<clever>
abathur: ah, the nix function just grabs whatever is after the last / and uses that as a name
<clever>
abathur: it should probably have a name= added to the fetchurl, to make it cleaner, and/or use fetchpatch
<lovesegfault>
cole-h: how's it going?
<superbaloo>
would anyone know how nix-shell works exactly?
<clever>
superbaloo: it evals a nix expression to a drv, builds every input to that drv, then puts the env from the drv into its own env, and launches a shell
<superbaloo>
I'm trying to reimplement it with nix-build
<superbaloo>
yeah, my question is mostly, what nix expression is used
<superbaloo>
is that just a nix-instanciate of stdenv.mkShell ?
<clever>
superbaloo: it depends on what arguments you ran nix-shell with
<superbaloo>
yeah of course
<clever>
superbaloo: what args did you use?
<superbaloo>
just nix-shell with a shell.nix
<clever>
superbaloo: then it uses whatever nix expression was in the shell.nix
<clever>
superbaloo: if you want nix-build to build the same expr, run `nix-build shell.nix`
<finnwww[m]>
I'm trying to install home manager with sudo nix-shell '<home-manager>' -A install but it says that it wasn't found in the Nix search path. Any ideas?
<clever>
qyliss: yeah, its also called _main, because EVERY binary has since been merged into a single one, and it uses argv[0] even more, to do the same logic
<superbaloo>
sweet
<clever>
finnwww[m]: you need to put home-manager into NIX_PATH or do -I home-manager=something
<selfsymmetric-mu>
Okay, so I'm installing `nix-direnv`. It requires a `.direnvrc` file. I want to make that declarative so I'm just going to add a `postInstall` that symlinks the file to `.direnvrc`. Is that the normal way to do it?
<selfsymmetric-mu>
I'm trying to guess at the correct pattern to replace something like manually creating an `.<app>rc` file.
sigmundv__ has quit [Ping timeout: 258 seconds]
teto has quit [Ping timeout: 252 seconds]
maxdevjs has joined #nixos
<selfsymmetric-mu>
,symlink
<selfsymmetric-mu>
,symlinks
cr4y1_ has quit [Ping timeout: 256 seconds]
h0m1 has quit [Ping timeout: 272 seconds]
h0m1 has joined #nixos
proofofkeags has quit [Remote host closed the connection]
justanotheruser has quit [Ping timeout: 252 seconds]
ddellacosta has joined #nixos
aditsachde has joined #nixos
markus1189 has joined #nixos
thc202 has quit [Ping timeout: 244 seconds]
m0rphism has quit [Ping timeout: 260 seconds]
markus1199 has quit [Ping timeout: 240 seconds]
CMCDragonkai1 has joined #nixos
cfinch has joined #nixos
proofofkeags has joined #nixos
cfinch has quit [Ping timeout: 265 seconds]
justanotheruser has joined #nixos
sarcasticadmin has joined #nixos
<aditsachde>
Hey folks, I wrote a quite mediocre post on how I'm using nixos based on my the little understanding i have of nix and nixos. Also, if there are any mistakes with the way I'm using nix/nixos, I'd love to know and fix them. https://aditsachde.com/posts/nixos-cicd/
CMCDragonkai1 has quit [Ping timeout: 260 seconds]
`_ has joined #nixos
<keithy[m]>
Hi guys, I am trying to build a path using code like so: (root + "/" + name) but the result is missing the / am I missing something?
<pjt_014>
aditsachde: minor thing, but you can take all the boot.* options and put them into one section for conciseness.
<pjt_014>
instead of boot.foo = ...; boot.bar = ...; , boot = { foo = ...; bar = ... }
<pjt_014>
keithy[m]: I think it's ++ for string concatenation
<selfsymmetric-mu>
It's not. String concatenation in Nix is `+`.
<keithy[m]>
I thought that was array... ok
<`_>
It's + if I recall correctly
<pjt_014>
yeah just checked, my bad
<pjt_014>
what are root and name here? strings
<pjt_014>
?
<`_>
Looks like they're variables
<keithy[m]>
root is ./.
<keithy[m]>
yes variables
<aditsachde>
pjt_014: I'm only writing the boot options once as that is in a file that all other configs import, but I will remember that tip for other places!
<aditsachde>
I got most of that infra working in the last week, so I'm probably going to be converting a lot of the other services I run to nixos based stuff in the next few weeks
<pjt_014>
aditsachde: It might save a bit of space elsewhere too. Like the virt. section could be rewritten as: virtualisation.vmware.guest = { enable = true; headless = true; };
<aditsachde>
ah i completely forgot that the keys can be all defined on a single line because of the semicolons
<pjt_014>
tis' handy
<pjt_014>
also, for your ssh key: you can define it within anoter file and import it so that the placeholder text is not needed
<keithy[m]>
ok that seems more useful, took me 3 hours to write what I got so far.
<keithy[m]>
to get all the lets and withs in the right place just to do a simple function
<pjt_014>
then simply don't share `secret.nix` or whatever you call it
selfsymmetric-mu has left #nixos ["gone to the land of dead hiccups and extinguished light bulbs"]
<`_>
Anyone have any experience creating custom scripts as shells for users? I got it going with `stdenv.mkDerivation` building some-shell.sh however attempting to access the user tells me it's expecting some-shell.sh to live in /run/current-system/sw/bin
sarcasticadmin has quit [Ping timeout: 264 seconds]
<aditsachde>
currently most of my infrastructure stuff is in a huge monorepo that is already cluttered with secrets, so I'm not really making an attempt to declutter my nix configs until i get a better hang of it
<aditsachde>
although yes, ideally I'd like to make everything public
<`_>
Basically, when I'd login as the user, I'd want it to echo `some stuff` then log out.
<tnks>
Is there a risk of a user installing an incompatible version of Nix commands than is run on system globally? I know for users it's easy to avoid this problem. But with recursive Nix now in, it seems like a concern larger than just user-level installations. I'm not sure how what happens if a recursive Nix expression calls an incompatible version of Nix.
user23939 has quit [Remote host closed the connection]
<jybs>
I'm patching a package, and I want to change the CFLAGS setting - it's already set to something, do I just have to copy that or can I add to it somehow?
niso has quit [Ping timeout: 244 seconds]
rose__ has joined #nixos
niso has joined #nixos
<rose__>
I'm having a very frustrating bluetooth issue: this morning my headphones were working properly, but then I connected a second bluetooth speaker, and now my computer won't connect to any bluetooth devices at all :(
<rose__>
When I type pair "MAC address", bluetoothctl acts like it's connected, but the device does not respond
aditsachde has quit [Remote host closed the connection]
adit[m] has joined #nixos
<rose__>
I just want to reset my bluetooth state to what it was this morning, is there a way to roll back?
georges has joined #nixos
rose_ has quit [Ping timeout: 272 seconds]
<rose__>
Sometimes my headphones say "device disconnected", which is odd because they never connected in the first place
<evanjs>
gah. do I _have_ to do tensorflow for a python packages nixpkgs-review run...
<evanjs>
It's failing due to ... I think OOM? I'll see after this nix build run...
<evanjs>
Might partly be due to /tmp on tmpfs but I have 16GB for tmpfs (32GB total) and 36GB of swap
Guest7 has joined #nixos
<evanjs>
**tensorflowWithCuda is what's running out of memory rn
<evanjs>
normal tensorflow (python37Packages, obviously) built fine
<Guest7>
Hi; I'm trying to setup a basic apache+php server and it wont find the index file. I am using services.httpd and in my virtualHost I have something of the form documentRoot = "/home/<user>/dir", and I chowned dir's user and group to "wwwrun", but I keep getting a 403 access denied error. Any thoughts?
cole-h has quit [Quit: Goodbye]
cole-h has joined #nixos
<rose__>
My headphones are turned off, but still bluetoothctl acts like it's connected
<rose__>
trying to disconnect powered off device: "Failed to disconnect device: Busy (0x0a)"
<rose__>
I guess that makes sense
<rose__>
oh now I'm getting some actual errors in the log
<jybs>
Is there a nice way to sub ar for gcc-ar to get -ftlo working?
<jybs>
sorry other way round
<rose__>
brb rebooting for the 20th time
<evanjs>
rose__: literally me when bluetooth doesn't work...
<Guest7>
does anyone have an example nixos config that uses an apache server?
rose___ has joined #nixos
rose__ has quit [Ping timeout: 264 seconds]
<rose___>
I vaguely remember it being massively painful the first time I had to setup bluetooth, but reading online, it looks like "pair" should just work?
<hyper_ch>
Guest7: any reason you want apache?
<Guest7>
hyper_ch: Yes, because that is what the hosting provider uses, and I cannot change that
<evanjs>
Guest7: I typically use github search with things like language:nix
<colemickens>
ottidmes: I have a set of nix files that actually pin nixpkgs. I'm experiencing an odd issue wehre nix-prefetch is getting the wrong cargoSha256 on my CI job (hard to see how this is possible), and `nix-prefetch` using NIX_PATH is the only place where it could differ.
proofofkeags has joined #nixos
<srhb>
rose___: Is this from bluetoothctl?
<rose___>
Yes, I am trying to use bluetoothctl to connect to my headphones
<colemickens>
if I run this same job in a `nix-shell --pure ...` it gets the correct 1c9pz cargoSha256 but in sr.ht it incorrectly gets 1lywr2 and then fails when the real fetcher gets 1c9pz at build time.
<srhb>
rose___: You probably also have to shutdown the bluetooth service when you clear that directory
* colemickens
had hoped putting this down for the afternoon would reveal a silly mistake (so far it just hasn't revealed itself)
<rose___>
srhb: yes, I rebooted the machine
<srhb>
rose___: Ah, okay.
<srhb>
colemickens: Err, how do you use nix-prefetch to find a cargo sha?
<srhb>
colemickens: Isn't that some composition of the deps?
<la-s>
colemickens: can't you specify the pinned nixpkgs in NIX_PATH too?
<rose___>
Ooh, another new error message: BUG: kernel NULL pointer dereference, address: 0000000000000020
<srhb>
rose___: Fun!
<Guest7>
What would be the equivalent of setting "Order deny,allow" in services.httpd config?
<colemickens>
yeah, I'd have to re-parse it back out of my pinned file though (after some other update script put it in there)
<srhb>
colemickens: Or have an expression that spawns a shell that has it set to that path
<srhb>
From which you can run prefetch.
<colemickens>
That's what I used to do, yeah, I just `NIX_PATH=... nix-prefetch` and left it unset otherwise.
<srhb>
Indeed..
<colemickens>
I guess I was hoping ultimately for an escape hatch where nix-prefetch would not need NIX_PATH
* colemickens
jumps in his time machine to the flakey future
sparogy has quit [Quit: authenticating]
sparogy has joined #nixos
jlv has joined #nixos
<jlv>
Is there a derivation builder for simple scripts? Something like `writeScript`, but for derivations?
<srhb>
jlv: I mean, that's pretty much mkDerivation. The lower level one is `derivation`
<cole-h>
colemickens: By chance, have you run into the issue where `swaymsg reload` will cause your keys to double?
<colemickens>
no, that sounds annoying/funny
<cole-h>
It's really annoying. Literally can't do anything until I restart my session :D
<srhb>
jlv: You probably want something like mkDerivation in conjunction with builder for simplicity, the lower level route gets pretty hairy pretty fast.
<cole-h>
Plugging in my spare keyboard to see if it happens for all of them...
<srhb>
jlv: See eg. the nix documentation for builtins.toFile
<cole-h>
TThhiiss iiss wwhhaatt hhaappppeennss
<cole-h>
llooll
<jlv>
srhb: Don't I still have to add `name`, `installPhase`, and whatnot for `srhb`. Every once in a while I want to add a simple shell script to my system as a binary. Right now, I'm looking at a script that is 5 lines of code.
<cole-h>
And this is much better
<jlv>
*`mkDerivation`
<jlv>
I have no idea how that turned into your username XD I am a little tired.
<srhb>
jlv: Name: Yes. That's fundamental for something to be in the Nix store. installPhase: No, that's not intrinsic to mkDerivation, but the default mkDerivation builder.
<jlv>
srhb: I think I'll write a `mkScript` function, like `writeScript`, but using `mkDerivation`.
<srhb>
jlv: I mean, that's already how writeScript is written, ultimately.
<srhb>
jlv: What are you trying to achieve exactly?
<rose___>
Well, after hours of debugging bluetooth, it's time to give up and move on with my life >:(
<srhb>
rose___: Sucks. :( Sorry, I've no idea why you're seeing that issue.
<jlv>
srhb: can I use `writeScript` like `systemPackages = [ (writeScript name script) ]`?
<srhb>
jlv: Yes.
rose___ has left #nixos [#nixos]
<jlv>
srhb: Oh XD I guess that's the answer to my original question then.
<srhb>
jlv: You may need writeScriptBin, actually, because iirc writescript produces a bare store path.
<srhb>
jlv: While writeScriptBin produces a storepath/bin/foo for proper linking into the system tree.
<clever>
jlv: writeScript puts the binary directly at $out, but systemPackages puts the thing in $PATH
<clever>
jlv: so your script is a literal file of /nix/store/hash-name, but systemPackages will try to do PATH=/nix/store/hash-name/dir, which doesnt exist
<clever>
jlv: writeScriptBin and writeShellScriptBin will solve that issue
agsdheidjd has joined #nixos
<jlv>
clever: I thought I had seen `writeShellScriptBin` before, but I couldn't find it in the nixpkgs manual.
<Guest7>
I may have found the root cause of the issue
<Guest7>
In the default httpd.conf, the following is set:
<srhb>
Hrmm, I feel like the top level nixpkgs/flake.nix is a little NixOS-centric. Overlays and config seem hard to get at without building a full config and relying on the nixpkgs module.
<jybs>
lovesegfault: Sorry to ping - but did you ever solve your gcc-ar issue?
nschoe has quit [Quit: No Ping reply in 180 seconds.]
drakonis_ has quit [Read error: Connection reset by peer]
Acou_Bass has quit [Ping timeout: 256 seconds]
drakonis_ has joined #nixos
nschoe has joined #nixos
Acou_Bass has joined #nixos
dingenskirchen1 has joined #nixos
dingenskirchen has quit [Quit: dingenskirchen]
dingenskirchen1 is now known as dingenskirchen
<Guest7>
hmm the logs say access denied "because search permissions are missing on a component of the path"
<srhb>
Guest7: Missing +x somewhere along the path?
reallymemorable has quit [Quit: reallymemorable]
<srhb>
Guest7: I would expect you can debug this by su'ing to a shell for the user running the server and listing the directory contents.
<hpfr[m]>
Anyone have a good approach for parsing a file with an attribute and value on each line into a Nix attribute set? Struggling to find builtins for something like this
simba1 has joined #nixos
drakonis_ has quit [Read error: Connection reset by peer]
<hpfr[m]>
I guess s/file/multiline string/ since file could mean nix or json
quinn has joined #nixos
distransience has joined #nixos
iqubic` has joined #nixos
<distransience>
Hi all, I have a Minisforum Z83-F with a Broadcom AP6255/AP6212 that fails to load the firmware on NixOS (brcmfmac mmc1:0001:1: Direct firmware load for brcm/brcmfmac43455-sdio.txt failed with error -2). I have the required file provided by the manufacturer, but it's unclear to me how to add it to my nix configuration; any help?
iqubic has quit [Ping timeout: 260 seconds]
<srhb>
distransience: Take a look at hardware.firmware and some of the existing firmware packages.
<srhb>
distransience: Actually, that one is already in firmwareLinuxNonfree
<distransience>
Any packages you might recommend? I see the "pathsToLink = [ "/lib/firmware" ]" part of hardware.firmware now
<srhb>
distransience: At least I see a brcmfmac43455-sdio.bin in there, so I would think it suffices
<distransience>
Neat. I'll test it out and if it gives me issues I'll give writing my own hardware.firmware package a shot.
<srhb>
distransience: Yeah, it shouldn't be too hard. :)
<distransience>
My manufacturer requires signing up to their forum, waiting for an admin to approve you after a day or two, and then you can download the sdio from the forum post 😅
growpotkin has joined #nixos
<distransience>
So definitely not "Free" licensing
<srhb>
Old-school non-free at that.
iqubic` has quit [Ping timeout: 265 seconds]
<distransience>
At least their Linux support is existent. Low bar, I guess
ddellacosta has quit [Ping timeout: 258 seconds]
<srhb>
Yeah. But better than nothing I suppose.
iqubic has joined #nixos
endformationage has quit [Quit: WeeChat 2.6]
nschoe has quit [Ping timeout: 246 seconds]
<cole-h>
How can I make an overlay like nixpkgs-wayland accessible via something like `pkgs.nixpkgs-wayland.<packages>`?
<cole-h>
And what if it's an import? e.g. `import sources.nixpkgs-wayland`?
<cole-h>
Oh
<cole-h>
That works too
<srhb>
cole-h: :)
<cole-h>
I was trying `wayland = ..` which doesn't work because that's already a package lol
<cole-h>
:D
thonkpod has quit [Ping timeout: 260 seconds]
<cole-h>
srhb++ Thanks for redirecting me in the right direction
<{^_^}>
srhb's karma got increased to 107
thonkpod has joined #nixos
<distransience>
Lol who else ever just reboots without even rebuilding
<srhb>
cole-h: Looks like its default.nix is already an overlay (but the name is waylandPkgs instead of nixpkgs-wayland)
<cole-h>
srhb: The thing is, I don't want the entire overlay set -- e.g. I want to run sway from nixpkgs, not git (since I've had several problems I have no intention in debugging/reporting)
drakonis has joined #nixos
<angerman>
is there some inline version o f makeWrapper?
<srhb>
cole-h: Ah.
<angerman>
say I want to just wrap the output of some binary. And get some path back? ${makeWrapperInline ... ...} would produce the wrapper?
<srhb>
angerman: I'm not quite sure what you mean by "wrap the output" but it sounds like you should take a look at wrapProgram
<angerman>
srhb: Let's see, thanks!
<shyim>
How can I run nixos-rebuild without caches?
<shyim>
my cache server is down
<shyim>
i cannot run it anymore xD
<srhb>
shyim: --option substituters ''
<srhb>
I think --option substitute false would work too
drakonis_ has quit [Ping timeout: 265 seconds]
<angerman>
srhb: hmm... ok. So I have `prop=${cc}/bin/cc`, and I want to turn that into `prop=${mywrapper "${cc}/bin/${cc}" "-L{bleh}/lib"}`
<shyim>
@srhb, does not work :(
<angerman>
*mkwrapper
<srhb>
angerman: Ahh, I understand no. No, I don't know of a version that's a nix expression and gives back a separate drv.
<srhb>
s/no./now.
<angerman>
Hmm... guess I'll have to do that by hand then.
<srhb>
angerman: It sounds like that would require recursive nix in some form
<srhb>
angerman: Since if you're using a file inside the build sandbox, we would have to re-exec to do _another_ bulid and then pop back up.
<srhb>
Would be totally cool though.
drakonis_ has joined #nixos
<srhb>
If it's not a file from the build sandbox I guess it's doable. :)
<angerman>
I believe the rust bootstrapper is nuking the env, so any env var set outside of the bootstrap process, won't make it inside the bootstrap calls. And my cc needs some -L flags. I also fail to find any way to inject -L flags into the rust bootstrap process. So I'm thinking about wrapping my $CC, such that it's supplying that arg by default.
<angerman>
srhb: well I can build a derived CC from the original CC, and inject that one.
<srhb>
Right right. That should work. :)
<angerman>
We can't be the only ones trying to use nix excessively for cross compliation, can we?
tsrt^ has quit [Ping timeout: 258 seconds]
<srhb>
I would guess not. ;-)
drakonis has quit [Ping timeout: 265 seconds]
drewr has quit [Ping timeout: 260 seconds]
hmpffff has joined #nixos
lorimer has quit [Read error: Connection reset by peer]
<angerman>
maybe to windows... 🤷♂️
charukiewicz has quit [Quit: leaving]
<srhb>
angerman: :O
gxt has quit [Ping timeout: 240 seconds]
<angerman>
that's been one of the primary drivers behind haskell.nix
<graf_blutwurst>
Good Morning. I have a very odd issue. Some buttons in web UIs are not clickable. This seems to be across 3 browsers the same story (qutebroweser, FF and chrome) and seems to be the case since my last nixos upgrade. has anyone encountered something similar?
opthomasprime has quit [Remote host closed the connection]
opthomasprime has joined #nixos
<srhb>
graf_blutwurst: No, that sounds super odd. Got an example website?
eoli3n_ has joined #nixos
<colemickens>
fascinating. I have ensured nix_path is unset, except when executing nix-prefetch, and when executing nix-prefetch it's set to $(nix-instantiate ...) to get the same nixpkgs. And yet, the job passes locally and fails in CI! In a nix-shell --pure!
<srhb>
colemickens: Sandboxes > shell shenanigans is the conclusion, I guess :P
<srhb>
But we already knew that.
<graf_blutwurst>
@srhb the problem ... went away? I have no clue why
<srhb>
graf_blutwurst: um. :)
<graf_blutwurst>
it was an internal site where it occured and i couldn't reproduce it so far. magic of computers i guess
growpotkin has quit [Ping timeout: 256 seconds]
<graf_blutwurst>
maybe it wasn't nixos related at all and just coincidence that it happened on multiple browsers and after me tinkering with the system. sorry about that
<srhb>
No worries :)
<colemickens>
srhb: something like that, I guess. I don't do anything weird to the daemon and the sr.ht nixos image has always worked perfectly for me previously
<colemickens>
srhb: I was wondering if it could be like an architecture, cpu feature sort of thing?
<colemickens>
I don't really know that much about the rust or fetchers in general
<srhb>
colemickens: cpu feature _probably_ not, since we allow that impurity in general and it shouldn't affect our nix expressions
<srhb>
colemickens: architecture.. Um. :) I mean, plausibly? You should be able to compare those from system.
<srhb>
colemickens: In fact, if you can get the non-matching drvs, nix-diffing them might be helpful.
eoli3n_ has quit [Remote host closed the connection]
fabianhjr has quit [Quit: Leaving.]
<colemickens>
srhb: the thing is, everything else is fine. This script updates dozens of packages. The nix-prefetch correctly updates the package's sha256 but the computed cargoSha256 is only calculated correctly... well, so far on both of my local machines
<srhb>
colemickens: Filesystem impurity or something?
<srhb>
colemickens: But yeah, confirm that the drvs are in fact identical first, I guess.
<{^_^}>
[nixpkgs] @gnidorah opened pull request #88049 → reaper: update to 6.10, fix libjack2 missing dependency → https://git.io/JfEuc
<colemickens>
hm, i'd have to think about it, i don't really know
<colemickens>
the derivations for this particular app are always cache hits
<colemickens>
(which assuming I don't have a huge flaw in my nix knowledge means the pkg drvs are the same)
coco has joined #nixos
<colemickens>
I am sort of worried that I'm treating nix-prefetch like too much of a black box. looks like it's got some opportunities for printf debugging/learning
<colemickens>
the way I was invoking nix-prefetch to update the sha256 used my build.nix which hardcoded nixpkgs+overlay
<colemickens>
the way I was invoking it to update the cargoSha256 wasn't going through the pinned nixpkgs and wasn't getting the overlay
<colemickens>
and my local system... apparently... has the overlay in my current user nixpkgs config.
<srhb>
colemickens: Yeah, when pinning, being explicit about config and overlays is important.
<srhb>
It's the only way to stop top-level/impure.nix from doing all sorts of dark voodoo.
* colemickens
was trying to...
<srhb>
Right.. :D
<srhb>
What I meant to say was: It's easy to accidentally get that wrong.
* colemickens
nods
eoli3n_ has joined #nixos
orivej has quit [Quit: No Ping reply in 180 seconds.]
<colemickens>
I wish I would've listened to you earlier and retraced the hashes a bit more closely.
<colemickens>
And that flakes (really magically solves everything I keep handwaving that it does and) comes soon. :)
orivej has joined #nixos
zupo has joined #nixos
orcus has joined #nixos
<srhb>
colemickens: I still have troubles getting overlays to work in a nice way with flakes, but maybe soon. :)
drakonis_ has joined #nixos
<srhb>
colemickens: "legacyPackages" is a very optimistic name for nixpkgs. :P
<srhb>
And a number of other small grievances. (like, having to manually .follows on each input to avoid multiple dependency trees). But at least it's usable.
drakonis has quit [Ping timeout: 244 seconds]
lsix has joined #nixos
NeoCron has joined #nixos
<{^_^}>
[nixos-homepage] @garbas merged pull request #427 → Fix comma-separated list on /learn.html → https://git.io/Jf0NN
<{^_^}>
[nixos-homepage] @garbas pushed 2 commits to master: https://git.io/JfEzw
shafox has joined #nixos
<distransience>
Sadly, flipping hardware.enableRedistributableFirmware didn't load the brcmfmac43455-sdio, although I've discovered linuxPackages.broadcom_sta to add to my boot.extraModulePackages
<distransience>
I just need to build the package for adding to hardware.firmware now; I have the txt file in the same directory as my configuration.nix, I just need to write it to /lib/firmware/... in the derivation, right?
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
coco has quit [Ping timeout: 260 seconds]
ozychhi has quit [Read error: Connection reset by peer]
<bqv>
srhb: whats going wrong with overlays? They work fine for me, but they probably won't take effect if you just drop them in config.nixpkgs.overlays or something
<srhb>
bqv: No, that part works just fine. The issue is with legacyPackages in non-NixOS contexts.
<bqv>
Ahh. I reexport pkgs as legacypackages and have all inputs in the system registry, so self becomes my "nixpkgs" and all overlays are applied systemwide
slack1256 has quit [Remote host closed the connection]
alexherbo2 has quit [Ping timeout: 240 seconds]
fling has joined #nixos
orivej has joined #nixos
<kayg04>
is it possible to replace docker-compose by `docker.containers` ?
hmpffff has quit [Ping timeout: 265 seconds]
eoli3n_ has quit [Remote host closed the connection]
<adisbladis>
They are not exactly equivalent
<adisbladis>
kayg04: So the answer is, it depends.
b42 has joined #nixos
<kayg04>
as far as I understand, `docker.containers` creates a systems service that runs `docker run` ?
<adisbladis>
Exactly
eoli3n_ has joined #nixos
<adisbladis>
I just mean docker-compose is not typically used to run production workloads, it's IME typically used for development
<adisbladis>
That's obviously somethingg that declarative containers is not well-suited for
hmpffff has joined #nixos
<kayg04>
well, TIL
<kayg04>
docker-compose is not used for production?
orivej has quit [Quit: No Ping reply in 180 seconds.]
<adisbladis>
kayg04: From my personal experience I haven't seen it much
<kayg04>
well if so, afaik, docker run has more flexibility over docker-compose, I should be getting more mileage out of my workload using `docker.containers` ?
<kayg04>
am I missing something?
orivej has joined #nixos
<LnL>
I've seen it used for deployment in multiple places
eoli3n_ has quit [Remote host closed the connection]
<adisbladis>
kayg04: I don't think you're missing anything.
hmpffff_ has quit [Ping timeout: 265 seconds]
rauno has quit [Remote host closed the connection]
<kayg04>
I use docker-compose only so that I don't forget how I deployed a certain container. think I'll try `docker. containers` then. thank you.
quinn has quit [Read error: Connection reset by peer]
quinn has joined #nixos
o1lo01ol1o has joined #nixos
hmpffff_ has joined #nixos
hmpffff has quit [Ping timeout: 260 seconds]
eoli3n_ has joined #nixos
orivej has quit [Ping timeout: 258 seconds]
sigmundv has joined #nixos
<adisbladis>
LnL: I was trying to make the point that docker-compose is something you can easily run in your shell for development or whatever, but NixOS declarative containers is not suited for the development case
o1lo01ol1o has quit [Remote host closed the connection]
<adisbladis>
Not that docker-compose is unsuited for production =)
<LnL>
heh yeah, never said it was nice :)
knupfer has joined #nixos
<{^_^}>
[hydra] @basvandijk merged pull request #762 → Fix declarative jobset missing type value → https://git.io/JfRzc
<infinisil>
pie_: Nix attrsets are strict in keys, so to evaluate self.lib it also needs to evaluate all other attributes, including the one you set there
linarcx has quit [Read error: Connection reset by peer]
<pie_>
so the problem is it needs to be strict in self/super?
<infinisil>
Not sure what that means
<pie_>
by proxy of using them to compute kernel'
<pie_>
?
linarcx has joined #nixos
<pie_>
i guess the exact issue isnt completely clear to me yet
<infinisil>
Still don't know what that means
<pie_>
is there any way to do what I want to do?
<bqv>
put it at a lower level
<pie_>
infinisil: youre saying it tries to evaluate this strictly: "linux_${kernel'}" = "";
<bqv>
pkgs.kernels.{blah}
<infinisil>
Just don't use self.lib
<infinisil>
Use super.lib or the lib from the module arguments
<bqv>
Or that
<pie_>
ok maybe this is unrelated to my real issue hold on
<bqv>
Also why is {^_^} making jokes now
<bqv>
Lmao
<pie_>
not being able to use self there seems pretty dumb though
<pie_>
anyway as you can see the second example only uses super
<bqv>
pie_: that also doesn't terminate?
<pie_>
its probably gonna reduce to this somehow though lol
<infinisil>
Strictness for sure can cause things to infrec
<pie_>
infinisil: yeah nevermind
<pie_>
anyway here's most of a trace for the current problem https://bpa.st/YAVA
teto has quit [Ping timeout: 256 seconds]
dansho has joined #nixos
cfinch has joined #nixos
<infinisil>
pie_: And that's the only code you need to reproduce this?
<pie_>
yeah
<pie_>
mm let me guess readFile is strict something something
<bqv>
probably a limitation from how nixpkgs sets up pkgs, not inherent to nix, does it work if you give it a constant name?
<pie_>
bqv: if i give it a constant name it the problem becomes trivial and nonexistent - the rest of the code doesnt even get evaled
<infinisil>
Can't think of a reason this wouldn't work
alexherbo2 has quit [Ping timeout: 265 seconds]
* pie_
looks up how to call the module system
<pie_>
ok i dont actually need the readfile
<bqv>
well, you might again have better luck if you drop it a level down
eoli3n___ has joined #nixos
<pie_>
to cause the infrec i mean <pie_> ok i dont actually need the readfile
dansho has quit [Remote host closed the connection]
dansho has joined #nixos
<pie_>
wait...what? <bqv> put it at a lower level<bqv> pkgs.kernels.{blah}
<pie_>
does that work?
<bqv>
Don't see why not
<pie_>
"pkgs.kernels.linux_${svf}" = ""; still infrecs tho :p
<bqv>
Wat
fendor has joined #nixos
<{^_^}>
[nixos-hardware] @Mic92 opened pull request #168 → tests: hide evaluation command behind verbose flag → https://git.io/JfEib
eoli3n___ has quit [Remote host closed the connection]
eoli3n___ has joined #nixos
eoli3n___ has quit [Remote host closed the connection]
iyzsong has joined #nixos
<pie_>
here's a minimal repro
<pie_>
> (import <nixpkgs/nixos> { configuration = let o = (se: su: let svf = su.runCommand "" {} ""; in { "linux_${svf}" = ""; }); in { nixpkgs.overlays = [ o ]; }; }).system
<{^_^}>
infinite recursion encountered, at /var/lib/nixbot/nixpkgs/master/repo/pkgs/top-level/stage.nix:74:29
<pie_>
infinisil: ^
zupo has quit [Ping timeout: 265 seconds]
zupo has joined #nixos
<infinisil>
Odd
<infinisil>
Maybe I'll take a closer look when I'm on my pc (currently phone)
jkarni_ has quit [Ping timeout: 246 seconds]
jkarni_ has joined #nixos
ok2` has quit [Ping timeout: 240 seconds]
knupfer has quit [Quit: knupfer]
knupfer has joined #nixos
knupfer has quit [Read error: Connection reset by peer]
knupfer has joined #nixos
knupfer has quit [Client Quit]
knupfer has joined #nixos
knupfer has quit [Client Quit]
alp has quit [Ping timeout: 260 seconds]
est31 has quit [Remote host closed the connection]
est31 has joined #nixos
knupfer has joined #nixos
magnetophon has quit [Ping timeout: 265 seconds]
<pie_>
infinisil: that would be cool thanks
knupfer1 has joined #nixos
knupfer has quit [Client Quit]
knupfer1 is now known as knupfer
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<pie_>
infinisil: this variant also infrecs:
<pie_>
> (import <nixpkgs/nixos> { configuration = ({pkgs,...}: let o = (se: su: let svf = pkgs.runCommand "" {} ""; in { "linux_${svf}" = ""; }); in { nixpkgs.overlays = [ o ]; }); }).system
<{^_^}>
infinite recursion encountered, at /var/lib/nixbot/nixpkgs/master/repo/lib/attrsets.nix:344:7
<pie_>
the difference is pkgs.runCommand
<pie_>
not necessarily the same infrec though idk
<hadrian[m]>
Am l good to execute?
<pie_>
my hunch is this is probably related to nixpkgs.overlays but that doesnt help much
alp has joined #nixos
<hadrian[m]>
l have hostiles incoming. Back in a few. Sorry.
<bqv>
What
litschi has quit [Quit: WeeChat 2.7.1]
litschi has joined #nixos
<pie_>
probably opened the wrong irc window for the drone task list
<pie_>
jk not enough jargon i dunno lol
<MichaelRaskin>
I would assume that «hostiles» means a video call either with some incredibly useless agenda or using some of the more broken apps
chiastre has quit [Remote host closed the connection]
itorres has quit [Ping timeout: 258 seconds]
tsruser[m] has joined #nixos
chiastre has joined #nixos
itorres has joined #nixos
<bqv>
cannot import '/nix/store/gj372l5y2c1vdzc3zcbgz1m9xvzib6wa-supportedtypes', since path '/nix/store/fwsqxfsafbkr03vj1b8bijjpxkpmdm1d-supportedtypes.drv' is not valid, at /nix/store/z5avpibqqzb18180a0klqqpxwfsx6mfy-source/nixos/modules/services/misc/dysnomia.nix:161:25
<bqv>
That can't be good
<infinisil>
bqv: What's the command you used?
<bqv>
uhh, bog standard nixos-rebuild
cfinch has quit [Ping timeout: 265 seconds]
<infinisil>
pie_: Ah, it inf recs because of attrset strictness after all
<infinisil>
Because runCommand (even super.runCommand) takes dependencies from self
<infinisil>
And to evaluate self.<anything>, self."linux-${svf}" needs to be evals as well because strictness
<pie_>
yeah i guess that makes sense
<pie_>
so everything sucks, now what :D
teto has joined #nixos
<bqv>
Which means it should be fine if its a level lower
<pie_>
reimport nixpkgs
<bqv>
Like I suggested
<pie_>
bqv: oh hm i guesssss
<bqv>
Why do people never take my advice :p
<pie_>
but i dont really get how your level lower thing is supposed to work
<pie_>
bqv: well because it didnt work when i tried it :P
<pie_>
like, how does super.pkgs.thing even affect super.thing
<pie_>
also i just realized i did it wrong because i quoted it
<pie_>
"super.pkgs.${thing}" is not super.pkgs."${thing}"
<bqv>
Just wrapping the dynamic name it a constantly named attrset, so the keys of pkgs are constant again
<bqv>
Heh
<infinisil>
bqv: Do you have anything regarding dysomnia in your config?
<bqv>
I think I have it enabled
<pie_>
bqv: so what if i want to be stubborn and still override the conceptually correct attribute? any way to make it work?
orivej has quit [Ping timeout: 256 seconds]
orivej has joined #nixos
<bqv>
infinisil: I have disnix enabled, thats it it seems
<pie_>
i.e. with your thing i think self.thing != self.pkgs.thing
<bqv>
pie_: I just mean in your override, rather than have (self: super: { linux-bleh = ""; }) have (self: super: { kernels = { linux-bleh = ""; }; })
<bqv>
So bleh can vary all it likes with next to no effect
<bqv>
Plus, better namespacing
<infinisil>
bqv: So this error can happen when IFD is done, but the config is evaluated with a `nix-instantiate --eval` *without* --read-write-mode (which would enable IFD support for nix-instantiate --eval)
<infinisil>
And dysomnia indeed does IFD
<infinisil>
dysnomia*
<bqv>
Oh god
<pie_>
bqv: but i want to override the kernel in linuxpackages :P
<bqv>
I did do a dry build just before
<pie_>
bqv: since its already there and its what other thigns will use
<infinisil>
Ah yes a dry build would do that too
<bqv>
pie_: oh! Yeah, thats more complex then..
magnetophon has joined #nixos
m0rphism has joined #nixos
<pie_>
so the point is attrnames need to be resolved by the time anything in pkgs is accessible in the overlay, right? ...
o1lo01ol1o has joined #nixos
* pie_
scratches head
<pie_>
wat
vidbina has quit [Ping timeout: 272 seconds]
* pie_
goes to look at the runcommand source
alp has joined #nixos
drakonis has joined #nixos
<pie_>
right thats just called with callpackage
aveltras has joined #nixos
<bqv>
Just, conceptually, the dependency injection mechanics rely on pkgs having a pretty constant toplevel set of keys, so avoid making that toplevel dynamic
<bqv>
If you override something in linuxPackages should be fine
<bqv>
And in other subsets
<pie_>
it just seems crappy that that i cant call lib from an overlay
<pie_>
obviously thats ot the actual problem
drakonis1 has joined #nixos
<pie_>
like yes the implementation is broken but im not convinced yet that what i want to do is wrong
drakonis_ has quit [Read error: Connection reset by peer]
<metheflea>
anyone using wireguard? i'm a nixos/linux noob and i'm having troubles setting it up the way i want it to... i've got a wg0.conf file that i'd like to use, but installing the wg-quick package and doing "wg-quick up wg0 " doesn't seem to work (can't reach any websites). i then tried using the "networking.wg-quick.interfaces.*" nixos option and while it does work, it creates a systemd service that starts on boot and i'd much
<metheflea>
rather have it default to off when i start my system
_viz-- has joined #nixos
medvid has quit [Ping timeout: 240 seconds]
eoli3n___ has joined #nixos
_viz-- has quit [Remote host closed the connection]
_viz-- has joined #nixos
_viz-- has quit [Remote host closed the connection]
_viz_ has joined #nixos
eoli3n___ has quit [Remote host closed the connection]
_viz_ has quit [Remote host closed the connection]
eoli3n___ has quit [Remote host closed the connection]
eoli3n___ has joined #nixos
<ivegotasthma>
hello, how can I downgrade a dependency
<ivegotasthma>
I'm facing an incompatibility between biber and biblatex from nixos-unstable
orivej has quit [Ping timeout: 258 seconds]
<betawaffle>
can anyone here help me with lorri?
<ivegotasthma>
betawaffle: what's up with it?
<betawaffle>
i've got my editor (vscode) set up to use direnv and ccls, direnv set up to use lorri, and a shell.nix with my dependency in `buildInputs`
<betawaffle>
and *everything is working* in my editor
<betawaffle>
like, the IDE features and such
<betawaffle>
but as soon as i try to introduce a default.nix which actually builds my binary, the ccls functionality stops working
<betawaffle>
my default.nix has the same `buildInputs` as my shell.nix
<betawaffle>
and nix-build works
<ivegotasthma>
have you tried running the commands from vscode manually to see the output? perhaps there's an error vscode is hiding
eoli3n___ has quit [Remote host closed the connection]
eoli3n___ has joined #nixos
<betawaffle>
yeah, so when i start the language server in vscode with only shell.nix present, it shows that it has the correct includes passed to clang
<betawaffle>
but when i start it after bringing my default.nix back... sec
<{^_^}>
[nixpkgs] @SCOTT-HAMILTON opened pull request #88065 → zxing-cpp: init at latest commit from march-20-2019 → https://git.io/JfEy6
<betawaffle>
hmm, ok let me restart vscode...
Rusty1 has joined #nixos
<betawaffle>
ok, i think i've figured it out... it has something to do with direnv
<betawaffle>
everything continues to work if i start vscode inside the project's directory
<betawaffle>
but vscode has a direnv plugin that's supposed to load in that stuff correctly, and it was working yesterday when i *only* had a shell.nix
<betawaffle>
let me try that again to make sure it still works
<betawaffle>
hmm, nope. it's not even working with only shell.nix now.
<betawaffle>
must be something wrong with the direnv plugin
cosimone_ has joined #nixos
cosimone has quit [Read error: Connection reset by peer]
dugur has joined #nixos
<betawaffle>
ok, well i guess i just figured out the plugin isn't very good
<betawaffle>
could have sworn it was working yesterday, but maybe not
rauno has quit [Remote host closed the connection]
_viz_ has quit [Remote host closed the connection]
knupfer has quit [Remote host closed the connection]
knupfer has joined #nixos
eoli3n___ has quit [Remote host closed the connection]
KeiraT has quit [Remote host closed the connection]
eoli3n___ has joined #nixos
jmeredith has joined #nixos
KeiraT has joined #nixos
chaoflow has joined #nixos
chaoflow has quit [Client Quit]
chaoflow has joined #nixos
<{^_^}>
[nixpkgs] @flokli merged pull request #88026 → linux: Enable fbcon deferred takeover when possible → https://git.io/JfETy
<lux2>
Is anyone developing C# / dotnet core with vscode on nixos?
<lux2>
I cant get the debugger working
<lux2>
I've omnisharp-roslyn installed and set `omnisharp.path` in vscode. And the language server seems to work fine
<lux2>
just debugging, well, doesnt
<lux2>
dotnet-sdk_3 is installed as well and `dotnet run` works fine
reallymemorable has joined #nixos
<non>
Hi, does anyone have a moment for an odd issue? Sorry for the noob question, I did check as many nix pills as I could find. I'm getting complete system hangs from running `systemctl restart systemd-udevd-control`, but only during a `nixos-rebuild` (or after a `nixos-rebuild dry-activate`). I am pretty sure I know when I `nix-env -i`d the problem,
<non>
but am new to generations and not sure how to roll everything back (does `nix-env -G` really cover everything other than the contents of `/etc/nixos`?)
<non>
I must be misunderstanding a lot because I boot into an os where a lot of packages from my configuration are missing, and the mentioned `systemctl` command doesn't hang everything. :(
<{^_^}>
[nix] @edolstra pushed 6 commits to auto-uid-allocation: https://git.io/JfEQP
<srhb>
non: let's step back a bit. Normally, unless you've done something not very beginner-like, the system generations are not manipulated directly with nix-env by the user
<srhb>
non: So none of the nix-env commands you've used (again, unless you specifically pointed at the system profile, which is probably quite dangerous) have affected your nixos system
<srhb>
non: So the rollback generations you see for your systems don't care at all about what you may have nix-env -i'ed
<srhb>
non: (In general, they'll correspond to at most one generation per nixos-rebuild switch/boot command)
dugur has left #nixos ["Leaving"]
<srhb>
non: This is probably why rebooting into an earlier generation seemed more bare to you than you expected.
<non>
srhb: Thanks, I don't mind a super bare environment. If there is a way to uninstall all `nix-env -i`'d things and perform a rebuild from there, that might work. I'm 99% sure that the issue is with some kind of `udevd` conflicts (in which case I know the date range), but there's no mention in `/etc/nixos`, which means I've left some config
<non>
somewhere, or I need to get rid of something I installed. Is this presumptuous of me?
cosimone_ is now known as cosimone
<srhb>
non: I would personally be surprised if your nix-env'ed things are the culprit, but you can nix-env -e '*'
<srhb>
non: That will get rid of all of them
D_ has quit [Quit: No Ping reply in 180 seconds.]
<non>
srhb: Cool, I saw that in a pill, but would that be "scratch" or less than scratch? The docs seemed to indicate that it would also remove `nix-rebuild`, and running it from the store path seems like I'm misusing something...
<srhb>
non: OK, just to clarify, you're on a standard NixOS system, right?
D_ has joined #nixos
<srhb>
non: Moreover: You're not specifying --profile /nix/var/nix/profiles/system to any of your nix-env commands, right?
<non>
srhb: Standard nixos, (sorry, should have mentioned sooner) I'm on `19.09` installed using a nixos iso. I have not been using `--profile` for `nix-env`.
<srhb>
non: In that case, none of the nix-env commands you have been using have been manipulating the system profile (NixOS)
<srhb>
non: That means none of them will be able to remove the system nixos-rebuild command.
o1lo01ol1o has joined #nixos
<srhb>
non: Furthermore, none of those commands remove anything from the store. They just remove profile symlinks, (eg. what you see in ~/.nix-profile)
<bqv>
is there a convenience function for the `overlay: arg: let super = f arg; in super // (overlay super)` pattern
jkarni__ has joined #nixos
<bqv>
might be clearer what i'm intending with `overlay: (arg: let super = f arg; in super // (overlay super))`
jkarni_ has quit [Ping timeout: 265 seconds]
teto has quit [Ping timeout: 244 seconds]
<srhb>
bqv: It's almost extends.
<bqv>
yeah
<bqv>
but not quite, as far as i can tell
<srhb>
It's probably extends with some trivial argument. squints harder :P
<bqv>
:D
<{^_^}>
[nixpkgs] @ryantrinkle opened pull request #88070 → find-libs: extract from stage-1 and turn into its own package → https://git.io/JfE5l
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<non>
srhb: Fascinating, thank you. I will try that if all else fails. The reason I am thinking it might be something `nix-env -i`d is because I managed to get `stm8flash` working with much flailing (which I push share to repay this great community), but ended up with a mess of packages.
<srhb>
non: Okay. If you want to try and keep things a bit cleaner, you could try and avoid using nix-env for now (that is, work only with configuration.nix)
<srhb>
non: That way, each system you build will be bootable from the boot menu.
teto has joined #nixos
asbachb has joined #nixos
<non>
Would uninstalling chromium (rebooting several times and other debugging of the issue if it isn't fixed, but not calling any garbage collection), then installing with `configuration.nix` preserve my browser profile and tabs? Sorry again for the noob question, nixos is just so unintuitively powerful.
<srhb>
non: Your browser profile is stored somewhere in your homedir. NixOS itself does not touch that in general.
<non>
Ah, my bad, of course
<srhb>
This is very much a channel for learning, no need to be apologetic about it. :)
<bqv>
that does remind me though of that plan i had to have packages come with their own homedir each
<bqv>
well, half-baked plan
<bqv>
as all my plans go, tbh
<non>
I always feel bad for giving up on search engines and asking for help, so thanks :D This community is great
fusion809 has quit [Remote host closed the connection]
<srhb>
bqv: hrmmm is it `extends (const overlay) f arg` ?
linarcx has quit [Quit: WeeChat 2.8]
<srhb>
no, silly f is in the way.
<srhb>
Ack, can't be done :P
<bqv>
yeah, i'm thinking i'll just explicitly do it
vidbina has quit [Ping timeout: 258 seconds]
<srhb>
bqv: Oh, no, it was right though. :)
<srhb>
Squinted the wrong way.
<bqv>
ahaha
cfricke has quit [Quit: WeeChat 2.8]
<bqv>
hmm, ok, i'll try that
<non>
It still hangs as soon as I run the rebuild. It must be something to do with the state at boot plus the changes made by `nixos-rebuild`. Is it the "system profile" what I should be looking at for that initial state at boot? Also, I see `cannot connect to daemon` during the rebuild. Client and daemon seem to be same version, and am running as root.
<srhb>
bqv: `extends (const overlay) = let super = f arg; in super // (const overlay) arg super;`, `(const overlay) arg super = overlay super` -- if I got my substitutions right.
<bqv>
yeah that looks right to me
<srhb>
non: It sounds like your daemon is indeed not in a good state. what does systemctl status nix-daemon say?
<non>
CPU also seems to be going hard, which makes me think to wait, but it's 1AM and last time I waited for 20 mins with no luck.
Cdric[m] has joined #nixos
<srhb>
non: Is it "active (running)" ?
<srhb>
non: The processes maxing cpu may also be a clue.
<non>
Gotta reboot to check. `journalctl` is interesting but nothing obvious
<srhb>
non: Yeah it might be time to pastebin some general system logs and things to figure out the state of your system.
<non>
Hmm, looks like I must've `nix-env`d my terminal, and `xterm` flickers and dies instantly. What a mess :D
<non>
daemon is `active (running)`
<srhb>
_flickers_ -- okay, that's.. :)
<srhb>
It sounds like you're currently living a computing failure from Hollywood.
<non>
The cameras are well hidden, when do I get paid? :P
<srhb>
I'm not on the board, sorry. :-)
<non>
nix-daemon and nix both report `--version` of `2.3`
<srhb>
So far so good.
m0rphism has quit [Quit: WeeChat 2.7.1]
<non>
I wonder if it is worth noting that my grub menu only has one `nixos` - I always assumed that this was just because I needed to regenerate it.
<srhb>
non: I don't remember the default grub config -- it might just be a "folder" defaulting to the newest one, but which you can open to reveal each generation
<srhb>
non: If that's not the case, then yes, that sounds wrong. It _should_ have one for every sudo nixos-rebuild switch/boot (at least until you garbage collect system generations)
<srhb>
non: Try scrolling around with it next time you boot.
<la-s>
Does anyone know how to fix errors like /nix/store/rkj5n3p7v7mff4j4n6f2nld7mpfb550r-binutils-2.31.1/bin/ld.gold: fatal error: $WORK/b001/exe/a.out: No space left on device ?
<la-s>
I do have space left on my device, and I have a large swap
<non>
`df`? Might be that the partition you have that path in is full
alexherbo2 has quit [Ping timeout: 265 seconds]
cr4y1_ has quit [Ping timeout: 272 seconds]
<la-s>
that's a good point
<non>
Someone smarter than me can probably advise if a nix garbage collection is the right approach to reclaim that space
<la-s>
My $XDG_RUNTIME_DIR has only 296M of space!
<la-s>
wtf
<MichaelRaskin>
df -i
<la-s>
well, that is also very low
<la-s>
I assume it's due to my $XDG_RUNTIME_DIR being so small, I wonder why
<MichaelRaskin>
Maybe it is in tmpfs?
morr_ has joined #nixos
<la-s>
It is
<la-s>
I am just wondering why my tmpfs is so small when I have so much swap
<la-s>
I do only have 4G of RAM
lunik1 has quit [Quit: :x]
<srhb>
la-s: systemd/logind/something defaults to some fraction of your total ram for each user
<la-s>
how do you set this in the nixos configuration?
<la-s>
ah I found it
<la-s>
services.logind.extraConfig
lunik1 has joined #nixos
<non>
srhb: Something interesting is happening for sure, now my CPU is going super sayin even after booting and doing little else (`st` installed just fine using `nix-env -i` but flickers and dies much like `xterm`), however `i3` and `i3lock` seem to work (until the CPU goes insane). I might try to catch it with htop.
<srhb>
non: That would likely help, yes.
<non>
Definitely no options on boot
<srhb>
la-s: But if you have that little RAM you may just want to not have it on tmpfs at all.
<srhb>
non: And I assume you have done nixos-rebuild switch since you installed NixOS?
<la-s>
srhb: that makes a lot of sense, but do you know how to make the runtime directory a normal directory?
<srhb>
la-s: Actually no, might have been one of the casualties of the modern age.
<non>
Yep, many. It's been about 5 months and I have used `-p` at times to name them. I can see some of these names as I `ls` around `/nix/var/nix`
<la-s>
I have always made my /tmp a normal directory, but I have never found out how to make the runtime directory a normal directory
<la-s>
maybe I can just use mount --bind
morr has joined #nixos
<srhb>
la-s: I think this is something you should research carefully first, it sounds like potential dangerzone.
<non>
`rcu_sched detected stalls on CPUs/tasks` =L too busy/hot to run `htop` now, even straight after boot.
<la-s>
best way to research it is to try it lol
<srhb>
la-s: Sure, just, beware the potential data loss :-)
morr_ has quit [Ping timeout: 272 seconds]
<srhb>
non: I think we could use a dmesg dump on some pastebin
<srhb>
non: And it sounds like you're suffering a very low level issue.
cr4y1 has joined #nixos
<srhb>
non: I have no explanation for your system profiles missing from boot other than the fact that you've renamed(?) some system profiles. Which should at the very least give you multiple submenus in grub.
<non>
That might be tricky, I can't even get a terminal up now. I wonder if these issues are even related, though. Seems like the lower issue was cased by the `nix-env -e`. I recall having issues with displaylink early on, which I may have resurrected, but I am single monitor now and have a backup of the `nix-env -e` stdout.
<srhb>
Oh, displaylink
<non>
Cooling the laptop in the hopes that it will give me more time after boot :P
<srhb>
That'll probably break everything.
<non>
HAHA!
<srhb>
Get rid of that and I bet things start behaving normally.
<non>
Is there a good way to purge it? I have since seen the light and discarded every piece of cursed hardware, but the software spirits remain, it seems
<srhb>
Since you seem unable to rebuild and your system generations are borked, your best bet is probably to remount your system from an installer and rebuild from there.
<srhb>
non: With every trace of the cursed thing removed from your config files.
<non>
I remember weird BIOS level voodoo, I hope I don't need to go back there
<srhb>
non: You _could_ also try running an old generations switch script with boot manually
<srhb>
non: eg. find an old system-generation that is prior to displaylink in /nix/var/nix/profiles/system*, and run ./that-generation/bin/switch-to-configuration boot
<srhb>
But since your boot enu is weird I don't know :)
<srhb>
s/enu/menu
<evanjs>
Sooo it looks like ntfy is broken for python27, which is what it is currently (statically/implicitly) set to (all-packages reference is ntfy = callPackage ../tools/misc/ntfy {};)
<evanjs>
ntfy = python3Packages.callPackage ../tools/misc/ntfy {}; seems to remedy this issue
<evanjs>
The failure is due to "AttributeError: 'module' object has no attribute 'test_xmpp'"
jakobrs has joined #nixos
zupo_ has joined #nixos
<bqv>
move it from all-packages.nix to python-packages.nix maybe? afaik python-packages is then spread across all python versions
<evanjs>
bqv: yeah that was my first thought!
<bqv>
just means rather than ntfy it'd be pythonPackages.ntfy
<evanjs>
like why is it in all-packages??
endformationage has joined #nixos
<evanjs>
Okay so that makes sense. the fact that it is in all-packages with pythonPackages imported, means that all-packages is then only ever passing in python2, right?
<evanjs>
and then when we move it to python-packages, it will allow us to get whatever python version we want, and should then also build correctly on each that it supports?
<srhb>
Should^tm
<evanjs>
for sure haha
<bqv>
wow that was hard to lex
<evanjs>
should I leave it in tools/misc or move it to development/python-modules btw?
<bqv>
should'v'nt'd
dermetfan has quit [Ping timeout: 240 seconds]
orivej has joined #nixos
<non>
srhb: Hm, how easy do you think it would be to boot onto a (newer) installer, cut off a new partition, and mount my old home, copy (some of) my old config.nix, etc. I've never tried this with nixos, but it's usually pretty painless, right?
zaeph3 is now known as zaeph
dansho has quit [Quit: Leaving]
jakobrs has quit [Quit: leaving]
<srhb>
non: Extremely easy
<srhb>
non: You don't even have to cut off a new partition imo
<srhb>
non: Just skip the partitioning steps and remount your existing system
<srhb>
non: Possibly nuking /boot as the exception
<non>
!?
<srhb>
non: (Since it seems to be in a weird state -- and maybe avoid using the profile names)
<{^_^}>
[nixops] @adisbladis opened pull request #1347 → Run --pdb check early → https://git.io/JfENG
justanotheruser has quit [Ping timeout: 244 seconds]
<srhb>
non: NixOS won't touch your existing things. Installing is almost the same as nixos-rebuild
<srhb>
non: So on a system with no changes, redoing the install is basically a no-op.
<srhb>
non: But fixing your bootloader sounds desirable, so I'd redo that.
<la-s>
srhb: well it works, but logind recreates the tmpfs when I log in a second time, so an acceptable solution for me. I just did it using filesystems."/run/user/1000"
<non>
But that's the opposite of what I want, I'm thinking that I've managed to install some displaylink spirits deep in the system somewhere
<srhb>
non: My guess is that you're just unable to switch to a system that doesn't have displaylink configured, but we can try to confirm that
<non>
Oh, so you mean to manually delete `/boot` and then just use the installer as a way to boot in the meantime?
<srhb>
non: That you'd somehow be able to load it from outside the NixOS config sounds unlikely
<evanjs>
ahhhhh
<evanjs>
they also import pythonPackages in the build expression
<srhb>
non: Unless you've done some really invasive stuff to get displaylink to, um "work"
<evanjs>
I think that's step two of the problem
<evanjs>
remove all references of that and import the deps in the { }: thing (always forgot what that's called lol)
<srhb>
non: No, I meant "install on top of your existing system except wipe boot"
<non>
I don't know, I'm just smart enough to install things where they don't belong. I vaguely remember installing some package the non-nix way and having to reboot to do some secure key stuff with bios.
<srhb>
eeee. :)
alexherbo2 has joined #nixos
KeiraT has quit [Ping timeout: 240 seconds]
<srhb>
non: Well, I'd redo the install.
<srhb>
non: Whether or not you want a new partition is your call
<srhb>
non: If it's _necessary_ you've done some very funky things. :-) if not, no harm done anyway.
<non>
Yep, I'd rather cherry pick in than the opposite at this stage, I was hoping to get some work done tonight but no luck haha. At least I've learned a lot.
<srhb>
non: The reason I'm being adamant about this is that "things installed the non-nix way" _should not_ affect your nixos system in the end.
<non>
Well I'm happy to do some diagnostics if you can think of any commands to check with. First I need to burn a USB to boot into.
teto has quit [Ping timeout: 260 seconds]
simba1 has joined #nixos
<evanjs>
hrm... it's still getting angry with the 2.7 version
<srhb>
non: Well, I'd like to see you switch into an older system configuration first. Your first one, if you haven't gc'de that, for instance.
<srhb>
non: In my mind that should be equivalent to re-installing, re. displaylink (if you didn't already have it set up there)
teto has joined #nixos
tsrt^ has joined #nixos
<non>
Any way to do that pre boot? `?` at GRUB gives me: `(d)efault, (t)imeout, (e)dit, (v)ersion, (Q)uit, (P)rint, (h)elp`
<srhb>
non: Not without knowing the store path of the system in question. actually, nevermind, too much hassle. Go the installer route.
cole-h has joined #nixos
<non>
initrd thing that `edit` drops me to is full of hashes :(
<srhb>
Easier on a local machine :-)
<srhb>
non: Yes, it points to exact versions.
KeiraT has joined #nixos
<non>
Well it is reassuring to know that you really believe nothing can change without lots of funk, I just think that in this case I have more funk than either of us anticipated
jakobrs has joined #nixos
<srhb>
non: Yes, and I don't have to tools to prove you wrong. :-P So you win this round.
proofofkeags has quit [Remote host closed the connection]
orivej has joined #nixos
<infinisil>
pie_: You can just look at the dependencies of your system or so
proofofkeags has joined #nixos
<pie_>
infinisil: i want to script stuff though
jakobrs has left #nixos [#nixos]
<pie_>
the kernel config option tooling sucks so i need to do some diffing between no options set and calling the kernels option merging script insteadof the nix option merging script
<pie_>
and once i do that i need to feed the output to the nix stuff
alp has quit [Ping timeout: 246 seconds]
<pie_>
well, two things, one is i want to see whats missing the other is i want it to work, its probably sufficient to just do the latter if i just want it to work hm
<pie_>
namely, call the kernels config thing instead of nixos'
virus_dave has quit [Read error: Connection timed out]
<pie_>
bleh
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<cole-h>
bbigras: Ping; you around?
<bbigras>
cole-h: yes
<evanjs>
Or wait, maybe this is the issue that's causing the python2 build to fail... hrm...
<cole-h>
That's what I get after `nix-build -E '(import ./nixpkgs {}).waybar.override { withMediaPlayer = true; }'`
proofofkeags has quit [Ping timeout: 256 seconds]
Whaze has quit [Ping timeout: 260 seconds]
<bbigras>
cole-h: oh I think my initial PR had the stuff for playerctl but a review made me think it was only part of my waybar config file and that I had to remove it.
<cransom>
gchristensen++ for one-at-a-time in hydra.
<{^_^}>
gchristensen's karma got increased to 297
<cole-h>
bbigras: Adding playerctl to propagatedBuildInputs shows output as expected
<gchristensen>
:) you're using it, cransom?
<bbigras>
cole-h: thanks. I'll add it back.
MmeQuignon has joined #nixos
<cransom>
as of now, i am.
<gchristensen>
nice! what for?
<bqv>
pie_: you could make nixpkgs's version better instead :p
<cransom>
our hydra does some deploy-y things and the processes get upset if someone ships a change before the previous job is finished.
<cole-h>
Can anyone tell me: why does `groups` report only wheel and users in sway systemd unit, but on TTY it shows wheel, users, audio, and input (as it should)???
gagbo has quit [Quit: I'm out !]
<cole-h>
s/in sway/in my sway/
<cransom>
hasn't happened in a while, but now it definitely won't happen
<ikwildrpepper>
gchristensen: yeah, excellent feature, especially for some longrunning benchmarks
<gchristensen>
:) cool!
<pie_>
bqv: im writing up an issue right now but that seems lik more of a dig than im up for right now
matthewcroughan has quit [Quit: Connection closed]
proofofkeags has joined #nixos
cole-h has joined #nixos
<{^_^}>
[nixpkgs] @mroi opened pull request #88075 → libmilter: fix compilation on Darwin → https://git.io/JfExb
fabianhjr has joined #nixos
<bqv>
they won't be garbage collected during build, and if you use out-link (i think it's on by default) then the results symlink prevents them being gc'd too
proofofkeags has quit [Ping timeout: 260 seconds]
<dave87>
request for help: I am setting up my new computer (catalina), i've made some baby steps (https://dev.to/louy2/installing-nix-on-macos-catalina-2acb + milti-user install) & not I wish to declaratively have vim installed (later to be reconfigured with an overlay) but i'm struggling to take this next step. which file should I be creating/editing?
<dave87>
default.nix, config.nix, configuration.nix & what would the most basic example of this file look like? thanks
amanjeev has left #nixos [#nixos]
<dave87>
*& now I wish
<bbigras>
bqv: sorry yes. I mean the next time I run `nix-collect-garbage`. But even with a `result`, it seems the deps are gone after `nix-collect-garbage`. If I run `nix-build` again, it fetches a lot of stuff. I wonder if it's because I'm using crate2nix.
<emily>
,locate par
<{^_^}>
Found in packages: go, par, opae, mcrl2, gecode, go_1_12, go_1_13, gap-full, python37Packages.pybids, python38Packages.pybids
<bqv>
bbigras: true, `result` only keeps the closure of the result, other stuff might be required for the build itself
<bqv>
there's no real way to keep those without pinning them manually or making the result depend on them
<emily>
gah
<emily>
the amount of times I see error: flake 'flake:nixpkgs' does not provide attribute 'par'
<emily>
and don't understand that it means "nix run is now nix shell", not "nixpkgs#par isn't a package"
<bbigras>
bqv: gotcha. thanks!
orivej has quit [Quit: No Ping reply in 180 seconds.]
dm9 has quit [Quit: WeeChat 2.8]
amanjeev has joined #nixos
orivej has joined #nixos
<infinisil>
dave87: home-manager is probably the best way
<infinisil>
Or nix-darwin actually
<infinisil>
I'd go for home-manager
dermetfan has quit [Ping timeout: 256 seconds]
drakonis has joined #nixos
<bqv>
emily: i never see that
<emily>
bqv: perhaps your nixFlakes is old
<infinisil>
dave87: Once you have home-manager installed, you can set the `programs.vim` or `programs.neovim` options in home.nix as documented in https://rycee.gitlab.io/home-manager/options.html
Whaze has joined #nixos
<bqv>
emily: it's new enough that i've been using `nix shell` for days now
Whaze has quit [Client Quit]
<bqv>
i'm using the nix flake itself, not nixFlakes
<emily>
bqv: well, nix run nixpkgs#foo doesn't work on any foo that isn't a magic nix app thing now
<emily>
so I run into the error a lot because of cached muscle memory
<evanjs>
Alright, looks like I really only need to test pytorch (in terms of tf packages to test) for 77714 as it has a direct dep on pillow. Hopefully it will succeed in building on my NVIDIA computer...
<evanjs>
poor laptop... lol
<lovesegfault>
eoli3n: I am omnipresent :D
<dave87>
infinsil: awesome, thanks. is there a reason why a second project is the preferred method than just a pure nixpkgs solution? due to the different OS's?
<bqv>
emily: yeah, i don't know how you've managed to get a nix like that, nixpkgs#blah still works fine for me
<evanjs>
But you can integrate it into your nixos-configuration as a submodule, which allows you to keep everything "managed" by your nixos configuration
selfsymmetric-mu has joined #nixos
<bqv>
and i can't see anything in the commit history to do with that at a glance
<emily>
bqv: I promise this is a deliberate change and not me misunderstanding things. nix run now actually runs things; you can do nix run blender-bin and it pops up some random Blender binary eelco packaged.
<emily>
see the issue I linked
<selfsymmetric-mu>
Once a nixpkgs PR is approved, who merges it into master?
<cole-h>
lovesegfault: Like, I start it from TTY and I get wheel, users, input, audio, but I do `systemctl start --user sway` (after doing the env stuff) and `groups` only says users and wheel...
<emily>
by necessity those things therefore have to be the weird flake app things and not random derivations
<emily>
nix shell is what nix run once was (and is distinct from nix dev-shell)
<evanjs>
dave87: obviously the module isn't usable on darwin, but home-manager is what allows you to get the level of configuration nixos configurations typically provide on platforms that are not on nixos, yeah
<bqv>
emily: oh! you mean you've been using `nix run nixpkgs#blah` and expecting it to not run things
<bqv>
i get you
<bqv>
yeah, i never used nix run, so i don't have that issue
<emily>
bqv: I've been expecting it to do anything at all
<emily>
it doesn't even look up packages in the same way as nix shell
<emily>
try it for yourself, pass anything you want after nixpkgs# and you'll get the same error
teto has joined #nixos
<emily>
it's looking at stuff on the flake, not in legacyPackages
<dave87>
ok, thanks all. :)
<emily>
but yes, it's just my muscle memory being broken, I mean to use nix shell
<lovesegfault>
cole-h: starting sway from a system user unit is not worth it IMHO
<bqv>
i just see:
<bqv>
error: unrecognised flag '--command'
<lovesegfault>
too many rough wedges
<bqv>
Try 'nix --help' for more information.
<emily>
the error message is just very confusing (also, it seems weird and terrible for nixpkgs#<foo> to mean different things depending on command)
xcmw has joined #nixos
<emily>
bqv: sounds like you have a broken wrapper
shibboleth has joined #nixos
<emily>
if it's for fish, nix already looks at $SHELL now so you don't need it
<cole-h>
lovesegfault: But I realllly like `journalctl -u sway --user` :( Yeah, I might follow in your footsteps and just `exec sway` instead of `systemctl start` it :(
<bqv>
oh, heh
<lovesegfault>
cole-h: you can direct sway output to journalctl
<lovesegfault>
although I don't remember how
<cole-h>
lol
selfsymmetric-mu has left #nixos ["gone to the land of dead hiccups and extinguished light bulbs"]
<evanjs>
You don't get any of the abstractions home-manager offers, and end up needing to e.g. override packages, and configure vim in one of the billion ways listed on https://nixos.wiki/wiki/Vim, etc x_x
<cole-h>
lovesegfault: ALSO, do you like your cursor theme? Mine is kinda janky (no hand when hovering clickables), but I don't know if it's because I have `seat * xcursor_theme default 24` in my sway config...
<dave87>
evanjs that looks fine on first look. I prefer one dep to two… I'll checkout home-manager though. Thanks for the options
<lovesegfault>
cole-h: it's meh, I don't put much thought into it
<cole-h>
But do you have a hand when hovering over clickables? That's the important part :P
<evanjs>
dave87: yeah it can be fine for smaller things, it depends on how much you want to do and if you want to extend/break things apart for ease of maintenance and etc later
<lovesegfault>
cole-h: I do not
<cole-h>
Darn.
qyliss has quit [Quit: bye]
<cole-h>
It reallllly bothers me... so I'll need to figure out how to fix that.
<finnwww[m]>
arg xserver keeps saying "failed to start session'
<dave87>
I'll see how the non home-manager way goes and report back sometime this week. 👍
qyliss has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<fresheyeball>
anyone have experience with cachix?
<fresheyeball>
join #cachix
<finnwww[m]>
anyone have experience with bspwm?
numkem has joined #nixos
teto has joined #nixos
<numkem>
I'm starting to look at hydra for both at home and at my work. Was wondering if there was a way to copy the built store path to a remote host after the builds are done? I'd like to use a NAS to store the results.
<buffet>
hey, im trying to drop into a nix-shell with texlive.combined.scheme-basic, with /nix being compressed via btrfs, however nix tries to remove the compression for /nix/store/...-texlive-combined-basic/bin/fmutil (not sure why) and fails to do so (missing permissions)
<numkem>
srhb: is there documentation regarding a input of type git ssh checkout?
<buffet>
is there any workaround, or can i stop it from trying to remove the compression?
teto has joined #nixos
<superbaloo>
is there a way to inline a shell script in "src" of a mkDerivation?
<ajs124>
pkgs.writeShellScript?
<srhb>
numkem: Barely. What do you want to know?
nixbitcoin has quit [Ping timeout: 240 seconds]
<numkem>
srhb: I think I might have found it. I have a private gogs repo and i need to clone using SSH, if I put the value of a git clone as `ssh://root@<server>:<por>t/folder/project.git` does that makes sense?
asbachb has quit [Ping timeout: 245 seconds]
nixbitcoin has joined #nixos
roconnor has joined #nixos
<roconnor>
what are the flags that stdenv adds to gcc on linux?
T0pH4t has joined #nixos
<superbaloo>
ajs124: looks like it works
<superbaloo>
thanks
vidbina has quit [Ping timeout: 264 seconds]
<srhb>
numkem: I think so, yes. The main issue is usually figuring out who should have the key and ssh config. Probably the hydra user, in its homedir.
<numkem>
srhb: that's exactly what I did, I'd have to provision that key through configuration.nix at some point
<numkem>
is there an option to write any file anywhere? I know there is one for `/etc/...`
<T0pH4t>
so does nixpkgs.config.allowUnfree not work for the primary configuration.nix on nixos-rebuild? I keep having to use NIXPKGS_ALLOW_UNFREE=1
<non>
trying to install fresh, and at the last stage: `File system "/dev/block/..." has wrong type for an EFI System Partition (ESP).` Does anyone know what this means? Maybe my 4AM google foo isn't so good.
<srhb>
numkem: No. And you probably should not try too hard. You can generate the key as a one-off via eg. a systemd-unit which may place a file anywhere, but anything generated in a build gets into the nix store, and thus becomes world readable.
<numkem>
srhb: very true, that sounds like a better approach
<srhb>
non: Sounds like it's not a FAT filesystem
<srhb>
non: An EFI install should have an efi system partition which is formatted via eg. mkfs.vfat
<srhb>
non: Either that or it's complaining about the partition table type associated with it.
<numkem>
trying to get back to my newly created jobset i get this error message in the logs for hydra-server: [error] Couldn't render template "date error - bad time/date string: expects 'h:m:s d:m:y' got: ''"
<non>
FAT32, set up via gparted (perhaps not formatted completely?). No boot flag, pursuing how to set that up, and I found something about setting a specific GUID.
o1lo01ol1o has quit [Remote host closed the connection]
<srhb>
non: OK, probably the partition flag, yeah. It's probably easier to change from the term, eg. with gdisk
<maddo>
hello, I have a question: I have two SSD drives (SATA) (MX500 250GB and Sandisk Ultra II 240GB), I have two different OS on each but now I can discard windows entirely.
magnetophon has quit [Read error: Connection reset by peer]
<maddo>
That being said, I wanted to do a RAID 0 btrfs with those two drives (the size mismatch shouldn't matter right?) and use the remaining space for ESP and swap
<maddo>
is it viable? Are there strong cons I should be aware of?
<srhb>
maddo: I think people fall in rather extreme camps re. btrfs -- either people think it's great or that it eats their data. ymmv. Other than that, just the usual raid 0 warnings. And I don't know to what extent one drive dead = all data gone with btrfs.
<maddo>
I'm more on the "it just works" camp with btrfs in general, but I never really attempted raid0
magnetophon has joined #nixos
<non>
srhb: Note for future explorers: Solved with `fdisk` using `t`
<srhb>
non: Good :)
magnetophon has quit [Read error: Connection reset by peer]
<cole-h>
Partitioning is one of those things that I just follow instructions from install guides... It's a bit esoteric to me (though if I spent some time learning the tools I'd imagine it would be easier).
<srhb>
cole-h: I'm one of those old school (or just old) people that think it's actually pretty simple once you get rid of all the fluffy gui tools that just confuse the matter. :-P
magnetophon has joined #nixos
<srhb>
But with practice etc., for sure.
<cole-h>
Maybe I'll learn {c{,f},g,f}disk some day...
<cole-h>
I've only partitioned twice ever (in recent memory) -- my initial Arch install ~2 years ago, and yesterday when I began setting up my test system :D
o1lo01ol1o has joined #nixos
orivej has quit [Read error: Connection reset by peer]
<srhb>
cole-h: cfdisk iirc is really useful for mass deployment formatting.
<srhb>
cole-h: Ah yes, okay :)
<srhb>
Before NixOS it was like a pastime for me.
<srhb>
Distrohopping, reformat, rinse and repeat.
orivej has joined #nixos
<cole-h>
Well, it is a test disk, so I'll be doing it again sometime soon
<srhb>
:)
zupo has joined #nixos
waleee-cl has joined #nixos
zupo has quit [Client Quit]
xd1le has quit [Quit: Quit]
<bqv>
in case anyone else here has issues migrating their ipfs repos, i got it to work by linking /var/lib/ipfs to /tmp/.ipfs and manually running `sudo -u ipfs env HOME=/tmp /nix/store/*ipfs-migrat*/bin/fs-repo-migrations`
<pbb>
My dhcpd fails to assign new IP addresses after a new system configuration with new networking scripts was activated, because dhcpd is not restarted but ISC DHCP can not deal with the interface being gone and appearing again. What systemd serviceConfig option could I set so that the dhcpd4 service is restarted when the network (in this case a specific bridge) is restarted?
teto has quit [Ping timeout: 260 seconds]
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
orivej has quit [Quit: No Ping reply in 180 seconds.]
morgrimm_ has joined #nixos
morgrimm has joined #nixos
orivej has joined #nixos
xd1le has joined #nixos
<jtojnar>
emilazy just noticed there is applyPatches in trivial builders
xd1le has quit [Client Quit]
xd1le has joined #nixos
<jtojnar>
emily^
proofofkeags has quit [Remote host closed the connection]
proofofkeags has joined #nixos
morgrimm_ has quit [Ping timeout: 246 seconds]
morgrimm has quit [Ping timeout: 246 seconds]
sherlock has joined #nixos
proofofkeags has quit [Ping timeout: 265 seconds]
ixxie has joined #nixos
astrofog has joined #nixos
sherlock has left #nixos ["WeeChat 2.8"]
teto has joined #nixos
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<non>
srhb: Just wanted to let you know that everything is working now, all ported, chrome starts where it was left, a long overdue reboot has been performed (in a roundabout way), all issues seem to be resolved. Thank you for your help and praise be to NixOS!
zupo has joined #nixos
<non>
Oh and, GRUB menu works :D
xd1le has quit [Quit: Quit]
dave87 has quit [Remote host closed the connection]
xd1le has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
<numkem>
srhb: do you happen to have a job example that is with custom packages? I'm not sure how to write the expression to make hydra happy
<srhb>
non: Great to hear! :)
<srhb>
numkem: I'm afraid it's been too long since I used Hydra -- I don't have any at hand, sorry. Perhaps take a look at the simpler ones on hydra.nixos.org
<srhb>
The ones that haven't been flakified at least
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<srhb>
Sure thing. Good luck. Heading off for the night. o/
<numkem>
srhb: have a good one!
xantoz has quit [Quit: WeeChat 2.8]
<T0pH4t>
is there a way to have nix-rebuild print out why it would need to fetch a resource from git. EG I'm guessing it can't find one of the deps in a binary cache and i would like to know which one.
<{^_^}>
[nixpkgs] @jtojnar opened pull request #88086 → gdk-pixbuf: move xlib into separate package → https://git.io/JfuUJ
proofofkeags has joined #nixos
<MichaelRaskin>
T0pH4t: before doing anything, it should print the list of things to build and things to fetch
xantoz has joined #nixos
<T0pH4t>
it doesn't, just says executing 'git'
<T0pH4t>
which it can't find, because i didn't install it
<T0pH4t>
it shuld have no reason to install git, if it has all bins in cache
<T0pH4t>
or am i missing something?
<{^_^}>
[nixpkgs] @jtojnar pushed 431 commits to staging-next: https://git.io/JfuU8
orivej has quit [Read error: Connection reset by peer]
<emily>
ah
<T0pH4t>
hmm I'm assuming binary cache settings also are not inherited (from host) by containers, so i need to set those again. Maybe thats related to why it needs git
<emily>
I think nixpkgs might actually try to execute host git in a couple of places :(
<emily>
to do things like get the nixpkgs version
orivej has joined #nixos
<emily>
hm, maybe not. I know it reads .git but I don't know about executing git itself
dingenskirchen has quit [Ping timeout: 246 seconds]
<asbachb>
Any iptable pros? I've got a setup with ufw + ubuntu (host) + lxd + NixOS (guest). I setup a mail server. Incoming smtp traffic is working outcoming. I guess the iptables is somehow screwed.
<bqv>
:)
domogled has quit [Remote host closed the connection]
<asbachb>
+ NixOs firewall ^^
Neo-- has joined #nixos
<ajs124>
asbachb: What is working and what isn't?
<asbachb>
ajs124: For know everything beside outgoing smtp traffic. But it seems a problem with the guest. The host is able to telnet other smtp servers, but the guest is not.
<quinn>
is anyone using the asterisk module in production? i'm looking to discern whether it's worth testing out
<ajs124>
that sounds more like a problem with your lxd/ubuntu setup. can the guest do any other outgoing traffic?
<asbachb>
ajs124: Let me check. I thought it does.
zupo has joined #nixos
<asbachb>
ajs124: I guess the setup is somehow broken. Now there's no connectivity at all inside the container. It seems that some process is screwing up iptables configuration.
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
<ajs124>
I don't know much about LXD, so I don't think I'll be of much help. The NixOS firewall doesn't do any outgoing filtering (at least by default), as far as I know, so it really shouldn't be a problem.
zaeph1 has joined #nixos
<asbachb>
I guess there're currently two problems. One problem I described before and another that the entire networking is getting somehow screwed.
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<asbachb>
I saw something like `valid_lft 3591sec preferred_lft 3141sec` in `ip addr`. What happens afterwards? dhcp is queried for a new ip?
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
virus_dave has joined #nixos
Maxdamantus has joined #nixos
zaeph1 is now known as zaeph
<virus_dave>
hey folks! on one machine, while trying to move /nix/store onto another device, i seem to have missed a few entries somehow, so now some commands like `nix-env -iA nixpkgs.bashInteractive` fail because /nix/store/aaaaaaaa-texinfo/ is missing. I tried to `nix-store --repair-path /nix/store/zzzzzzz-texinfo` it but that fails. Nothing i had should have been locally built, so how i can i repair my local store? Seems like just repopulating locally from
<virus_dave>
or perhaps a better way to ask: It seems like nix is locally keeping some kind of "expected store contents" independent of just "what's in the /nix/store". How can I nuke this, so everything that gets built in the future just attempts to reinstall?
brandonhamilton has quit [Client Quit]
oborot has joined #nixos
<energizer>
virus_dave: (not an expert) when that's happened to me it was solved by nix-collect-garbage
<oborot>
I can't seem to install steam using nix, it seems it's not able to download dependencies.
<oborot>
Getting this error: error: cannot download steam_1.0.0.56.tar.gz from any mirror
<energizer>
virus_dave: specifically, `nix-collect-garbage -d`. but i've heard if that goes wrong it can leave the system unbootable
asymptotically has quit [Ping timeout: 240 seconds]
<bbarker>
what's the simplest way to import a nix expression from a github repository via http(s)? I wouldn't be against avoiding the need to set up a proper nix package repository (like NUR)
<lewo>
bbarker: you can use builtins.fetchurl
<hyperfekt>
virus_dave: nix-store --verify --repair could do it maybe
<bbarker>
lewo, ah, thanks, think i found an example of that, will try it
<bqv>
how does one override python packages
<lewo>
bbarker: for a less simple but more convenient way, niv is pretty nice
<asymmetric>
error is: error: the string '/nix/store/ljap4xgjw4gimgzj88dxs41nc672d0nn-yq-2.7.2/bin/yq . < /nix/store/sk5n35njn0fpi3fzdkvgx7pwbp7djgd4-solarized-light-256.yml' is not allowed to refer to a store path (such as '!out!/nix/store/y0371y4cv7i7xsp0m2if9m8933aa2rw8-yq-2.7.2.drv'), at /home/asymmetric/test.nix:5:7
<kiouoni>
hiya. I'm trying out nix. I had some issues with declarative package installing (as from https://nixos.org/nixpkgs/manual/#sec-building-environment) and am just installing packages with nix-env -i, now. Is there a way to generate a configuration file based on what you have installed with nix-env? I am using elementaryos, if it matters.
Maxdamantus has quit [Ping timeout: 258 seconds]
o1lo01ol1o has joined #nixos
Maxdamantus has joined #nixos
orivej has quit [Ping timeout: 260 seconds]
orivej_ has joined #nixos
<infinisil>
kiouoni: Nope that's not possible
o1lo01ol1o has quit [Remote host closed the connection]
zeta_0 has joined #nixos
<chloekek>
You can scan PATH for executables, resolve the symlinks to store paths, then look up which derivations they come from, hope that you can recover the nixpkgs attribute names, and then subtract system packages. It’s not so trivial.
<asymmetric>
adisbladis: so passing the contents of the file to the yaml parser, instead of the path?
<romildo>
How to check in a script if it is being run in NixOS or not?
cole-h has joined #nixos
drakonis1 has quit [Ping timeout: 258 seconds]
<T0pH4t>
is there a way to temporarily disable req of package signing in configuration.nix? nix.conf seems to use require-sigs=false, but can't seem to set that in configuration.nix
drakonis1 has joined #nixos
<adisbladis>
asymmetric: Well, yes and no. The important bit here is to avoid string context where it's not allowed.
kiouoni has quit [Remote host closed the connection]
<adisbladis>
T0pH4t: 1. Why do you think you want that?
MmeQuignon has quit [Ping timeout: 260 seconds]
<T0pH4t>
because i am testing a proof of concept right now and dont' want to deal with that
<adisbladis>
Well, the reason why you can do so and how they can be built from a reference in a config file, or systemd unit or something similar is because that string actually retains a reference to the derivation
<{^_^}>
the string '{ "hello": "/nix/store/df15mgn0zsm6za1bkrbjd7ax1f75ycgf-hello-2.10" }' is not allowed to refer to a store path (such as '!out!/nix/store/y40xd3jgvsc01wazsipq0shqlfhx76p5-hello-2.10.drv'), ...
<adisbladis>
Because attrsets do not carry any context there isn't a valid way to go from a string with context to an attrset
<{^_^}>
[nixpkgs] @CorbanR opened pull request #88091 → Do not strip binary when darwin(OSX) → https://git.io/JfutG
<infinisil>
If a build is failing, it wouldn't be cached
<T0pH4t>
well, i mean for a local binary cache
<T0pH4t>
i've published binaries to an s3 from one machine and then when i go to another it won't use those binaries
<T0pH4t>
both running linux
<infinisil>
Ah
<T0pH4t>
infact i have ah ost machine linux, which spins up a nixos vm
<T0pH4t>
the nixos vm can't seem to use the binaries i just built/published from host
<infinisil>
T0pH4t: If you have some expression you can `nix-instantiate`, check the output of that command on both machines
<hexa->
i've split up my nftables rules into multiple files that are included from networking.nftables.ruleset. How would I get nftables.service to reload while keeping the firewall rules in separte module?
virus_dave has quit [Quit: virus_dave]
<T0pH4t>
but would that show arch/build setting diffs?
<T0pH4t>
i'm assuming something in vm is different from my host box
<infinisil>
T0pH4t: I'd first verify that nix-instantiate gives the same result on both machines
<infinisil>
If it does, then it's a problem with the cache. If it doesn't, then it's a problem with the nix evaluation being impure
<infinisil>
Ah yes, that's nice for getting an idea where to look if it differs
<asymmetric>
adisbladis++
<{^_^}>
adisbladis's karma got increased to 77
<adisbladis>
<3
<T0pH4t>
emily: thx for that!
<cole-h>
If I have `hardware.pulseaudio.enable = true;` but I don't want the service to start, can I just mask it? Or will NixOS overwrite that masked unit?
xcmw has joined #nixos
<cole-h>
Oh, I can probably just `systemd.user.services.<name>.enable = lib.mkForce false;`, right?
<infinisil>
cole-h: Yup, though that will turn it off completely. You can set `serviceConfig.wantedBy = lib.mkForce []` to only prevent it from starting automatically
<cole-h>
I want to turn it off completely, so that's perfect :P
<cole-h>
The jack service doesn't really work for me, so I use home-manager and cadence+jack2 -- maybe I'll look into that once I've got this completely set up
mallox has quit [Quit: WeeChat 2.8]
<cole-h>
But jack needs to manage pulseaudio itself
CptCaptain has joined #nixos
<iwq>
are anyones fans significantly louder on nixos than on windows? it's strange, on my previous hardware it used to be the opposite
<CptCaptain>
Can someone explain this to me? Why does "\'" evaluate to "'" while "\\'" evaluates to "\\'"? How do I get "\'"?
<simpson>
iwq: You could check your governor and powertop. In general, you'll use the same techniques on NixOS as you would on other Linux distros.
<MichaelRaskin>
> builtins.stringLength "\\'"
<{^_^}>
2
<MichaelRaskin>
> builtins.stringLength "\\"
<{^_^}>
1
<MichaelRaskin>
CptCaptain:
<infinisil>
,escape "\'"
<{^_^}>
Escape this in '' strings with: "\'"
<{^_^}>
Escape this in " strings with: \"\\'\"
<infinisil>
CptCaptain: ^
<CptCaptain>
{^_^}: Thinking if it this makes sense now
<infinisil>
,escape \'
<{^_^}>
Escape this in " strings with: \\'
<{^_^}>
Escape this in '' strings with: \'
astrofog has quit [Quit: Quite]
<cole-h>
,escape '
<{^_^}>
Escape this in " and '' strings with: '
<MichaelRaskin>
CptCaptain: basically the point is that the _output_ escapes the backslash
<MichaelRaskin>
> ''\''
<{^_^}>
"\\"
<CptCaptain>
Yes, ty
<infinisil>
Ahh
xcmw has quit [Ping timeout: 260 seconds]
lemsip has quit [Read error: Connection reset by peer]
<MichaelRaskin>
Because it _has to_ use the foo group
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
<cole-h>
infinisil: As one of the resident Nix masters, I'm hoping you can help... I have a folder full of scripts, in `script.sh` and `script.py` format. Is there any way I could generate an attrset based on those files?
<CptCaptain>
What regex will yield [ "foo" "foo" "foo" ]
<cole-h>
(Assuming no conflicts) e.g. go from `script1.sh` to `script1 = toString ./scripts/script1.sh;` and `script2.py` to `script2 = toString ./scripts/script2.py;`
<cole-h>
Is there an easy way to do this?
<infinisil>
cole-h: builtins.readDir is your friend
romildo has left #nixos ["Leaving"]
<cole-h>
Ooh
selfsymmetric-mu has joined #nixos
oborot has quit [Remote host closed the connection]
<selfsymmetric-mu>
What's the Nix way to say "after this package, symlink $HOME/<this-path> to $HOME/<that-path>". It sounds like one HOOK and one FUNCTION but I don't know what they are.
<srid>
Some ryzen love on the orange site. Makes me lust after a desktop machine for building Haskell Nix projects: https://hackerweb.app/#/item/23223147
<cole-h>
infinisil: How would I "iterate" over the generated attrset?
<infinisil>
selfsymmetric-mu: None, because packages are pure in Nix, they can't change anything except $out. The only thing you can do is to put a script in $out/bin/setup that executes some commands. Related to that are setup-hooks, but those wouldn't be used for this kind of thing
<hodapp>
given it was merged only 14 days ago and such
roconnor has joined #nixos
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
CptCaptain has quit [Quit: WeeChat 2.8]
<notgne2>
hodapp: should do, iirc once it reaches that channel on github, it will be available once Hydra finishes building/testing, which you can see the status of here https://status.nixos.org/
<emily>
is there a better way than coinflip to figure out whether you should be using checkPhase or installCheckPhase?
<emily>
hodapp: 14 days should be more than enough unless the channels are very stuck
<emily>
hodapp: channel latency is more on the order of hours to days than weeks ideally
<emily>
hodapp: btw, an easy way to check yourself is to just check whether a commit is on the nixos-20.03 (or nixos-unstable etc.) branches; those are kept in sync with the channels
<emily>
(as opposed to master and release-20.03, which are the upstreams for those channels)
<emily>
> (import <nixos-20.03> {}).nvidia-docker.libnvidia-container.version # a little simpler
<{^_^}>
attribute 'libnvidia-container' missing, at (string):310:1
<emily>
hm, or not
<emily>
oh it's in a let
<emily>
that's gnarly
<hodapp>
emily & cole-h, thanks :)
<cole-h>
emily: It's let-bound :(
<hodapp>
still want to stab CUDA in the face every time I try to do anything with it
<adisbladis>
That's proprietary software for ya
<adisbladis>
It's actively hostile
<hodapp>
proprietary is one thing, but CUDA just consistently fails at playing nice with absolutely *everything*
chloekek has quit [Ping timeout: 265 seconds]
<willghatch>
Hey, I have a computer set up as a router running nixos 19.03. To get an IP address from my ISP, I have to use a vlan (vlan2) on the external interface. On 19.03 this is working great, but if I try to upgrade to 19.09 or 20.03 I can't get an IP address on the vlan interface. I've tried dhcpcd, network-manager, and systemd.network configuration, and I haven't been able to get an IP address with any of
<kaliumxyz>
I have my udev rules setup correclty and everything seems to be working fine, though I cannot get waveforms working correctly, it has all its dependencies and seems happy but refuses to talk to adept.
<kaliumxyz>
From what I can tell there should be a /dev/ entry for the analog discovery 2 USB oscilloscope I'm trying to use, though it does not seem to be created, however I have no way of confirming this, the unit itself works fine on a windows computer.
<kaliumxyz>
waveforms can detect it but only from the device manager and not as an option to use it (same as my FPGA from the same brand (digilent)).
<kaliumxyz>
which is plain weird.
dermetfan has quit [Ping timeout: 256 seconds]
<kaliumxyz>
I get the following error on start of the program:
<kaliumxyz>
Adept NOK
<kaliumxyz>
dmgrEnumDevicesEx 0xC12
<kaliumxyz>
and if using "find devices" under "my device is not listed" while I'm not using an FHS I get the following error: "FTDI library loading failed"
chiastre has quit [Ping timeout: 240 seconds]
<kaliumxyz>
I'm using autoPatchelfHook, except for the FHS version where I just copy all of the libs into one folder using a build script.
<kaliumxyz>
Also of note is that the non-FHS version does not show the audio card as a possible oscilloscope to use.
<kaliumxyz>
Which leads me to believe that I am running into some isolation, does nixos isolate programs from direct hardware access in some way?
gustavderdrache has quit [Quit: Leaving.]
chiastre has joined #nixos
<kaliumxyz>
Running the program as root does not make any difference.
hooo has quit [Quit: Connection closed for inactivity]
o1lo01ol1o has quit [Remote host closed the connection]
DOWJONEZ has joined #nixos
maddo has quit [Quit: See ya]
<emily>
kaliumxyz: FHS stuff runs in some kind of namespace but I don't think it includes /dev, and normal programs run unsandboxed
<emily>
it's more likely due to a failure to even get the drivers loaded at the kernel level (I'm guessing you've checked dmesg/journalctl already) or a missing library
<kaliumxyz>
how would I fix the drivers not getting loaded emily?
<kaliumxyz>
its supported by the current linux kernel, and this software runs on most linux distributions.
DOWJONEZ has left #nixos [#nixos]
<kaliumxyz>
Also I can flash the internal FPGA of the oscilloscope and detect it, which made me think it was an issue with waveforms rather than a driver issue.
<kaliumxyz>
Though drivers are voodoo to me.
<{^_^}>
[nixpkgs] @adisbladis opened pull request #88101 → nixos/display-managers: Use dbus socket unit shipped by upstream → https://git.io/Jfu3d