<{^_^}>
[nixpkgs] @Ma27 opened pull request #80582 → nixos/nixpkgs: add override/merging capabilities from the module system to the `nixpkgs.config` option → https://git.io/JvBwd
jgeerds has quit [Ping timeout: 240 seconds]
justanotheruser has quit [Ping timeout: 272 seconds]
<newuser99>
Normally, the way I can tell whether my files are broken is by running nix-shell and it'll tell me. Is there a way to find out whether my files are broken using lorri?
mbrgm_ has joined #nixos
werner291 has quit [Quit: werner291]
ng0 has joined #nixos
mbrgm has quit [Ping timeout: 260 seconds]
mbrgm_ is now known as mbrgm
Synthetica has quit [Quit: Connection closed for inactivity]
<some_nick>
I tried installing nixos on a computer with an amd Ryzen 5 3400g apu and I'm not able to use two monitors. The hdmi port works but not the displayport. Any ideas of how to fix this?
casaca has quit [Quit: leaving]
Soo_Slow has joined #nixos
<Ashy>
some_nick: can you boot an Ubuntu live USB and test that quickly?
<Ashy>
also, are they 4k monitors? maybe 1 4k monitor is the max res for a 3400g?
<Ashy>
I'm curious about this as I'm planning a 3400g build soonish
drakonis has quit [Ping timeout: 248 seconds]
<some_nick>
Ashy yeah ubuntu works. They're not 4k monitors!
<{^_^}>
[nixpkgs] @jonringer opened pull request #80595 → python: execute pythonImportsCheckPhase in subshell → https://git.io/JvBiT
<some_nick>
Do multiple monitors require special configuration or should it be working out of the box? I see some config parameters related to multiple monitors under 'services.xserver.xrandrHeads
drakonis has joined #nixos
notgne2 has joined #nixos
<Ashy>
they don't need anything special on my machines, tried gnome or kde?
drakonis_ has quit [Ping timeout: 272 seconds]
<clever>
some_nick: for me, it just works in xfce4-display-settings
<clever>
but the settings only take effect after login, so the login screen is a little wonky
Fare has joined #nixos
<cole-h>
Does anybody here use pacman on NixOS? I saw that there's a package for it, but wanted to know how well it actually works.
<cole-h>
`pacman -F` + `nix-locate -w 'path'` are a fantastic pairing for finding missing headers/binaries/what have you, so I'd love for it to still work when I (eventually) switch over
<lovesegfault>
cole-h: I doubt it's meant to be used as a real pkg manager
<cole-h>
Of course not
<cole-h>
As long as `pacman -F` functions properly, that would be a great boon (to me, at least)
drakonis1 has joined #nixos
<bhipple>
If you give me an example cmd I can try running it on NixOS
<newuser99>
What did I do wrong? The Python environment doesn't have access to the all the dependencies. https://bpaste.net/BIBA
<bhipple>
pacman -F python
<bhipple>
(could not find or read directory: /var/lib/pacman/)
<bhipple>
error: failed to initialize alpm library
<bhipple>
^ Doesn't look promising :)
<cole-h>
RIP.
<bhipple>
but from what I find `nix-index` / `nix-locate` is all you really need
<newuser99>
or, How do I go about diagnosing this?
<cole-h>
Only thing I dislike is all those results that actually aren't what I want lol
<cole-h>
The ones surrounded in parens... bleh
<pbb_>
Hi, is it possible to add a new top-level attribute to the NixOS module system that I can access from any other NixOS module by adding it to the module parameters?
<cole-h>
Though `--top-level` does filter most of those away
<pbb_>
In my case I want to pass some external sources as a variable into my NixOS config derivation, and I would like to have access to them via a sources attribute
justanotheruser has quit [Ping timeout: 272 seconds]
<cole-h>
bhipple++ Thanks for testing that though :^)
<{^_^}>
bhipple's karma got increased to 4
drakonis1 has quit [Quit: WeeChat 2.7]
some_nick has quit [Remote host closed the connection]
kleisli has quit [Ping timeout: 260 seconds]
wiml has quit [Quit: wiml]
reallymemorable has joined #nixos
<newuser99>
I do `nix-build` and it makes a `result/bin/myscript`. How do I make the bin/ directory get added to PATH?
<pbb_>
newuser99: are you on NixOS or on another distribution using Nix?
<newuser99>
pbb_ ubuntu
<pbb_>
Then probably "nix-env -i result/"
<cole-h>
`nix-env -i -f default.nix` (or whatever its name is)
reallymemorable has quit [Client Quit]
<pbb_>
yeah that works too
<iqubic>
Is there some way for me to figure out what name I need to pass in to Home Manager's xsession.pointerCursor.name option? I'm trying to use the Bitata Dodgar Blue from pkgs.bibata-extra-cursors.
<cole-h>
Unless you want to use home-manager, in which case you could create an overlay for it
<cole-h>
iqubic: `man home-configuration.nix` and search for `xsession.pointerCursor`
<iqubic>
I've done that already.
phirsch has quit [Read error: Connection reset by peer]
<iqubic>
And it doesn't really help me identify the proper name.
<cole-h>
The docs make it sounds like you need to specify the package with `xsession.pointerCursor = { package = ...; name = ...; };` or something similar
<iqubic>
Correct.
<cole-h>
`defaultCursor` is "The default cursor file to use within the package."
<iqubic>
I'm using "package = pkgs.bibata-extra-cursors;"
mexisme has quit [Ping timeout: 240 seconds]
<cole-h>
Tried `ls`ing the derivation's directory and finding the cursor file?
<iqubic>
No.
<cole-h>
That's what `defaultCursor` (seems to) want, so try that maybe?
<cole-h>
And if still nothing, try asking in #home-manager as well
<iqubic>
Right. But it's not defaultCursor that I need to set. It's name.
<newuser99>
cole-h: https://bpaste.net/HWRA how do I get the bin/ dir added to the nix-shell's PATH?
drakonis1 has joined #nixos
<newuser99>
or, get that script on the nix-shell's PATH
<cole-h>
newuser99: Look up :-)
<cole-h>
19:18 cole-h │ `nix-env -i -f default.nix` (or whatever its name is)
<cole-h>
`nix-env --install --file default.nix` is the long hand, if that clues you in to what it's actually doing ;)
<newuser99>
cole-h: I just want it installed into the nix-shell, not everywhere
<cole-h>
add `let org = pkgs.callPackage ./org.nix { };` and put `org` in your `buildInputs`?
<newuser99>
cole-h: which buildInputs is that?
gustavderdrache has left #nixos [#nixos]
drakonis1 has quit [Ping timeout: 258 seconds]
kloenk has quit [Remote host closed the connection]
felixfoertsch has quit [Ping timeout: 265 seconds]
xkapastel has joined #nixos
drakonis_ has joined #nixos
<{^_^}>
[nixpkgs] @ryantrinkle opened pull request #80601 → nixos/printers: allow supplying raw ppd file → https://git.io/JvBPi
<cole-h>
What's the unquoted homepage URL RFC again?
ryantrinkle has joined #nixos
<cole-h>
Ah, 45
reallymemorable has quit [Quit: reallymemorable]
<tnks>
I just ran `nix verify --all` and saw `[7523 paths verified, 241 untrusted]`. Are the untrusted packages just part of life? Or is it reasonable to expect all packages to be trusted?
<iqubic>
So, I have an issue here. I'm trying to run "home-manager --switch" to rebuild my new HM config. I'm getting this error: The option `services.lorri' defined in `/home/avi/.config/nixpkgs/home.nix' does not exist.
<iqubic>
Oh, but "man home-configuration.nix" pulls up a similar list, but lacks that configuration option. So I think I need to update home-manager somehow.
Fare has quit [Ping timeout: 240 seconds]
Supersonic has quit [Disconnected by services]
Supersonic112 has joined #nixos
Supersonic112 is now known as Supersonic
<{^_^}>
[nixpkgs] @cole-h opened pull request #80602 → rls: add llvm to buildInputs → https://git.io/JvBXk
mexisme has quit [Ping timeout: 258 seconds]
<iqubic>
So, I don't know what the issue is here.
<{^_^}>
[nixpkgs] @eraserhd opened pull request #80603 → plan9port: fix linker flags for macosx → https://git.io/JvBXY
<cole-h>
Also, shouldn't it be `home-manager switch`, not `--switch`?
<iqubic>
Yeah.
<iqubic>
I was running `home-manage switch`
<iqubic>
Alright... turns out that I needed to be running the unstable version of home-manager.
mrus has joined #nixos
<iqubic>
So all I had to do was switch channels, update channel, and then it all works.
drakonis_ has quit [Remote host closed the connection]
drakonis_ has joined #nixos
<newuser99>
I can do `myOrgEnv = import ./../modules/org;` and then `nix-env -iA nixpkgs.myOrgEnv`. That successfully installs the package. But when I try putting `home.packages = [pkgs.myOrgEnv ] ` it says that entry `is not of type `package'.`
<newuser99>
How can I make it work?
<clever>
newuser99: what does the default.nix return?
<clever>
newuser99: if you eval `myOrgEnv` in `nix repl '<nixpkgs>'` what is the type?
<newuser99>
clever: lambda @ path/default.nix
<newuser99>
looks like I can do `myOrgEnv = self.pkgs.callPackage (import ./../modules/org) {};` and that works
<clever>
newuser99: you dont need the import, callPackage imports for you
<clever>
newuser99: you likely want myOrgEnv = self.callPackage ../modules/org {};
<zeta_0>
hello there, i am using visidata to view the data of a csv file, but i cannot figure out how to get it to copy a whole link? the links cut off at ...
<zeta_0>
clever: hello again, a while back you helped me setup ssh-agent and gpg-agent. anyways gpg-agent seems to automate gpg password pretty well, but for ssh i keep having to manually type my password every time, which is a hassle, any tips?
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
nekroze has joined #nixos
<nekroze>
hi all, I am writing some custom options and am setting some default stuff in the config block with "mkDefault" however when I try to use the value of that option from "config" input at the top of the file I am not getting the value but a set that appears to define the override option. How do I get my custom options to collapse this down to the final value?
cole-h has joined #nixos
veske has joined #nixos
veske2 has joined #nixos
ixxie has joined #nixos
c0c0 has joined #nixos
chimay has joined #nixos
leothrix has quit [Remote host closed the connection]
bvdw has quit [Read error: Connection reset by peer]
bvdw has joined #nixos
leothrix has joined #nixos
chimay has quit [Ping timeout: 255 seconds]
<{^_^}>
[nixpkgs] @xavierzwirtz opened pull request #80609 → mssql-server: init at 15.0.4003.23-3 → https://git.io/JvByr
<{^_^}>
[nixpkgs] @FRidh merged pull request #73819 → Multithreaded perl on darwin → https://git.io/Je6Gj
<{^_^}>
[nixpkgs] @FRidh pushed commit from @knl to staging « perl: Enable threading on darwin »: https://git.io/JvByK
lovesegfault has quit [Quit: WeeChat 2.7]
<{^_^}>
[nixpkgs] @xavierzwirtz opened pull request #80610 → azuredatastudio: init at 1.15.1 → https://git.io/JvByS
nekroze has quit [Remote host closed the connection]
philr has joined #nixos
<CMCDragonkai>
I have a service that I'm writing a custom nixos module for. This service has a command that can be executed multiple times to create multiple instances of it. What is the right way to create the nixos config with the number of instances specified? Also the same service can be ran multiple times with different options as well.
m0rphism has joined #nixos
<sshow>
CMCDragonkai: maybe a named systemd service, where you start them off with `systemctl start myservice@option1.service`, where e.g. option1 might be a reference to a config file
<CMCDragonkai>
Sure, but I want to know how best to write the nixos module config to express this
<CMCDragonkai>
For example... `systemd.services.? = { ... }`, should I generate multiple of these instances
<Taneb>
srk: the thing I'm using writes to the HOME directory, and later reads what it wrote back
andymandias_ has quit [Quit: ZNC 1.7.4 - https://znc.in]
<srk>
you can mkdir it as well
logzet has joined #nixos
<srk>
it doesn't matter if it's /something or mktemp as build is isolated and what's not copied to $out is discarded anyway
KeiraT has quit [Remote host closed the connection]
KeiraT has joined #nixos
werner291 has joined #nixos
wavirc22 has joined #nixos
toppler` is now known as toppler
f0i[m] has joined #nixos
mojjo has joined #nixos
f0i has left #nixos [#nixos]
newuser99 has quit [Remote host closed the connection]
<mojjo>
how can I eval a nix expression that does not result in a derivation.. Something like `nix-eval '1+1' # -> 2`
<mojjo>
(not from inside `nix repl`)
<srk>
nix-instantiate --eval -E '2+2'
<mojjo>
nice, thx!
<srk>
nix eval '(1+2)'
<srk>
mojjo: better ^^
<tilpner_>
srk: Is it? AFAIK the nix CLI is still unstable, and the parentheses are a weird hack
tilpner_ is now known as tilpner
<niksnut>
the parentheses are gone on the flakes branch
cosimone has joined #nixos
<niksnut>
you now have to say 'nix eval --expr "1+2"'
<srk>
haha lol, didn't realize it doesn't work without parentheses
<srk>
niksnut: cool
<tilpner>
That's much better, an extra flag instead of cramming two separate modes into one via weird syntax!
<niksnut>
yes
philr has quit [Ping timeout: 255 seconds]
<niksnut>
fwiw the reason for the parentheses is that it allowed combining different kinds of installables, e.g.: nix run nixpkgs.hello '(with import <bla>; foo.override { ... })'
<niksnut>
so you can't do that anymore
<{^_^}>
[nixpkgs] @rileyinman opened pull request #80618 → iosevka: disable update check in build → https://git.io/JvBQn
werner291 has quit [Remote host closed the connection]
<CMCDragonkai>
a question about `nixos-rebuild` and `-I nixpkgs=./pkgs.nix`
<CMCDragonkai>
this overrides the search path with `nixpkgs`
<CMCDragonkai>
set to a file that exports the whole package set
<CMCDragonkai>
the build succeeds but there's this error
<CMCDragonkai>
`/nix/store/x7hj5hp24flxbx6jnnws5rnjs31g0ybw-bash-4.4-p23/bin/bash: /etc/nixos/pkgs.nix/nixos/modules/installer/tools/get-version-suffix: Not a directory`
<tilpner>
I get that too, it should be harmless
sigmundv has joined #nixos
<CMCDragonkai>
tilpner: you're also using `-I nixpkgs=./pkgs.nix`
<CMCDragonkai>
?
<tilpner>
You can work around it, if you really want to
<tilpner>
The trick is to: mkdir nixpkgs; mv pkgs.nix nixpkgs/default.nix
<CMCDragonkai>
tilpner: why does that solve the problem?
<tilpner>
And then -I nixpkgs=$PWD/nixpkgs
<{^_^}>
[nixpkgs] @edolstra pushed to release-19.09 « nixUnstable, nixFlakes: Update to latest »: https://git.io/JvBQC
<tilpner>
It doesn't by itself, but now your <nixpkgs> is a directory, not a file
<tilpner>
So you can place the file it expects, where it expects it
<CMCDragonkai>
it tries to use the actual script `$nixpkgs/nixos/modules/installer/tools/get-version-suffix`
<CMCDragonkai>
how does putting `pkgs.nix` into `nixpkgs` directory allow that part of the script to run?
<tilpner>
Yes, you can place that now, because now your <nixpkgs> is a directory again
<CMCDragonkai>
oh you mean
<CMCDragonkai>
for me to actually copy that file
<CMCDragonkai>
into that directory
<CMCDragonkai>
for me to manually do this
<CMCDragonkai>
`get-version-suffix`
<tilpner>
Not necessarily, let me check
<tilpner>
It runs that script with bash, so you can place an alternative bash script at that location, which resolves the version number differently
<tilpner>
Presumably, you have pinned your nixpkgs sources somewhere, so you would use that to get the revision
<CMCDragonkai>
hmm i'm afraid of that its accepted that <nixpkgs> is meant to point to a directory
<CMCDragonkai>
will there be problems in the future?
__monty__ has joined #nixos
<tilpner>
Yes, there will be
<tilpner>
You can work around those too
<tilpner>
At the cost of your setup becoming so weird that nobody can help you debug it
iyzsong-x has joined #nixos
iyzsong has quit [Ping timeout: 240 seconds]
<tilpner>
But the only other thing you really need is nixpkgs/nixos/default.nix
veske has joined #nixos
veske2 has joined #nixos
<CMCDragonkai>
tilpner: yea because of <nixpkgs/nixos> paths used everywhere
<tilpner>
Some profiles you find in the wild (or from nixos-generate-config) also use <nixpkgs/...> in imports
<tilpner>
That will be a problem, and you'll have to either provide them, or fix the imports
mexisme has joined #nixos
<tilpner>
You can replace those imports with { modulesPath, ... }: { imports = [ "${modulesPath}/foo" ]; }
vidbina has joined #nixos
zupo has joined #nixos
<CMCDragonkai>
hmm it seems for nixos at least, `<nixpkgs>` is expected to be a directory, and also it's expected that this directory is at least the original main nixpkgs, or a forked representation
<gordon>
Hi. I’m still new to Nix, and since I used a bunch of scripts in my previous system, i’m trying to package those. So I did the python part (https://github.com/gordonzola/mbsync_tools), the repository builds in a venv, but when i write my derivation, I have a python error "ERROR: Could not find a version that satisfies the requirement SecretStorage (from mbsync-tools==0.1.0) (from versions: none)"
<gordon>
i’m trying to understand how `buildPythonPackage` works differently than `python setup.py install`
<tilpner>
CMCDragonkai: Correct, but it works well enough for me. I expect to have to rediscover the whole thing when flakes land, so I've stopped changing things for now
<gordon>
oh. It works after adding my python dependencies in the derivation
<CMCDragonkai>
tilpner: how will flakes affect nixos modules?
<tilpner>
CMCDragonkai: I haven't played with them yet, but AFAICT (and correct me if I tell wrong) our pinning will become unnecessary thanks to flake.lock
o1lo01ol1o has joined #nixos
<tilpner>
CMCDragonkai: nix-channel will be replaced by that, and there'll be no need for "<nixpkgs> is not an actual nixpkgs tree", or so I hope
<tilpner>
CMCDragonkai: I don't have the time to redo my setup right now, but if you're just setting it up for the first time, you might as well give flakes a try?
<CMCDragonkai>
tilpner: do you have any idea why `nixos-rebuild switch -I nixpkgs-overlays=/etc/nixos/nixpkgs-overlay/overlays.nix` doesn't work?
<CMCDragonkai>
I tested it with `nix repl -I nixpkgs-overlays=...`
<CMCDragonkai>
and the overlay applied
<CMCDragonkai>
not sure why rebuild switch doesn't apply it?
<gordon>
now i’m trying to add my derivations to the system. I created a git repository (https://github.com/gordonzola/nixpkgs), but i still don’t really understand how I should add it to the system. `nix-channel` doesn’t seem to support git, so it won’t find the nixexpr.tar.bz2. What should I do to make my packages available system-wide?
jojojojojo has quit [Remote host closed the connection]
jojojojojo has joined #nixos
<tilpner>
CMCDragonkai: Uh oh, you need to apply those manually
<CMCDragonkai>
i thought that the overlay would be applied automatically and i could use custom `pkgs.*`
loyon[m] is now known as ircfreenodenetlo
<CMCDragonkai>
inside the `configuration.nix`
<tilpner>
CMCDragonkai: In your nixpkgs/default.nix, interpret <nixpkgs-overlays> and pass them to nixpkgs while importing
<tilpner>
CMCDragonkai: Well, it depends on how you exactly you do the pinning
<emilazy>
tilpner: I guess "ideally" <foo/bar> should work even if foo is a .nix file dierctly?
<emilazy>
*directly?
gxt has quit [Ping timeout: 240 seconds]
<tilpner>
CMCDragonkai: What did you put in nixpkgs/nixos/default.nix?
<emilazy>
meaning, like... `import¯¹ ((import <foo>).bar)`, or something.
<CMCDragonkai>
tilpner: I've gone back to what I had before
<emilazy>
if you get what I mean.
<CMCDragonkai>
tilpner: which was `nixpkgs=./nixpkgs`
<CMCDragonkai>
and `./nixpkgs` is the actual nixpkgs repo
sigmundv has quit [Ping timeout: 240 seconds]
<tilpner>
emilazy: I guess that might work, import^-1 might just be unsafeGetAttrPos in this particular case
<tilpner>
CMCDragonkai: And you're passing both with -I to nixos-rebuild?
<CMCDragonkai>
the only way i was able to get the `<nixpkgs-overlays>` working in configuration.nix was with reevaluating pkgs like this
<tilpner>
I've never used <nixpkgs-overlays>, but I think it might only apply when importing <nixpkgs>, not when importing <nixpkgs/nixos>
<CMCDragonkai>
but i need the overlay applied for `environment.systemPackages`
<CMCDragonkai>
as in i need the overlay applied at the point of evaluating the `configuration.nix
<tilpner>
Yes, so use nixpkgs.overlays instead
dm9 has joined #nixos
<CMCDragonkai>
tilpner: why doesn't `nixpkgs.overlays` just default to using `<nixpkgs-overlays>` if it is specified in NIX_PATH?
<emily>
tilcreator: right, but I mean you could make <foo/bar> desugar to that directly
<emily>
er, tilpner: sorry
<tilpner>
emily: No, you can't
<CMCDragonkai>
this whole part in the manual seems to be missing information, as in what is this `overlays.nix` file, and where to put it
<CMCDragonkai>
> Because overlays that are set in NixOS configuration do not affect non-NixOS operations such as nix-env, the overlays.nix option provides a convenient way to use the same overlays for a NixOS system configuration and user configuration: the same file can be used as overlays.nix and imported as the value of nixpkgs.overlays.
<{^_^}>
error: syntax error, unexpected IN, expecting ')', at (string):281:31
<CMCDragonkai>
I'm using this now: `nixpkgs.overlays = import <nixpkgs-overlays>;`
<CMCDragonkai>
tilpner: just to confirm, the `$NIX_PATH` doesn't change until you reboot right?
<tilpner>
emily: You can't translate every <foo/bar> to import^-1 ((import <foo>).bar), not without changing nixpkgs entirely. There is no rule that makes sure <foo> will evaluate to an attrset with a bar attribute
<emily>
tilpner: er, right.
<tilpner>
emily: So it would have to be some weird fallback, and I don't think that's worth the complexity
<emily>
tilpner: I guess what I want is for <foo/bar> to be an actual module path and for Nix modules to work filesystem-hierarchically like that.
<emily>
I expect it on my desk by Monday, thanks.
<tilpner>
CMCDragonkai: The option nix.nixPath only takes effect if you login again
<tilpner>
CMCDragonkai: Other means of changing NIX_PATH may take effect immediately
sigmundv has joined #nixos
Synthetica has joined #nixos
iyzsong has joined #nixos
iyzsong-x has quit [Ping timeout: 240 seconds]
<alexarice[m]>
Does anyone know why haskell packages seem to be compiled twice?
<Taneb>
alexarice[m]: they're compiled both without and with library profiling
<alexarice[m]>
Taneb: is there a way to disable one of them?
<alexarice[m]>
just takes a long time
<Taneb>
alexarice[m]: haskellPackages.lib.disableLibraryProfiling, I think
<alexarice[m]>
Taneb: thanks Taneb++
<{^_^}>
Taneb's karma got increased to 14
lsix has joined #nixos
<{^_^}>
[nixpkgs] @Mic92 merged pull request #59414 → lumo 1.9.0 -> 1.10.1 plus darwin support → https://git.io/fjm8d
<aaa_>
Hi, how can I change this http://dpaste.com/1SYD14Y to use "let unstable = import <nixos-unstable> {}; in { ... }" and then use unstable rustc/cargo ?
<aaa_>
Like using unstable.rustup
<ikwildrpepper>
aaa_: the example already uses unstable right? what do you want to do exactly?
justanotheruser has quit [Ping timeout: 240 seconds]
LysergicDreams has joined #nixos
<ikwildrpepper>
aaa_: or do you mean something like this: let nixpkgs = fetchTarball channel:nixos-unstable; unstable = import nixpkgs {}; in unstable.rustc
<nihh>
Hi I am confused as to why my computer wants to build pandas (python package) and not just fetch it from some binary cache? E.g.: nix-shell -p "python3.withPackages (ps: [ ps.pandas ])"
magnetophon has quit [Remote host closed the connection]
<hoppfull>
I'm trying to understand how to properly use NixOS. There seems to me to be three levels of environments. The first is the environment described in configuration.nix which describes the OS. Everyone has access to what's described there always. The next level is with nix-env, which seems to be tied to users. The third level seems to be in the shell
<hoppfull>
with nix-shell and nix-build. Setting up a development environment at the shell level is attractive but some build tools has some global state that doesn't seem possible to limit to the shell. For example, say I installed a global package with npm (I know I can avoid this by installing as a dev dep), then other shells using npm would get access to
<hoppfull>
that global package as well. Is it possible to prevent this? Or should I set up a dev environment on the user level and then just change user?
<hoppfull>
nix-env can produce profiles but are they removed when invoking gc?
<gordon>
you should avoid to use npm packages directly, and create nix derivations that can use npm themselves. This way, even with nix-shell, you’d specify your dependencies, that would be available in your shell, in your case your derivation representing your npm package
<gordon>
hoppfull | nix-env can produce profiles but are they removed when invoking gc? < no, because those packages are present in profile generations that still exist (even if you’re on a newer generation). The garbage collector only removes unreferenced data
<hoppfull>
gordon: ok, so say I have a database installed on multiple profiles, they would all share the same state?
<hoppfull>
Am I thinking wrong about this? Because I could just go with docker but I want to give nix a shot.
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<simpson>
hoppfull: What is "a database"? What can be installed in profiles/environments are the *binaries*, the tools themselves. That is not the same as running services, like starting a database binary with a given database state.
<hoppfull>
If I use npm I can install all dev dependencies locally and if I upgrade a dependency version, I just push to my repo and my team mates will automatically use the new compiler, linter, formatter whatever.
drakonis has joined #nixos
<simpson>
hoppfull: There are similar techniques using Nix; in particular, direnv and nix-shell can be combined to give an experience that includes auto-reloading of tools.
<simpson>
The concept that must be let go of, I think, is *installation*.
<hoppfull>
simpson: ok
nihh_ has joined #nixos
<Profpatsch>
the Pandoc binary is 130MB; I wonder if we can make a statically compiled stripped-down version that’s fixed to Markdown as input and only a few MB big
zupo has joined #nixos
<hoppfull>
Also, is it correct that nix-shell doesn't have access to other channels than nixos? Like nixpkgs-unstable
<{^_^}>
[nixpkgs] @Infinisil pushed 2 commits to release-20.03: https://git.io/JvReJ
<simpson>
Yeah. Channels are part of the system and user configuration, not part of each individual Nix invocation. They're passed to Nix via $NIX_PATH. One can customize which nixpkgs is used in their shell.nix; lemme find an example.
<tilpner>
That will make hello from nixos-unstable available in a new shell
mexisme has joined #nixos
<tilpner>
-I effectively adds to NIX_PATH
nschoe has quit [Ping timeout: 255 seconds]
<hoppfull>
tilpner: I want to avoid having a long readme explaining how to set up a dev environment. I want to be able to clone repo and open a shell inside it and have everything you need to get going. If we need to upgrade to a newer linter, compiler etc, it should be defined in the repo such that everyone use the same versions.
orivej has joined #nixos
<tilpner>
hoppfull: No problem, just fetch nixos-unstable in your shell.nix
<tilpner>
hoppfull: let unstable = import (builtins.fetchTarball "channel:nixos-unstable") {}; in ...
<simpson>
hoppfull: The way I've set up my repository above is one particular way to do this. When combined with an .envrc (which I don't include for safety reasons) with `use nix` and etc., direnv will incant nix-shell, which will load default.nix, which will build all of the tools and local stuff in the repository, and put it into the environment.
<simpson>
nix-build is used to do builds of the main artifacts.
<tilpner>
hoppfull: Or pin it to a specific version of unstable, which is a better idea
<hoppfull>
interesting
<simpson>
hoppfull: You've mentioned npm, which is a hateful and selfish tool when it comes to package management. I hear that the best strategy there (but not one I've used personally) is to use nix-shell to load npm and all of the build tools, but then to use a local npm sandbox to build your specific project.
mexisme has quit [Ping timeout: 240 seconds]
fresheyeball has joined #nixos
<fresheyeball>
good [local time of day] folks!
<hoppfull>
I mentioned npm but it's not a huge problem. stack is a little bit wierder. It can integrate with nix somehow and haskell-ide-engine seems to have its own instructions on how to work with nix. Then there is postgresql and pgadmin. Right now I start the database with docker which is nice because I can easily clear everything by just restarting the
<hoppfull>
image.
<fresheyeball>
I turned on distributed builds linking my current workstation to the build slave in my basement
Mateon1 has quit [Remote host closed the connection]
Mateon1 has joined #nixos
<fresheyeball>
and everything was working perfectly. Very satisfying to see all my hardware get used at once for something
<fresheyeball>
but now, I have a problem
<fresheyeball>
while distributed builds still work in most cases
kaliumxyz has quit [Ping timeout: 268 seconds]
<fresheyeball>
for some reason one of my nix-builds just hangs
nihh_ has quit [Quit: Leaving]
<fresheyeball>
copying path '/nix/store/b3zxdyhp140gz0xixwnfhcmhsmd8jnmy-hashable-1.2.7.0' to 'ssh://goldwater'...'
<hoppfull>
I'm hoping there is a nugget of information that gives me the ability to always contain a dev environment in a repo without hoping that the specific tool has some integration instructions predefined.
<fresheyeball>
goldwater is the name of the build slave
<fresheyeball>
it just sits there like that
<fresheyeball>
how can I debug this?
<simpson>
hoppfull: Docker should be available, if you need to smooth your transition.
<gchristensen>
fresheyeball: can you SSH to the remote builder?
<fresheyeball>
gchristensen: yes
kaliumxyz has joined #nixos
<fresheyeball>
I can do both pre-reqs in the docs
<evanjs>
clever: ^ any idea how I can accomplish that kernel config import stuffs? I figured extending the first linuxPackages set I made and then overriding the kernel would work but ...
<fresheyeball>
I can ssh to it both as my account and as root
<evanjs>
wait.. is overrideAttrs + // {} all I need?
<fresheyeball>
I can ping the nix store as either account as well
<gchristensen>
fresheyeball: on the remote builder, do you see the connection which nix has made?
<hoppfull>
simpson: Yeah, that was plan B. I wanted to give nix a chance but I fear nix isn't designed to do what I want.
<simpson>
hoppfull: But, like, mostly it's about either having Nix set everything up for you, or having direnv set everything up for you, or something along those lines. The combination is quite potent.
<hoppfull>
simpson: thanks for the help!
<fresheyeball>
gchristensen: what do you mean?
jakobrs has joined #nixos
<evanjs>
like structuredExtraConfig = o.structuredExtraConfig // { more_options }
orivej has quit [Ping timeout: 258 seconds]
<evanjs>
that... seems to work?
<gchristensen>
fresheyeball: well Nix has just run `ssh goldwater nix-store ...` can you see that nix-store?
<simpson>
hoppfull: TBH what you want is probably wrong and horrible. That's how humans are. And, worse, no human really understands computers. So the best that we can do is to try to convince the software that we have that our desires are reasonable.
<fresheyeball>
gchristensen: yes that works
<fresheyeball>
ssh goldwater nix-store --version
<hoppfull>
simpson: I'm going to look up direnv. At first I thought it was just a mechanism to automatically load nix-shell with the local environment but maybe it can do more.
<gchristensen>
fresheyeball: no, do you see the one which Nix is currently running
<hoppfull>
simpson: I'm definitely open to the fact that I'm trying to achieve the wrong thing.
<fresheyeball>
gchristensen: I don't understand what you are asking
<simpson>
hoppfull: It's very possible to do almost all of this with direnv and Nix. Starting Pg, even, could be done with direnv. The main problem is the entire concept of doing Web development in a box, IMO, when the Web is such a non-hermetic and integrated environment.
<clever>
evanjs: probably
<gchristensen>
fresheyeball: the nix daemon is currently running nix-store on goldwater. can you see it in the process list?
shafox has joined #nixos
<fresheyeball>
gchristensen: if I ssh into goldwater
<fresheyeball>
and run systemctl status nix-daemon.service
<fresheyeball>
it shows it's running
<gchristensen>
what is the name machine are you running nix-build on?
<fresheyeball>
dunlap is the name of my workstation where I issued the command
<gchristensen>
are dunlap and goldwater nixos?
<evanjs>
clever: ahh so close. Doesn't look like I can do that for extraConfig... which make sense IIUC
<fresheyeball>
yes both computers are nixos
<gchristensen>
fresheyeball: is dunlap connecting to goldwater over SSH as the root user?
<fresheyeball>
I just ran systemctl restart nix-daemon.service on goldwater, and the problem persists
<evanjs>
the structuredExtraConfig bit seems to work fine, though
<fresheyeball>
gchristensen: I believe so, like I said distributed builds are currently working for most things
<gchristensen>
fresheyeball: you're still misunderstanding my questions, so I'm getting as much information as possible to make my actual question more clear
<fresheyeball>
gchristensen: fair
nschoe has joined #nixos
<gchristensen>
fresheyeball: in your SSH log on goldwater, do you see evidence of root logging in?
<fresheyeball>
that is the bottom of the sshd log on goldwater durring a hung build
cole-h has joined #nixos
<gchristensen>
fresheyeball: okay, so it looks like, then, that nix-daemon on dunlap is connecting to goldwater as the "nix-ssh" user?
<evanjs>
On another, more generic note, is there an easy way to query attributes I would normally be able to override? in this case, structuredExtraConfig, extraConfig, etc
<fresheyeball>
gchristensen: I am not sure about that
<fresheyeball>
the nix-ssh user is being used for substitution
<evanjs>
Can't figure out how I would query them from the repl or etc (if it is possible)
<fresheyeball>
but the build is going over root I think
<phaebz>
When booting from USB I am put into GRUB, manual says it should boot straight to nixos. USB stick dd'ed and UEFI boot.
<gchristensen>
fresheyeball: what does your /etc/nix/machines file contain?
<jakobrs>
phaebz: That sounds like the manual is wrong. I've always booted into GRUB when using USB.
<gchristensen>
okay, then it is probabl yconnecting over root. if you cancel your build on dunlap, run journalctl -fu sshd on goldwater, and start a new build on dunlap do you see nix connecting as root?
<gchristensen>
looks promising, fresheyeball, I think it is connecting as root.
<phaebz>
ok
<gchristensen>
fresheyeball: okay, now I can clarify the earlier question.
ryantrinkle has quit [Ping timeout: 260 seconds]
<gchristensen>
fresheyeball: nix-daemon on dunlap is running `ssh goldwater nix-store ...`. on goldwater, that `nix-store` is being run as the `root` user which causes nix-store to skip the nix-daemon, and run the build right there.
<gchristensen>
fresheyeball: on goldwater, do you see in the process tree the root user having logged in over SSH, running nix-store?
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
phaebz` has joined #nixos
cr4y1 has quit [Ping timeout: 258 seconds]
phaebz has quit [Read error: Connection reset by peer]
<fresheyeball>
I see root with nix-store many many times
<fresheyeball>
hard to tell if a new one popped up
erasmas has joined #nixos
<gchristensen>
what are they all doing?
<fresheyeball>
all different stuff
fendor has quit [Remote host closed the connection]
<fresheyeball>
nix-store --serve --write
<gchristensen>
building?
<fresheyeball>
gchristensen: is there a way to sort htop by most recently spawned?
<gchristensen>
dunno
<cole-h>
Is there a way to use a single NixOS module from master rather than nixos-unstable? Especially one that applies a `.patch`? Is adding `master` as a channel and using `<master/nixos/modules/programs/fish.nix>` all that's needed?
<gchristensen>
does it look like they're building?
<clever>
fresheyeball: does that nix-store have any child proccesses? `ps -eH x`
<clever>
fresheyeball: does its pid appear in the args of any other process?
opthomasprime has joined #nixos
<gchristensen>
fresheyeball: but anyway, it sounds to me like it is getting stuck because your remote is building as many things as it can, and your system is having to wait for build slots
<fresheyeball>
gchristensen: but if I build other things it works fine
<fresheyeball>
can I kill all nix things on goldwater?
<fresheyeball>
see a clean slate?
<gchristensen>
sure
<fresheyeball>
how to do?
<fresheyeball>
reboot?
shafox has quit [Remote host closed the connection]
<gchristensen>
up to you :)
<fresheyeball>
gchristensen: do you have a recommendation other than rebooting?
<gchristensen>
sometimes I'll `pkill nix-store`
<gchristensen>
I have a unit file being provided by «systemd.packages = [ package ];» and I want to disable it. can I just do systemd.services.the-service-name.enable = false; ?
<{^_^}>
[nixpkgs] @cole-h opened pull request #80651 → [20.03] Backport `rls: add llvm to buildInputs` → https://git.io/JvRvZ
<{^_^}>
[nixpkgs] @jonringer pushed commit from @xfix to master « pythonPackages.can: remove pytest-timeout check input »: https://git.io/JvRvC
<fresheyeball>
ok it appears no nix-store processes are running
shafox has joined #nixos
ng0 has joined #nixos
<fresheyeball>
running the build on dunlap spawned one nix-store process on goldwater
<fresheyeball>
nix-store --serve --write
<fresheyeball>
and that's it gchristensen
<fresheyeball>
if I run pkill nix-store on goldwater
<fresheyeball>
the build on dunlap get's unstuck with error: interrupted by the user
<fresheyeball>
gchristensen: I did pkill on both nix-store and nix-build
<fresheyeball>
then ran from dunlap
<fresheyeball>
no nix-build get's spawned
<fresheyeball>
the build appears to hang on dunlap
<infinisil>
cole-h: Yeah mostly, you should also disable the old module though, with `disabledModules = [ "programs/fish.nix" ]`
<fresheyeball>
and we get one nix-store --serve --write process on goldwater
<fresheyeball>
I am rebooting goldwater now
<fresheyeball>
there is a ton of hung nix stuff on here
phaebz`` has joined #nixos
<cole-h>
infinisil++ Thanks, I'll suggest that to them.
<gchristensen>
anything in dmesg?
<{^_^}>
infinisil's karma got increased to 217
ryantrinkle has joined #nixos
<jakobrs>
How do I prevent closing the lid from disconnecting wifi? I tried setting `networking.networkmanager.powersave` to `false` but it didn't work.
phaebz`` has quit [Read error: Connection reset by peer]
<fresheyeball>
gchristensen: it's still rebooting I tink
<grp>
hi. I know I can normally find the pkg a file belongs to by just following the symlinks to the store, but I've found an exception to the rule and I've yet to find a way. The file in question is nixos-rebuild. there's no such package
<cole-h>
infinisil: And just to be super clear: using the module from that specific channel will also use that channel's tree (e.g. the patch used will be from that channel and not from nixos-unstable)?
<{^_^}>
[nixpkgs] @jonringer pushed commit from @filalex77 to release-20.03 « cargo-update: 1.5.2 -> 2.5.0 »: https://git.io/JvRvF
<jakobrs>
grp: realpath $(which nixos-rebuild)
<infinisil>
cole-h: If you mean the pkgs from that channel, the answer is no. Only the module itself will be used from there. To use some pkgs from that channel you need to use overlays or so in addition
<fresheyeball>
gchristensen: rebooting seems to have solved it
<fresheyeball>
good enough for me, but I would rather understand
<gchristensen>
cool
sgrunert has joined #nixos
<fresheyeball>
anything else I should check?
<fresheyeball>
how to check dmesg
<gchristensen>
I would have checked dmesg before rebooting
<cole-h>
infinisil: Alright, thanks. Then I won't suggest that and just let them wait for unstable to be bumped ;)
<grp>
jakobrs: /nix/store/11dm9ivx65mb2adavbsws5qqzh4bpqjl-nixos-rebuild/bin/nixos-rebuild This is indeed what I'd do (and did), but there's no nixpkgs.nixos-rebuild
<gchristensen>
I'm going to leave that up to you as a learning exercise :)
<fresheyeball>
yeah I got your message to late
<fresheyeball>
ok :)
<infinisil>
cole-h: Alright :)
<fresheyeball>
thanks!
<fresheyeball>
gchristensen: while I got you one question (docs are not helping me)
<{^_^}>
[nixpkgs] @vbgl pushed commit from @Ptival to master « coqPackages.coq-ext-lib: 0.10.3 -> 0.11.1 »: https://git.io/JvRfR
c0c0 has quit [Quit: WeeChat 2.6]
phaebz``` has joined #nixos
<aaa_>
ikwildrpepper, when I use this http://dpaste.com/0MY25ZN, and I enter nix-shell, my rustup --version is 1.21.1, but I would like the unstable one.
phaebz`` has quit [Read error: Connection reset by peer]
<aaa_>
cole-h, the problem is that that first entry found in PATH is that symlink to rustup
<cole-h>
Then comment out `rustup` ;-) You shouldn't need it on Nix anyways. It's for downloading and updating Rust versions, which we manage with Nix.
<cole-h>
Unless I'm misunderstanding something (which is entirely possible)
<aaa_>
Ok done that, now the version of rustc is 1.40.0, not 1.41
<aaa_>
I'm using unstable.rustc
<aaa_>
in buildInputs
<cole-h>
Have you updated your channels recently?
<aaa_>
cole-h, thank you, works now :)
<aaa_>
I needed to update the channels.
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<cole-h>
Yep, I don't think they update automatically.
<{^_^}>
[nixpkgs] @Mic92 pushed 2 commits to release-20.03: https://git.io/JvRUm
<{^_^}>
[nixpkgs] @hlolli opened pull request #80655 → release-20.03-cherry-pick: lumo Merge pull request #59414 from hlolli/lumo-bump → https://git.io/JvRUG
Tucky has quit [Ping timeout: 258 seconds]
ardumont has quit [Ping timeout: 272 seconds]
drakonis has quit [Ping timeout: 240 seconds]
stites has joined #nixos
<stites>
hey all! I'm trying to overwrite a systemd ExecStart script -- is this something I can do?
phaebz``` has quit [Read error: Connection reset by peer]
<qyliss>
stites: set systemd.services.<name>.serviceConfig.ExecStart
phaebz``` has joined #nixos
<qyliss>
if it's being set by another module already, and you want to override it, you'll have to use lib.mkForce
casaca has joined #nixos
<stites>
niiice! that's exactly what I needed!
<stites>
thank you!
drakonis1 has joined #nixos
orivej has joined #nixos
ardumont has joined #nixos
<mikky>
so I have this Go package (traefik) which is built using `buildGo112Package` and I need to overlay it (newer version) and build it using `buildGo113Package`. I know how to override the version but I'm stuck with forcing build113
drakonis1 has quit [Ping timeout: 248 seconds]
waleee-cl has quit [Quit: Connection closed for inactivity]
<Philonous>
I have 2 machines, both have the same channels, I ran nix-channel --update on both, but when I run »nix-shell -p keepassxc« one of them fetches the closure from cache, the other tries to build it and fails
<newuser99>
I have `nix-repl> base` -> `{ home = { ... }; manual = { ... }; programs = { ... }; }` How can I expand the "..." to see what's inside?
opthomasprime has left #nixos [#nixos]
ris has joined #nixos
<sehqlr[m]>
> newuser99: I have `nix-repl> base` -> `{ home = { ... }; manual = { ... }; programs = { ... }; }` How can I expand the "..." to see what's inside?
<sehqlr[m]>
try base.home or similar
<{^_^}>
error: syntax error, unexpected $undefined, expecting ')', at (string):281:19
<newuser99>
sehqlr[m]: is there a way to expand all of it?
<Philonous>
Never mind, figured it out. Turns out I didn't have the same channels after all
<sehqlr[m]>
Not that I'm aware, Nix is a lazy language, so not everything gets evaluated
<sehqlr[m]>
you could do something like a map to get all the attrs, but I don't know how to do that off the top of my head
<gustavderdrache>
you can use the :p command
<sehqlr[m]>
ooo, TIL
<newuser99>
nice
<gustavderdrache>
nix-repl> [{ x = 1; }] ==> prints [ { ... } ]
<gustavderdrache>
nix-repl> :p [{ x = 1; }] ==> prints [ { x = 1; } ]
<kiloreux>
It's not getting cached along with other derivatives. Why not?
<clever>
asbachb: and if you try to `ip link set IF up`, what happens?
<jared-w>
kiloreux: do you have a sha256 for it? If you don't have a sha for the link, it'll redownload after nix's internal timer expires (which defaults to an hour)
<clever>
kiloreux: nix isnt aware of things it depends on at IFD time, you have to make a derivation that downloads the url, and nix-build it
<jared-w>
if you do, it can check the sha and see if it needs to download it again (which it will if it doesn't match)
<clever>
kiloreux: this lets me `nix-build release.nix -A ifd-pins` to make a single drv that depends on all of those sources
<clever>
and as long as that result symlink exists, and your using sha256's, nix can reuse the products
<jared-w>
clever: tbh I feel like I've already gotten enough use out of that thing that it'd be worth its weight to be in the lib somewhere
<kiloreux>
I actually already have a sha256 for it. with fetchTarball. And I have it baked into my machine images. But sometimes it will still download stuff from Github.
<clever>
jared-w: the api is somewhat similar to the initrd and system-tarball stuff, so maybe something more like those would be suited for nixpkgs/lib
<kiloreux>
The image I am talking about is an AMI of course.
o1lo01ol1o has quit [Ping timeout: 260 seconds]
<jared-w>
kiloreux: definitely do the ifd pin thing then. I needed that to build some stuff in a ultra sandboxed environment a few days ago
<kiloreux>
I actually use nix-env in my environment. along with buildEnv
HugoDaniel has joined #nixos
asbachb has quit [Ping timeout: 260 seconds]
drakonis has joined #nixos
ATuin has joined #nixos
domogled has joined #nixos
<kiloreux>
ANd this was working until a few days back.
<kiloreux>
I am wandering also what could have broke it.
<peelz>
how can I find a package by its sha256 hash?
chimay has quit [Ping timeout: 240 seconds]
<clever>
peelz: nix needs both a name and a sha256, to compute the $out path
HugoDaniel has quit [Ping timeout: 260 seconds]
<peelz>
ah
<peelz>
well I can't find anything beginning with the sha256 hash in /nix/store
<peelz>
also it looks like the hash in /nix/store is too short to be sha256
<peelz>
or are they truncated?
<simpson>
Zooming out a bit, what are you trying to do and how is it that you only have the hash but not the name?
<peelz>
oh just trying to see if I have the package downloaded already
drakonis_ has quit [Ping timeout: 240 seconds]
<clever>
peelz: the hash in /nix/store/hash-name, is a truncated hash of both the sha256 and the name, and some other strings
<clever>
peelz: the simplest option is to just call pkgs.fetchurl with the name and sha256 again, and look at the path it produced
andreas303 has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @vcunat pushed to master « lkl: fixup build after PR #78994 »: https://git.io/JvRIM
<clever>
kiloreux: lines 2 and 7 are IFD, if you want things cached better, you must move the result of both fetchTarball calls into a let block, and then reference them in the output of something
<clever>
kiloreux: the mkPins function i linked to can generate that output, just pass it the path to both nixpkgs
<AmandaC>
So, https://bugs.freedesktop.org/show_bug.cgi?id=111805https://bugs.freedesktop.org/show_bug.cgi?id=111805 mentions a "drmtip" branch, is that packaged by NixOS, or has it been merged to master, or similar? I'm getting some pretty frequent issues with i915 lately. Currently building against linuxPackages_5_5 but not sure if it'll work, and I'm not very in-touch with the kernel scene
Ariakenom has quit [Ping timeout: 240 seconds]
<samueldr>
we are not packaging drmtip, though it's interesting to know it seems to have a fix for the issue
<samueldr>
hopefully it'll come soon in stable versions
<samueldr>
though, the current recommendation, coming for 20.03, is to use the previous LTS, 4.19 with intel GPUs... though that is not possible for all machines
domogled has quit [Quit: domogled]
<samueldr>
I figure relatively recent hardware could have issues with the older LTS
<AmandaC>
Hrm. I'm on 4.19 at the moment, and stil getting issues
kiloreux_ has joined #nixos
dansho has joined #nixos
<AmandaC>
IT's not a straight-up hang, but eventually anything using opengl will just crash, and refuse to come back up with "i915: input/output error"
<samueldr>
oh
<samueldr>
sorry, I assumed it was the issue that's been common starting ~5.2/5.3
<AmandaC>
It's like a less-volatile version of that
<samueldr>
in that case I don't know
mexisme has joined #nixos
<AmandaC>
I'm not sure if my CPU is dying somehow, or if a lite-version of the hang in 5.3/5.3 ended up in a patch release of the 4.19 line
leah2 has quit [Remote host closed the connection]
drakonis has joined #nixos
o1lo01ol1o has joined #nixos
johnw has quit [Remote host closed the connection]
leah2 has joined #nixos
<gchristensen>
I'm working on a Python application packaged with python3Packages.buildPythonApplication, and during setuptoolsBuildPhase I get ModuleNotFoundError: No module named 'setuptools'. I added setuptools to buildInputs / propagatedBuildInputs but still get this error. this is using a nixos-unstable-small from the last few days
<gchristensen>
does this ring any bells for anyone?
<gchristensen>
interestingly, my application doesn't use setuptools at all.... hrm
tjg1 has joined #nixos
mexisme has quit [Ping timeout: 258 seconds]
Fare has joined #nixos
leah2 has quit [Remote host closed the connection]
leah2 has joined #nixos
drakonis has quit [Ping timeout: 260 seconds]
drakonis has joined #nixos
HugoDaniel has joined #nixos
<das_j>
hmm, is there a way to get a stdenv for aarch64 on my x64 machine?
<das_j>
I need this to build OVMF for an aarch64 VM
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
botrbot has left #nixos [#nixos]
<AWizzArd>
I typed `nix-env -iA nixpkgs.haskell.compiler` and this triggered a download of several things. I aborted this via Ctrl+c. How can I remove the files and directories that have been created so far via this command?
<{^_^}>
[nixpkgs] @worldofpeace pushed commit from @r-ryantm to release-20.03 « ephemeral: 6.2.0 -> 6.2.1 »: https://git.io/JvRq9
<{^_^}>
[nixpkgs] @wucke13 opened pull request #80669 → webdis: init at 0.1.7 → https://git.io/JvRqd
<Arahael>
After a power failure, I noticed that the AC 9560 wifi I have became extremely unreliable... When I investigated, I discovered that I was running a new kernel, which makes sense - the forced reboot resulted in me running an updated kernel. I then updated to the latest nixos (20.03), and the latest kernel but the problem remained...
<Arahael>
I've taken advantage of nixos's generations to select an older build (which is freaking awesome!), but now I'm wondering: How can I run the latest nixos but keep this older kernel - alternatively, how should I investigate this wifi issue?
srid-- has quit [Quit: Connection closed for inactivity]
reallymemorable has quit [Quit: reallymemorable]
<Arahael>
So far it seems to only reliable on 4.19.69, the slightly newer kernel 4.19.80 is unreliable, and the latest 5.x series is similarly unreliable. :(
<amanjeev>
Why do only some of the Jetbrains packages give me "unfree" error. I have the "nixpkgs.config.allowUnfree = true;" set already and clion, goland and webstorm install just fine.
<andi->
arahael: likely reporting it upstream to the stable developers is the best long term fix. Short term you can import the kernel from a previous nixpkgs generation by setting `boot.linuxPackages` to something from a previous channel version
<amanjeev>
but idea-ultimate, pycharm-pro do not
<andi->
amanjeev: because the pro and ultimate are unfree (e.g. require a license)
<Arahael>
andi-: My suspicion or gut feeling is it's a non-free firmware issue, how do I rule that out?
<zeta_0>
zeta waits patiently for a response
<cole-h>
If you have a home-manager question, your best bet is -> #home-manager
<andi->
arahael: you could also rollback to a previous linux-firmware version?
<cole-h>
(It's also generally considered rude to ping people for help; just ask your question and if/when somebody has an answer, they will help you)
maddo has quit [Quit: See ya]
judson has joined #nixos
<zeta_0>
apologies, it is just a quick question
<cole-h>
Aren't they all ;)
<Arahael>
andi-: I'm not really sure how I'd do that - do I use an overlay for that or can I select a particular version in the same way as the kernel?
<qyliss>
arahael: boot.kernelPackages
<Arahael>
qyliss: My assumptioni s that the boot.kernelPackages is "just" the kernel - it's not also including the non-free firmware, isn't it?
<qyliss>
oh, right
<qyliss>
I think it includes more than that
<superbaloo>
how would someone sign data in nix?
orivej has quit [Ping timeout: 255 seconds]
<qyliss>
Because there's a distinction between linux and linuxPackages
<superbaloo>
(signing a kernel for secureboot)
<Arahael>
qyliss: Hmm, quite likely - but I still suspect that the non-free firmware is separate. :)
<cole-h>
Does NixOS support secureboot? I was under the impression it didn't.
<superbaloo>
I'm not sure how nixos has to "support" secureboot
<superbaloo>
it's merely a signature that's checked by the uefi
<Arahael>
andi-: I'll check it out anyway - if it's purely a kernel issue, then just using a previous kernelPackage should do the trick and effectively rules out the firmware as an issue.
<samueldr>
if you want to support it properly, all artifacts that will be touched, so, kernel, initramfs, need to be signed, too, at build time
<superbaloo>
yeah, no
<samueldr>
otherwise you're still left with a wide open gap
<Arahael>
superbaloo: I assume that a chain of trust is involved, too.
<superbaloo>
you can build a single efi with kernel + initramfs