<xenophile>
Hello. So I'm currently trying to setup startx as my display manager and I seem to be doing everything correctly but for some reason when I run startx I get a .serverauth file does not exist. Any suggestions?
anders^ has joined #nixos
Supersonic has quit [Ping timeout: 276 seconds]
Supersonic112 is now known as Supersonic
toppler has quit [Read error: Connection reset by peer]
<gchristensen>
for what it is worth, I think it would be highly valuable for you to describe your goal and the end result of what you want to use hydra for
<matthewcroughan>
Well, I have actually, in both threads.
<matthewcroughan>
I want to use Hydra. But not on my tiny 256GB boot disks.
<matthewcroughan>
I want to use it, on my hard drive.
<matthewcroughan>
I do not want the main /nix/store that everything else uses to be on the hard drive, though.
<gchristensen>
I'm not totally sure you can do that
<matthewcroughan>
If I declared a NixOS container and ran Hydra, why does it still write to the /nix/store on the host?
<matthewcroughan>
It's because there's a link to the daemon. I think.
<clever>
matthewcroughan: nixos containers share the /nix/store by default
<gchristensen>
matthewcroughan: I'd encourage you to experiment more! it is hard to answer so many questions, and I think a lot of your questions are accessible via experimentation -- and you'll learn more that way too:)
<matthewcroughan>
Fair enough, I enjoy the challenge.
<matthewcroughan>
Hydra is the first thing I've wanted to "just work" though, because it's a big thing.
<gchristensen>
not many people expect hydra to do that
<matthewcroughan>
It actually opens the doors for me to do the experimentation I wanted to do.
<matthewcroughan>
Because if I get hydra working, (on a big disk) I can do all the cross-compiling I want to do.
eisen has joined #nixos
<gchristensen>
hydra is often regarded as "unreleased software"
<matthewcroughan>
This problem of not being able to get its data stored on a different disk is a distraction from what I wanted to do, but you're right.
<matthewcroughan>
So? Nix is regarded as "experimental software"
<gchristensen>
it isn't
<gchristensen>
the unstable nix is experimental
<gchristensen>
just setting expectations for hydra to "just work"
<tomberek>
matthewcroughan: is it that you want outputs to end up on another disk, or to solely use another disk?
<matthewcroughan>
gchristensen: pretty sure `nix --help` says 'Warning: This program is experimental and its interface is subject to change.' on every single version.
<matthewcroughan>
tomberek: is that not the same thing?
<gchristensen>
yes, the `nix` program is part of the unstable, experimental interface
<matthewcroughan>
I want my SSDs to be devoid of hydra.
<matthewcroughan>
I want the large HDD to be the only place hydra does its work.
<matthewcroughan>
gchristensen: That's not what it says though. It says that the program is experimental. "AND" its interface is subject to change :D
<matthewcroughan>
If that "and" were a comma instead.. Maybe I'm bad at English though.
<gchristensen>
fine, I don't care, don't listen.
<matthewcroughan>
I did. I'm tired too, you don't have to waste your time with me.
<gchristensen>
please don't waste the community's time either
<matthewcroughan>
Harsh.
<euank>
matthewcroughan: above when you said "Nix is regarded as experimental", that was interpreted as nix, the whole, not nix, the command. `nix-build --help` doesn't say it's experimental, nix the ecosystem isn't experimental, just the nix experimental command.
<matthewcroughan>
euank: Makes sense, but I don't see why feelings were hurt :P
<euank>
The naming is confusing, and there was a miscommunication there because of that. I think that was the confusion, the fact that 'Nix' and 'nix' were two different things (ecosystem, single cli tool)
<gchristensen>
no hurt feelings here
<euank>
I'm jumping in a bit late, so I'm missing most of the context, but it does sound like you might need to spend some time hacking on things yourself to come back with more specific and clear questions.
<matthewcroughan>
I saw some meme earlier about Nix/NixOS/Nix DSL
<matthewcroughan>
euank: How more specific can I get?
<matthewcroughan>
What exactly is it that I have gotten wrong?
<gchristensen>
instead, I'm concerned about behavior that to may border on help-vampire behavior
<matthewcroughan>
I want hydra to store its paths on a different disk than the host's /nix/store.
<matthewcroughan>
gchristensen: Am I a help-vampire if I help others quite often?
<gchristensen>
it describes more than a ratio of give and take
<matthewcroughan>
Well if you want to help me out in PMs, do. But I don't see why you have to try and embarass me publicly in an IRC channel.
<gchristensen>
I have no such intent
<matthewcroughan>
Loaded terms like 'help-vampire' say otherwise from my perspective. So harsh :D
<tomberek>
okay... let's just move on? eh? I'm actually interested in figuring out the question
<matthewcroughan>
Nobody has to help me. I'm in a bit of a mood now, sorry for the time wasting.
<jonringer>
you might be able to launch hydra in a chroot'd environment, where the nix store is mounted from another location
<jonringer>
For my server, I just have all of /nix on a seperate zfs pool
<euank>
One other thought would be `unshare -m -- mount --bind /hdd/nix /nix`
<matthewcroughan>
jonringer: i.e a container, no?
<euank>
Maybe you'd have to bindmount over the nix-daemon's socket too.
<matthewcroughan>
euank: is it possible to declare that behavior?
<matthewcroughan>
I don't see any options, I guess you could make a systemd service do this at startup with a target for hydra's service.
<jonringer>
could be, but it would have to do it at the kernel level. Nix doesn't like symlinks. And I'm not sure how docker's volumes would play
<gchristensen>
docker volumes are bind mounts
<euank>
systemd + BindPaths I think would let you do that.
<jonringer>
bind mounts also work :)
<colemickens>
extra nix-daemon systemd unit with a bind-mount, or whatever, with a non-default socket location, use hydra modules extra env args to override the NIX_DAEMON it points at
<euank>
iirc systemd creates its own mount namespace for the service if you set BindPaths
<matthewcroughan>
Is it not possible to tell nixos-containers not to use the host's /nix/store? Wouldn't this be easier?
<colemickens>
or NIX_REMOTE, as it were
<matthewcroughan>
euank: that sounds even better.. will play.
<tomberek>
A NixOS container would put things into /var/lib/containers/name which can just preemptively be bind_mount'd somewhere
<euank>
caveat emptor of course, I haven't run hydra yet
toppler has quit [Read error: Connection reset by peer]
<colemickens>
a quick peek at the nixos containers makes them seem quite tied to sharing the store (which makes sense as you think/read through it)
toppler has joined #nixos
<jonringer>
I like to think of the nix store as a cache as well, and sharing should be encouraged
<euank>
Yeah. From my reading of the above, it sounds like there's a desire for a fast cache for the stuff needed for boot (ssd) and a slow cache for the stuff needed for the build server (hdd)
<tomberek>
colemickens: good point.... perhaps copy/modify it to create a new "non-sharing" systemd-nspawn based thing? Where the host's store and daemon are not shared?
<tomberek>
an option like "disableSharing" could even do it?
<jonringer>
you could also leverage something like zfs, which will have an in-memory arc. And helps mitigates subsequent accesses
<matthewcroughan>
colemickens: and they're just a wrapper around nspawn containers.
<matthewcroughan>
Could create a new module that is the same but removes the attachment to the host's /nix/store.
<euank>
Yeah, I was just thinking of a zfs l2arc, but zfs has its own sets of caveats
<matthewcroughan>
Because you can't declare an nspawn container, you have to bootstrap it.
jgt_ has joined #nixos
proofofkeags has joined #nixos
stree has quit [Ping timeout: 260 seconds]
<matthewcroughan>
euank: there's no desire for a fast cache.
<matthewcroughan>
It's just that I don't have a lot of money, but I have an SSD as a boot drive, and a hard drive as a storage device for state.
<matthewcroughan>
I want hydra's build artifacts to live on the hard drive, that's all.
jgt_ has quit [Ping timeout: 252 seconds]
<matthewcroughan>
same as if I did `nix build --store /mnt/my-hdd nixpkgs#thing`
AlpineLlama has joined #nixos
<euank>
well, not slow/fast then, but you don't want to boot from the hdd and just use the ssd for.. I guess for /home or for nothing.
<matthewcroughan>
yeah, I just want a different state directory, for hydra :P
<matthewcroughan>
Like you would have the state for any other application, but this is strange because of how special /nix/store is.
<matthewcroughan>
euank: to clarify, I want every other application on the system to use the /nix/store on the SSDs. I just want a separate /nix/store for hydra to use that is independant of the host, and a whole VM seems wasteful. But a VM actually seems like it is less effort at the moment.
<tomberek>
matthewcroughan: There are several things to try that have been suggested. Can you try some of these out and let us know? (i'm curious myself)
<matthewcroughan>
But that cannot be declared. (I hope you don't mind the repetition :P)
<matthewcroughan>
I do not fully understand the suggestion euank made about systemd and BindPaths, but I'm excited to try that either way and I will learn something.
<MysteriousSilver>
Hi! - I tried installing lxappearance but launching it says 'segmentation fault core dumped'. Any idea what causes it? strace output: http://ix.io/3m1M
<MysteriousSilver>
NixOS 20.09
rajivr has joined #nixos
<delroth>
I'm trying to package a shell script that 1. needs some binaries in PATH (-> wrapProgram); 2. looks at "basename $0" to figure out which operation to run (busybox-style symlinks). Any idea how to make that work? :)
<delroth>
wrapProgram even with --argv0 doesn't fool the basename $0
<delroth>
I could patch fairly easily but I'm wondering if there's a more common solution to this
<tpw_rules>
i had to do that once and i ended up just patching the programs
<euank>
MysteriousSilver: you probably have a coredump in `coredumpctl`, which might let you get slightly more information
<euank>
My thought is that we might get a function name / line number out of it, and at the very least that'll make it easier to check if it's fixed upstream
ahmedelgabri has joined #nixos
<mounty>
Can someone help with an incomplete LaTeX installation? Error is: "LaTeX Error: File `tikz.sty' not found" What should I add to configuration.nix please?
<hrnz>
the medium schem should be more than sufficient
<hrnz>
scheme*
ddellaco_ has joined #nixos
ahmedelgabri has quit [Ping timeout: 260 seconds]
ddellaco_ has quit [Ping timeout: 246 seconds]
toppler has quit [Read error: Connection reset by peer]
toppler has joined #nixos
<MysteriousSilver>
euank: sudo lxappearance works in gnome but not in sway
<SumnerEvans[m]>
I don't think that lxappearance should work at all in sway. You have to configure your theme through other methods.
justanotheruser has quit [Ping timeout: 276 seconds]
<MysteriousSilver>
oh ok
<MysteriousSilver>
lxappearance installation is mentioned in sway page of Nix wiki
<danderson>
I feel like I've asked this before, but... Is there some supported way of making fetchFromGitHub or other similar fixed output derivations into a regular derivation that doesn't enforce checkout hashes?
<euank>
MysteriousSilver: I wasn't clear on that at all before, but if you want to dig more, `coredumpctl info lxappearance` will have the path to the actual coredump, and `coredumpctl gdb ...` can open a debugger to it (if you're familiar with gdb)
<danderson>
the use is for a set of linked private repositories, where we have confidence that just the commit hash fully describes the code we want. Having the fixed-output hashes makes the derivations more annoying to update.
<euank>
I dunno if you want to dig more, but if you do, gdb + 'backtrace' to get the function call stack that led to a segfault is I think the next thing I'd try.
<MysteriousSilver>
oh ok
xenophile has joined #nixos
<euank>
Sorry for the vagueness before, that was my bad!
<xenophile>
I really need help with this. I would appreciate it if some could help me with this problem
<euank>
if you asked a question, it didn't come through xenophile (or at least my irc client didn't get it)
eisen has quit [Ping timeout: 246 seconds]
<xenophile>
I'm trying to setup startx as my display manager. I seem to be doing everything correctly but when I run startx I get an error that says serverauth does not exist
<xenophile>
Out of curiosity do you use startx as a display manager?
<euank>
I do not. Mine's `services.xserver.desktopManager.xterm.enable`, and then I use home-manager to run xmonad.
<euank>
But my vague memory of x/startx is that (EE) logs in xorg.log/x.log are usually the reason it's failing to run, and I'm less confident in startx's output. Could be wrong ofc
<Graypup_>
so I want to install neovim on macOS with nix, but it's trying to use python 2.7 and failing. is there a config i have to do or something?
<Graypup_>
"pynvim-0.4.3 not supported for interpreter python2.7"
evanjs has quit [Read error: Connection reset by peer]
Diogo93013 has joined #nixos
evanjs has joined #nixos
cub3s_ has joined #nixos
<albanick>
hello, what would be the best place to ask for help regarding a derivation for a pre-compiled binary software that needs fuse? Basically, I try to create a derivation based on the following but the app doesn't find the fuse library: https://github.com/apeyroux/tresorit.nix/blob/master/default.nix
beertoagunfight has quit [Ping timeout: 260 seconds]
<pennae>
albanick: can you patchelf an rpath to libfuse into the binary in fixup?
<euank>
notice the preFixup / libpath / --set-rpath bits
<albanick>
for patchelf, I'll need the exact path (/nix/store/jqshdfkjqsdhfldjksq/...therightfuselibfile.so). how can I find out its path from the .nix script?
hyiltiz has quit [Ping timeout: 240 seconds]
<euank>
If you look up at the step-by-step stuff above, it walks through it, like with 'nix-locate -1 -w lib/libQt5Svg.so.5
<albanick>
fundamentally, what is the issue with NixOS and fuse?
legendofmiracles has joined #nixos
jgt_ has joined #nixos
<albanick>
is this explained somewhere?
<euank>
Thinking about this a second longer, I might be sending you down the wrong path. I think autoPatchelfHook actually does the whole rpath dance for you, and doing it by hand might make it clearer what's going wrong.
<albanick>
libfuse is missing as a dependency (shown by "nix-store -q --tree result")
<euank>
And the binary includes a reference to a libfuse.so if you run `ldd` on it?
boxscape has joined #nixos
<euank>
It's possible to do things like run the fuse binary directly or use dlopen, and both of those things would cause autopatchelf to ont know about the fuse dependency.
<albanick>
well it seem that the binaries don't use fuse directly, but I get the following message in the logs, which was confusing: [["LibraryNotFound","dynamic_library.cpp",82,{"Library":"libfuse.so.2"},[["libfuse.so.2: cannot open shared object file: No such file or directory"
<albanick>
so, indeed, it seems that the soft is either running the fuse binary or uses dlopen (whatever this is), but doesn't find it
<euank>
I give it pretty good odds that it's using dlopen
vidbina has quit [Ping timeout: 252 seconds]
<albanick>
is there a way to fix the derivation to work with dlopen?
<albanick>
if I understand well how dlopen works, it is probable that the link to libfuse.so.2 is hardcoded
<euank>
yes. I think adding it to the rpath still fixes it, and it's just autoPatchelf which won't manage to do that automatically for you.
<albanick>
ok good to know
<albanick>
I'll try to add it to the rpath then
<pennae>
add fuse to the rpath of the binary and set dontPatchELF
<euank>
It looks like `runtimeDependencies` might also exist for this.
<albanick>
so you suggest that I don't autopatch it
<albanick>
where is runtimeDependencies documented?
<heyarne[m]>
i'm used to the german keyboard layout with a dead tilde, where one press on `~` and one on `N` results in `Ñ`. I tried configuring this via `services.xserver.layout = "de"; services.xserver.xkbVariant = "deadtilde";`, but I can't find the xkbVariant that does what I want. I tried figuring them out from `etc/X11/xkb/rules/base.lst` to no avail. I'm on 20.09 - can somebody give me advice how I'd figure that out?
__monty__ has joined #nixos
Stphan[m] has joined #nixos
fendor has joined #nixos
Chris26 has joined #nixos
<Chris26>
Hi, i was wondering if anyone new of any plans to hold the nixos conference in 2021 or 2022?
muppeth has quit [Remote host closed the connection]
muppeth has joined #nixos
redkahuna has joined #nixos
exondation has quit [Ping timeout: 250 seconds]
exondation has joined #nixos
<redkahuna>
Hey all, trying de build my first python38Package.buildPythonApplication, so I add some depencendis in propagatedBuildInputes, but let the other in setup.cfg because doesn't exist in the nix store, but when building I have an error in the Executing pipInstallPhase, paste here `https://paste.debian.net/1196603/` some thing like can't pip install orjon while the build wihtout nix work (pip install)
<eyJhb>
`{ pkgs ? import <nixpkgs> { }, ... }:` in a build file gives me `error: while evaluating anonymous function at /state/home/projects/nixos/nixus-setup/deploy/default.nix:1:1, called from undefined position:`
<eyJhb>
Using - nix-build -A config.nodes.srtoffee.deployScript deploy/
<eyJhb>
I am not sure what the issue is here
ddellacosta has joined #nixos
mikeplus64 has quit [Ping timeout: 245 seconds]
meh` has joined #nixos
ddellacosta has quit [Ping timeout: 240 seconds]
<cub3s_>
To unbreak a broken Haskell package in nixpkgs, is it enough for me to get it to compile on haskellPackages (haskell.packages.ghc8104) or do I also have to get it compile on the other GHC versions?
arjen-jonathan has joined #nixos
priv_drgn has joined #nixos
zupo has joined #nixos
<priv_drgn>
Is it normal that Nix-env uses 70-80% of available ram?
<priv_drgn>
(i am trying to install lumina using Nix)
<yorick>
ajs124: php 8.0.6 got released
teto has joined #nixos
<viric>
priv_drgn: higher % the less ram you have
<priv_drgn>
Yes, I don't have much ram
<priv_drgn>
431 mb
<priv_drgn>
(total)
<viric>
then I'd say that it'd be normal you cannot even run nix-env
<srhb>
priv_drgn: evaluation of nix can be very expensive, usually it's mostly a problem when evaluating complex NixOS expressions, but 431MB is getting close to the limit of what is even usable with barebones nixpkgs.
<priv_drgn>
hmm. So I guess that is then also the reason it takes quite long without outputting anything..
<viric>
if 438MB are total and nix-env takes 80% of that, I'd say linux can't give more of 80% to it
<srhb>
priv_drgn: Are you using -i without -A ?
<priv_drgn>
yes
<srhb>
priv_drgn: Don't do that. :)
<srhb>
priv_drgn: nix-env -iA nixpkgs.lumina
<viric>
"nix-env -i" is for patients or for those who have a supercomputer
<ThomasSanchez[m]>
yes sorry I was a bit concise :) I can start a shell by doing nix-shell -p cmake and I've cmake available.
<ThomasSanchez[m]>
Now, I'd like to have cmake and its companion, ccmake which is a curses (or qt) GUI. Unfortunately, I don't know what is the actual package name and I wondered what was the procedure from cmake-cursesUI to the actual name I should use on the `nix-env -p cmake cmake-XXXX` command line
NieDzejkob has quit [Remote host closed the connection]
NieDzejkob has joined #nixos
<ThomasSanchez[m]>
ok, while reading the conversation above, I think I got my answer. Better put: I needed the derivation name of cmake curses tool
<ThomasSanchez[m]>
It seems that if I give -P argument, I get it
<ThomasSanchez[m]>
yup, that works: `nix-shell -p cmake cmakeCurses`
<maralorn>
cub3s_: hm, sorry, what are you pointing at?
<cub3s_>
maralorn, ah it's a few pages up. i'll just repeat it here :) "To unbreak a broken Haskell package in nixpkgs, is it enough for me to get it to compile on haskellPackages (haskell.packages.ghc8104) or do I also have to get it compile on the other GHC versions?"
<maralorn>
Yeah, we normally care only about our "current" ghc for that.
<cub3s_>
maralorn, great to know! also, do i manually have to remove a fixed package from the broken list, or will it get automatically removed from the broken list next time?
<dminuoso>
atralheaven: And regarding pipes, nix is not an imperative general purpose language. The closest analogy would be infinite lists, but we dont have those in nix since they are strict in length.
<maralorn>
cub3s_: We are considering to do a "scan" for needlessly marked broken packages before branch of. But in general yes, you need to open a PR unmarking it broken.
<dminuoso>
maralorn: The code that automatically detects/marks broken packages, is that available somewhere?
<cub3s_>
maralorn, ah, so currently it's more of a resource-saving mechanism for your servers?
<{^_^}>
[nixpkgs] @sternenseemann pushed 2 commits to haskell-updates: https://git.io/J3DND
<cub3s_>
(to avoid rebuilding broken packages over and over)
<pennae>
dminuoso: there is lib.pipe, but not sure whether that's a good name for function composition
<dminuoso>
pennae: That's something different entirely.
<maralorn>
dminuoso: Not really no. There is basically none. We just throw away the list of broken packages in the config let hydra built all of them, paste the list of broken packages from hydra and put them back into the config.
<{^_^}>
#121713 (by maralorn, 2 days ago, open): maintainers/scripts/haskell/hydra-report.hs: init
<atralheaven>
in this code: http://p.ip.fi/Di3n I'm getting an error at line 84
<maralorn>
cub3s_: Resource-saving is a big part of it. But it’s also important for users so we prevent them from accidentally trying to compile packages.
<dminuoso>
cub3s_: Consider, since nix is deterministic, if hydra finds out that haskellPackages.foo is broken, it will be broken for you too. No need for you to wait 20 minutes and figure it out then.
<dminuoso>
(hence what I said earlier about convenience feature)
<sterni>
atralheaven: you seem to missing the { to start the attribute set
<atralheaven>
and how can I skip "pkgs" in namespace?
<dminuoso>
Use `with`
<cub3s_>
this is really cool :)
<cub3s_>
but i'm also glad i only have to worry about default compiler
<cub3s_>
when approx do you estimate ghc901 will become default?
<maralorn>
cub3s_: We track stackage nightly. So when they switch not switching would become a burden for us, so I hope they hold of until it’s reasonable. I personally don‘t count 9.0.x ready for prime time until haskell-language-server supports it.
<dminuoso>
Besides, there's nothing wrong with sticking to older GHC versions. 8.6.5 is *very* solid. :)
lsix has quit [Ping timeout: 258 seconds]
<cub3s_>
maralorn, excellent
<legendofmiracles>
A friend of mine is trying to add a package to nixpkgs, but when he runs it locally it always prints this: `error: cannot evaluate a function that has an argument without a value ('buildGoModule')`, When I try to recreate it, it doesn't print this. What could be the possible cause for this?
<cub3s_>
yeah 9 is currently PITA
<cub3s_>
just sticking with 8.10.4 now
<dminuoso>
legendofmiracles: Different nixpkgs?
<legendofmiracles>
It should be fairly recent, no older than yesterday
<dminuoso>
What invocation do you both have?
<dminuoso>
If you want proper determinism, you should both use the same pinned nixpkgs *and* use a pure nix-shell (e.g. nix-shell --pure)
<legendofmiracles>
invocation? We tried both `nix-build <some flags i don't remember>` and `nix-env -f . -iA <package>`
<redkahuna>
Hey all, trying de build my first python38Package.buildPythonApplication, so I add some depencendis in propagatedBuildInputes, but let the other in setup.cfg because doesn't exist in the nix store, but when building I have an error in the Executing pipInstallPhase, paste here `https://paste.debian.net/1196603/` some thing like can't pip install orjson while the build wihtout nix work (pip install). Any help
<redkahuna>
plz ?
ddellacosta has joined #nixos
<legendofmiracles>
dminuoso: it turns out they were being dumb and not running the commands from the root of the dir :p
<legendofmiracles>
* dminuoso: it turns out they were being dumb and not running the commands from the root of the nixpkgs dir :p
<atralheaven>
How hard is creating a nix package? I couldn't find a package for "https://elkowar.github.io/eww/main/", and the installation seems to be just few commands, I thought it might be simple enough to create it myself
ahmedelgabri has quit [Ping timeout: 260 seconds]
<dotlambda>
redkahuna: orjson will have to be added to Nixpkgs. But that is nontrivial because it uses Rust. I might make a PR for you.
<phalrax>
Hi, recently updated to tailscale 1.8.0 and did nix-collect-garbage -d, noticed by chance that version 1.6.0 is still in my nix store, but I don't get why... why-depends says this: '/nix/store/z0vh7l4ak8p44253ym571ps2pci8mmib-nixos-system-lain-21.05pre287333.63586475587' does not depend on
<phalrax>
gchristensen: cool, exactly what I looked for, thank you:)
solene has quit [Remote host closed the connection]
lsix has joined #nixos
<dotlambda>
redkahuna: Pip will download a (precompiled) wheel from PyPI.
sangoma has quit [Read error: Connection reset by peer]
<redkahuna>
dotlambda: I can't download the same wheel from pypi ?
<legendofmiracles>
For programs that have unstable-dd--mm-yyyy as a version, `nix run` (apparently the unstable gets identified as the name as well) doesn't work, is that a nix bug? Or should every program that uses that versioning scheme set `meta.mainProgram`?
<redkahuna>
dotlambda: I have the same trouble with pypika and it's not use rust
<__monty__>
legendofmiracles: I know there's been talk about this recently. Not sure what the conclusions were though if any have been reached.
stree has quit [Ping timeout: 246 seconds]
johnnyfive has joined #nixos
Dotz0cat has quit [Ping timeout: 260 seconds]
johnnyfi- has quit [Ping timeout: 268 seconds]
PorkySpine has joined #nixos
jonathanx_ has quit [Remote host closed the connection]
Alddar has joined #nixos
<Alddar>
Hi, I am trying to install morph https://github.com/DBCDK/morph but I don't understand the install instructions. When I run the command it says make-build, command not found. Can someone please point me to the right documentation to understand the installation? Thanks
ahmedelgabri has joined #nixos
<dotlambda>
redkahuna: Make sure you add all dependencies to propgagatedBuildInputs.
<ronthecookie>
hey I'm working on a deriv for a rust project that has some c++ code built in too and there is a header file that is trying to include <stdint.h> but it isn't found (re-running it for log)
<ronthecookie>
will post once it's done
SkyFox[m] has joined #nixos
fortuneteller2k[ has joined #nixos
ahmedelgabri has joined #nixos
sangoma has quit [Read error: Connection reset by peer]
<redkahuna>
Hey all, no python package based on rust are available in python3Packages, like rtoml orjson, is this more difficul to add rust based package ? why? it's excepted to have them ?
stree has quit [Ping timeout: 240 seconds]
<hexa->
redkahuna: cryptography is
<hexa->
the problem with orjson was, that it relies on rust unstable features iirc
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<a12l>
I'm currently writing a Nix expression for a program, and need to run a ruby script to generate manpages before CMake begins to check for dependencies. When I try to run `ruby $sourceRoot/doc/docgen.rb mangen` in `preConfigurePhases` and in `preBuildPhases` I get error messages saying that the dir doesn't exist.
<a12l>
lordcirth: Both seem to work. I'm using `srcs` because I need to download an external dependency.
<a12l>
So I've `srcs.program = { ... }; src = srcs.program;` in my expression. So I define `src`.
boxscape has quit [Quit: Connection closed]
beertoagunfight has joined #nixos
im590 has joined #nixos
sangoma has joined #nixos
thongpv87 has quit [Remote host closed the connection]
<justanotheruser>
how can I do `nix-repl> :l <nixpkgs> nix-repl> nixpkgs.hello` without going into nix repl?
chloekek has joined #nixos
<chloekek>
Requesting at a URL such as “https://nixos.org/channels/nixos-20.09” responds with a redirect. Is it possible to see which URL it redirected to when nix-channel --update was last invoked?
ddellaco_ has quit [Remote host closed the connection]
ddellacosta has joined #nixos
yoctocell has quit [Remote host closed the connection]
<chloekek>
Seems like I can run file /nix/var/nix/profiles/per-user/root/channels/nixos and then take the version number from there and then substitute it into the current redirect for that URL :)
<clever>
[clever@amd-nixos:~]$ nix-instantiate --eval '<nixos>' -A lib.version
<clever>
"21.05pre274021.d496205cf22"
<clever>
chloekek: this will open <nixos> from $NIX_PATH, then eval lib.version within it
<triplekaykay>
Hello, channel. If anyone could kindly help me, I'm trying to build something, but the final stage is failing because there's alterations being made to a read-only file system. My user is of the "nixbld" group, and also fails under root, so I do believe it's some kind of configuration error. The error specifically looks something like this.
<triplekaykay>
chmod: changing permissions of '/nix/store/y502vpypwa850q4klfp9799y7z2rb79n-wrapped': Operation not permitted
proofofkeags_ has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @Ma27 opened pull request #122060 → maintainers/check-hydra-by-maintainer: use `pkgs.hydra-check` from package set itself → https://github.com/NixOS/nixpkgs/pull/122060
<ToxicFrog>
So I tried to `nix-shell -p haskellPackages.hs-twitterarchiver`
<ToxicFrog>
And the build failed with an error from the haskell compiler: "Module ‘System.IO.Error’ does not export ‘try’"
<ToxicFrog>
And I have no idea how to even start debugging this
<matthewcroughan>
The programming language that seems to be used for the site.
<matthewcroughan>
"Less is a dynamic preprocessor style sheet language that can be compiled into Cascading Style Sheets and run on the client side or server side."
<matthewcroughan>
I've used fdisk on the drive since discovering this, and it doesn't matter whether I format as gpt or dos, both situtations end up being unable to be booted
<colemickens>
huh I just got this from nix-build-uncached but it built fine with `nix build` : error: derivation '/nix/store/brwghfxy26scz71071bzv8cadf8078cv-wlroots-8008d21f5b81c705bb14d168aeda508c975125c9.drv' does not have wanted outputs 'examples,out'
<evils>
if a test exists for the module who's options you changed, nix-run -A nixosTests.<test name>
<evils>
shla: you can also do `nix-build <path to nixpkgs> -A <attribute to build>` (it'll drop a symlink to the build result in your working directory)
<shla>
so build or run?
<evils>
nix-build, sorry about that typo
sangoma has joined #nixos
furrymcgee has joined #nixos
mdash has quit [Remote host closed the connection]
<winterqt>
dotlambda: re: `jesec-libtorrent` tests, been playing with it for a while and can't seem to get it to work either. should I just give up and commit the enable rtorrent tests change, or do you think i should ask the upstream maintainer if they have any clue?
<mica[m]>
I'm trying to update the dateutils package, but when I run nix-build, it tells me it doesn't know how to deal with .tar.xz files... but the current drv uses a tar.xz file... what gives?
VideoGameEnjoyer has quit [Remote host closed the connection]
VideoGameEnjoyer has joined #nixos
juanfra__ has joined #nixos
VideoGameEnjoyer has quit [Remote host closed the connection]
<dotlambda>
winterqt: Either is fine with me. Running rtorrent's tests also means testing libtorrent.
griff__ has joined #nixos
mdash has quit [Ping timeout: 276 seconds]
<{^_^}>
[nixpkgs] @maralorn pushed to haskell-updates « haskellPackages.haskell-gi: Bump versions manually to still build with used C libs »: https://git.io/J3SQT
<matthewcroughan>
And by the way. Since the overlay is declared inside of desktop-sway-unstable.nix, does that mean that the whole entire configuration of anything that imports desktop-sway-unstable.nix will have the contents of this overlay? Or are these overlay packages only available to the things imported by desktop-sway-unstable.nix?
<matthewcroughan>
My current understanding is that yes, everything in the system will be effected by this overlay, if this file is imported.
<__monty__>
qyliss: I've *finally* managed to build the expression after removing the "allowedReferences = []". But I think why-depends doesn't support expressions. So how would I use it to figure out why bootstrap-tools is pulling in llvm 7?
<valerii_leontiev>
Hey guys. I've build my old flake config and got the result folder on Nixos stable. How can I switch to it?
<superbaloo>
the steps in the wiki only works with the default builder
<tejing>
valerii_leontiev: note that they're talking about the unstable version of nix, not nixos. you can stay on stable nixos with a flake-based config just fine
<superbaloo>
if a derivation uses a custom builder, they don't necessarily work
<winterqt>
superbaloo: would stdenv work?
<superbaloo>
huuum ... it has to with whether 'builder =' is defined in your derivation or not
<chloekek>
Is it possible to run node2nix without package.json? I have package-lock.json, which I think should be sufficient to generate a Nix file, but I can’t find how to run node2nix without package.json file.
<apache8080>
I am trying to run a command in boot.postBootCommands but I get a command not found error even though that command is a package in environment.systemPackages
<apache8080>
is there something different with the path for postBootCommands?