known to the buildPhase.
<aleph->
Hmm, how does one pass changes to the docker daemon to docker? Need to set it to using an insecure registry
<clever>
nschoe: ah yeah, *looks*
never_released has joined #nixos
eoli3n__ has quit [Ping timeout: 256 seconds]
<clever>
nschoe: can you gist your nix exprs?
vykook has quit [Ping timeout: 265 seconds]
<nschoe>
clever: just so you know: I've tried including 'binutils' ,'binutils-unwrapped', both in 'buildInputs' and 'nativeBuildInput' ,to no avail: there is just never an unpreficed 'ar' in the PATH.
<nschoe>
Then I found this thing to put in makeFlags on the Internet, I tried it and now it builds.
<nschoe>
I tried to grep everything I could in the source of bcm2835, nothing.
<nschoe>
I did notice that there was a prefixed 'ar', (something with 'linux-gnu-aarch64' as a prefix in ${binutils}/bin/.
<{^_^}>
[nixpkgs] @mweinelt opened pull request #86250 → [19.09] openldap: add patch for CVE-2020-12243 → https://git.io/Jfmic
<clever>
nschoe: looks like automake detected the wrong ar
<nschoe>
clever: oh I never managed to get that AR = <something>
<nschoe>
clever: but how could it select the "right ar"? The only 'ar' is the on ein binutils-unwrapped/bin, and it's prefixed.
<energizer>
https://bpaste.net/N7RA could not locate fortran compiler, but i have pkgs.gfortran in buildInputs. How can i get it to notice the fortran compiler?
<clever>
nschoe: the stdenv already set $AR the env var
o1lo01ol1o has quit [Ping timeout: 240 seconds]
morgrimm has quit [Ping timeout: 240 seconds]
bastion-tester has joined #nixos
<clever>
nschoe: preConfigure = ''makeFlags="$makeFlags AR=$AR"''; is enough to fix it as well
<nschoe>
clever: it seems you're right: I've just removed my makeFlags and entered the nix-shell, echo $AR deos print "aarch64-unknown-linux-gnu-ar"
<clever>
nschoe: that forces make to set AR (the make var) to $AR (the bash env var)
<nschoe>
clever: so I guess just putting AR=${AR} would work, then?
<energizer>
answer: it needed scikit-build
<nschoe>
Ah, well you bascially wrote this :)
<clever>
the problem is that the makefile is force-setting AR, not providing a default and fallback
erictapen has joined #nixos
<clever>
the example i linked from open-firmware, just assumes $AR (the env var) is already set, and fails otherwise
<clever>
use nix or go away :P
eoli3n__ has joined #nixos
<iqubic>
So, I'm running afoul of this issue: https://github.com/p11-glue/p11-kit/issues/275 Basically my flatpak applications can't properly work now. Anyone know if there's a fix for this in nixpkgs yet?
<clever>
nschoe: your likely generating the same string as what normal $AR contains
<drozdziak1[m]>
I'm trying to install an overridden vimHugeX I defined in `~/.config/nixpkgs/config.nix`. It does what I want in nix-shell, but it gets in a conflict with another vim_configurable on man pages. Setting the priority on vim_configurable coesn't do the trick as it still complains again
<drozdziak1[m]>
Even when I uninstall vim altogether it's all the same
<nschoe>
clever: ok
<nschoe>
clever: I've just tried your preConfigure=makeFlags=... and it build :)
<nschoe>
Thanks
<bqv>
nix does full evaluation from scratch every time, doesn't it? just checking cause i've seen times vary sometimes, guessing that's just from dumping paths to store though
<clever>
drozdziak1[m]: what does `nix-env -q` list?
<iqubic>
I wish flatpak and NixOS weren't such a pain to integreate.
<iqubic>
*integrate
Darkmatter66 has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @veprbl pushed commit from @matthuszagh to master « pythonPackages.pylibftdi: init at 0.18.1 (#71286) »: https://git.io/JfmiF
<nschoe>
Let's see if copying the clojure works now.
<bqv>
iqubic: appimage works a lot nicer, if available
<iqubic>
I see.
<iqubic>
Thank you for that.
<{^_^}>
[nixpkgs] @erictapen opened pull request #86253 → nixos/printers: fix example for hardware.printers.ensurePrinters.*.model → https://git.io/Jfmip
vykook has joined #nixos
<iqubic>
I'm not looking to switch what I'm using, I'm looking to see if this current issue can be fixed.
Darkmatter66 has joined #nixos
<simpson>
Hopefully the issue would be fixed if your target package were built using Nix; I'm assuming that whatever you're working with is unfree.
o1lo01ol1o has joined #nixos
<{^_^}>
[nixpkgs] @Ericson2314 opened pull request #86254 → mkDerivation mesonFlags: Fix cross file logic to handle more cases → https://git.io/JfmPT
<iqubic>
simpson: How will we fix this p11-kit bug?
<iqubic>
How can we fix it?
vykook has quit [Ping timeout: 260 seconds]
<simpson>
iqubic: It sounds like they'll release a new version rolling back the RPC changes. So we'd just have to update to that new version in nixpkgs and possibly leave a comment for the next person, warning them that that particular release version is bogus.
<nschoe>
clever: this looks to be working: clojure is copied on the rPi and executed. I do get a "test-hpi: I2C: : illegal operation (Received an unexpected NACK.)" error.
<nschoe>
But I think it's an issue with my wiring or the arduino polluting the I2C bus
<clever>
nschoe: id double-check things with a scope
<nschoe>
clever: yeah. A lot of unpilling to do :)
<iqubic>
simpson: How long will I have to wait to for this new version to be made?
<nschoe>
But at least the cross-compilation (and bcm packaging) worked.
<nschoe>
Now I'll try *not* using HPi :D but after I fixed my I2C wiring :=)
<simpson>
iqubic: I don't know how to answer that question. I recommend not waiting.
<simpson>
Well, who are you waiting for, what changes are you going to integrate, and what is your definition of working? Perhaps finding another path forward will be more fruitful. I don't know; I have no idea what this package is used for or why it's there.
nDuff has quit [Read error: Connection reset by peer]
vykook has joined #nixos
linarcx has quit [Read error: Connection reset by peer]
linarcx has joined #nixos
mallox has quit [Ping timeout: 246 seconds]
kumikumi_ has quit [Ping timeout: 250 seconds]
vykook has quit [Ping timeout: 260 seconds]
jluttine has quit [Ping timeout: 264 seconds]
<energizer>
i need to figure out why this build fails to find the fortran compiler that should be available. how can i diagnose it? https://bpaste.net/AW3Q
<energizer>
can i somehow get a shell inside the build environment?
<nschoe>
energizer: if you are writing your own derivation, you can 'nix-shell default.nix` yes
<simpson>
Yeah, you can use nix-shell *without* -p.
jasom has joined #nixos
<energizer>
nschoe: i dont think that does what i need. the problem is that the build fails, so nix-shell refuses to enter the environment because the environment is not successfully built
jluttine has joined #nixos
zemm has quit [Ping timeout: 246 seconds]
zemm has joined #nixos
kumikumi has joined #nixos
<energizer>
maybe i misunderstood your suggestion
<energizer>
when a build fails, how do you figure out why?
<cole-h>
Read the logs. Add `-K` to keep the output so you can explore the files it was operating on. Add `breakpointHook` to its `buildInputs` and use `cntr` to enter the sandbox as soon as it fails.
ramses_ has joined #nixos
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<iqubic>
simpson: Basically, I'm just trying to debug this issue where running Steam in flatpak gives me ssl certificate invalid errors when trying to do any networking at all.
<nschoe>
energizer: entering teh nix-shell will not initiate the build: this is how you debug a failed build, precisely.
morgrimm has joined #nixos
vykook has joined #nixos
<bqv>
breakpointHook
<bqv>
I'm gonna remember that one
<bqv>
that sounds like a lifesaver
<cole-h>
It's pretty nice.
zupo has joined #nixos
<{^_^}>
[nixpkgs] @jonringer opened pull request #86258 → libbfd: remove stale patch → https://git.io/JfmX2
markus1189 has quit [Ping timeout: 260 seconds]
markus1189 has joined #nixos
<{^_^}>
[nixpkgs] @globin pushed 2 commits to structured-attrs: https://git.io/JfmXw
<energizer>
ok if i do --keep-failed, looking in env-vars shows that PATH only includes the basic stdenv stuff plus python, and it's missing the programs that I passed buildInputs. aren't the buildInputs programs supposed to be on PATH?
<nschoe>
energizer: have you pasted your nix expression (default.nix) for this?
asbachb has quit [Ping timeout: 240 seconds]
<clever>
energizer: run `nix-shell` directly on the .drv that failed, then just `echo $PATH`
<esclear>
Huh, for some reason I can't start my pulseaudio user unit anymore, since it can't find pulseaudio.socket O_o
<esclear>
Although there is a pulseaudio.socket file right next to the pulseaudio.service unit file
linarcx has quit [Ping timeout: 240 seconds]
vykook has quit [Ping timeout: 244 seconds]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<{^_^}>
[nixpkgs] @Ericson2314 merged pull request #86254 → mkDerivation mesonFlags: Fix cross file logic to handle more cases → https://git.io/JfmPT
<{^_^}>
[nixpkgs] @Ericson2314 pushed 2 commits to staging: https://git.io/Jfm1t
<{^_^}>
[nixpkgs] @jonringer opened pull request #86259 → glibcCross: add arm entry for meson flags → https://git.io/Jfm1z
<energizer>
is there anybody doing little one-off packaging contracts? i like using nix but i dont have time to spend a hours figuring out how to package my deps every couple days, so maybe i can outsource that?
<{^_^}>
[nixpkgs] @grahamc pushed 2 commits to release-20.03: https://git.io/Jfm1j
<Kyndig>
could someone please explain to me how to install packages from a closure which I've just imported? All the documentation shows export and import, but then doesn't say how to actually get a user environment setup to reflect the imports...
vykook has joined #nixos
gustavderdrache has quit [Quit: Leaving.]
<infinisil>
Kyndig: Where do you want to install it/
<infinisil>
?
<Kyndig>
in my user profile
<{^_^}>
[nixpkgs] @robaca opened pull request #86261 → webcamoid: init at 8.7.1 → https://git.io/JfmMI
<Kyndig>
i want my 'nix-env -q' output to match that of the user that exported the closure
<Kyndig>
sorry forgot to mention the closure is all of the original user's ~/.nix-profile
<infinisil>
Well there is a way to install raw /nix/store paths with nix-env, but it's really not something you usually want to do
<Kyndig>
right, that's the problem
<infinisil>
And it only works with nix-env, not NixOS or other uses of Nix
<Kyndig>
it doesnt work
orivej has quit [Quit: orivej]
<infinisil>
What you should rather do is to write a nix expression that produces the environment you want, then anybody can install that however they like
<infinisil>
,declarative Kyndig
<{^_^}>
Kyndig: There are multiple ways of managing declarative profiles. 1) Attrset, compatible with imperative use of nix-env https://git.io/fAQHW ; 2) buildEnv, providing more control over the paths that are linked into the profile https://git.io/fp0aU ; 3) home-manager, providing nixos-like config for your ~ https://github.com/rycee/home-manager
<infinisil>
These are all ways of achieving this
<Kyndig>
i'm basically trying to do a nix-copy-closure, but I have no direct ssh connection
<Kyndig>
so I used nix-store --export --requisites ~/.nix-profile
<Kyndig>
and imported to destination
orivej has joined #nixos
<Kyndig>
so the store objects are there
<Kyndig>
but how to get my profile to match
infandum has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @flokli pushed 2 commits to staging-20.03: https://git.io/JfmM4
<infinisil>
As already said, this is not something one usually wants to do, in fact it's probably never the right thing to do
<infinisil>
Lemme think about this for a bit
<Kyndig>
people never want to copy their built store ?
<simpson>
Kyndig: Folks usually declare whatever they want to copy, then copy the declaration. If both boxes have Internet access, then this is usually easier than anything else.
<infinisil>
No I mean installing paths with nix-env
<infinisil>
Kyndig: So you can just do `nix-env -i /nix/store/...` I believe
<Kyndig>
yes
<Kyndig>
that does work
<Kyndig>
but
morgrimm has quit [Ping timeout: 244 seconds]
<infinisil>
Kyndig: What's the problem with doing it declaratively?
<Kyndig>
lets say I do a 'nix -q' on the original computer and my output is: tmux-3.0a
lukash_ is now known as lukash_|away
proofofk_ has quit [Remote host closed the connection]
proofofkeags has joined #nixos
<Kyndig>
then i do 'nix-store -q --references ~/.nix-profile' in order to see the store items that I could directly install with nix-env -i /nix/store.... , but what I see is:
<Kyndig>
nix/store/xy3nfgkc3dpc91ims3bf3zxkkgj12v1c-tmux-3.0a and nix/store/xy3nfgkc3dpc91ims3bf3zxkkgj12v1c-tmux-3.0a-man
orivej has quit [Ping timeout: 244 seconds]
<Kyndig>
so now on target machine a 'nix -q' shows the latter
<Kyndig>
not the former
vykook has joined #nixos
<Kyndig>
target machines sees two different packages
<Kyndig>
instead of the one tmux
<infinisil>
Looks like a problem you wouldn't have by doing it declaratively
<simpson>
Yeah, this is the sadness of nix-env. Package names are just not enough information for reconstitution. I think that a lot of folks have gotten used to the disappointment of going by package names and versions, because that's all most other package managers offer.
orivej has joined #nixos
<Kyndig>
i'm trying to compile on one machine and copy to another
<Kyndig>
to make declarative config on target machine it goes through the whole nix-defexpr process
<Kyndig>
i don't want that
<simpson>
Hm. So don't you want nix-copy-closure?
<Kyndig>
i believe i do want nix-copy-closure, but I don't have direct ssh
<Kyndig>
need two steps
<Kyndig>
export ; then import
<infinisil>
Kyndig: ~/.nix-defexpr? That won't have anything to do with that
<infinisil>
You mean that it would evaluate Nix?
<Kyndig>
let's just say I'm using a thumbdrive to transfer
<Kyndig>
I want it to use the store items that are built already for the install rather than doing a new query which references ~/.nix-defexpr
proofofkeags has quit [Ping timeout: 240 seconds]
<Kyndig>
i don't have a channel on the target machine
maddo has quit [Quit: See ya]
<infinisil>
Yeah that's no problem if you do it declaratively
<Kyndig>
even without a channel?
<simpson>
I'm searching through the manual, but there is a way to copy parts of the store to a thumb drive. The main obstacle is that Nix has a database tracking the store, and not just the store itself, so you've gotta update that database too. I don't know offhand how that's done.
<Kyndig>
no NIX_PATH
<simpson>
You can always copy the *whole* store, of course, but that's probably not what you want to do.
<infinisil>
Yes, NIX_PATH is just the default way to get a nixpkgs, but it's not the only way
<Kyndig>
right; but I mean I have no set of nixpkgs expressions
<Kyndig>
at all
<Kyndig>
on target machine
<Kyndig>
just want the bins
<infinisil>
Do you have Nix installed?
<infinisil>
Because that's all you need
<Kyndig>
yes just nix
<Kyndig>
but if I tried "nix-env -iA tmux" for example, it starts looking for .nix-defexpr
<Kyndig>
how do i point it do the store items
<Kyndig>
but intelligently so that it knows that tmux is actually 2 store items
Desetude has quit [Quit: WeeChat 2.7.1]
<Kyndig>
not just one
vykook has quit [Ping timeout: 260 seconds]
opticnerve has joined #nixos
<simpson>
You can give nix-env paths into the Nix store, and it'll do what you want it to do. $(nix-env -i /nix/store/...-tmux-...)
<Kyndig>
sorry; typing is not quickest way to explain
<Kyndig>
yes that's what I did at first nix-env -i $(cat list_of_references_of_original_nix-profile)
<simpson>
infinisil: So, I suspect that that thumb drive is important. I can't find anything on this; how *does* one copy some packages to a thumb drive, and add them offline to a store?
<Kyndig>
yep; that's the documentation i've been talking about
<Kyndig>
they all say to export and then import
<Kyndig>
but don't show how to get your user-env profile created
<Kyndig>
to match the original one
<infinisil>
Okay so there appears to be a way
<Kyndig>
I even tried manually creating a symlink in nix/var/nix/profiles/my-user
<simpson>
Well, how did you make your original one!?
<infinisil>
Again, I don't recommend it, but
proofofkeags has joined #nixos
<infinisil>
(wait, writing it out)
<Kyndig>
original one just using nix-env -iA for all packages
<Kyndig>
they all built
<Kyndig>
from source
<cole-h>
Why don't you have ssh access, aagin? If you have thumbdrive access, surely ssh access would be easier to set up...
<simpson>
Okay. That's not a reproducible action; if you do it again (say, after a channel update), you'll get a different result. Does that help explain why your current route cannot give you what you desire?
<Kyndig>
it's not really relevant didn't want to distract, but I'm building in ~/nix/...
<Kyndig>
don't have root on target system
<cole-h>
But you have Nix and the nix-daemon, OK, understandable.
<Kyndig>
not nix daemon just single user
<Kyndig>
same username
<Kyndig>
both sites
<Kyndig>
same prefix
<Kyndig>
so yes, I could simply copy all of ~/nix
<Kyndig>
and did for starters
<Kyndig>
but going foward
<Kyndig>
want to use store --export and --import for updates
<simpson>
Going forward, *declare* the packages that you want, using Nix expression language, and then you'll be able to *reproduce* the setup on the other machine.
<cole-h>
(^ That's kinda our shtick...)
pamplemousse has quit [Quit: Lost terminal]
<infinisil>
Yeah very much
<Kyndig>
you mean just do a "nix-env -iA packageA packageB" on the target machine right?
<cole-h>
g/nix-env/d
<Kyndig>
with a facy declarative config though
<{^_^}>
[nixpkgs] @flokli merged pull request #85772 → Use Open Watcom to build VirtualBox BIOS → https://git.io/JfkHL
<simpson>
No, nix-env actions are *not reproducible*, for the most part. The same text, typed on different machiens, will make nix-env do different things.
<Kyndig>
it still looks for nixpkgs expressisons
<Kyndig>
I don't have nixpkgs on target machine
<simpson>
The trick is precisely to avoid making nix-env do any lookups or hard work. Instead, nix-env will only install *precisely* what you declare in a single file.
civodul has quit [Quit: ERC (IRC client for Emacs 26.3)]
alp has quit [Ping timeout: 272 seconds]
<bqv>
generally a decent rule of thumb: if you're using nix-env, you're probably doing something wrong, or recovering from doing something wrong
<infinisil>
For reference, here's how you *could* transfer a profile to another machine (but again I don't recommend it): Get the profile path with `p=$(realpath /nix/var/nix/profiles/per-user/$USER/profile)`. Export its closure with `nix-store --export $(nix-store -qR $p) > out`. Import it on the other machine with `nix-store --import < out`, then set the profile with `nix-env -p
<cole-h>
Well, nixpkgs is really easy to get... It's just a `git clone github.com/nixos/nixpkgs ~/vcs/nixpkgs` away
<simpson>
You don't need a nixpkgs channel on the target. nixpkgs can be referenced inline from your single Nix file.
<simpson>
infinisil++ for answering the question as asked
<{^_^}>
infinisil's karma got increased to 271
<MichaelRaskin>
I would say that «doing something wrong» is about nix-env -iA (and even more so -i)
vykook has joined #nixos
<energizer>
must nix-env exist at all?
<MichaelRaskin>
-e is undoing, --set is a complicated story (and low-level)
<fresheyeball>
I like it
morgrimm has joined #nixos
<simpson>
It's a useful transitional crutch.
<Kyndig>
nix-env is for those who want imperative management i thought
o1lo01ol1o has quit [Remote host closed the connection]
<Kyndig>
I used declarative for system wide nixos, but use imperative for user level
<Kyndig>
i like it that way
<Kyndig>
system uses stable, user uses unstable
<cransom>
nix-env is a temporary bandaid for me where i'm exploring a tool temporarily and don't want to add it to anything permanent. just slightly longer lasting than a `nix-shell -p`
o1lo01ol1o has joined #nixos
<Kyndig>
but that's a side track
<cole-h>
I use imperative for user level, and much prefer it, precisely because all I need to do is copy a few files, `home-manager switch` and bang
<cransom>
if, as a user, all you are doing is nix-env, you are making life harder than it needs to be. you might as well be doing homebrew/etc
<MichaelRaskin>
For undecided I use a nix-shell-like tool
<energizer>
it would be be better to `nix-install() { echo "$@" >> ~/.config/nixpkgs/my-packages ;}` and have nix import from that file
<bqv>
i genuinely only reach for nix-env when i want to pin a package so it doesn't accidentally get gc'd away
<bqv>
other than that, declarative or nix-shell
<Kyndig>
apparently there's something more about a user's environment than just the store, that needs to match because after just importing the store paths directly, I got errors about missing nix/var/share paths
<Kyndig>
that were not needed for the nix package itself
<{^_^}>
[nixpkgs] @worldofpeace pushed 6 commits to master: https://git.io/JfmDV
CMCDragonkai is now known as Guest55208
<simpson>
Kyndig: That's the database that I was trying to tell you about. If you use the NAR path, and export and import along the documented lines, then you'll not have that problem.
niksnut has quit [Remote host closed the connection]
zupo has joined #nixos
<simpson>
bqv: Note that Kyndig does want to only perform the compilation on one of two machines, so a copy-closure, export, or cache would be required, and their setup precludes copy-closure and caches.
trubi has joined #nixos
<bqv>
that can still be done declaratively, with build machines
<bqv>
my first question is why, my second question is can this be solved with a bind mount? but feel free to ignore me cause i'm just sating my curiosity at this point
<MichaelRaskin>
Not using /nix does not preclude you from running a local nix-serve cache, BTW
<simpson>
Right, but it does not make the problem simpler, since now a closure has to be copied to the cache.
<Kyndig>
the rabbit hole grows
<Kyndig>
i would 've loved to do proot
<Kyndig>
but the kernel doesn't allow user namespaces
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/JfmyY
<bqv>
simpson: how exactly?
<bqv>
using substituters is just an automatic method of doing that exact copy-closure
<bqv>
if anything it's an oversimplification
<bqv>
i think you're confusing build substitution for cachix
<bqv>
they are not the same thing
<simpson>
bqv: In the sense that the problem's already well-described enough that the solution is $(nix-store --export) to thumb drive, $(nix-store --import) on the target, $(nix-env -i) the desired packages or use infinisil's trick.
<bqv>
i can understand that that might be simpler if this is a one-time issue and you won't be sharing packages between these machines again
<bqv>
otherwise i don't see how
est31 has quit [Remote host closed the connection]
<simpson>
Okay, and how does it not work?
<simpson>
You'll need to be more specific, now that we know which commands you're running and what the expectations are.
<Kyndig>
ok; So machine A is the builder B is the target which has no nixpkgs tree
user_0x58 has quit [Ping timeout: 240 seconds]
erasmas has quit [Quit: leaving]
<infinisil>
commas!
<Kyndig>
B does have a simple ~/nix bootstrap with nix installed
<simpson>
You mean that it has no channels configured? Or you mean that somehow there's no nixpkgs.tar.gz at all in the Nix store and there can't be?
est31 has joined #nixos
user_0x58 has joined #nixos
<Kyndig>
no nixpkgs.tar
<bqv>
Kyndig: does it have network access?
<Kyndig>
so I can do nix-store --export on machine A, transfer via thumbdrive to machine B and nix-store --import
<Kyndig>
then nix-store -i ~/nix/store/path/to/a/store/item
<Kyndig>
that works
<infinisil>
nix-env?
<Kyndig>
so i figured I need a list of all those store items
<bqv>
this is going to give me an anyeurism
<Kyndig>
so back on machine A, I do: nix-env -q --references ~/.nix-profile
<Kyndig>
gives me a list
<infinisil>
Kyndig: Didn't you take my advice earlier?
vykook has quit [Ping timeout: 264 seconds]
philr_ has joined #nixos
<Kyndig>
what advice about the exporting per-user profile?
<Kyndig>
was getting to that
<infinisil>
Ah
<Kyndig>
i did essentially the same thing by finding where my Machine A profile pointed in the store and creating the same symlink on machin B
<Kyndig>
the user-env link
<infinisil>
Sorry I was confused because you switched to using ~/.nix-profile and --references
gustavderdrache has joined #nixos
<{^_^}>
[nixpkgs] @zowoq opened pull request #86266 → minikube, docker-machine-{hyperkit,kvm2}: use makefile → https://git.io/Jfmyz
bhipple has joined #nixos
jbrock has quit [Quit: jbrock]
<Kyndig>
sorry; I really should just so a little write up an post on ix.
jbrock has joined #nixos
jbrock has quit [Remote host closed the connection]
jbrock has joined #nixos
<infinisil>
I'm still waiting for the actual problem!
<bqv>
i'm sure there is a better solution to whatever you're trying to achieve, including the things you've already done, using declarative files and nix-serve
<bqv>
but it is incredibly hard to tell without you explaining from first principles, what you're trying to a chieve
wolfshappen has quit [Ping timeout: 260 seconds]
<bqv>
and i understand if that's nosey but we won't be able to un-overcomplicate this otherwise
<simpson>
bqv: It's not hard to imagine, if one has worked in a datacenter before: Kyndig is trying to get their preferred set of tools onto an airgapped/LAN-only machine in the NOC, and the admin doesn't know/trust Nix.
trubi has quit [Read error: Connection reset by peer]
<Kyndig>
i'm not worried about privacy; just didn't want to distract from issue; ultimately, after getting correct profile symlink pointing to imported store environment, nix-env then gave me errors about non-existen dirs under ~/nix, but outside of ~/nix/store.
<Kyndig>
i'm working on a writup
<infinisil>
Kyndig: And the error is..?
wolfshappen has joined #nixos
trubi has joined #nixos
<infinisil>
And what command did you run specifically?
est31 has quit [Remote host closed the connection]
<Kyndig>
that's what I'm putting together
<Kyndig>
starting from scratch and dcoumenting
jbrock has quit [Quit: jbrock]
<infinisil>
Oh, well I can already tell you now that if your nix store is at /nix/store on the source machine, the target one has to have the same nix store path
jbrock has joined #nixos
<infinisil>
Otherwise you can't reuse builds from the source machine
est31 has joined #nixos
<Kyndig>
the errors are gone now, because i copied the entire ~nix dir
<Fare>
What need I put in my nix recipe to use `sysctl(1)` on a mac?
o1lo01ol1o has joined #nixos
<cole-h>
What problem are you trying to solve with sysctl?
<Fare>
finding the number of CPUs. But I suppose I can try to override this with the number of cores that the recipe is supposed to use... how do I get that number from my recipe?
eoli3n__ has quit [Ping timeout: 246 seconds]
<cole-h>
,locate bin sysctl
<{^_^}>
Found in packages: procps, sysctl, toybox, busybox, unixtools.procps
<cole-h>
Fare: What is a "recipe" in this context?
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
bhipple has quit [Ping timeout: 260 seconds]
<energizer>
iirc that's the conda term for what derivations
infinisil is now known as infi
<energizer>
s/what/what nix calls/
freeman42x has joined #nixos
fenedor has joined #nixos
vykook has joined #nixos
<{^_^}>
[nixpkgs] @mweinelt opened pull request #86268 → coturn: apply patch for CVE-2020-6061/6062 → https://git.io/JfmSO
<{^_^}>
[nixpkgs] @mweinelt opened pull request #86270 → [20.03] coturn: apply patch for CVE-2020-6061/6062 → https://git.io/JfmSl
andymandias has joined #nixos
<Fare>
simpson, bingo. That's what git grep buildCores suggested, too
vykook has quit [Ping timeout: 260 seconds]
justanotheruser has quit [Ping timeout: 246 seconds]
zeta_0 has joined #nixos
<jumper149>
Hi, I want to install libreoffice in my nix-env, but it started compiling instead of just downloading from cache.nixos.org and then it failed miserably. I checked https://hydra.nixos.org/job/nixpkgs/trunk/libreoffice.x86_64-linux and it seems it's not building on the git commit I'm on. So my question: How can I still install it? Do I need to install it from another channel?
<Fare>
(thanks!)
<Fare>
jumper149, take it from a slightly more stable channel?
trubi has quit [Read error: Connection reset by peer]
<infinisil>
jumper149: As a simple fix you can do `nix-env -f channel:nixos-20.03 -iA libreoffice` to install from the stable channel, as Fare suggested
<zeta_0>
there are no syntax errors in that shell.nix file, so i'm not sure why that terminal error is being thrown
<cole-h>
I was gonna suggest `nix-store -r <hydra's successful path> && nix-shell -p <hydra's successful path>`, but that works too
<{^_^}>
[nixpkgs] @mweinelt opened pull request #86271 → [19.09] coturn: apply patch for CVE-2020-6061/6062 → https://git.io/JfmSV
bhipple has joined #nixos
aveltras has quit [Quit: Connection closed for inactivity]
<{^_^}>
[nixpkgs] @flokli merged pull request #85998 → nixos/nsswitch: Make databases more configurable → https://git.io/JftkV
<jumper149>
Ok I will go with using another channel.
<jumper149>
cole-h: What would be that path?
<bqv>
check hydra.nixos.org
<zeta_0>
it's telling me: `value is a set while a list was expected`, i copied it directly from home.nix to shell, so i'm guessing i need to change it a bit so `buildInputs` can recognize it