<ebzzry>
the equalizer controls in pulseeffects can’t be enabled :(
sigmundv_ has quit [Ping timeout: 260 seconds]
<jtojnar>
works for me
<jtojnar>
any warnings when running from console?
<jtojnar>
(might need to kill it first)
<jtojnar>
((in case it is running in background))
<tnks>
I rattled of a question earlier, but I think it didn't get answered, and I just got busy with work. Does anyone know much about how `nixos-20.03-darwin` relates to `nixos-20.03` as a channel? I can't find documentation for it like I can see for other channels.
CMCDragonkai1 has joined #nixos
<jtojnar>
tnks you can click the hydra job link on https://status.nixos.org/ and then view the constituents
<jtojnar>
(all of them have to be green for the channel to advance)
<tnks>
jtojnar: thanks. I think that really helps.
growpotkin has quit [Ping timeout: 256 seconds]
ebzzry has quit [Read error: Connection reset by peer]
CMCDragonkai1 has quit [Remote host closed the connection]
<cole-h>
And also, there needs to be no queued jobs for that evaluation (e.g. the evaluation needs to have finished, with or without build errors).
<jtojnar>
only the jobset is nixpkgs-20.03-darwin and the special job that needs to succeed is darwin-tested
<jtojnar>
tnks^
boxscape has quit [Quit: Connection closed]
felixfoertsch23 has joined #nixos
wak-work has quit [Quit: authenticating]
felixfoertsch has quit [Ping timeout: 256 seconds]
felixfoertsch23 is now known as felixfoertsch
wak-work has joined #nixos
waleee-cl has quit [Quit: Connection closed for inactivity]
spacefrogg has quit [Quit: Gone.]
aw has quit [Quit: Quitting.]
aw has joined #nixos
spacefrogg has joined #nixos
knerten2 has joined #nixos
lord| has quit [Read error: Connection reset by peer]
knerten1 has quit [Ping timeout: 240 seconds]
sbetrium has joined #nixos
lord| has joined #nixos
lord| has quit [Read error: Connection reset by peer]
lunaa has joined #nixos
<tnks>
I'm trying to figure out how to navigate Hydra better (I get lost really easily). From status.nixos.org, I'll just use nixpkgs/trunk/unstable as an example. In this path, I see that "nixpkgs" is "project." And I see that "trunk" is a "jobset." Looking at the configuration tab for "trunk" it looks like the jobsets are configured as a specific Nix expression from a branch of a Git checkout. How then is the special
<tnks>
"unstable" job configured? Otherwise, "jobs" look like packages built by the Nix expression for the jobset.
lord| has joined #nixos
luna9 has quit [Remote host closed the connection]
<gchristensen>
"unstable" is defined in the Nix expression just like the other builds
<tnks>
gchristensen: okay, I'm going to find that definition in Nixpkgs... or try to.
<lunaa>
How does one get the hash of a fetched git repo (such as for the nixpkg protonvpn-cli-ng)?
<qyliss>
,tofu lunaa
<{^_^}>
lunaa: To get a sha256 hash of a new source, you can use the Trust On First Use model: use probably-wrong hash (for example: 0000000000000000000000000000000000000000000000000000) then replace it with the correct hash Nix expected. See: tofu-vim
<T0pH4t>
so curious question, why would a nixpkgs update cause gcc to no longer work with cmake :/ https://pastebin.com/wwVjg4Yk
<T0pH4t>
fails to perform the simple prog check
<T0pH4t>
kind of feel like others should hit this...
Supersonic112 has joined #nixos
<tnks>
Okay, so maybe rather than getting too lost in `nixos-org-configurations`, I'll make sure to read Hydra's manual. I forgot that was there, and I've never set up Hydra myself.
lunatera has joined #nixos
Supersonic has quit [Ping timeout: 260 seconds]
Supersonic112 is now known as Supersonic
lunaa has quit [Ping timeout: 240 seconds]
lunaa has joined #nixos
<energizer2>
will nixos-install overwrite existing files? will it erase the contents of existing directories?
lunatera has quit [Ping timeout: 264 seconds]
<clever>
energizer2: what would it overwrite?
<clever>
it basically just runs nix-build in a chroot, and then updates the target /boot/
lord| has quit [Ping timeout: 260 seconds]
<energizer2>
clever, i have an installation on a disk, it's not booting for some reason, so i want to reinstall it but i dont want to lose data
<clever>
energizer2: nixos-install never wipes anything, it just runs nix-build in a chroot
<pjt_mpt>
while installing tor (with services.tor), I noticed that /etc/tor/torrc is not created--so programs that expect it there fail. I do have one in /nix/store/...-torrc though, which seems to reflect configuration.nix.
drakonis has joined #nixos
<pjt_mpt>
Is there a 'correct' way to fix this?
<pjt_mpt>
I'm tempted to just hardlink it to the store, even if temporarily
growpotkin has quit [Client Quit]
reanimus has quit [Ping timeout: 260 seconds]
<comrandroxaos[m]>
<pjt_mpt "while installing tor (with servi"> programs installed using nix? if so, the "correct" way to fix would be to add the dependancy to the expression for the program in question
<pjt_mpt>
okay yeah I see it: both execstart lines do /nix/store/...tor -f /nix/store/...torrc
<clever>
and that lets nixos services work without /etc/tor/torrc
Orbstheorem has joined #nixos
<pjt_mpt>
unfortunately, it looks like my client program can't accept other locations for the torrc file
<clever>
pjt_mpt: why does the client program need to read torrc?
<pjt_mpt>
git-annex uses tor for p2p repo syncing. I'm assuming it needs it for initial setup
cyphase has joined #nixos
simba1 has joined #nixos
<pjt_mpt>
...I'm just now realizing it'll try to write to it.
<pjt_mpt>
hm
<pjt_mpt>
ultimately I think I'd want to put all that in services.tor.hiddenServices
<pjt_mpt>
once I figure out what ports it uses
<pjt_mpt>
If hardlinking won't work, guess I'll give it an exact copy of the file in the expected place
lunaa has quit [Quit: Leaving]
<pjt_mpt>
*then* write it as a submodule
<clever>
pjt_mpt: you can also symlink
drakonis_ has joined #nixos
<pjt_mpt>
to the torrc in the store?
<clever>
yeah
detran has joined #nixos
santiweight52 has joined #nixos
<pjt_mpt>
Isn't it still going to error out? I'm fairly sure the process will try to modify the version in the store. What difference does symlinking make?
<santiweight52>
And the error I'm receiving. Running nix-build in the exploitSrc directory works just fine. And the core of the application also appears to build correctly. But it seems to expect a URL for the location of the exploit-poker build. Is this due to my use of import?
<santiweight52>
(I'm a big noobie at Nix)
gxt has joined #nixos
<srhb>
santiweight52: I don't think this is nix specific, it sounds like the "project" function expects each "packages" attribute to be a tarball
<srhb>
santiweight52: I guess that's from obelisk
<santiweight52>
Ah okay - I'll go post over at reflex then
<laduke-132>
hey, i'm setting up home-manager for the first time and I can't figure out how to get it to manage bash.. The readme says like If you do not plan on having Home Manager manage your shell configuration then you must... but what do I do if I do want home manager to manage my shell config?
<nate48423>
america government and denaro are getting my human brain i am nathan biernatt and i dont want the america government to get my brain pleae tell your leaders and my brain goes through things at times and i dont want the americans to make money off my brain.
dhess_ has joined #nixos
<laduke-132>
oh never mind I got er
nate48423 has quit [Killed (Sigyn (Stay safe off irc))]
andi- has quit [Ping timeout: 272 seconds]
dhess has quit [Ping timeout: 272 seconds]
liff has joined #nixos
morr has quit [Ping timeout: 272 seconds]
<srhb>
santiweight52: As far as I can see, a source is expected, so yeah, your import is likely wrong.
<santiweight52>
Weird - the reference is being picked up
morr has joined #nixos
<santiweight52>
A build is being triggered by import ../. {} that is
heijligen has joined #nixos
<srhb>
santiweight52: Yes, sure, but I don't think that a package is the expected form of "packages"
<srhb>
santiweight52: I'm basing this purely off the default packages being sources, not packages.
<srhb>
santiweight52: (That is frontend, backend, etc.)
<srhb>
I'm using the term "package" here quite loosely.
<srhb>
The point is, I think you're _not_ supposed to provide the built package.
omnibs has quit [Remote host closed the connection]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<fps>
hmm, i'd like to get julia 1.4.2 running as well as cuda and cudnn. for julia i might be successful with a FHS environment i guess. but what about cuda and cudnn?
<jybs>
It's like it's completely ignoring the patch
<s1341>
hi.
bbb has joined #nixos
<srhb>
jybs: Aside from the broken coment it looks right to me. How are you concluding that the patch is being ignored?
bbb is now known as Guest18331
alp_ has joined #nixos
<jybs>
When I run it from a nix-shell in a similar way I can see it doing the (long) rebuild process. It doesn't do that from the container
<jybs>
postgresql-12.2.drv is not built
<s1341>
so i'm trying to bump the version of stf to https://www.npmjs.com/package/@devicefarmer/stf. I need some help. I checkout out 24e623f3f13, which has node-v8 support, and ran pkgs/development/node-packages/generate.sh after patching node-packages-v8.json, but I can't seem to figure out how to install the updated package after that...
<s1341>
any help would be appreciated.
<sephii>
I'm having trouble building my first Python package. Here are my questions in order of importance: 1) I'm trying to build a Python application that has a "plugins" system (which are regular Python modules). The plugin depends on the application, so I can't build the plugin with `buildPythonPackage` and then add it to `propagatedBuildInputs` of the application because the plugin depends on the application. 2)
<sephii>
What's the recommended way of creating a derivation for an application that can have multiple plugins, so the user can choose which ones they want? 3) The application uses pytest to run the tests. I tried adding `checkInputs = [ pythonPackages.pytest ];` but the check phase fails because it cannot find pytest. Any idea why?
<jybs>
And then if I change the patchfile I don't even get a rebuild
jmarhee has joined #nixos
<srhb>
s1341: Are you asking how to build it from a local checkout?
<s1341>
srhb: yes... because I can't see a way to upstream the change, as tehre is no longer a nodePackages_8_x in upstream...
<s1341>
and stf requires nodejs 8
ris has quit [Remote host closed the connection]
<srhb>
s1341: Ah yes. Then I guess you're better off with an older nixpkgs that hasn't dropped node8 support yet. I find it doubtful that it will be reintroduced. Either that or build it out of tree using one of the standalone node2nix tools.
ris has joined #nixos
<s1341>
srhb: so what i've done is pulled the old tree, patched the node-packages-v8.json and reran generate... the question is how to now go about installing from that patched tree?
<srhb>
s1341: nix-build path-to-checkout -A nodePackages.attribute? Assuming nodePackages corresponds to the node8 package set in your checkout
<s1341>
there is no default.nix in the
<s1341>
pkgs/development/node-packages/ folder
<srhb>
s1341: There is in the top level folder of nixpkgs though
<s1341>
just default-v8.nix and default-v10.nix
<s1341>
ok. i need to look for that.
<s1341>
srhb is there any way to enable autocomplete for the current tree?
kleisli has quit [Remote host closed the connection]
kleisli has joined #nixos
<s1341>
ok. now it appears to be building...
<s1341>
srhb: will that nix-build command INSTALL it? or just build it?
<srhb>
s1341: Just build it
<s1341>
so how do i install?
<srhb>
s1341: You can nix-env -i the result
santiweight52 has quit [Remote host closed the connection]
fpob has joined #nixos
<srhb>
s1341: (Or if you want to do it declaratively, you'll import your customNixpkgsChcekout in your configuration.nix and add nodePackages.yourPackage to environment.systemPackages.
fpob has left #nixos ["Leaving."]
civodul has joined #nixos
fpob has joined #nixos
fpob has left #nixos ["Leaving."]
<s1341>
srhb: how would I declaratively patch node-pakcages.json?
cfricke has joined #nixos
cr4y1 has joined #nixos
<srhb>
s1341: Is that the only modification in your tree?
<srhb>
s1341: If so, you can probably get away with just replacing the file with a checked-in copy.
<s1341>
so that it applies to nodePackages.@devicefarmer/stf ?
<srhb>
s1341: Without knowing the node ecosystem I assume you'll want overrideAttrs, not override
<karantan>
I have this default.nix config: https://pastebin.com/YF5B5qdt. Build works but if I want to run this program as systemd service it doesn't have access to packages like `tree`, `wget`, etc (listed in `nativeBuildInputs `). How do I tell nixos that these packages (`nativeBuildInputs`) are required when it runs this program?
o1lo01ol1o has quit [Ping timeout: 240 seconds]
<srhb>
s1341: The general pattern will be foo.overrideAttrs (oldAttributes: { nativeBuildInputs = oldAttributes.nativeBuildInputs ++ [ newstuff ]; })
<s1341>
no. my problem is that i cannot specify nodePackages.@devicefarmer/stf.override... taht is bad syntax
<s1341>
how do i put in the '@devicefarmer/' so that it does what I want?
<srhb>
Oh, the package actually contains an "@"? I assume you'll want nodePackages."@foo" then
<bahamas>
hello. I have the following nix file: https://bpa.st/PH6Q. I'm trying to build it with a remote builder found inside a Docker container. when I run `nix build -f server.nix`, I get "assertion failed at /nix/store/apnn10fiqg0gz1xf87vg0qnm6gk6sd7c-nixos-20.03/pkgs/os-specific/linux/kernel/generic.nix:56:1"
<s1341>
thanks @srhb!
<bahamas>
the command `nix ping-store --store ssh://nix-docker` succeeds
<bahamas>
in ~/.nix/remote-systems.conf I have `nix-docker x86_64-linux /etc/nix/docker_rsa 4
<bahamas>
so it looks like nix is not using the remote builder. can anyone tell why?
<symphorien>
bahamas: the assertion is while evaluating
<symphorien>
Even before remote builder can be used
<bahamas>
symphorien: ok, so what do I need to change?
<symphorien>
No idea. Your nix file probably
<bahamas>
I would have assumed that all operations needed for a certain architecture would be performed on that architecture
<{^_^}>
[nixos-weekly] @domenkozar merged pull request #124 → add status blog post → https://git.io/JfxIt
<{^_^}>
[nixos-weekly] @domenkozar pushed 4 commits to master: https://git.io/Jfx0i
__monty__ has joined #nixos
FRidh has joined #nixos
alp_ has quit [Ping timeout: 272 seconds]
alp_ has joined #nixos
domogled has quit [Quit: domogled]
zupo has joined #nixos
bahamas has quit [Ping timeout: 246 seconds]
<regnat>
bahamas: if you're building for another system than your local one you need to set it explicitely
smatting_ has joined #nixos
fpob has joined #nixos
<regnat>
(But I don't remember how to do that tbh)
<sb0>
how do I set boot.shell_on_fail=1 on a system that won't boot and therefore where I cannot run nixos-rebuild?
<srhb>
sb0: Edit the bootloader kernel command line, usually using 'e'
<sb0>
I tried simply editing the kernel command line in systemd-bootloader to add "boot.shell_on_fail=1" but that did not work
<srhb>
sb0: Is =1 really correct?
<sb0>
is the syntax different there?
<srhb>
sb0: Used boot.shell_on_fail just yesterday, and that worked.
<srhb>
Not sure if =1 should be equivalent
<sb0>
without the =1 ?
Thra11 has quit [Ping timeout: 240 seconds]
<srhb>
Yes.
<sb0>
it worked thanks
<srhb>
Great :)
mac10688 has joined #nixos
Thra11 has joined #nixos
bahamas has joined #nixos
<kraem[m]>
Has anyone noticed that natural scrolling in xorg.conf isn't respected when having updated nixos-unstable (which introduced linux 5.7.4)?
Guest91668 has quit [Quit: WeeChat 2.7.1]
bbb has joined #nixos
bbb is now known as Guest9016
asymptotically has joined #nixos
kenji has joined #nixos
thc202 has joined #nixos
alp_ has quit [Ping timeout: 272 seconds]
sigmundv_ has joined #nixos
bahamas has quit [Read error: Connection timed out]
bahamas has joined #nixos
pingiun has joined #nixos
<karantan>
how do I export additional env. vars to systemd service? I want to add things to PATH but if I do this: "Environment = "PATH=${pkgs.tree}/bin:${pkgs.openssl}/bin:$PATH";" I just overwrite the old PATH
<makefu>
karantan: you could create a small wrapper around the executable or update the derivation of your executable
<{^_^}>
[nixpkgs] @Mic92 opened pull request #91340 → home-assistant: add hexa as maintainers → https://git.io/Jfxzv
<srhb>
Either way -- via some extra in the right location :)
<bahamas>
srhb: I have several vhosts
<srhb>
bahamas: If you can set it globally I' sure that's nicer. If not, at least you have a programming language to template the thing into each vhost ^^
est31 has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @magnetophon opened pull request #91341 → bschaffl: init at 0.1 → https://git.io/Jfxzz
<{^_^}>
[nixpkgs] @danieldk opened pull request #91342 → python3Packages.tokenizers: init at 0.7.0 → https://git.io/Jfxzw
est31 has joined #nixos
ris has quit [Ping timeout: 246 seconds]
ris has joined #nixos
Thra11 has quit [Ping timeout: 240 seconds]
sebass[m] has joined #nixos
<bahamas>
srhb: I set in `services.nginx` errorLog="syslog:info", but in the generated nginx.conf I only get "error_log syslog;" do you have any idea why?
<srhb>
bahamas: I think you left out part of the option name there?
<bahamas>
also, the stuff I put in services.nginx.appendConfig doesn't appear at all
nikita` has joined #nixos
<bahamas>
which part?
<srhb>
I remember there's some funky shenanigans with priorities of these options, but I don't recall appendConfig being ignored
_ris has quit [Read error: Connection reset by peer]
ris has joined #nixos
orivej_ has joined #nixos
orivej has quit [Ping timeout: 256 seconds]
fps has joined #nixos
o1lo01ol1o has joined #nixos
Neo-- has quit [Ping timeout: 258 seconds]
<kiwiirc>
srsly?
<patagonicus>
bahamas: I think services.nginx.commonHttpConfig can be used for that.
<patagonicus>
bahamas: Or just *.appendHttpConfig
orivej_ has quit [Ping timeout: 246 seconds]
orivej has joined #nixos
ckauhaus has joined #nixos
<bahamas>
patagonicus: yes, I used appendHttpConfig
<bahamas>
btw, I'm trying to install lsof on nixos. I get multiple "couldn't resolve host name" the last being "unable to download 'https://ftpmirror.gnu.org/glibc/glibc-2.30.tar.xz': Couldn't resolve host name". then the build fails
<elvishjerricco>
Well it works if I set `useBootLoader = true`
teto has quit [Ping timeout: 260 seconds]
<elvishjerricco>
But I'm trying to use `systemd-ask-password-plymouth.service` (which has to be manually enabled), and it's not creating `/run/plymouth/pid`
o1lo01ol1o has quit [Remote host closed the connection]
energizer2 has quit [Ping timeout: 256 seconds]
orivej has quit [Ping timeout: 265 seconds]
orivej_ has joined #nixos
<{^_^}>
[nixpkgs] @rnhmjoj merged pull request #91224 → terminus-font: make build reproducible by using `gzip -n` → https://git.io/JfNCW
<mudri>
I'm trying to free up some space on my hard disk, and I want to know whether these statistics sound at all unusual. I'm using NixOS with a moderately light setup, and using several Haskell programs. I recently did `nix-collect-garbage --delete-older-than 14d` and set `nix.autoOptimiseStore = true;`. Most of the statistics come from `ncdu`:
<mudri>
- /nix/store/.links reports 318.3 GiB of disk usage, and contains over 5 million items.
<mudri>
- /nix/store uses 337.7 GiB of disk space. It contains over 25 million items.
<mudri>
- /nix/store contains over 100 copies of GHC, at 12 different versions. Oldest is 7.10.3. Each has a size over 1 GiB.
<mudri>
- `time ls /nix/store >/dev/null` reports `1.66s user 1.86s system 3% cpu 1:44.77 total` (my guess is that this slows down `nix-build`, `nix-shell`, &c).
<vika_nezrimaya>
Let's say I have a Nix cross-compiled for a new architecture. I've seen references to a "native stage" in Nixpkgs code. Does this mean that having only a Nix (cross-compiled) and a toolchain installed on a native device this device can build a native stdenv and nix for itself?
agsdheidjd has joined #nixos
orivej_ has quit [Quit: No Ping reply in 180 seconds.]
<betaboon>
can i allow unsupportedplatforms for specific packages ?
orivej has joined #nixos
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
<vika_nezrimaya>
in ~/.config/nixpkgs/config.nix: { allowUnsupportedSystem = true; } or NIXPKGS_ALLOW_UNSUPPORTED_SYSTEM=1 as an environment variable.
<fps>
mudri: maybe there's a way to calculate the space of the closure of one of your profile generations
<vika_nezrimaya>
betaboon: I don't see a predicate for this in nixpkgs' config.nix but the abovementioned ways might help you
<betaboon>
vika_nezrimaya: i dont want to do that systemwide, but for a specific package
<betaboon>
vika_nezrimaya: thanks anyway :D
<vika_nezrimaya>
A hacky way would be reevaluating nixpkgs with config modified
<{^_^}>
[nixpkgs] @edolstra pushed to release-20.03 « Update nixUnstable and nixFlakes »: https://git.io/JfxrP
Neo-- has joined #nixos
<mudri>
fps: I haven't done it for a while, but my impression was that `nix.autoOptimiseStore = true;` made that happen automatically for each build. I think last time I tried `nix-store --optimise`, it churned away for ages and then said that no more space had been saved (because everything was already linked).
<fps>
mudri: oh ok.
<{^_^}>
[nixpkgs] @ElvishJerricco opened pull request #91344 → ZFS: Ask for stage 2 encryption passwords using systemd-ask-password → https://git.io/JfxrN
<kiwiirc>
if you can declaratively document your entire system in a config lang and with that file automatically set up a fresh system, why would anyone want to do it through commands (imperative)?
<kiwiirc>
i guess maybe if you use something else for the same purpose already like if you used ansible for auto cm?
<vika_nezrimaya>
My opinion is: people use Ansible because people use Ansible. Others use Nix because it's useful and cool
<vika_nezrimaya>
I don't know why Ansible has become so popular, and why Nix hasn't taken its place as the most popular tool
<dminuoso>
vika_nezrimaya: With Ansible it's quite obvious "what happens" and you can add crutches easily, because thats the imperative mind model.
<vika_nezrimaya>
but Nix is even more than Ansible.. It's a build system, a CI environment, a whole operating system builder, even a static site generator
<dminuoso>
Also, Ansible can easily be used ontop of existing systems, instead of building up something well principled from the beginning.
<dminuoso>
It's pretty much the same rationale why imperative languages are so popular and declarative ones are not.
<kiwiirc>
i think nixos needs better marketing. like showing ppl setting up a whole system with 1 simple file vs going and typing and clicking all this stuff. and maybe a gui configurator for nix config files would make it more accessible to noobs?
<fps>
mudri: you might try following the links of your current profile until you find the store path of it and then use nix-copy-closure or nix-store --export to find out its closure size
<dminuoso>
Also, nix has some severe disabilities.
<dminuoso>
It has quite poor documentation left and right.
<fps>
mudri: if it's 300G in size then there's nothgin you can do. if it's much smaller you might try throwing away generations ;)
<dminuoso>
It suffers from lack of discoverability of things, and you usually have to just "read the source, and more transitive sources" to understand whats going on in some non-trivial code
<vika_nezrimaya>
how can a signle NixOS generation could be 300G
<vika_nezrimaya>
that's not scientifically possible
<kiwiirc>
i read a comment that discoverability of what you can put in configs is a problem. like there's no comprehensive index of keys and values or smth
<{^_^}>
[nixpkgs] @raboof opened pull request #91345 → k4dirstat: init at 3.2.1 → https://git.io/JfxoO
<vika_nezrimaya>
kiwiirc: we have `man configuration.nix`
<vika_nezrimaya>
it's very useful to me when I want to look something up or check if a service has a NixOS config module
<kiwiirc>
read what dminuoso just said. said discoverability at same time i did
<mudri>
fps: Okay, that sounds like a good course of action. I was hoping to avoid needing to copy stuff, but I do at least have another hard disk to use for that.
<kiwiirc>
might be a good place to add some web tool or smth
<fps>
mudri: i'm guessing a little bit here though..
<dminuoso>
kiwiirc: No the problems are more fundamental than that. When I want to write some nix code, what primitives are available?
<fps>
possibly there's a smarter way to find out the closure siz
<dminuoso>
Where's the documentation on the lib functions?
<dminuoso>
How can I figure out what nix primitives are available?
<kiwiirc>
ya that's what i was thinking
<dminuoso>
Or nixos primitives
<kiwiirc>
what's the best way to solve that?
<kiwiirc>
cuz i agree. nix concept is powerful and simple, but it requires knowing what to do so you can do that powerful and simple thing, and that's like a logic error
<kiwiirc>
needs an easier onramp
<kiwiirc>
or portal
<mudri>
Oof, I did `nix-store -q --roots /nix/store/535yvd58cyiv9sba2v3vykrg80kbi089-ghc-7.10.3`, and the output was:
<mudri>
No clue why these haven't been garbage-collected...
rk04[m] has joined #nixos
<vika_nezrimaya>
dminuoso: I think the Nixpkgs manual has been improving a bit under the hood though. I recently peeked at the latest Hydra build of the manual and it was a lot bigger than before.
<fps>
mudri: i just did: nix-store --export $(nix-store -qR /nix/store/ahw7z4szvcjh29m5vjgz8vya3i0dnabv-user-environment) > environment-export
<symphorien>
mudri: do you know that nix-collect-garbage --delete-older-than must be run as root ?
orivej has joined #nixos
<symphorien>
Otherwise it only deletes user profiles
<mudri>
Ah, that might have been my problem. 😲
<{^_^}>
[nixpkgs] @samdroid-apps opened pull request #91348 → pulsar: init at 2.6.0 → https://git.io/JfxKH
<mudri>
Oof, yeah, looks about right. It looks like I've never run it as root, so even the oldest system generation is still there (and now about to be deleted).
<mudri>
Thanks everyone!
bahamas has quit [Ping timeout: 260 seconds]
barm0 is now known as barm
<{^_^}>
[nixos-homepage] @garbas pushed 8 commits to master: https://git.io/Jfx6I
<{^_^}>
[nixos-homepage] @garbas merged pull request #487 → Implement local link checking as part of build → https://git.io/JfAiw
<{^_^}>
[nixos-homepage] @garbas pushed 0 commits to link-checking: https://git.io/Jfx6L
noudle has quit []
adisbladi is now known as adisbladis
_ris has quit [Ping timeout: 246 seconds]
<{^_^}>
[nixpkgs] @sternenseemann opened pull request #91350 → ocamlPackages.{ipaddr,macaddr}: 3.1.0 → 5.0.0 and update all affected packages (tuntap, conduit, cohttp, …) → https://git.io/Jfx6p
snajpa has joined #nixos
_ris has joined #nixos
orivej has quit [Quit: No Ping reply in 180 seconds.]
o1lo01ol1o has joined #nixos
orivej has joined #nixos
cr4y1 has joined #nixos
ris has joined #nixos
_ris has quit [Ping timeout: 265 seconds]
omneh has joined #nixos
omneh has joined #nixos
omneh has quit [Changing host]
orivej_ has joined #nixos
OmnipotentEntity has quit [Read error: Connection reset by peer]
_ris has joined #nixos
orivej has quit [Ping timeout: 256 seconds]
ris has quit [Ping timeout: 264 seconds]
morr has quit [Ping timeout: 260 seconds]
_ris has quit [Ping timeout: 260 seconds]
meh` has quit [Ping timeout: 240 seconds]
erik-4levels has joined #nixos
_ris has joined #nixos
<erik-4levels>
Hi all, just a general quick question: is there a nix function to generate an md5 hash from an arbitrary string that can be used in .nix configuration files?
<erik-4levels>
I'm hoping for something like lib.md5sum or builtins.md5 or something similar.
Guest9016 has quit [Quit: WeeChat 2.7.1]
<srhb>
erik-4levels: builtins.hashString "md5"
<erik-4levels>
The reason being is that vagrant in combination with libvirt uses md5 hashes for it's shared folder feature via a p9 mount..
<erik-4levels>
Oh, great!
<erik-4levels>
Thanks a million!
<srhb>
Note that it's the nix base16 representation.
<srhb>
.. I think?
<srhb>
So merely being "md5" might not be enough for you
<erik-4levels>
let's see if this does what it's supposed to ;-)
<srhb>
I think there's an issue on providing the --to-somebase functions as nix builtins, but iirc it's not there yet.
morr has joined #nixos
<erik-4levels>
Ah, so it's hashString "md5" "value-to-hash"
<srhb>
Yes.
<erik-4levels>
And is there a way to access the current directory's path in a nix expression? I know this is hacky..
<srhb>
erik-4levels: Not the way you mean, I think.
bbb has joined #nixos
drdee has quit [Ping timeout: 260 seconds]
<erik-4levels>
You're right, I shouldn't be doing that anyway ;-)
bbb is now known as Guest9998
<srhb>
Pretty sure that's the spirit :P
<erik-4levels>
And access to ENV vars is neither possible, right?
<srhb>
Not if you want deterministic builds.
<srhb>
(there's builtins.getEnv -- but the cost of using it is probably much too high)
<erik-4levels>
It's just for convenience, to avoid my colleagues to have to adjust a nix expression up front
sb0 has quit [Quit: Leaving]
<srhb>
erik-4levels: Why not pass arguments instead?
<erik-4levels>
I need the current working directory when invoked from a vagrant up (using libvirt)
<srhb>
Ah.
<erik-4levels>
It's quite a challenge actually, since there's little control over vagrant..
<erik-4levels>
(I hate vagrant, prefer scripted virsh much more, but that's what the team went for)
<srhb>
erik-4levels: Sorry, I don't know enough about vagrant to come up with sensible suggestions, I think.
<vika_nezrimaya>
is vagrant kinda like "Docker for VMs"?
orivej has joined #nixos
<erik-4levels>
It's supposed to help setup a VM without going through the various provider interfaces
orivej_ has quit [Ping timeout: 260 seconds]
<snajpa>
repeated question: Realtek 8822BE PCI wireless network adapter at /nix/store/g96fjafw73gdsh3lvm6bz9q1jsxikvbq-generate-config.pl line 81.
<snajpa>
sadface
<erik-4levels>
So you can focus on how the VM is supposed to work and not on how to setup a VM with virtualbox, vmware, virsh, ..
<patagonicus>
I've only looked at it briefly a few years ago, but I'd describe it as a sort of ansible, but targeted directly at VMs, i.e. it'll set up your VM and the system running in the VM.
boxscape has joined #nixos
<erik-4levels>
But they only support the regular providers (VirtualBox, VMWare, ..) and especially libvirt is missing
<snajpa>
anyone other than me in here who's got beef with generate-config.pl? :D that thing doesn't like my (maybe it doesn't like Linux 5.8-rc2 more)
<patagonicus>
But the use cases are similar to docker, mostly getting somewhat reproducible environments for dev work or running production.
<vika_nezrimaya>
so Docker for VMs
<erik-4levels>
We're using it to setup our local dev environment
<vika_nezrimaya>
I think NixOS can do it to, but without Vagrant
<erik-4levels>
I'm sure nixos does it, I've been doing it in the past with nixops and that has worked great!
<erik-4levels>
Reproducible builds every time, never failed on me
<boxscape>
what's the canonical way of running "nix build" on a default.nix file that lists dependencies as arguments (not all of which are directly under nixpkgs but might be under, e.g., xorg)?
<vika_nezrimaya>
nixos-rebuild build-vm builds a VM from a config, and nixos-shell (forgot the author) even automatically dumps you into the shell in this VM
<boxscape>
s/arguments/parameters I guess
<erik-4levels>
At one point I was managing +15 servers with nixops, till the point that building the list of derivations would trash the nixops machine
<erik-4levels>
so I could only deploy in batches
<vika_nezrimaya>
boxscape: I'd use something like `nix build -E "with import <nixpkgs>; callPackage ./default.nix { inherit (xorg) xlib; };"`
<erik-4levels>
But still, very big fan
<boxscape>
vika_nezrimaya cool, thanks
Chiliparrot has joined #nixos
<vika_nezrimaya>
that's how Nixpkgs does it in all-packages.nix
<{^_^}>
boxscape: If you're updating a package file in nixpkgs that starts with something like `{ stdenv, cmake }:`, use `nix-build -A the-package-attribute` in the nixpkgs root with the corresponding package attribute to build it. The mapping from package attributes to package files is in pkgs/top-level/all-packages.nix
<qyliss>
Ugh, I always think that message is more helpful than it is
<qyliss>
Never mind
Guest78 has quit [Ping timeout: 260 seconds]
ris has joined #nixos
_ris has quit [Ping timeout: 272 seconds]
<gchristensen>
could it be updated?
<qyliss>
Maybe!
<gchristensen>
what should it say?
<qyliss>
,callPackage = If you're updating a package file in nixpkgs that starts with something like `{ stdenv, cmake }:`, use `nix-build -A the-package-attribute` in the nixpkgs root with the corresponding package attribute to build it. If it's not in nixpkgs, try `nix-build -E 'with import <nixpkgs> {}; callPackage ./file.nix { }'. The mapping from package attributes to package files is in
<{^_^}>
callpackage redefined, was defined as If you're updating a package file in nixpkgs that starts with something like `{ stdenv, cmake }:`, use `nix-build -A the-package-attribute` in the nixpkgs root with the corresponding package attribute to build it. The mapping from package attributes to package files is in pkgs/top-level/all-packages.nix
<{^_^}>
If you're updating a package file in nixpkgs that starts with something like `{ stdenv, cmake }:`, use `nix-build -A the-package-attribute` in the nixpkgs root with the corresponding package attribute to build it. If it's not in nixpkgs, try `nix-build -E 'with import <nixpkgs> {}; callPackage ./file.nix { }'. The mapping from package attributes to package files is in
<jared-w>
Need a bit of help with my current nerdsnipe of the day. Trying to figure out how to a) run a zsh function (designed to be ran in an interactive shell with my zshrc) inside a oneshot systemd service, b) how tf do I run gsettings inside a systemd oneshot service so that it actually works. wrapGAppsHook is close, but not quite.
<qyliss>
,callPackage = If you're updating a file in nixpkgs that starts with something like `{ stdenv, cmake }:`, use `nix-build -A` in the nixpkgs root with the corresponding package attribute to build it. If it's not in nixpkgs, try `nix-build -E 'with import <nixpkgs> {}; callPackage ./file.nix { }'. The mapping from package attributes to package files is in pkgs/top-level/all-packages.nix.
orivej_ has joined #nixos
<{^_^}>
callpackage redefined, was defined as If you're updating a package file in nixpkgs that starts with something like `{ stdenv, cmake }:`, use `nix-build -A the-package-attribute` in the nixpkgs root with the corresponding package attribute to build it. If it's not in nixpkgs, try `nix-build -E 'with import <nixpkgs> {}; callPackage ./file.nix { }'. The mapping from package attributes to package files is in
<qyliss>
,callPackage
<{^_^}>
If you're updating a file in nixpkgs that starts with something like `{ stdenv, cmake }:`, use `nix-build -A` in the nixpkgs root with the corresponding package attribute to build it. If it's not in nixpkgs, try `nix-build -E 'with import <nixpkgs> {}; callPackage ./file.nix { }'. The mapping from package attributes to package files is in pkgs/top-level/all-packages.nix.
<qyliss>
That's better :)
<qyliss>
Thanks for the encouragement gchristensen
<gchristensen>
thank you for shaving down this sharp edge :)
orivej has quit [Read error: Connection reset by peer]
<jared-w>
bikeshed incoming: make a `nix callPackage` subcommand that is just `nix-build -E ...` but less painful
<gchristensen>
`nix` doesn't know about `callPackage`
<jared-w>
yeah, I'm not saying it's a great idea, I'm just saying it's a common annoyance I run into when debugging stuff and wanting to have nice { a, b, c } files but also not have to remember the `with import; callPackage` wrapper ceremony
<qyliss>
You could write a nix-call-package script
<jared-w>
honestly, I should just do that
<{^_^}>
[patchelf] @domenkozar pushed to aarch64 « CI for aarch64 »: https://git.io/JfxXA
<jared-w>
gchristensen: I remember seeing some musing from you earlier on the next generation of what a dev environment looks like (in the lorri repo). Any newer thoughts on what a (nix-shell v2 + lorri + buildEnv + cloud CI cache + ???) might look like?
<TheSirC[m]>
I tried passing it as a `nix-shell` (like `nix-shell -p linuxPackages.v4l2loopback`) but then I realized my stupidity...how to load it ?
<evils_>
TheSirC[m]: i have `boot.initrd.kernelModules = [ "pcspkr" "usbmon" "your module here?" ];`
raimu-san has quit [Quit: WeeChat 2.7.1]
<evanjs>
Is there any guidance in the wiki or elsewhere on when we should use hard vs soft deprecations? Not implementing anything myself, but got curious after I noticed youcompleteme has a hard deprecation notice (i.e. update to YouCompleteMe)
<TheSirC[m]>
evils_: So there is no my way to plug it like `modprobe -i v4l2loopback` ?
<evanjs>
In contrast to things like... oh. aliases for live programs
<evils_>
TheSirC[m]: i'm not familiar with kernel modules that aren't available by default, maybe install the package and then modprobe it?
o1lo01ol1o has quit [Remote host closed the connection]
<evanjs>
Mrm but e.g. compton -> picom seems to work
<TheSirC[m]>
<evils_ "TheSirC: tried `boot.extraModule"> So you need to reboot and it is loaded on reboot right ?
<TheSirC[m]>
I wanted to have a symlink the .ko file to load and unlaod it as I wish
o1lo01ol1o has joined #nixos
omneh_ has joined #nixos
omneh has quit [Ping timeout: 260 seconds]
doyougnu has joined #nixos
raimu-san has quit [Quit: WeeChat 2.7.1]
<evils>
TheSirC[m]: i suspect the boot.extraModulesPackage gets you the module available, probably best to install that in a way that ensures it matches your kernel version, the boot.initrd.kernelModules just loads it at boot
raimu-san has joined #nixos
doyougnu has quit [Remote host closed the connection]
doyougnu has joined #nixos
_ris has quit [Read error: Connection reset by peer]
<doyougnu>
hello all, I have a scala project I'd like to run locally. I've looked over the nixos and nixpkgs manual but didn't see any information on running scala projects. Does anyone have a good example repo I can read through?
<{^_^}>
[nixpkgs] @romildo pushed 2 commits to release-20.03: https://git.io/Jfx91
<raimu-san>
Hi ! Do you know how to get dolphin thumbnail working outside full KDE ? I get this error : "serviceType "ThumbCreator" not found". I have kio-extras and kdegraphics-thumbnailers installed on the system.
<TheSirC[m]>
evils: 🆗 I am fairly curious : why did you pick the namespace `config.boot.kernelPackages.[package]` and not `pkgs.linuxPackages.[package]` ?
doyougnu has quit [Read error: Connection reset by peer]
<evils>
TheSirC[m]: the example does that, i think it ensures you use the linuxPackages that matches the kernel set by your config
doyougnu has joined #nixos
<TheSirC[m]>
Cool !
wadkar has joined #nixos
orivej has quit [Quit: No Ping reply in 180 seconds.]
<wadkar>
Hi, I wanted to know how to go about installing nix on ubuntu where my root partition is already full. I have /mnt where there is plenty of space, but I don't think I can symlink /nix to /mnt/nix, right?
orivej has joined #nixos
<wadkar>
I am not comfortable doing a hardlink either - so maybe I can do `mount --bind /mnt/nix /nix`? Will that break nix?
mekeor has quit [Ping timeout: 258 seconds]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Guest9998 has quit [Quit: WeeChat 2.7.1]
<{^_^}>
[templates] @edolstra pushed to master « templates.simpleContainer: Use NixOS 20.03 »: https://git.io/JfxHw
<TheSirC[m]>
evils: Well... it did not go well at all
ddellacosta has quit [Ping timeout: 265 seconds]
ddellacosta has joined #nixos
<karantan>
I need to add `nice` tool to https://nixos.org/nixos/options.html#systemd.services.%3Cname%3E.path but I can't find this package. I have it by default on nixos but it's not available to systemd service
<evils>
TheSirC[m]: did something break?
<{^_^}>
[nix] @edolstra pushed to flakes « nix flake init: Use git add --force »: https://git.io/JfxQ0
<vika_nezrimaya>
if you think it's insecure then type out my password in here, it's ok, I'm allowing it
<vika_nezrimaya>
i'll probably change it if that happens tho :3
<bqv>
obviously i can't reverse it instantly, but i don't see how someone able to get onto my system and peek into my store to see my password can't also peek into it, see my hashedPassword and then just bruteforce it
<vika_nezrimaya>
well bruteforcing salted SHA-512 is not going to be easy for a casual snooper
<bqv>
easier than getting remote access to my system...
mallox has quit [Quit: WeeChat 2.8]
<vika_nezrimaya>
but
<vika_nezrimaya>
if someone got into your system hard enough to peek into the /nix/store, I don't think they're gonna need your passwrd
<vika_nezrimaya>
It inputs a one-time garbage string and PAM magically accepts it as a password
<adisbladis>
How do you unlock it?
<adisbladis>
I was also using this approach before but there was something I didn't like about it...
<adisbladis>
I can't remember what
<{^_^}>
[nixpkgs] @Profpatsch opened pull request #91363 → small treewide: his -> theirs/its → https://git.io/Jfx5f
<vika_nezrimaya>
adisbladis: you mean from the lockscreen? the same thing. I can insert the yubikey and press the button
_ris has quit [Ping timeout: 246 seconds]
<adisbladis>
No, I mean how do you lock/unlock the token
<vika_nezrimaya>
In case of PAM auth, I don't. My presence is enough for it
orivej_ has joined #nixos
<adisbladis>
Let's say you enable your screen locker and leave your token in the port
<vika_nezrimaya>
I dont
orivej has quit [Ping timeout: 265 seconds]
<vika_nezrimaya>
the only way you can touch my YubiKey is by prying it from my dead cold hands :3
<adisbladis>
I have one constantly plugged in to one of the ports
<vika_nezrimaya>
oh noes
<vika_nezrimaya>
that's... a bit more insecure than usual
<vika_nezrimaya>
I only do that at home
<adisbladis>
That's the entire reason why they have the small form factor
<qyliss>
That's what the nano yubikeys are for
<dbe>
yubikeys are pretty magical :)
<vika_nezrimaya>
qyliss: i got a normal one, it's big, I carry it on my keychain which never leaves my sights unless I'm home and the doors are locked
_ris has joined #nixos
<adisbladis>
I got three of them :3
<vika_nezrimaya>
I consider myself reasonably secure against a stolen yubikey attack unless you count my family as an adversary (they don't even know how to use the yubikeys let alone insert them correctly)
<dbe>
i never really understood the nano form factor yubikeys. we use yubihsm's at work to handle certificates, but that's not the norm for yubikeys even though they do support pkcs11
<qyliss>
I use my Yubikey with a passphrase and tap-to-sign
<qyliss>
Stealing the yubikey doesn't get you anything if you don't know the passphrase
<dbe>
for the yubihsms we're authenticating a device, not a user, so that's ok
<vika_nezrimaya>
I sign commits with my YubiKey, tap-to-sign is impractical for me
<qyliss>
I also sign commits with my yubikey
<vika_nezrimaya>
qyliss: it gives you OTP codes which can log you into my computer because I don't use it as a second factor, I use it as an alternative sign-in method
<qyliss>
Tap to sign is fine
<adisbladis>
Oh wow
<qyliss>
Yes, in that case you don't want to leave it unattended
<adisbladis>
They're gonna release a new yubikey with a fingerprint reader
<qyliss>
Oh finally!
<adisbladis>
So nice :)
<qyliss>
That's been a real missing piece of the puzzle
<vika_nezrimaya>
WOW
<vika_nezrimaya>
Yes, that's what we really needed
karantan has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<adisbladis>
I might actually pay for a yubi for the first time ^_^
karantan has joined #nixos
<vika_nezrimaya>
did you get yours for free?
<adisbladis>
Yep
<vika_nezrimaya>
yay! I got it as a gift from a friend
<adisbladis>
I have some old friends at yubico
<vika_nezrimaya>
one of my friends just got too many yubikeys and since I didn't have one he just sent me one
<vika_nezrimaya>
he got it declared at the customs as a "usb key"
<vika_nezrimaya>
well technically it is a USB device and a key... we just didn't say it's a cryptographical key
<vika_nezrimaya>
probably broke a law or two but nobody noticed
<vika_nezrimaya>
oops sorry this is offtopic
ddellacosta has quit [Ping timeout: 265 seconds]
fresheyeball has quit [Ping timeout: 264 seconds]
<bqv>
Taneb: fwiw i also have my config on github, and i make use of hashedPassword and password both in different places, but both encrypted
fresheyeball has joined #nixos
<bqv>
just frustrating cause it seems like an ultimately pointless restriction
<{^_^}>
[nix] @meditans opened pull request #3736 → Allow relative paths in --store option → https://git.io/Jfxd9
gungoman has joined #nixos
<hoverbear>
I'm getting
<hoverbear>
I'm getting `/nix/store/lh2sslfz4l6mac9xgwyy155hq3ab8y8x-gcc-9.3.0/include/c++/9.3.0/cstdlib:75:15: fatal error: stdlib.h: No such file or directory` on a `nix-shell` based Rust compile. Could anyone advise what I might be missing?
omneh_ is now known as OmnipotentEntity
OmnipotentEntity has quit [Changing host]
OmnipotentEntity has joined #nixos
<jtojnar>
hmm, journal contains switching to system configuration /nix/store/7pn3w0k9bm2v8z08zzlrrq7fjjvjiw7m-nixos-system-kaiser-20.09.20200623.90aeb80 failed (status 4)
o1lo01ol1o has quit [Remote host closed the connection]
hoverbear has quit [Quit: Connection closed]
hoverbear has joined #nixos
<jtojnar>
but the switch-to-configuration did not really complain, maybe that is why the new generation did not appear in the boot menu
billiozzi[m] has joined #nixos
adamblackwater[m has joined #nixos
<jtojnar>
Hayden^
medvid has quit [Ping timeout: 240 seconds]
<jtojnar>
right "warning: error(s) occurred while switching to the new configuration" would be printed by nixos-rebuild, not switch-to-configuration
drakonis_ has quit [Read error: Connection reset by peer]
<sephii>
(trying again now there seems to be more people online ^^): I'm having trouble building my first Python package. Here are my questions in order of importance: 1) I'm trying to build a Python application that has a "plugins" system (which are regular Python modules). The plugin depends on the application, so I can't build the plugin with `buildPythonPackage` and then add it to `propagatedBuildInputs` of the
<sephii>
application because the plugin depends on the application. 2) What's the recommended way of creating a derivation for an application that can have multiple plugins, so the user can choose which ones they want? 3) The application uses pytest to run the tests. I tried adding `checkInputs = [ pythonPackages.pytest ];` but the check phase fails because it cannot find pytest. Any idea why?
billiozzi[m] has left #nixos ["User left"]
drakonis_ has joined #nixos
erasmas has joined #nixos
orivej_ has quit [Quit: No Ping reply in 180 seconds.]
o1lo01ol1o has quit [Remote host closed the connection]
<regnat>
sephii: For 1 I think the usual practice is to package the application and the plugins separately, and then have an extra derivation that wraps the application with some plugins
<regnat>
(For a python thing probably by adding them to PYTHONPATH)
<{^_^}>
[nixpkgs] @NeQuissimus pushed 10 commits to master: https://git.io/Jfxbf
orivej has quit [Ping timeout: 256 seconds]
orivej has joined #nixos
<hyper_ch_>
sphalerite: not the ethernet
<{^_^}>
[nixpkgs] @NeQuissimus pushed 6 commits to release-20.03: https://git.io/Jfxbm
drakonis has joined #nixos
infandum has joined #nixos
drakonis_ has quit [Ping timeout: 264 seconds]
<KarlJoad>
Hey guys, what emacs major and minor modes are best for editing Nix files, both configuration and project? I already have nix-mode, but are there others that I should know about?
<bqv>
i also have nix-buffer and nix-update installed
<bqv>
protip: install epkgs then just M-x epkgs-list-packages and type nix
orivej has quit [Quit: No Ping reply in 180 seconds.]
<infandum>
Should I use my system configuration.nix or home manager to manage packages?
<infandum>
On those lines, on a non-nixos computer, should I use a my-packages.nix file or home manager to manage packages?
<bqv>
infandum: what do you think the difference is?
<hoverbear>
(Asking again) is there any tricks/tips to getting `stdlib.h` available in a `mkDerivation` based `nix-shell`?
<jtojnar>
teto1 works for me
<jtojnar>
teto1 nix (Nix) 2.4pre20200521_00b562c
<infandum>
bqv: Well, with home-manager it looks like I could use the same package and configuration across all systems, regardless of if they are nixos or not, but home-manager is not "official".
<KarlJoad>
bqv: I'm staying away from epkgs because my emacs config follows me through many different OSs, and Nix is not always installed. But nix-mode, nix-buffer, and nix-update are the 3 you use?
<hoverbear>
^ My error from inside a `nix-shell` trying to build some vendored snappy/rdkafka libraries.
<teto1>
jtojnar: I have the exact same version yet it says "fatal: ambiguous argument 'release-19.09': unknown revision or path not in the working tree." . `nix registry list` list the entry though
<bqv>
infandum: what i was gunning for is, if you install things using home-manager, they're only available to that user's environment - if you're happy with that and you get benefits then why not
<bqv>
KarlJoad: yeah, i don't use epkgs for package management, i just use it as a quick way of searching melpa :p
<teto1>
aka nix registry lists it under `global flake:nixpkgs/release-19.09 github:edolstra/nixpkgs/release-19.09`
daniel has joined #nixos
daniel is now known as Guest67603
<KarlJoad>
bqv: Ah. Searching wasn't my problem. I just wanted to know if there are other modes I should consider. I miss some kind of autocompletion, if there is any.
<bqv>
nix doesn't have an lsp server, nix-mode's kinda the best you'll get
<bqv>
especially since it's so painful to statically analyse
<Guest67603>
Hello, does anyone have any experience with apacheHttpd and mod_wsgi ? I can't get it to work
<KarlJoad>
RIP... Oh well.
<patagonicus>
On the topic of home-manager: is there something built into it for automatic updates? For NixOS itself I can use system.autoUpgrade.
noudle has joined #nixos
drakonis has quit [Ping timeout: 240 seconds]
<KarlJoad>
Also, how does users.mutableUsers work? I have it set to false, but I was still able to create a new user as root. Is that supposed to happen?
<teto1>
bqv there is rnix for an lsp server but I never tried. Just saw a default config for it in neovim-lsp
fresheyeball has quit [Ping timeout: 256 seconds]
<bqv>
oh, i stand corrected
bbb1 has quit [Quit: WeeChat 2.7.1]
<KarlJoad>
Melpa doesn't seem to have rnix.
<bqv>
it wouldn't, you'd install lsp-mode or eglot and set it up to run the rnix binary
fresheyeball has joined #nixos
johnw has joined #nixos
<{^_^}>
[nixos-homepage] @github-actions[bot] pushed commit from GitHub Actions to master « Update flake.lock and blogs.xml [ci skip] »: https://git.io/JfxNH
medvid has joined #nixos
orivej has quit [Quit: No Ping reply in 180 seconds.]
<KarlJoad>
Ah... I see what you mean now.
orivej has joined #nixos
<{^_^}>
[nixpkgs] @flokli merged pull request #91195 → extlinux-conf-builder: expose and use base builder command, allow a custom FDT to be specified → https://git.io/JfbpG
<KarlJoad>
How does users.mutableUsers work? I have it set to false, but I was still able to create a new user as root. Is that supposed to happen?
o1lo01ol1o has joined #nixos
<OmnipotentEntity>
Yes, you can create a new user, but on reboot the users are reset, if I understand the way it works KarlJoad
<cransom>
the description on `users.mutableUsers` from man configuration.nix describe what happens when it's false
gungoman has joined #nixos
orivej has quit [Ping timeout: 246 seconds]
orivej has joined #nixos
<KarlJoad>
cransom: Yes, I just wanted to find out when "system activation" is. The warning states: "the contents of the user and group files will simply be replaced on system activation", and I wasn't sure when system activation was.
Seebi has quit [Quit: Leaving]
<cransom>
after a nixos-rebuild switch
drakonis has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<adisbladis>
Or at boot
<KarlJoad>
Is it boot, or `nixos-rebuild switch`, or both?
boxscape has quit [Quit: Connection closed]
<OmnipotentEntity>
I know it happens at boot, so both.
<bqv>
i made the mistake of updating nixpkgs and now it's time for damage control >_>
<adisbladis>
KarlJoad: Both
<KarlJoad>
Ok, perfect.
<adisbladis>
activation happens when activating a generation, activating the selected generation happens at boot-time
<adisbladis>
Or when switching to a new generation
<bqv>
have you tried switching to nixFlakes first?
<arianvp>
yes
<bqv>
oh then rip
<arianvp>
nix flake show doesnt exist. nix flake list-inputs always crashes with "flake lacks attribute 'edition'"
<adisbladis>
Huh
<bqv>
oh, nixFlakes is old
<bqv>
put edition = 201909; in your flake.nix
<arianvp>
nix shell "is not a recognised" command
<OmnipotentEntity>
nix-shell
<bqv>
wow nixFlakes is very old
<bqv>
OmnipotentEntity: no
<arianvp>
the guide says "nix shell"
<adisbladis>
arianvp: `nix shell` -> `nix run`
<adisbladis>
Recently `nix run` was renamed
<OmnipotentEntity>
oic
<arianvp>
but this tutorial is not even a month old :/
<bqv>
yeah it was renamed the other way
<bqv>
nixFlakes is probably way older than a month old
<adisbladis>
OmnipotentEntity: We are talking specifically about flakes
<bqv>
it's set manually to a successful hydra build
<arianvp>
but the tutorial says I should get nixFlakes from the 20.03 channel
<OmnipotentEntity>
I was on the page, and the very first line was nix-shell, I didn't see lower where the nix shell was.
<bqv>
huh
<OmnipotentEntity>
My b
<OmnipotentEntity>
Considering arianvp claimed that "not a single worked so far" I thought he was stuck at the top.
<OmnipotentEntity>
sorry
<OmnipotentEntity>
not a single command, I accidentally a word.
m0rphism has joined #nixos
<arianvp>
ah hmm maybe this is my nix channel being weird
<bqv>
OmnipotentEntity: didn't mean to be rude, was just typing fast
<arianvp>
let me make sure this isnt PEBKAC =)
noudle has quit []
<cole-h>
arianvp: I wouldn't be surprised if it isn't -- I think someone else came in here recently with the same complaint.
cosimone has joined #nixos
lsix has quit [Quit: WeeChat 2.8]
lord| has quit [Ping timeout: 264 seconds]
<bqv>
arianvp: you are running 'nix shell' from within the nixFlakes env, right? cause I tested that shell here and it's fine, and it's using the correct nix
o1lo01ol1o has joined #nixos
<bqv>
so 21/05 nixFlakes does have the shell command
camsbury has joined #nixos
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
orivej has quit [Ping timeout: 246 seconds]
<bqv>
i could imagine what might have happened is your shell path is misordered and some other nix version is overriding it, try `which -a nix`
lord| has joined #nixos
<bqv>
top one should be a plain nix path
orivej has joined #nixos
<camsbury>
hey does anyone know how I can set `pkgs` for the entirety of `home-manager` without writing an overlay for everything
cosimone_ has joined #nixos
<camsbury>
tiptoeing my way into using it and I'd like to use a pinned version I have for all packages I pull in through `home-manager` instead of the default nixos config
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
<pylsan>
I got a linux-mix and it uses nix as secondary pkg manager. Cant seem to check for updates to those programs in a easy way?, supose to rebuild all ? But I dont have a "OS" definition file.
knupfer has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @vcunat merged pull request #91363 → small treewide: his -> theirs/its → https://git.io/Jfx5f
<jtojnar>
pylsan what is linux-mix?
<pylsan>
Just linux mint and lack of packages made me trow in nix
orivej has quit [Quit: No Ping reply in 180 seconds.]
orivej has joined #nixos
drakonis_ has quit [Ping timeout: 264 seconds]
<camsbury>
looking to set `pkgs` for all of `home-manager` to a value, open to tips!
Neo-- has quit [Ping timeout: 260 seconds]
drakonis_ has joined #nixos
cosimone_ has quit [Remote host closed the connection]
karetsu has joined #nixos
cosimone_ has joined #nixos
<vika_nezrimaya>
camsbury: override NIX_PATH for home-manager with `alias home-manager='env NIX_PATH=nixpkgs=/home/camsbury/nixpkgs home-manager'`?
Guest67603 has quit [Quit: Guest67603]
<camsbury>
I have it as a NixOS module, imported from a channel
<camsbury>
so idk if that makes sense in this case
davidv7 has joined #nixos
<vika_nezrimaya>
it does not
<ar>
/33
<vika_nezrimaya>
camsbury: home-manager doesn't allow you to do that when used as NixOS module, but by modifying <home-manager/nixos/default.nix> you're able to patch that ability in.
<vika_nezrimaya>
This file constructs the home-manager Nix module itself, and it has a pkgs argument. If you manage to make that argument a config option in config.home-manager attribute set, you'll be able to customize it.
FRidh has quit [Ping timeout: 256 seconds]
<vika_nezrimaya>
but you'll have to fork it
<camsbury>
So basically I should clone it locally, update the `default.nix`, and import it directly?
<vika_nezrimaya>
yep
<camsbury>
kk
<camsbury>
thanks
Smith[m]1 has joined #nixos
nuncanada has joined #nixos
cYmen has quit [Quit: bye]
zupo has joined #nixos
cYmen has joined #nixos
Thra11 has quit [Ping timeout: 246 seconds]
asheshambasta has quit [Remote host closed the connection]
evanjs has quit [Read error: Connection reset by peer]
alp has quit [Ping timeout: 272 seconds]
<sephii>
I ran `nix-build default.nix` in a directory and I now have a `result/` directory I cannot remove ("read-only filesystem"). How can I remove it?
evanjs has joined #nixos
vika_nezrimaya has quit [Ping timeout: 264 seconds]
<bqv>
rm result
<bqv>
note lack of trailing slash
acarrico has joined #nixos
<sephii>
d'oh -_- thanks!
<cole-h>
camsbury: You can set `useGlobalPkgs = true;` to make home-manager use your nixpkgs channel for all pkgs
<cole-h>
bqv: I usually do `unlink result` instead :P
<bqv>
also this
<camsbury>
cole-h: yea... I was just hoping I could actually take advantage of them being different - stable for non home-manager and unstable for home-manager
<cole-h>
`_module.args.pkgs = unstable-pkgs;` might work in your user's h-m config
<cole-h>
But really, if you want unstable packages, it's easier than that...
<simpson>
seven-eleven: Normally you'd add a Nix boot/install image to your host management system. Indeed you'd presumably use NixOS. What are you planning?
ris has joined #nixos
_ris has quit [Ping timeout: 272 seconds]
<lukegb>
glittershark: can you run nix in single-user mode?
orivej_ has quit [Ping timeout: 264 seconds]
orivej has joined #nixos
<simpson>
seven-eleven: Oh, or maybe another possibility is that you're imagining deploying some Nix-built package, and you're wondering how to deploy Nix in order to deploy that package. In *that* degenerate case, you only need to copy the package (and its closure) itself, and Nix doesn't need to be present on the target machine.
ashesham` has joined #nixos
<seven-eleven>
simpson, im using ubuntu and i want to install nix package manager for all users with the nix daemon. i think i have to run the nix install script so all necessary files are created in /etc/profile.d/
<seven-eleven>
simpson, yeah and I'd like to do that for many other hosts automatically if possible :-)
FRidh has quit [Ping timeout: 258 seconds]
FRidh has joined #nixos
<simpson>
seven-eleven: Can you be more specific than "automatically"? Which configuration management are you using? How do you normally do things automatically with Ubuntu?
ris has quit [Read error: Connection reset by peer]
<seven-eleven>
simpson, usually I use ansible to manage many hosts at once, but ansible either runs a script as root or as user, but nix install script requires to run some stuff as user and sometimes requires sudo, therefore im stuck there
ris has joined #nixos
<simpson>
seven-eleven: If you were to delegate NOPASSWD permissions to sudo (which I am *not* necessarily recommending, as it's dangerous) then I think, from a cursory reading, that the installation is otherwise non-interactive.
<seven-eleven>
I think that would work, let me try to set NOPASSWD before running the script and removing it after I run the script
<seven-eleven>
that could be a workaround
<simpson>
seven-eleven: However, you might find something like https://github.com/ktosiek/ansible-nix more useful; I think that this is an Ansible role for installing Nix declaratively.
<glittershark>
lukegb: I mean this is literally just the default nix config from buildImageWithNixDb
<seven-eleven>
simpson, oh yeah I saw that, the age of the last commit worried me, thats why I thought using the install script (which is well maintained) is safer :-)
<seven-eleven>
simpson, tried it with sudo NOPASSWD. i agree with you it should be avoided but it worked fine and in ansible i can configure it to automatically disable NOPASSWD right after nix was installed
orivej has quit [Ping timeout: 260 seconds]
orivej has joined #nixos
<FRidh>
pep 517/518 took quite some time to get fully implemented. That's more or less done now so now its time to move further on that basis
<simpson>
seven-eleven: Okay. I would definitely invest time in trying to get the community role to work, and contributing updates as needed, if I were to sink more time into the Ansible route.
<simpson>
But at least there's a route forward now.
<seven-eleven>
yeah, thank you :-)
o1lo01ol1o has quit [Ping timeout: 240 seconds]
<{^_^}>
[nix] @Ericson2314 opened pull request #3738 → WIP: Hash always has a valid type → https://git.io/Jfpeb
<sephii>
I think my package is now working :) I'd like to distribute it alongside my app so users can install it. How is this usually done (without adding it to nixpkgs)? If I distribute the default.nix file, how can other people use it? Also I had to use `{ pkgs ? import <nixpkgs> { } }:` at the top of my file, instead of the usual `{ pkgs, buildPythonApplication }:` etc. because it wouldn't build with this. What's
<sephii>
the proper way of doing it?
<gchristensen>
armin: sure
atopuzov[m] has quit [Quit: authenticating]
atopuzov[m] has joined #nixos
ashesham` has quit [Ping timeout: 272 seconds]
cosimone_ has quit [Ping timeout: 260 seconds]
cosimone has joined #nixos
<seven-eleven>
simpson, this is how the workaround looks in ansible, worked fine https://dpaste.org/NLoB :-)
Neo-- has joined #nixos
<simpson>
seven-eleven: Whatever latches your hatches. I personally would try to figure out why Ubuntu and Ansible are necessary, and work to deprecate them in the organization.
* simpson
radical
Spritzgebaeck has joined #nixos
cosimone has quit [Client Quit]
karetsu has quit [Quit: WeeChat 2.8]
gxt has quit [Quit: WeeChat 2.8]
justanotheruser has quit [Ping timeout: 265 seconds]
justanotheruser has joined #nixos
<Spritzgebaeck>
hello, i'm a new nixos user and come from archlinux. i've got a problem with a systemd user service. it's running a ruby script with following shebang #!/usr/bin/env ruby, it runs if i start it from a terminal, but systemd throws an error: /usr/bin/env: ‘ruby’: No such file or directory
<seven-eleven>
simpson, replacing ubuntu with nixos and ansible with? I think nixos can do some automatic installation/configuration through configuration.nix/config.nix but maybe it cant replace everything ansible can do
<cole-h>
Spritzgebaeck: You'll need to add ruby to its PATH. Typically done by `systemd.user.services.<service>.path = [ pkgs.ruby ];` or something similar.
Thra11 has joined #nixos
<simpson>
seven-eleven: NixOS's base tools, plus ssh, are often enough. There are several tools, like Nixops and Morph, which are largely wrappers around nix-copy-closure and ssh.
<Spritzgebaeck>
cole-h: ah, thanks
<simpson>
But, again, that's a long-term plan, and it's not necessary to be ready for that today. Just something to look forward to in the future.
<teto1>
to answer my own question: `LANG=C nix shell github:edolstra/nixpkgs/23f6739791ea779310ea6bae07f10775631582fd#vim` worked around my issue with the "ambiguous" issue
<kw25>
When I call 'xmonad --recompile' as user, ghc isn't finding the XMonad module. Any ideas why? I have xmonad, xmonad-contrib, and xmonad-extras in my configuration.nix.
FRidh has quit [Client Quit]
orivej_ has quit [Quit: No Ping reply in 180 seconds.]
<{^_^}>
[nixos-homepage] @edibopp pushed 2 commits to master: https://git.io/JfpUq
<{^_^}>
[nixos-homepage] @edibopp merged pull request #480 → Migrate the security teams GPG key download to keys.openpgp.org → https://git.io/JfbZJ
orivej has joined #nixos
<kw25>
Or alternatively, is there a way to completely wipe out my per-user nixos configuration and hope that all the important bits can be rebuilt from configuration.nix?
stepcut has quit []
benny_ is now known as benny
D_ has quit [Ping timeout: 244 seconds]
D_ has joined #nixos
kmein has quit [Ping timeout: 240 seconds]
endocrimes has quit [Quit: running from the computers]
kmein has joined #nixos
endocrimes has joined #nixos
inkbottle has joined #nixos
energizer_ has joined #nixos
agsdheidjd has quit [Ping timeout: 260 seconds]
zebrag has quit [Ping timeout: 264 seconds]
<{^_^}>
[nixpkgs] @Mic92 merged pull request #91327 → Fix RPATH for libraries downloaded by rustup → https://git.io/JfxGG
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
alp has joined #nixos
kw25 has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @benley opened pull request #91375 → discord: fix appindicator tray icon support → https://git.io/JfpTY
visl has joined #nixos
KarlJoad has joined #nixos
knupfer has quit [Quit: knupfer]
knupfer has joined #nixos
<KarlJoad>
When using `users.mutableUsers=false` are users that are created supposed to continue having a nix profile?
<gchristensen>
yeah
<gchristensen>
but the set of users and their passwd/groups/etc will not be modifiable
<KarlJoad>
So, from what I understand, root can still create users and groups, but they will be cleared after a system activation or reboot. The passwords of users/groups are NOT modifiable. And when a user that was created is removed, they keep their nix-profile in the store?
<gchristensen>
I don't think root can still create users
teto1 has quit [Ping timeout: 272 seconds]
<glittershark>
why would I be getting "file 'nixpkgs' was not found in the Nix search path" even though I have a channel called nixpkgs?
<KarlJoad>
I have `users.mutableUsers=false` and I was able to make a `test` user.
<gchristensen>
and yeah, a user which is no longer exists due to a system change will continue having their profile and their hom edirectory
<KarlJoad>
Which was what I wanted, but I'm sure that's expected behavior. However, why do they keep their profile and home directory?
<KarlJoad>
Shouldn't those be cleared after a system-level change and activation?
<gchristensen>
NixOS makes (almost no) attempt to manage user state, and for good reason
diekl has joined #nixos
<gchristensen>
NixOS configurations should roll forward and back and forward again leave you in the same state
<gchristensen>
if it deleted user data, that promise would not hold
<KarlJoad>
Ok. I that makes sense. Is it safe to just remove that user's directory in the `/nix/var/nix/profiles/per-user` then?
<gchristensen>
sure
<gchristensen>
an example of this is one time I deployed my router's configuration to a backup server. with most configuration management tools this would be a crisis, but with NixOS the damage was entirely undone by rolling back
<{^_^}>
[nixpkgs] @LnL7 merged pull request #90463 → clamav: enable build on darwin → https://git.io/Jf7Ra
<KarlJoad>
Understood. I was just not sure about all of the semantics that go along with `users.mutableUsers=false`
<gchristensen>
and it would have been a catastrophe if the router's configuration changed anything having to do with partitioning / data layout / user data
<gchristensen>
understood
<gchristensen>
the semantics are pretty much: passwd and groups are immutable
<gchristensen>
(I am pretty sure you should not be able to create a user with mutableUsers = false ...)
teto1 has joined #nixos
<KarlJoad>
Strange. I was able to make a `test` user for testing a home-manager module today while I have `users.mutableUsers = false`. It correctly removed the user and their groups, and left their information behind (nix-profile and home-directory).
<gchristensen>
interesting
<gchristensen>
good to know :)
<KarlJoad>
It might be because I haven't given a hashed password to root, though I doubt it.
<KarlJoad>
As far as I know, my user is the only one affected by the mutableUsers setting.
<KarlJoad>
On a related note, is there a way to find what is pointing to a particular file with a symlink? Essentially, go backwards from an item to possible symlinks to that starting point.
orivej has quit [Quit: No Ping reply in 180 seconds.]
<kw4775[m]>
&wg 4
orivej has joined #nixos
<KarlJoad>
energizer: That is actually super helpful. Now, the version of this package could have been installed by a user too, so I need to dig through their files as well.
davidv7 has quit [Remote host closed the connection]
camsbury has quit [Remote host closed the connection]
<wadkar>
Hi, I wanted to know how to go about installing nix on ubuntu where my root partition is already full. I have `/mnt` where there is plenty of space, but I don't think I can symlink /nix to /mnt/nix, right? What are my options here? Any tips? Suggestions? Thanks!
davidv7 has joined #nixos
<KarlJoad>
Is there somewhere I haven't considered? The user's home directory and /nix/var/nix/profiles/per-user/user are the 2 I've looked at so far.
<obadz>
energizer++
<{^_^}>
energizer's karma got increased to 21
<gchristensen>
wadkar: how about a bind mount from within /mnt/nix to /nix ?
<{^_^}>
[nixpkgs] @tshaynik opened pull request #91376 → Bump linode-cli version → https://git.io/JfpkN
<wadkar>
gchristensen: cool, that'll work `mount --bind /mnt/nix /nix` just need to make sure those dirs exists. Thanks a lot!
orivej has quit [Quit: No Ping reply in 180 seconds.]
teto1 has quit [Ping timeout: 260 seconds]
<itshaydendev>
Anyone?
xantoz has quit [Client Quit]
xantoz has joined #nixos
orivej has joined #nixos
<simpson>
What's the question?
<simpson>
Or do you actually need a GH admin?
teto1 has joined #nixos
<drakonis>
itshaydendev: try cloning the repository?
<itshaydendev>
drakonis: The issue isn't with nixpkgs, it's with how it's updating the config. Because it won't
<itshaydendev>
simpson: basically I have `services.xserver.windowManager.bspwm.configFile` set in my configuration, but once it's installed the first time, it never updates.
orivej has quit [Quit: No Ping reply in 180 seconds.]
absoluutely has quit [Ping timeout: 256 seconds]
orivej has joined #nixos
<simpson>
itshaydendev: I don't use bspwm, but just to double-check: You changed the configuration file, and then did a nixos-rebuild switch and logged out and back in, but it was still the old configuration?
<itshaydendev>
yeah.
<itshaydendev>
basically, i forgot to include my polybar startup script. so I added "sh ~/.config/polybar/launch.sh" to my bspwmrc, and then rebuilt, relogged (even tried rebooting) and nothing happened.
<itshaydendev>
i know the script works because if i drop into a tty and start alacritty with "DISPLAY=:0 alacritty", and run the same command, it works fine
<{^_^}>
[nixpkgs] @vcunat pushed 24 commits to release-20.03: https://git.io/JfpLW
<{^_^}>
[nixpkgs] @vcunat merged pull request #91226 → [20.03] Backport linux latest and testing → https://git.io/JfNCy
<simpson>
Can bspwm tell you its configuration file path? You could cross-reference to see which file is being used. It might be in the process arguments which you could retrieve with top or htop or /proc or etc.
sbetrium_ has quit [Read error: Connection reset by peer]
<betawaffle>
Can anyone think of a reason why it’d be a bad idea to have /nix be backed by overlayfs?
<b42>
what's the advantage?
<itshaydendev>
simpson: oh, sorry, weechat should really notify me when someone types here lol
<itshaydendev>
simpson: i'll have a look
itshaydendev has quit [Quit: WeeChat 2.8]
orivej has quit [Ping timeout: 265 seconds]
orivej_ has joined #nixos
<betawaffle>
b42: inside a container, allowing the store to be backed by both the store from the image and write new stuff to a bind-mounted volume from the host.
<chipb>
I don't suppose there's a pre-canned way to build all fixed output derivations contributing to an arbitrary derivation for, say, offline build of the arbitrary derivation later on?
itshaydendev has quit [Quit: WeeChat 2.8]
<chipb>
I gather that rust/golang/etc. are special beasts for their dependency handling, so ignoring those is fine for my purposes.
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « python37Packages.azure-mgmt-security: 0.3.0 -> 0.4.0 »: https://git.io/Jfptg
KarlJoad has quit [Quit: Konversation terminated!]
<Yaniel>
oh yeah, haskell
orivej_ has quit [Quit: No Ping reply in 180 seconds.]
<chipb>
of course, maybe I could muddle through it on my own by processing the dependency graph and looking for derivations with the outputHash attribute?
<chipb>
I suppose I could just live with setting up proxy and do the builds in that environment, but I've got far more compute at my disposal where there is no proxy. bah.
kenji has quit [Ping timeout: 256 seconds]
<Yaniel>
I mean, getting the nix expressions for them is the hard part
<Yaniel>
once you have that it is mostly quite straightforward
doyougnu has quit [Ping timeout: 264 seconds]
<Yaniel>
for rust, you generate a nix expression from the Cargo.toml and then it fetches the crates and builds them in the usual nix way
<Yaniel>
I believe the other languages mentioned work similarly
<hoverbear>
Yaniel Things kind of fall apart when you try to build for musl sadly
<Yaniel>
that's a separate issue, no?
<chipb>
if I was particularly married to musl, I'd probabably just use alpine. :-P
<seven-eleven>
how can I remove a package with nix? i tried `nix-env --uninstall tor-browser; nix-collect-garbage` but tor-browser is still installed
fendor_ has quit [Quit: Leaving]
<chipb>
hm. I at least recall seeing that as a candidate for a Rust crate FOD fetcher on a github issue somewhere.
<cole-h>
seven-eleven: That only works if you `nix-env --install`'d it. If it's part of your systemPackages or home.packages, you'd need to remove it from there
<Yaniel>
(and then rebuild your system)
<cole-h>
^
<seven-eleven>
hm installed it with `nix-env -iA tor-browser` in the user shell
<seven-eleven>
i have to remove "tor-browser-bundle-bin-9.5" not tor-browser
<{^_^}>
[nixpkgs] @veehaitch opened pull request #91380 → python3Packages.tlsprofiler: init at 1.0 → https://git.io/Jfpmm
KarlJoad has quit [Remote host closed the connection]
pjt_tmp has joined #nixos
<seven-eleven>
now its gone, thank you!
<infinisil>
Yeah, so I suggest the fzf way of uninstalling stuff, then you won't type the wrong thing :)
<seven-eleven>
yeah, im going to alias it right away
<infinisil>
Or alternatively, don't use nix-env at all. Many people like to declaratively install user packages instead
<infinisil>
,declarative
<{^_^}>
There are multiple ways of managing declarative profiles. 1) Attrset, compatible with imperative use of nix-env https://git.io/fAQHW ; 2) buildEnv, providing more control over the paths that are linked into the profile https://git.io/fp0aU ; 3) home-manager, providing nixos-like config for your ~ https://github.com/rycee/home-manager
Chiliparrot has quit [Ping timeout: 240 seconds]
<seven-eleven>
yeah thats why i wanted to uninstall it so I can install it with the declarative way :-)
<seven-eleven>
infinisil, i guess you are using home-manager?
<infinisil>
Yea
<seven-eleven>
i wanted to try nix without home-manager first, but I think i try home-manager now
<miguel_clean>
I have declared a simple gitolite/nginx/gitweb setup, but it seems that each time I push something to a branch, this branch becoms invisible in gitweb until I force the restart of the gitweb cgi script. I am completely new to nixos and wonder if I might be missing something obvious?
orivej has quit [Ping timeout: 246 seconds]
orivej has joined #nixos
slack1256 has joined #nixos
justanotheruser has quit [Ping timeout: 240 seconds]
NeoCron has quit [Ping timeout: 260 seconds]
<aleph->
Hmm, I'm curious has anyone built a python package that uses setuptools_scm? Seem to be having an issue packaging a package using it.
Vikingman has joined #nixos
erasmas has quit [Quit: leaving]
<b42>
there seems to be quite a lot of matches when grepping for setuptools_scm in pkgs/development/python-modules
<{^_^}>
[nixpkgs] @hmenke opened pull request #91382 → v2ray-plugin: init at 1.3.1 → https://git.io/JfpYS
Rusty1 has joined #nixos
<aleph->
Hmm, odd.
teto1 has quit [Quit: WeeChat 2.8]
<aleph->
Getting this error when attempting to build a local copy of dosage. https://paste.rs/eHV
<clever>
aleph-: its trying to do network at build time, which isnt allowed
<aleph->
Ah, huh. So what's the recommend approach again? Been ages since I've done some nix packaging.
<clever>
aleph-: you need to figure out why its doing network, and make it stop
<aleph->
Fair enough.
KarlJoad has quit [Quit: Konversation terminated!]
<infinisil>
aleph-: Change setuptools to setuptools_scm?
<qyliss>
It's not actually hung, it just takes a long time
<hoverbear>
Ahhhh okay great
<qyliss>
When garbage collecting, every deleted path is in fact just moved to /nix/store/trash
codygman has quit [Read error: Connection reset by peer]
<hoverbear>
That'd make sense
<qyliss>
And then at the end of the GC, that directory is deleted
codygman has joined #nixos
<hoverbear>
Ahhhh okay perfect. Thank you. :)
<qyliss>
It's basically a huge rm -r, and that takes forever because Unix doesn't give us a way to delete a non-empty directory
<hoverbear>
Since it's over 300gb it'll take a moment
<qyliss>
YYYYYYYYYYYYYYYYYYYYYYeah
<hoverbear>
I was just surprised since it's an optane and should in theory be able to wipe itself in a few seconds
<hoverbear>
Then again zfs isn't fast
<qyliss>
It's just so many syscalls
<qyliss>
As well as anything to do with the filesystem
<hoverbear>
Filesystems are bad
<qyliss>
Because it has to delete every file, one by one
<qyliss>
I always thought ZFS should have just implemented a directory delete operation
<hoverbear>
Agree
<qyliss>
Have some zfs rmr command or something to do it
<kiwiirc>
does linux support zfs mirror over 2 nvme drives?
<hoverbear>
Sadly I can't keep using zfs since we need to do kubernetes tests and the k8s support on nixos seems broken :(
<hoverbear>
(with zfs)
<hoverbear>
kiwiirc You can build a pool over multiple drives yes
absoluutely has quit [Ping timeout: 265 seconds]
slack1256 has quit [Ping timeout: 240 seconds]
<aleph->
Hmm, okay. So if I'm building a python package from GH. What would be the best way to include several scripts a $src/scripts dir in $out/bin/? Should I just completely overload installPhase?
<hoverbear>
Is there a way to say "The thing containing me" in a recursive set?
<aleph->
Indeed. It's a helper script to be called after to generate a cbz archive.
<qyliss>
Then they have to go in $out/bin, because that's what goes into PATH
<qyliss>
Oh I misread your quesiton
<qyliss>
use postInstall
<aleph->
Got it. Hmm, is there a debug I can turn on for building? Seems to not be copying my scripts to $out/bin/ and I'd like to gather introspection
orivej has quit [Ping timeout: 240 seconds]
gustavderdrache has left #nixos [#nixos]
orivej has joined #nixos
<qyliss>
NIX_DEBUG = true might help
<aleph->
Nope, and seems I can't just do an ls in my postInstallPhase either. Hmm