<freeman42x[m]>
it's not, not only it does not replace the value of `$nix_shell_info` in/outise of nix-shell but there is also no prompt change when entering a `nix-shell`
<zeta_0>
stack-to-nix generated a default.nix that is throwing an error when i try to g into a nix-shell: error: attempt to call something which is not a function but a set, at /home/zeta/TaskLite/default.nix:5:10 , https://dpaste.org/42E4
<zeta_0>
i'm trying to use stack-to-nix in a stack project of a git repo that i cloned, but i guess the nix-tools have a couple of rough edges
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/Jff9w
<morgrimm>
Does anyone know if ndrp's flake config repo is supposed to be a template repo on GH?
<jtojnar>
freeman42x I use --run fish (when not using direnv)
<morgrimm>
nrdxp*'s repo
<qy[m]1>
that's why the brnach is called "template" :p
drakonis1 has quit [Quit: WeeChat 2.8]
drakonis_ has quit [Ping timeout: 265 seconds]
<morgrimm>
qy[m]1 That's what I figured :P But shouldn't there be a 'use template' button or something near the finder to actually use it?
<morgrimm>
Do I need to opt into a feature preview for it or something?
<qy[m]1>
Jan Tojnar: you should check out any-nix-shell. then you can just use fish for your nix-shells
<qy[m]1>
everyone seemed to think that was impossible but it was already done :p
<cole-h>
morgrimm: It's not a GitHub template.
<qy[m]1>
morgrimm: it's not a github template, just clone/fork it and customize it
<gchristensen>
it is impossible and preserve the semantics / behavior of nix-shell
thc202 has quit [Ping timeout: 252 seconds]
<morgrimm>
Ahhh
<morgrimm>
Yup, fair enough lol
<jtojnar>
freeman42x the fish wiki page is not well written
<morgrimm>
Do you just rebase your template branch on his to keep up to date? and re-integrate with live?
<zeta_0>
that sucks that the default.nix that stack-to-nix generates does not work, what a bummer
<gchristensen>
zeta_0: sounds like you're trying to vampire your way to some support here
CrazyMind102 has joined #nixos
<zeta_0>
gchristensen: no, i never directly asked anyone for help, just seeing if anyone has ran into the same problems with stack-to-nix, and i tried to paste bin as much info as i could, for if someone wanted to take a look
pamplemousse has quit [Ping timeout: 250 seconds]
peelz has joined #nixos
Wulfsta has joined #nixos
<zeta_0>
gchristensen: sometimes when i post my questions on reddit, it takes a long time for people to respond, what other question and answer websites would you recommend for anything nix related?
<cole-h>
jtojnar: Heh, single quotes prevents shell expansion (re your answer to the fish nix-shell prompt)
<cole-h>
freeman42x[m]: If you change the single quotes to double quotes in `echo -n -s ' ...'`, it should work as you want it to
<cole-h>
zeta_0: Discourse
drakonis_ has joined #nixos
<Wulfsta>
Can someone explain the situation with neovim and vim-plug? I wanted to pull my config over from Arch but can't seem to figure out how NixOS's vim plugin system works
<DigitalKiwi>
cole-h: yeah that's what we do here is have discourse
<jtojnar>
not sure what to do about the wiki page, since fish_prompt.fish is usually pretty gnarly so just saying to add the snippet is not very helpful
<zeta_0>
so how does discourse compare to reddit and stackoverflow and others?
jkachmar has joined #nixos
<cole-h>
That's if you want to host it yourself AFAICT
<zeta_0>
oh i see
reallymemorable has quit [Quit: reallymemorable]
<jbox>
How can I do something like systemd.services.minecraft@test = {}; ?
<jbox>
Nix doesn't like the @
<gchristensen>
."minecraft@test" = { ...}
<jbox>
DUH thank you lol :)
<gchristensen>
no worries, it isn't so obvious sometimes :)
Wulfsta has quit [Ping timeout: 240 seconds]
Wulfsta has joined #nixos
reallymemorable has joined #nixos
markus1199 has joined #nixos
markus1189 has quit [Ping timeout: 258 seconds]
<zeta_0>
i've been using nix/nixos for about 6 months now, when will i get to the point where nix starts to make more sense? how long did it take all of you to get proficient at nix?
<zeta_0>
well, i guess i should ask this on reddit instead
<cole-h>
(Or Discourse...)
<zeta_0>
ya, that too
<DigitalKiwi>
have you read nix-1p? nix pills?
<atemu12[m]>
zeta_0: Took a few months for it to make sense, which parts don't in your case?
<DigitalKiwi>
i got a lot better at it by packaging stuff that didn't exist or updating or fixing broken stuff in NixOS/nixpkgs
<cole-h>
^
<cole-h>
I've been at it since February and I think I'm reasonably competent in Nix-as-a-packaging-language
<gchristensen>
and debugging stuff, searching code, experimenting, learning
<zeta_0>
i read the nix-pills a couple of months ago, but only some of it stuck, ya i think i'm eventually going to have to start packaging stuff for it to make more sense
Henson has joined #nixos
<Henson>
is there a way I can build multiple Haskell executables in a cabal file with Nix? I'm using the haskell mkDerivation function.
<{^_^}>
[nixpkgs] @worldofpeace pushed to treewide-broken-20.03 « treewide: mark broken wip »: https://git.io/JffHl
reallymemorable has quit [Quit: reallymemorable]
<zeta_0>
i just joined this open source project, packaging it as a nix pkg should be a good way to learn nix better: https://github.com/ad-si/tasklite
<zeta_0>
[zeta starts digging through the nix documentation]
<gchristensen>
I love the logo
reallymemorable has joined #nixos
matthuszagh has joined #nixos
<matthuszagh>
hello. I'm getting localhost connection errors when attempting to use syncthing
<matthuszagh>
I've configured the service to run from my user (not root)
<matthuszagh>
I get a curl: connection refused error. when i run sudo netstat --tnlp it seems that the local server is never created
<matthuszagh>
when i run this as root i don't get any such error
<matthuszagh>
has anyone else experienced this?
<atemu12[m]>
zeta_0: Yeah, definitely start working with Nixpkgs if you want to learn Nix; practical experience is the best way IMO.
<{^_^}>
[nixpkgs] @worldofpeace pushed 3 commits to master: https://git.io/JffQl
nwspk has quit [Quit: nwspk]
nwspk has joined #nixos
<energizer>
is there a shorter way to write this? `nix-build -E '(import <nixpkgs> {}).callPackage ./. {}'`?
mexisme has quit [Quit: WeeChat 2.7.1]
aw has quit [Quit: Quitting.]
spacefrogg has quit [Quit: Gone.]
bqv has quit [Quit: WeeChat 2.8]
aw has joined #nixos
Havvy has quit [Read error: Connection reset by peer]
Kyndig has joined #nixos
spacefrogg has joined #nixos
<Kyndig>
does anyone know if there is an equivalent of gentoo's KERNEL_DIR for nix?
<{^_^}>
[nixpkgs] @lopsided98 opened pull request #85418 → nixos/grub: support initrd secrets → https://git.io/JffQb
Havvy has joined #nixos
<drakonis_>
the kernel is in the store
<Kyndig>
well, a kernel is
<Kyndig>
not the one i'm running on
<Kyndig>
for user build in homedir
teto has quit [Ping timeout: 240 seconds]
<drakonis_>
wait what?
<Kyndig>
i suspect something isn't working right because packages may need to know about my local host's kernel config
<drakonis_>
uh
<drakonis_>
what are you doing?
<Kyndig>
I'm running nix in my homedir on an old red hat machine that I don't own
<Kyndig>
don't control the kernel
morgrimm has joined #nixos
<drakonis_>
what are you trying to build?
<Kyndig>
so I'm wondering if there's something I can set in the nix env to point to the kernel source dir
<drakonis_>
and why would they need to know about your kernel's config?
<Kyndig>
for packages to use
<drakonis_>
they don't consume that for builds?
<Kyndig>
for the same reasons as KERNEL_DIR exists for gentoo
<Kyndig>
any packages that need to know kernel config I assume
<Kyndig>
the running kernel's config
<drakonis_>
there aren't any that do?
<Kyndig>
there are in gentoo
<Kyndig>
I guess what I'm asking is do any nixpkgs look at the kernel that's built by nix for any info?
<Kyndig>
if so, then I need to make them look at my actual runtime kernel instead
<simpson>
Which packages are you worried about? When Nix isn't managing the bootloader, this isn't a worry; I think that we normally push kernel feature detection to runtime, while pushing kernel module requirements to NixOS configuration.
<Kyndig>
yeah; i don't think it's just for management
<Kyndig>
it's for compile parameters
blaggacao has joined #nixos
morgrimm has quit [Ping timeout: 264 seconds]
<Kyndig>
maybe i should find a good example from gentoo
<energizer>
is it possible to get the nixos-generate-config before nixos is running?
reallymemorable has quit [Quit: reallymemorable]
<energizer>
i just want to generate the hardware-configuration.nix
<Kyndig>
wireguard is an exmaple
<hyper_ch>
Kyndig: until 5.6
<Kyndig>
right
<simpson>
Kyndig: The package named `wireguard-tools` can be built and installed regardless of whether the running kernel supports it.
<drakonis_>
no and why would it need?
<Kyndig>
I can't say I'm familiar with all the reasons that KERNEL_DIR exists; but I just figured if I'm building everything from scratch, just like gentoo, then an equivalent parameter may need to be set
<drakonis_>
but why?
<hyper_ch>
wireguard <3
<Kyndig>
not for wireguard
<Kyndig>
that was just an example of a package which used it
<Kyndig>
at one point
<Kyndig>
my gentoo install is old
<simpson>
Kyndig: As a hint, I tested out not just `wireguard-tools`, but /nix/store/7l4j4671x81k7w520mqj5m1sgr7pij6h-wireguard-tools-0.0.20190123, and I didn't build it myself, but summoned it from cache.
<simpson>
Nix isn't Portage; it's a fair amount smarter about what's in a package, and about how packages are built. This does lead to some things that Gentoo might consider not great, because some optional features are enabled at compile-time, but the tradeoff is that packages decompose along the same lines as their dynamic linkage, including along kernel interfaces!
<{^_^}>
[nixpkgs] @jonringer pushed commit from @cgevans to master « ipe: fix Qt wrapping »: https://git.io/Jff7D
<Kyndig>
ahh; that's easy, because this old kernel does not support user namespaces, henc have to recompile
<Kyndig>
bootstrapped
<Kyndig>
with my home dir as prefix
<drakonis_>
right
<Kyndig>
~/nix
<drakonis_>
but uh
<Kyndig>
~/nix/store
<drakonis_>
you have to boot onto the kernel to do that
<peelz>
cole-h: having `wireshark` in a let statement at the top is really sneaky. I'm used to seeing `with pkgs; [ wireshark ]` for example... lol
<drakonis_>
to have access to those thigns
<peelz>
cole-h: I would've prefered `[ cfg.package ]` here... /shrug
<Kyndig>
i'm not following
<Kyndig>
i'm in a running kernel
<cole-h>
peelz: A fair complaint. But it's not that sneaky when the file is only 42 lines long :P
<drakonis_>
uh
<drakonis_>
you have to recompile everything because you're changing the store location
<peelz>
cole-h: true lol... sneaky at a first glance though :)
<simpson>
drakonis_: Kyndig doesn't own the machine that they're working on.
<Kyndig>
right
<drakonis_>
aight
<simpson>
But I guess that you've got permission to spend a lot of CPU rebuilding everything~
<Kyndig>
the more standard option is to use proot or the builtin version of it for nix
<Kyndig>
but I can't due to the old kernel with no user namespace
<drakonis_>
i've done that, yes.
xO1 has joined #nixos
Supersonic has quit [Disconnected by services]
Supersonic112 has joined #nixos
Supersonic112 is now known as Supersonic
kleisli_ has joined #nixos
<Kyndig>
i did the same with gentoo, setting KERNEL_DIR=/usr/src/kernels/blah and I did notice some parts of the build process looking at that dir
<Kyndig>
my gentoo zsh works fine; but I'm having issues with my nix-compiled one
<Kyndig>
was just making a wild guess
<Kyndig>
that the KERNEL_DIR was relevant
troydm has joined #nixos
cjpbirkbeck has quit [Quit: cjpbirkbeck]
<simpson>
Aha, the actual problem. What sorts of issues?
<jbox>
can I set a umask for a certain user? I have a system user account and would like the files it creates to be g+rw
waleee-cl has quit [Quit: Connection closed for inactivity]
morgrimm has joined #nixos
morgrimm has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @bcdarwin opened pull request #85420 → python3Packages.neo: init at 0.8.0 → https://git.io/Jff5k
<Kyndig>
random core dumps and ulimit reports all zeros
<Kyndig>
I did enable debug info in my zsh overlay and looked looked at various core files
<Kyndig>
seemed to be inconsistent where it failed
<Kyndig>
always something to do with memory though
<hyper_ch>
is it possible to install nixos with bios and uefi at the same time? Like the live isos can boot up from either?
orivej has joined #nixos
<Kyndig>
i also have a number of doCheck=false overrides due to not wanting to spend the time trying to figure out why various checks fail, but now I'm commenting those out and rebuilding.
<Kyndig>
i'll try to deal with them as they may be the cause of my zsh issues
<samueldr>
hyper_ch: it should be possible, but I don't think NixOS is setup to handle updating both sides of the equation at the same time as it is
<samueldr>
for best results you'd have to use grub, which can use the same configuration file to load the menu
<samueldr>
so at least the menu would be updated
<samueldr>
but the bootloader itself I don't think both can be updated as it is
<hyper_ch>
samueldr: well, I don't need updating... al I need it create base system and then be able to dd stuff
<samueldr>
for the iso we're not relying on the nixos modules, and it actually is an issue, it should probably rely on the modules so one can reproduce the same results on an installed system
<hyper_ch>
it's for a friend's small company. He wants to encrypt now all notebooks (7 of them)
<hyper_ch>
so I'd just thought I could have a 5TB external usb
<hyper_ch>
add nixos to it with both uefi and bios
<hyper_ch>
and just dd the harddisks onto the usb drive
<hyper_ch>
so in case something goes wrong with the system
<hyper_ch>
you can just boot up the xternal usb again and write it back
<hyper_ch>
but it's a mix of bios and uefi notebooks
<samueldr>
in that case, you could probably set it up for UEFI, and then rebuild for MBR afterwards using grub, ensuring you have a bios boot partition in your GPT for mbr grub
<samueldr>
without testing and checking, I thinkg it should work
<samueldr>
I don't know exactly, but the way I would try is to do a normal install on the usb drive, as UEFI, making sure to have a bios boot partition, an ESP partition, a /boot, the nixos partition
<hyper_ch>
esp and bios boot are different?
<samueldr>
yes
<samueldr>
the bios boot partition is used by grub on GPT disks to embed the bigger parts of grub, on MBR it (ab)uses the "empty" space between the partition table and the first partition
<samueldr>
there is no such "empty" space on a GPT formatted disk
<hyper_ch>
ah ok
<samueldr>
that would be the most tricky part of the setup I think, again without having tested it
<hyper_ch>
I think I see what you're doing
<samueldr>
and now you know about it
<hyper_ch>
beause uefi would be able to locate the esp
<samueldr>
yes, and it loads the grub program in the ESP
<samueldr>
while a legacy bios boot would read it from the protective mbr of the GPT
<hyper_ch>
and if it's bios, then it would ignore the esb and just but legacy
<samueldr>
yep
<hyper_ch>
awesome :) I'll give it a try
<samueldr>
good luck :)
<hyper_ch>
I mean that's what weekends are for, right?
<samueldr>
oh, and hyper_ch, --install-bootloader, it's a flag for nixos-rebuild
<samueldr>
not sure you will need it, but it may end up being useful
<hyper_ch>
:)
<hyper_ch>
samueldr: and pondering to add some bash script with newt) that will make it easy for him to take new images or restore them
<samueldr>
now that's not related to the boot process :)
<freeman42x[m]>
samueldr: under 20.03 Beta NixOS Alt + F2 shortcut is broken even though it is set https://i.imgur.com/pyOtGs7.png
<freeman42x[m]>
I would like to know what I could do to help fix this issue, and where the fix would be applied. Sounds like fix would be in nixpkgs probably
zeta_0 has quit [Quit: rcirc on GNU Emacs 26.3]
<samueldr>
all the work for NixOS indeed is contained in the nixpkgs repo
<samueldr>
though, sorry, I don't know what can be done to fix this
<freeman42x[m]>
samueldr: should I open an Issue on nixpkgs GitHub? I could not find anyone encountering this yet
<freeman42x[m]>
*anyone else
<samueldr>
it is an option, yes
<freeman42x[m]>
I know it's an option, asking for your opinion whether I should or not
<samueldr>
sorry, I have no strong opinion, but in case of doubt, I'd tend to go with opening an issue to at least have a trace for the next people having the issue
Kyndig has quit [Remote host closed the connection]
geekthattweaks has joined #nixos
<energizer>
why did nixos get pulled into the nixpkgs repo?
palo1 has joined #nixos
<hyper_ch>
nixos get pulled into the nixpkgs repo?
<sphalerite>
You'll still have a newer version in your store probably
<sphalerite>
unless you removed it from your config at some point
<peelz>
sphalerite: I don't think the version's changed. Lemme check the hashes
<srhb>
peelz: They don't need to, any dependency of the drivers count as well.
<sphalerite>
nix-collect-garbage --delete-old deletes older generations of your profile, then gc's the store — i.e. deletes anything that isn't in use by a profile or running process anytmore.
<srhb>
peelz: (eg. say libc got bumped, the drivers will change)
<peelz>
but the nvidia drivers are still referenced in my active system
<peelz>
oh
<peelz>
srhb: gotcha
<peelz>
srhb: why does it need to redownload binary blobs though??
<srhb>
It's all transitive, which is why nixpkgs packages change so much more often than other distros
<srhb>
peelz: You probably garbage collected the source because nothing was referencing it
<srhb>
peelz: (Your system likely does not depend on them at runtime)
<peelz>
hmm
<peelz>
that's annoying
<srhb>
peelz: You can add the src to your systemPackages if you like.
<srhb>
It's not like you don't have control of this :)
<peelz>
srhb: true, but I'm not sure how to do that for binary blobs :P
<peelz>
srhb: that seems a bit hacky... am I wrong?
<srhb>
peelz: Why does it seem hacky to you?
<sphalerite>
it also doesn't redownload them unless you've gc'd between updates.
<peelz>
right
<sphalerite>
Just re-unpacks, patches and builds
<srhb>
peelz: The linux package set is configurable, so the source _might_ be different for different versions of it. Therefore, it's good to refer to the dynamic one (config.linuxPackages) instead of hardcoding pkgs.linuxPackages.nvidia_x11? :)
<srhb>
peelz: Or at least that's my argument
<peelz>
srhb: I guess it's the only way. Is there no side effect to putting the src into my system packages?
<srhb>
peelz: It's not the only way but it's an easy way.
<srhb>
peelz: Your system just has to refer to the path in _some_ way, it doesn't matter how. :)
<sphalerite>
config.boot.linuxPackages* :)
<srhb>
sphalerite: Thanks :)
<sphalerite>
err wait
<srhb>
boot.kernelPackages :P
<sphalerite>
config.boot.kernelPackages?
<sphalerite>
hehe
<srhb>
peelz: Consider that /run/current-system "looks like a normal distro root partition" if you squint really hard
<srhb>
peelz: On these, it would be relatively normal to have sources lying around.
<peelz>
srhb: HMMMMM haha *squints harder*
<sphalerite>
well /run/current-system/sw even mroe :P
<srhb>
yeah :P
<peelz>
what does sw stand for?
<srhb>
software I guess?
<energizer>
systemwide
<peelz>
ahh
<srhb>
sandwich!
<srhb>
Never thought about it :P
<energizer>
(i think)
<peelz>
srhb: where would the sources end up?
<sphalerite>
might not actually work since it's a file, so it can't really be linked in?
<sphalerite>
system.extraDependencies might be more suitable
<srhb>
peelz: Well, you can decide that for yourself. If you wanted them in eg. share, you'd make a tiny derivation (runCommand is probably easy) to create an $out/share/nvidia-drivers that links to the source
<srhb>
Or that ^ :)
<peelz>
oh huh
<sphalerite>
for creating symlinks, I'm a big fan of linkFarm myself :D
<srhb>
Definitely.
<srhb>
peelz: However, I too think you should embrace "gc less often or never" ;-)
<srhb>
Much nicer experience.
<peelz>
oh?
<peelz>
but then whenever I review PRs or stuff that's unrelated to my system, I end up with a bunch of junk that I'm never gonna use
<srhb>
Well, only if you desperately need the space and already have reclaimed what you can via store optimization etc.
chagra has joined #nixos
<srhb>
peelz: Yes, but with gc you might end up with the same junk more than twice. Without gc, you will only end up with it twice.
<srhb>
peelz: What I'm saying is: It usually makes sense to optimize for redownloads and writes as long as you have enough space. Why care what /nix/store has lying around that is completely isolated from you? :)
<peelz>
srhb: I don't follow
<srhb>
It's only a pet peeve of mine, if you like garbage collecting, don't let me stop you. ;-)
<srhb>
I can see that it has a certain undefinable cathartic element.
<peelz>
hmm fair enough
<peelz>
btw "A list of packages that should be included in the system closure but not otherwise made available to users. This is primarily used by the installation tests."
<MichaelRaskin>
But generally Nix zeitgeist seems to be «we spend storage to save on thinking about irrelevant things»
<peelz>
system.extraDependencies sounds like exactly what I need :D
chagra_ has quit [Ping timeout: 258 seconds]
<peelz>
<3 sphalerite srhb
<{^_^}>
sphalerite's karma got increased to 83
<peelz>
<3 srhb
<{^_^}>
srhb's karma got increased to 102
morgrimm has joined #nixos
<sphalerite>
yeah my free space generally bounces between 100G and 10G every few weeks as I accumulate stuff and gc
<srhb>
Yeah, same. My work laptop does not have as much disk space as I'd like though. But hooray for auto-optimize.
alexherbo2 has joined #nixos
<srhb>
(But yeah, I certainly wouldn't mind if our closure were smaller by default, but meh)
drakonis_ has joined #nixos
<peelz>
srhb: `services.xserver.videoDrivers = [ "nvidia" ]` is the only nvidia-related thing in my config. How can I find out what derivations this option is pulling in?
<srhb>
peelz: I'd just cd into /nixos/modules in a checkout and grep for videoDrivers and see how it treats that string.
<srhb>
peelz: morally it says: if elem "nvidia" config.services.xserver.videoDrivers then kernelPackages.nvidia_x11
drakonis has quit [Ping timeout: 256 seconds]
drakonis1 has joined #nixos
ent has joined #nixos
<srhb>
And as said earlier, your kernelPackages are configured via boot.kernelPackages, so you can refer to it with: config.boot.kernelPackages.nvidia_x11
<nixuser>
srhb: I am using hydra on a debian VM. hydra fetches a project from a private gitlab repository. The first fetch of the project threw an error "nix hydra memory allocation of x bytes failed". jobset failed due to signal 6 (aborting)
<hyper_ch>
but rsync is better if you already have a batch of files on the receiving end
<hyper_ch>
(or one might just use zfs)
<nixuser>
Afterwards the error "failed due to signal 11 (Segmentation fault)" is thrown for every retry
zacts has quit [Quit: WeeChat 2.7.1]
<srhb>
peelz: I see you're a fan of lisp :P
<peelz>
srhb: I've actually never done any lisp, but that's exactly what I was thinking haha (it looks like lisp!)
<hyper_ch>
jbox: i'd use rsync over scp... can't you install rsync on the remote server?
<etu>
peelz: That could almost work in guix I guess ;)
<srhb>
I so want an $ operator.
<peelz>
^ that would be great
<srhb>
I just end up making useless names in let bindings instead.
<jbox>
hyper_ch: I just did that. Everything's working fine now :)
<peelz>
etu: I didn't actually know about guix until I switched to nix. Is guix basicaly nix but in lisp?
<hyper_ch>
jbox: :)
<hyper_ch>
there's a difference if you use . or ./ - just use what you prefer
jmeredith has quit [Quit: Connection closed for inactivity]
_viz_ has joined #nixos
stteevveen has joined #nixos
<stteevveen>
Hi! it is not clear for me the difference between nativeBuildInputs vs buildInputs . When do you choose to use nativeBuildINputs ?
<symphorien>
for things which are executed during build
<symphorien>
as opposed to things which are executed after installation
<symphorien>
the distinction is only sensible for cross compilation
<symphorien>
cmake and pkg-config go to nativeBuildInputs whereas usually libfoo goes to buildInputs
<stteevveen>
symphorien thanks, that makes sense.
<siers>
what does native mean there?
<Profpatsch>
betaboon: oh wow, this is great!
vidbina_ has joined #nixos
<stteevveen>
@siers after symphorien explanation I would say: native as in "used during a nix build" . anyone confirm ?
<betaboon>
Profpatsch: i wrote some reasoning in the issue. i think this is the right way to go about it
<Profpatsch>
betaboon: Now I wonder how I should test it.
<Profpatsch>
I don’t want to destroy our staging instance :)
<Profpatsch>
Well, I guess I might as well.
<Profpatsch>
taint the ec2 instance, reapply
<Profpatsch>
see if it still hangs
<symphorien>
siers: native is the opposite of cross iirc. in autoconf parlance
oida has quit [Quit: byez]
<symphorien>
because these inputs are not cross compiled
<betaboon>
Profpatsch: i dont think it will destroy it. if i understand it correctly, it would just create a new ssh-connection when there is no multiplex-master-process
klntsky has quit [Remote host closed the connection]
<siers>
symphorien, ah! I bet most people have no idea what that means, so no one can remember
<siers>
I think I will now
<siers>
but wouldn't something like compileBuildInputs/runtimeBuildInputs make more sense?
<symphorien>
ahah there is a more complex and more precise taxonomy
<symphorien>
depsBuildBuild, depsBuildHost etc.
<siers>
or that
<{^_^}>
[nixpkgs] @symphorien opened pull request #85436 → paperwork: unmark as broken → https://git.io/JfJTO
oida has joined #nixos
gulplante has quit [Ping timeout: 265 seconds]
morgrimm has joined #nixos
c0c0 has quit [Quit: WeeChat 2.6]
morgrimm has quit [Ping timeout: 256 seconds]
<{^_^}>
[nixpkgs] @Mic92 pushed to master « python38.pkgs.python-jsonrpc-server: disable for python38 »: https://git.io/JfJkO
<vika_nezrimaya>
I just deleted my nixops files
<vika_nezrimaya>
they weren't in git sadly
reallymemorable has joined #nixos
<vika_nezrimaya>
I deleted like the whole directory with all the expressions
<{^_^}>
[nixpkgs] @Mic92 pushed to release-20.03 « python38.pkgs.python-jsonrpc-server: disable for python38 »: https://git.io/JfJku
<vika_nezrimaya>
good thing I updated my RPi recently
<vika_nezrimaya>
lol
arjen-jonathan has quit [Ping timeout: 272 seconds]
<stteevveen>
Trying to do my first nix derivation for c/c++. My result directory is empty after I use nix-build. What should I add in the Makefile install rule ? I've tried cp myApp out and cp myApp result , but that doesnt work
<vika_nezrimaya>
Assuming a standard GNU build system, your PREFIX should be equal to $out when you call make and make install
<vika_nezrimaya>
and you should install it to $PREFIX/bin/
<vika_nezrimaya>
that way it'll work on both FHS and Nix
<betaboon>
Profpatsch: yeah thats what i think happends.
<betaboon>
just trying to write nomad-jobs in nix with nomad-provider and json job-specs. super-phun-thyme
xO1 has joined #nixos
<stteevveen>
vika_nezrimaya Not sure I understand. It's nix-build that is calling make install, not me. In the Makefile for install : I have writen cp myApp ${PREFIX}/bin/ but it seems PREFIX is empty in that case, since it tries to copy to /bin/myApp when I execute nix-build.
<vika_nezrimaya>
$(PREFIX) maybe?
<vika_nezrimaya>
you might be confusing Make's syntax with Nix's one
arjen-jonathan has joined #nixos
relaxmax has quit [Ping timeout: 256 seconds]
_viz_ has quit [Ping timeout: 258 seconds]
pause123 has joined #nixos
<pause123>
Hey, anybody having issues with alacritty using nix? It crashes during start
<pause123>
Searched the issue threads but wasn't able to find a solution
<pause123>
It shows mostly fontconfig error
<vika_nezrimaya>
Are you by any chance used Noto Mono for your terminal?
<pause123>
I don't think so, I am using default ubuntu font
<{^_^}>
[nixpkgs] @jappeace opened pull request #85438 → cut-the-crap: init at 1.0.0 → https://git.io/JfJIB
<vika_nezrimaya>
Change it to something else. It might help, I've encountered the same after *somebody* broke the Noto font package by moving stuff to a subdirectory. Might've been an impurity in my system that made it search for the font in the default directory, but I doubt it
cosimone has joined #nixos
<vika_nezrimaya>
too lazy to actually git blame it
<aanderse>
is there any reasonable way to have a declarative nixos container and pin the nixpkgs version in just that container?
<aanderse>
my actual goal is to be able to take a users configuration.nix and boil it down to a minimal reproducible container so i can reproduce a crash this user is having
cosimone has joined #nixos
mvnetbiz_ has quit [Quit: Bye!]
mvnetbiz_ has joined #nixos
cosimone has quit [Remote host closed the connection]
pause123 has quit [Remote host closed the connection]
<jtojnar>
aaron I just build the repro configurations using `nixos-rebuild build-vm -I nixpkgs=… -I nixos-config=…`
<aanderse>
Jan Tojnar: oh yeah, build-vm... i have never actually used that and forgot it existed >_<
<noonien>
hmm, does `gnomecast` work? i get `gi.repository.GLib.Error: gdk-pixbuf-error-quark: Unrecognized image file format (3)` whenever trying to run it.
<noonien>
i've tried the several ways of fixing gdx-pixbuf to load svg images, but none worked
<stepcut>
I am using nixops to deploy to a server and I have enableRollback enabled. But eventually the hard drive fills up. I could set, nix.gc.automatic = true; but I do not think that will help because the generation roots will still be active. How do I configure nixops so that I only keep the last 2-3 generations around on the server?
xfix has joined #nixos
knupfer has quit [Read error: Connection reset by peer]
knupfer has joined #nixos
<{^_^}>
[nixpkgs] @adisbladis merged pull request #82651 → mopidy: Create a mopidyPackages set → https://git.io/JviSL
<immae>
stepcut: you gc automatic will remove lingering objects, but they’s not lingering: you need to remove them first from /nix/var/nix/profiles/system (in the remote system)
<{^_^}>
[nixpkgs] @fgaz opened pull request #85442 → castor: init at 0.5.0 → https://git.io/JfJ3l
<immae>
for instance: `nix-env -p /nix/var/nix/profiles/system --delete-generations +3`
<stepcut>
immae: right.. but I just want to run 'nixops deploy' and not have to run other things as well
<immae>
It’s not supported currently
<immae>
You may add that as activation script if you want (might not be a good idea)
<immae>
(Also note that the command above concerns remote, not the local one from which you’re running the command)
<stepcut>
I kind of wish it was an option to `nix.gc`
<immae>
in my nixops configuration I have a `make cleanup` which cleans up local and remote, I don’t think you can do better for now saddly (profiles defined via nix-env are quite stateful actually I think, it never get removed automatically)
<b42>
enableRollback keeps the old generations on the local machine where nixops is executed, right?
<immae>
b42: both
<immae>
both local and remote have the generation
<immae>
(remote has his own, local has all the machines ones)
<hyper_ch>
I think I figured out why my new ECC setup isn't working - I have no gpu
<b42>
right, but without enableRollback the remote still keeps the old profiles?
drakonis_ has joined #nixos
<{^_^}>
[nixpkgs] @Mic92 opened pull request #85443 → st: copy config file in 'prePatch' instead of 'preBuild' → https://git.io/JfJ3D
nschoe has joined #nixos
<immae>
b42: you’re right, it does (as far as I see from the code, don’t take my word for it)
<{^_^}>
[nixpkgs] @Mic92 closed pull request #85055 → st: copy config file in 'prePatch' instead of 'preBuild' → https://git.io/JvjkD
<{^_^}>
[nixpkgs] @Mic92 merged pull request #85443 → st: copy config file in 'prePatch' instead of 'preBuild' → https://git.io/JfJ3D
<stepcut>
well, close to what I'd like. It would be nice if I could use the syntax like `nix-env --delete-generations +5` instead of just a time period.
<stepcut>
Also, I normally have to use the -p flag to remove the generations that nixops creates, `nix-env --delete-generations old -p /nix/var/nix/profiles/system`
drakonis_ has quit [Ping timeout: 264 seconds]
<stepcut>
I think the activation script is probably the approach that will work for me.
<stepcut>
I don't really need a cron job, since nixops deploy is the only thing that changes the garbage on that machine anyway
<nschoe>
Hi, has anyone else an issue with KiCad? Adding `kicad` to my `systemPackages` fails with "No space left on device" (I still have 44GB on `/`). When I use kicad.override {with3d = false'} it builds fine, though. I now there are a lot of 3d model packges, but it's a few GB at most. Should not be an issue.
<gchristensen>
stepcut: maybe open an issue on the NixOps repository describing the problem, and the problem with the available solutions? if you're having this problem, many people are -- and it'd be nice to have a good fix
jeromelanteri has joined #nixos
<immae>
(this reminded me that I was very late in my cleanup too... `104229.18 MiB freed`)
<stepcut>
gchristensen: an issue about needing `-p`? or about automatic generation deletion?
zupo has joined #nixos
<gchristensen>
stepcut: the awkwardness of having it fill up with profiles an no clear/coherent way to deal with it
noudle has quit []
<srk>
nschoe: is that nixos system?
<nschoe>
srk: yes, I'm on 20.03
<srk>
weird, maybe check df while nix is working? I've seen /run/user tmpfs filled few times - but haven't had this problem with kicad
<nschoe>
b42: ah! I did enable `boot.tmpOnTmps` in my configuration.nix. Does that mean I have to remove it?
phreedom has quit [Ping timeout: 240 seconds]
<nschoe>
srk: well I'd like to understand what is going on :/ And preferably don't use any hack.
phreedom has joined #nixos
pbogdan has quit [Remote host closed the connection]
<srk>
nschoe: sure, I've switched to local library for different reasons (fp-lib-table changes)
<srk>
nschoe: it could use nix! :D
<nschoe>
srk: I used to do this too: I had the symbols, footprints and packages3D cloned manualyl in my home dir and set the correct paths inside KiCad.
cfricke has quit [Quit: WeeChat 2.8]
<nschoe>
But this is a fresh install and wanted to start vanilla.
<nschoe>
I don't understand why it cannot download, nor why my /run is only 1.9G.
<srk>
size is probably fine, mine is 1.6G on a desktop a only 400Mb laptop
chagra has quit [Ping timeout: 240 seconds]
<srk>
you can try remounting it a bit larger
stepcut has quit []
vidbina_ has joined #nixos
cosimone has quit [Remote host closed the connection]
jco has joined #nixos
cosimone has joined #nixos
acarrico has quit [Quit: Leaving.]
_viz_ has joined #nixos
pbogdan has joined #nixos
<pbogdan>
it
pbogdan has quit [Client Quit]
<nschoe>
Ok.
<nschoe>
if you say it's enough, then it should do it. So... I'm wondering why it fails with no space left then :/
<najib>
Hi, I am looking for OMR (optical mark recognition) software. I am not found any when search in nixos packages. FormScanner (www.formscanner.org) look nice. Hope there will be some OMR software in nixos.
<gchristensen>
anyone here have that there XPS 7390 and happy with it?
rardiol has joined #nixos
<ultranix>
does anyone know how I can include androidsdk_9_0 like so in a shell.nix: ANDROID_HOME = "${androidsdk_9_0}/libexec/android-sdk"; and set 'android_sdk.accept_license = true' for that derivation in the Shell ?
<jluttine>
Would it be ok if packages contain debug?false argument so that debugging can then be enabled easily without needing to figure out how to do it for that particular package? Definitely, most packages wouldn't have this argument, but would it be considered ok/good if such is added if someone just does the effort of figuring out how to enable debugging for the package?
<{^_^}>
[nixpkgs] @cdepillabout pushed 2 commits to haskell-updates: https://git.io/JfJZx
<jluttine>
Just thinking if I'd make a pull request for one package where I did this
vidbina_ has quit [Ping timeout: 265 seconds]
<ultranix>
i think the value is actually ' config.android_sdk.accept_license = true; ', i'm still trying to understand how config works
growpotkin has joined #nixos
najibpuchong has joined #nixos
<najibpuchong>
Hi, I am looking for OMR (optical mark recognition) software. I am not found any when search in nixos packages. FormScanner (www.formscanner.org) look nice. Hope there will be some OMR software in nixos.
o1lo01ol1o has quit [Remote host closed the connection]
Ariakenom has quit [Ping timeout: 256 seconds]
mallox_ has quit [Ping timeout: 265 seconds]
justanotheruser has joined #nixos
zeenk has quit [Quit: Konversation terminated!]
o1lo01ol1o has joined #nixos
<duckfullstop>
super dumb question: if I have a systemd unit that's being instantiated by a derivation in nixpkgs, and I want to add some packages to its path, is my only choice to override the package?
<jakobrs>
I'll try to remember to try that when I figure out why I needed to do it in the first place, and on what
pbogdan has joined #nixos
pbogdan has quit [Remote host closed the connection]
<cole-h>
lol
pbogdan has joined #nixos
pbogdan has quit [Remote host closed the connection]
pbogdan has joined #nixos
ris has joined #nixos
<jluttine>
does nix strip debug symbols? my meson build doesn't contain debug symbols although i built it with mesonBuildType="debug" and i can see that this indeed gets passed to the meson builder
<jluttine>
if nix does strip debug symbols, can i somehow avoid that?
<{^_^}>
nschoe: 1 hour, 33 minutes ago <srk> can reproduce (kicad) but I'm running out of disk space for real - it quickly eats 10G during download/unpack
duckfullstop has quit [Quit: Why do you need...?]
<nschoe>
srk: hi again. I've just received your email. 10GB? Wow that's huge. Is the unpack done in /run ?
<nschoe>
Regarding libinput: I've enabled natural scrolling (because that's how I like it for trackpad), but it __also__ enabled it for th emouse, which ... sucks. I know it's possible to disable it with xinput, but is there a way to do it via option?
knupfer has quit [Quit: knupfer]
pluppens has joined #nixos
knupfer1 has joined #nixos
dansho has joined #nixos
_viz_ has quit [Quit: Lost terminal]
_viz_ has joined #nixos
knupfer1 is now known as knupfer
cosimone has quit [Quit: Terminated!]
<srk>
nschoe: no, it's happening in /nix/store (I've asked whether you were on nixos as I've seen this problem with /run/user with nix on other $distro)
<nschoe>
srk: hi back. Ah ok. No no I'm running full NixOS (and won't go back !)
<srk>
nschoe: checked-out repo is 6.2G tho! hydra won't cache stuff larger than >2G iirc
<nschoe>
srk: hum.... that is a huge repo :/
knupfer has quit [Quit: knupfer]
jco has quit [Ping timeout: 252 seconds]
knupfer has joined #nixos
<srk>
indeed, it means you need at least ~12G to install :D
<b42>
i think the unpack is done on /tmp and later installPhase copies it to /nix/store?
<ajs124>
just skip the unpackPhase
<nschoe>
b42: that would certainly explain my issue then: my /tmp is 3.8G only
<b42>
right, sources are downloaded to /nix/store, unpacked to /tmp, and installed to /nix/store so i guess you need 12G in /nix/store and 6G in /tmp
<sphalerite>
nschoe: add an entry to services.xserver.inputClassSections
<nschoe>
sphalerite: oh nice! I wnated to use additionalOptions but it is included in the section with MatchIsTouchpad. I did not know about services.xserver.inputClassSections thanks
LambdaDuck has joined #nixos
<nschoe>
sphalerite: just to double check: setting services.xserver.inputClassSections won't erase what's in there already?
<LambdaDuck>
Is there a way to look at the logs for the last failed build of a specific package on hydra?
<emily>
when using flakes, can a NIX_PATH entry point to a flake?
<sphalerite>
nschoe: no, it won't — the only other thing that seems to set that option is the trackpoint module
<emily>
NIX_PATH="nixpkgs=flake:nixpkgs:..." breaks on the :
<emily>
(needed to get some legacy stuff to work... like nix-shell :') )
<nschoe>
sphalerite: awesome, thanks.
<sphalerite>
nschoe: and even that sets it at regular priority, so it would only be added if you were using it
<nschoe>
Now I have to figure out what goes into this section input ^^
<nschoe>
Thanks
<LambdaDuck>
For example, haskellPackages.free-theorems is marked as broken and I want to find out why it failed.
<sphalerite>
LambdaDuck: packages marked as broken won't get built by hydra, your best bet is to try building it yourself (you can use the NIXPKGS_ALLOW_BROKEN environment variable to bypass the error)
<snicket>
can I use 'gcc -L/some/path/with/library' in nixos, or is it neutered somehow so that I don't do sth stupid? I am not packaging anything, just playing around. and trying to figure out if I'm doing sth wrong or is it nixos blocking me
<sphalerite>
snicket: should work,
<sphalerite>
snicket: though if you're running your gcc after installing it with nix-env…
<LambdaDuck>
sphalerite: Ooh, nice I didn't know about the environment variable. I just assumed that the old logs from before it was marked is available
<emily>
LambdaDuck: that's from 2016 though, so it's been broken for a long time
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to release-20.03: https://git.io/JfJ4I
<emily>
LambdaDuck: (and I don't know if there's an easy way to go through the entire "propagated from ..." chain when there's a bunch of them, which is my usual problem)
<emily>
LambdaDuck: np, hydra UI is pretty arcane ^^
<emily>
supports like 8x more languages when it's offline
<ajs124>
the best part is the austrian localization
<flokli>
ajs124: about #43607, no, I didn't further dig into it. If there was an example config with some samba server, so I could easily tinker around, it'd be easier
<flokli>
We could even then make this a nixos vm test to ensure it doesn't break in the future
<ajs124>
flokli: I'm 99% sure this is bigger than samba
<ajs124>
The systemd example you posted only works because it's run in "systemd's context" (for lack of a better description).
mananamenos has quit [Remote host closed the connection]
<flokli>
ajs124: yes, but the samba example would be a real-world example in how this breaks
<flokli>
it's just the server config missing, isn't it?
<ajs124>
So is the systemd example. getent passwd shows systemd dynamic users if it can find the nss plugin.
seku has joined #nixos
<ajs124>
Idk what's missing. What I'm observing is that glibc isn't trying to connect to the nscd socket or trying to read any config files that seem relevant to nscd.
fendor has quit [Read error: Connection reset by peer]
<colemickens>
bqv: that feels like trick questions sorta :P
<colemickens>
I don't really know. But I know pipewire is how screen/surface sharing in wayland/webrtc/browsery apps works
<colemickens>
so if you want to test the latest Sway / screen share stuff, you need a pipewire-enabled-webrtc-enabled-firefox, and because of deficits in Firefox's build system, it requires patching to enable... phew.
<colemickens>
so anyway, I'm trying to get a patch from fedora or somewhere else working, I don't think it ought to be hard.
<qy[m]1>
ah, i see
<colemickens>
gchristensen: you merged the obs PR? or more than that :)
<gchristensen>
well, it works :)
<colemickens>
(i was still sorta trying to tell if it was actually working properly for you or not, I couldn't decipher the screenshot)
<colemickens>
haha, good
<gchristensen>
I was chromakey'ing my way to Art
_viz_ has quit [Quit: Lost terminal]
_viz_ has joined #nixos
<qy[m]1>
gchristensen: 85453 is a trivial fix if you're free
<{^_^}>
#85183 (by Xe, 3 days ago, open): zig: upgrade to 0.6.0
<{^_^}>
[nixpkgs] @matthuszagh opened pull request #85456 → kicad: add option to override footprint, symbol and 3d model locations → https://git.io/JfJ0y
<{^_^}>
[nixpkgs] @peti pushed 4 commits to haskell-updates: https://git.io/JfJ0S
o1lo01ol1o has quit [Remote host closed the connection]
dermetfan has joined #nixos
knupfer has quit [Quit: knupfer]
knupfer1 has joined #nixos
<Cadey>
thanks gchristensen!
knupfer1 is now known as knupfer
<kaliumxyz>
srk I was just being stupid
<kaliumxyz>
its not supposed to show up at all, you can only flash it from arduino IDE not see it status
<kaliumxyz>
Also I spent the last half hour patching every single binary arduino pulls in.
<kaliumxyz>
lol
luc65r has joined #nixos
reallymemorable has joined #nixos
arjen-jonathan has quit [Ping timeout: 265 seconds]
mr_sm1th has joined #nixos
codygman has quit [Read error: Connection reset by peer]
codygman has joined #nixos
codygman has quit [Read error: Connection reset by peer]
<aterius>
Anyone gotten nvidia-docker working/tried it on nix 20.03 yet?
codygman has joined #nixos
cosimone has quit [Quit: Quit.]
<energizer>
on nixos, is my user supposed to have `nix-channel`s or not?
<luc65r>
energizer Your user can have nix-channels
<energizer>
luc65r: when are they relevant?
<gchristensen>
by default you should be subscribed to the root channels
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<luc65r>
energizer unless you want a different channel that the root one, no
<energizer>
luc65r: that wasn't a yes-or-no question :)
<energizer>
like, if i `nix-channel --remove` all of my channels, when will i notice?
<aterius>
I follow the latest stable release in my root channel, I add nixpkgs-unstable in my user channel, such that when I use home-manager or nix-shell, my userspace packages are kept up to date with nixpkgs-unstable
<euank>
iiuc, they'd let you install a package from a different channel for your user if you wanted. i.e. if you wanted to install `nixos.foo` vs `my-channel.foo`
<euank>
So you'd only notice if you tried to install packages from a removed channel (which I guess could happen from a few different operations)
<drakonis>
but i suppose the standalone install will default to that after a upgrade
drakonis_ has quit [Ping timeout: 240 seconds]
<hr[m]>
Hi all, I have a PR, https://github.com/NixOS/nixpkgs/pull/85248 with only one reviewer assigned, but the reviewer has been inactive on GitHub for > 1 month. Are there any other reviewers able to take a look at it.
<Raito_Bezarius>
I just updated all my nixpkgs using nix-channel --update && nix-env -u '*'
<bqv>
This is a paste of nix-diff's output between nixpkgs-unstable and what i have from my config: https://gateway.ipfs.io/ipfs/QmdGvaU3R63diYjUKmAx6A5ynm9enZxjUTnaqSWxCsSUqo - Can anyone decipher what exactly is different here? It looks like it's saying i've made a completely new stdenv or something, just for this package
<bqv>
(which is patently not true, cause otherwise everything else would be missing cache and rebuilding too)
<bqv>
try running whatever command you're running with env GC_DONT_GC=1
<Raito_Bezarius>
energizer: is poetry2nix supposed to work on dependencies which are themselves supporting poetry?
morgrimm has joined #nixos
<energizer>
Raito_Bezarius: what is it that you're trying to do?
<Raito_Bezarius>
I'm just packaging a library
<Raito_Bezarius>
but I'm having errors with a poetry.lock on some not-really-maintained library
<Raito_Bezarius>
> list index 0 is out of bounds, at /nix/store/ji0i27arn83r3251is7hx10hwlvnig1h-nixpkgs-20.09pre221715.5f814bf9cc4/nixpkgs/pkgs/development/tools/poetry2nix/poetry2nix/mk-poetry-dep.nix:67:25
<{^_^}>
error: syntax error, unexpected ',', expecting ')', at (string):304:30
<{^_^}>
[nixpkgs] @adisbladis merged pull request #85448 → mesa-glu: use HTTPS instead of FTP → https://git.io/JfJck