<{^_^}>
dhall-lang/dhall-haskell#902 (by matthew-piziak, 50 weeks ago, closed): Is `dhall-to-nix` suitable for NixOS system administration?
Pwnna has joined #nixos
<selfsymmetric-mu>
^ I wonder if there's a way to get types into the mix for users who want them while still allowing a shared common Nix language and workable NixOS administration.
<selfsymmetric-mu>
Hey, a reboot fixed steam. I have a file that I put previously fixed errors into. This one said `just reboot, usually`. Thanks, file!
ddellacosta has joined #nixos
gustavderdrache has quit [Quit: Leaving.]
vika_nezrimaya has quit [Ping timeout: 256 seconds]
<niso>
zeta_0: you can't (incase you are refering to the generated file) -> it's in the nix store, thus public for everyone
<jlv>
Does anyone know why alsa keeps dropping the volume on my usb dac when it is plugged in, and how I can get it to remember the volume? The weird thing is, sometimes it works and sometimes it doesn't.
<niso>
zeta_0: that's why you shouldn't store secrets in there
<huevo5050>
notgne2 finally i cant get to build trought release.nix but solved the lfs way fetching the 3 lfs files sideways with fetchurl and composing new source dir. Building well now. Thanks a lot for your support!
aleph- has quit [Read error: Connection reset by peer]
aleph- has joined #nixos
scalar has quit [Ping timeout: 240 seconds]
<tA_>
what package would provide libssl for the sec-fhs environments? openssl and libressl didnt seem to
silver_hook has quit [Ping timeout: 250 seconds]
<zeta_0>
niso: in the An easy alternative: MSMTP section(last line), it says that i need to do that http://darcs.net/Using/Send
<zeta_0>
this msmtp has been to much of a hassle, i think i might give up on it
reallymemorable has quit [Quit: reallymemorable]
michalrus has left #nixos ["WeeChat 2.6"]
reallymemorable has joined #nixos
silver_hook has joined #nixos
<tA_>
ok update: i have libssl.so.1.1 and i need libssl.so.1.0.0 in order to run this, is there a way to symlink one to the other or something?
jlv has quit [Remote host closed the connection]
<niso>
zeta_0: did you try it? does it error out? also: the configuration i showed you does not invlove a ~/.msmtprc file (it generates the config in ~/.config/msmtp/conf)
<{^_^}>
rycee/home-manager#371 (by matthew-piziak, 1 year ago, open): mbsync.service not found
<selfsymmetric-mu>
zeta_0: I have `protonmail-bridge` installed and for some reason I have `tmux` installed with the comment `protonmail-bridge daemon workaround support`.
<selfsymmetric-mu>
I wonder what I was doing there…
<selfsymmetric-mu>
Oh, huh. `tmux new-session -d -s mail 'protonmail-bridge --cli'`
<selfsymmetric-mu>
zeta_0: Darcs the version control system? How is that related to mailing?
<zeta_0>
niso: it doesn't recognize the host, so i'm thinking i'm going to have to use my gmail email address instead, and change host to smtp.gmail.com
mexisme has joined #nixos
<niso>
zeta_0: well, selfsymmetric-mu mentioned a bridge before - i assume it's required?
<zeta_0>
i think so, but i'd have to pay for that bridge right selfsymmetric-mu ? anyways, i have a gmail accound so i'll just change it to that
<zeta_0>
or i have to be a subscriber to protonmail in order to use it, right now i'm using the free version
<zeta_0>
niso: i'll make those gmail changes real quick, i'll be back in a bit
peelz has quit [Remote host closed the connection]
<selfsymmetric-mu>
What does it _mean_, "doesn't exist". It's literally in `commonUnitOptions`.
<selfsymmetric-mu>
Those aren't common to all units?
<niso>
zeta_0: is the plain the issue? not the google-settings? or a wrong password? did you look it up?
dansho has quit [Quit: Leaving]
<zeta_0>
niso: i'm pretty sure that's not the problem, but let me double check with my gmails password
<zeta_0>
niso: i think i'm going to call it a day, i'll try again tomorrow, i'm going in circles
<zeta_0>
niso: thank you for the help
<zeta_0>
niso++
<{^_^}>
niso's karma got increased to 7
<niso>
zeta_0: alright, yw, gn8
hoplaahei has joined #nixos
<zeta_0>
niso: i'll contact the darcs people again tomorrow, maybe they'll be able to point me in the right direction
<niso>
zeta_0: you have a problem with sending an email - this isn't darcs related
<niso>
zeta_0: in fact since you configured everything so far in msmtp.extraConfig it's not even nixos related - you could just write the config file on any Linux and copy&paste it into extraConfig
<zeta_0>
niso: the only option i did not set was this: The ~/.msmtprc file needs to have permissions set to 0 for group and others, otherwise msmtp will refuse to send any mail.
<niso>
zeta_0: you got an authentication error - means msmt works fine, and the permission didn't matter
<niso>
zeta_0: you can re-try darcs once you know that it's dependencies (in this case: msmtp) work
<zeta_0>
niso: ok, i understand that, but does that dbus error mean?
pjt_014 has joined #nixos
<niso>
zeta_0: i'm not sure, for now i'd just assume that it will be resolved once you manage to loginto your google account
bridge[evilred] has quit [Remote host closed the connection]
<niso>
zeta_0: you might want to tripple-check your google-settings (or maybe try to use another mailclient to see if msmtp works with them)
bridge[evilred] has joined #nixos
red[evilred] has joined #nixos
felixfoertsch23 has joined #nixos
<hoplaahei>
hi. KDE5 is overriding the keyboard variant set in configuration.nix. I uncheck 'Configure layouts' and click apply, but the checkbox gets ticked again. For now, I can set the variant through kde, but does anyone have a fix for this?
<zeta_0>
niso: i need to replace username@domain.com with my gmail address, correct?
felixfoertsch has quit [Ping timeout: 256 seconds]
<zeta_0>
hoplaahei: i tried a custom keyboard variant with kde and gnome, and it overrided the custom keyboard layout, but i got it to work with xmonad by itself, in home.nix, i set the option: keyboard = null; so it would not get overrided
<zeta_0>
niso: ok i'm troubleshooting now
<zeta_0>
niso: i give up for now, i'll try again tomorrow
<hoplaahei>
zeta_0: thanks for the info
spacefrogg has quit [Quit: Gone.]
aw has quit [Quit: Quitting.]
<zeta_0>
hoplaahei: your welcome, and i can paste bin my custom keyboard configuration, if you want to see how i setup mine
aw has joined #nixos
ddellacosta has quit [Ping timeout: 264 seconds]
<{^_^}>
[nixpkgs] @jtojnar pushed to haskell-updates « haskellPackages.llvm-hs: Use LLVM 9 »: https://git.io/Jvxte
spacefrogg has joined #nixos
<hoplaahei>
zeta_0: yes thanks
reallymemorable has quit [Quit: reallymemorable]
<zeta_0>
home.nix(home-manager): sessionVariables.keyboard = null; and here's my custom keyboard layout setup in configuration.nix: https://dpaste.org/aJRz
<zeta_0>
hoplaahei:
jumper149 has quit [Quit: WeeChat 2.7]
<zeta_0>
hoplaahei: i have it as a separate keymap.nix file, which i import into configuration.nix
<selfsymmetric-mu>
How do I get Nix to do the equivalent of `systemctl --user start plex-mpv-shim`?
vanillaicecream has joined #nixos
wavirc22 has joined #nixos
<vanillaicecream>
Hi all! I'm trying to build a package of mine which uses the npm module `katex`. `katex` isn't found in nodePackages, and I've been having no luck trying to generate a build file for it using node2nix.
<evils>
selfsymmetric-mu: you want `services.plex-mpv-shim.wantedBy = [ "multi-user.target" ];`
<{^_^}>
[mobile-nixos] @samueldr opened pull request #125 → Fix evaluation failure where aarch64-linux outputs are accidentally x86_64-linux → https://git.io/Jvxt2
<evils>
selfsymmetric-mu: unless there's something about `--user start` that's not covered by that
<vanillaicecream>
The end goal is to be able to run the `katex` binary while building another package (a website). I've been using an overlay to define both the website package and the (dysfunctional) katex package.
<selfsymmetric-mu>
evils: Oh I see, I need to _define_ `options.services.plex-mpv-shim.wantedBy` and then _invoke_ `services.plex-mpv-shim.wantedBy` to the same thing?
<selfsymmetric-mu>
How did you know this? I'd have a hard time figuring that out without your help.
<selfsymmetric-mu>
evils++
<{^_^}>
evils's karma got increased to 3
<evils>
selfsymmetric-mu: i looked at existing modules that define services and imitated them
<selfsymmetric-mu>
A true programmer.
<peelz>
Is there a way I could make a module that changes some home-manager config options for all `home-manager.users` entries? I'm using `<home-manager/nixos>`
<evils>
selfsymmetric-mu: lol i missed that you had a `wantedBy` line in your gist, you'll want to move line 9 to the block that contains line 14
<selfsymmetric-mu>
Do I still need `services.plex-mpv-shim = {enable = true; wantedBy = [ "multi-user.target " ];}`?
<evils>
selfsymmetric-mu: no, you'll only need `enable = true;`
<selfsymmetric-mu>
evils: Fabulous. Okay I'll try that now.
<evils>
selfsymmetric-mu: if you wanted that attribute user configurable (kinda seems odd), you'd want to make an option for it, and use that option's value in the service definition, around line 13
<selfsymmetric-mu>
Oh I see. I agree, I don't want that.
<peelz>
rycee: sorry for the ping but I thought you might know the answer to my question :P
<selfsymmetric-mu>
So I just need the `mkEnableOption` and that's it for configurability
<selfsymmetric-mu>
Rebooting, brb.
selfsymmetric-mu has quit [Remote host closed the connection]
<{^_^}>
[mobile-nixos] @samueldr merged pull request #125 → Fix evaluation failure where aarch64-linux outputs are accidentally x86_64-linux → https://git.io/Jvxt2
<{^_^}>
[mobile-nixos] @samueldr pushed 3 commits to master: https://git.io/Jvxqe
anirrudh has joined #nixos
selfsymmetric-mu has joined #nixos
waleee-cl has quit [Quit: Connection closed for inactivity]
<evils>
selfsymmetric-mu: if starting the unit with systemctl works, you shouldn't have to reboot to start it with nixos-rebuild
<selfsymmetric-mu>
evils: Hm. It's not working.
<selfsymmetric-mu>
I have `services.plex-mpv-shim.enable = true;`.
<selfsymmetric-mu>
And `systemd.user.services.plex-mpv-shim.wantedBy = ["multi-user.target "];`
<selfsymmetric-mu>
Hm, whoops, I have an extra bit of whitespace there.
<selfsymmetric-mu>
The service is still loaded but inactive.
<{^_^}>
[nixpkgs] @worldofpeace pushed 3 commits to master: https://git.io/Jvxqs
<blibberblob>
hi, can anyone give me advice on patching an ELF file? I have a pre-built binary in a node package which I need to run. When I try to run it directly, it says "no such file or directory: <path to file>" even though the file exists. When I run `ldd` on it, the output is here: https://gist.github.com/adnelson/b1dfcd93b2e7a3eacdf6cfad202fa8af
<blibberblob>
I'm not really familiar with how dynamic linking works, but from what I gather from ldd, it seems like three out of four dynamic libraries referenced in the ELF are able to find a path, except for `linux-vdso.so.1`. I can't find much mention of that in nixpkgs, so I'm not sure what this is supposed to point to.
<evils>
selfsymmetric-mu: i'm quite unfamiliar with channel pinning, sure you don't mean nixpkgs.unstable? and maybe (though i don't think this is wrong) it shouldn't be lib.mkIf since that block is already `with lib;`; maybe check systemctl status plex-mpv-shim after a nixos-rebuild test?
<selfsymmetric-mu>
evils: For `systemctl status plex-mpv-shim` I get `Unit plex-mpv-shim.service could not be found.` as expected.
<selfsymmetric-mu>
And for `systemctl --user status plex-mpv-shim` I get loaded but inactive.
<selfsymmetric-mu>
So maybe I'm just not enabling it properly?
<selfsymmetric-mu>
`services.plex-mpv-shim`…does that work for user services?
Supersonic112 has joined #nixos
Supersonic has quit [Disconnected by services]
<selfsymmetric-mu>
Or do I need like `services.users.<my_user>.plex-mpv-shim` or something like that?
Supersonic112 is now known as Supersonic
<selfsymmetric-mu>
Right now I just have `services.plex-mpv-shim.enable = true`;
anirrudh has quit [Quit: Leaving.]
<selfsymmetric-mu>
Huh, you're right, it builds with just `mkIf`. Weird that `lib.lib.mkIf` was apparently working. :/
<selfsymmetric-mu>
`nixos-rebuild only runs a daemon-reload for each user with running user services.`
zeta_0 has left #nixos ["rcirc on GNU Emacs 26.3"]
<selfsymmetric-mu>
I don't know what I'm doing wrong…this just points me back to wantedBy.
<evils>
i don't think it was trying `lib.lib`, it was just using `lib.mkIf`, but it finds `mkIf` since attributes from `lib` are in scope
<selfsymmetric-mu>
Ah, okay.
<evils>
i'm not familiar with user services...
<selfsymmetric-mu>
Ah well. I guess I can just drop this into a @reboot cron job and call it a day.
zeta_0 has joined #nixos
<evils>
selfsymmetric-mu: does it have to be a user service?
<zeta_0>
what is the linux command to give full permissions to a file?
<evils>
zeta_0: `man chmod`
<selfsymmetric-mu>
evils: Nope, it doesn't. It can be system-wide, but it needs to launch mpv in such a way that my user can see it.
<zeta_0>
the manually can be very cryptic, i just need a quick command, oh well
<zeta_0>
uhh, lets see
<zeta_0>
i think this is the correct command, right? chmod 777 my_app
<zeta_0>
i didn't think that this msmtp pkg would give me so much trouble, i really need that functionality so darcs can automatically email patches through the terminal
<AmandaC>
That'll let anybody on the system read, write, and execute it, zeta_0
<zeta_0>
AmandaC: An easy alternative: MSMTP, bottom section:
<zeta_0>
The ~/.msmtprc file needs to have permissions set to 0 for group and others, otherwise msmtp will refuse to send any mail. , http://darcs.net/Using/Send
<zeta_0>
AmandaC: i'm just trying to troubleshoot an error i'm getting
<AmandaC>
So you want x00
<zeta_0>
i'm not even sure if that chmod command will help with my issue
<AmandaC>
With x, being the right bit mask got what you need
<selfsymmetric-mu>
I made it a system-wide service but with `User = "matt"` (yours truly) in the `serviceConfig`. Now it's starting with `services.plex-mpv-shim.enable = true`, and it's connecting as intended. Thanks for your help!
kleisli has quit [Remote host closed the connection]
<selfsymmetric-mu>
Yuck, or so I thought.
kleisli has joined #nixos
<selfsymmetric-mu>
It appears that although it's starting and Plex is discovering it, it's not actually launching mpv. Bleh.
<vanillaicecream>
Say I do `services.nginx......root = pkgs.websitePackage`, will that have unintended effects, or will the root be set to the store path of websitePackage?
<pjt_014>
sooo I have a pr that has not yet been merged
<pjt_014>
and I have a bug that's ready to be fixed elsewhere
<pjt_014>
I want to make another nixpkgs fork so that I can make a PR for it, but Doing so just redirects me to the existing fork I already have
<evils>
you want another branch on your fork
<pjt_014>
Hm. Guess that makes sense
<pjt_014>
Btw any idea how long till 20.03? I'm using that as a gauge for when I should bother people about my pr again
<pjt_014>
and I got maybe two more in the works
<pjt_014>
one option I wanna implement that's within my skillset, and one package that's broke that should be removed
<evils>
seems like the few blocking issues for 20.03 are fairly specific, so i think you're safe bothering people, though these things don't go very fast
casaca has quit [Remote host closed the connection]
ehmry_ has joined #nixos
<pjt_014>
If I were to ask for a package to be marked as broken that'd be comparably quick right?
<pjt_014>
I don't know how likely it'll be fixed and there's a similar package that ofers the same functionality anyway sooooo :shrug:
<pjt_014>
If you/anyone else wants to test it, the broke one to maybe remove is greg and the one that'd supersede it is castget
drakonis has quit [Quit: WeeChat 2.7.1]
<pjt_014>
woah
<pjt_014>
other python packages have this exact problem
<{^_^}>
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
morgrimm has quit [Ping timeout: 256 seconds]
<vanillaicecream>
any idea why I can't connect to a nixos machine created by nixos-infect on DigitalOcean? curl etc gives me timeouts.
ehmry is now known as ehmry[m]
ehmry_ is now known as ehmry
ehmry[m] has left #nixos ["User left"]
<steell>
how can i override the package set for a NixOS module listed in my imports in configuration.nix?
<steell>
just for the one module, i know setting nixpkgs.overlays will affect *all* modules
c0c0 has joined #nixos
<hyper_ch>
you're trying to get a different (older) package?
knupfer has quit [Read error: Connection reset by peer]
knupfer has joined #nixos
<srhb>
steell: If the module doesn't expose a package option, this is not trivially possible
<srhb>
steell: Depending on the module it _may_ be possible by overriding just an ExecStart or something in the systemd service, if it's only used there, say.
linarcx has quit [Ping timeout: 265 seconds]
<srhb>
vanillaicecream: Did you open the firewall?
<vanillaicecream>
srhb: I did not; quick hint on how to do that?
<srhb>
oranginal: That will create /etc/inputrc containing "foo"
<notgne2>
on one of my servers, there seems to be an issue with `boot.extraModulePackages = [ config.boot.kernelPackages.wireguard ];` and I can't seem to figure out why
<notgne2>
it works on everything else
<notgne2>
error: The option value `boot.extraModulePackages.[definition 1-entry 1]' in `/home/notgne2/ops/nixfiles/modules/ezwg.nix' is not of type `package'.
<oranginal>
srhb: I'll give it a try, thanks.
<srhb>
oranginal: But again, that's only for stuff that's readline-determined, like say tab completion in bash.
anirrudh has quit [Quit: Leaving.]
<srhb>
oranginal: For disabling all bells, it's your terminal emulator as etu said.
<oranginal>
I think that is okay, then. etu was right. My terminal also has a config for it.
<notgne2>
it looks like on this machine, linuxPackages_latest.wireguard is null
<srhb>
notgne2: Instead of "this machine" we probably need a nixpkgs commit or channel to know what to say about that. :)
<srhb>
notgne2: Unless you suspect architecture issues.
<cole-h>
lovesegfault: I don't know, but maybe it has a flag like there is for Rust: `-C link-arg=-fuse-ld=lld`
<cole-h>
Something to look for
<lovesegfault>
Yeah, there is that; but the clangStdenv doesn't even come with lld
<lovesegfault>
which imo is odd
<{^_^}>
[nixpkgs] @vbgl opened pull request #84691 → ocamlPackages.dns-client: init at 4.4.1 → https://git.io/JvxsJ
<srhb>
peelz: Your configuration is not callPackage-able. There's a function called eval-config.nix that's necessary to do the module system shenanigans
<peelz>
srhb: ah...
<peelz>
srhb: how can I call eval-config?
<cole-h>
lovesegfault: Hope you find a solution, I'm heading out
<srhb>
peelz: you may also be able to use nixos-option home-manager.users.peelz
revtintin has quit [Ping timeout: 240 seconds]
cole-h has quit [Quit: Goodbye]
<evertedsphere>
okay i went back to the default kernel package and it still won't work
<evertedsphere>
is there a --jobs 1 for nixops
gulplante has joined #nixos
gulplante has quit [Client Quit]
gulplante has joined #nixos
cjpbirkbeck has quit [Quit: cjpbirkbeck]
<peelz>
srhb: I was trying to debug why my expression wasn't working as expected
<peelz>
srhb: turns out I had a typo somewhere... idk; rewrote it and it works now lol
<peelz>
srhb++ thanks :D
<{^_^}>
srhb's karma got increased to 98
<srhb>
peelz: eval of nixos works something like this: nix eval -f '<nixpkgs/nixos>' config.environment.etc.bashrc.text
<srhb>
Just as an example.
<peelz>
ah
mexisme has joined #nixos
<srhb>
(the nixos default.nix will do the eval-machine stuff on your current configuration.nix
revtintin has joined #nixos
<peelz>
ahhh that makes more sense
<srhb>
You can look at it if you want to dive deeper into how eval-config.nix is used
<peelz>
insightful
lsix1 has quit [Ping timeout: 260 seconds]
<evertedsphere>
why is my computer compiling what appears to be glibc, does nixos-unstable not have binary caches ready yet
<srhb>
evertedsphere: Yes, if the channel is bumped, it means the evaluation is complete. Everything that will build has been built (barring transient failures that are not blocking the channel)
<srhb>
Which means: The only things that paths that can be added are either manual reruns or identical paths that suddenly succeed in some other evaluation.
<srhb>
So to a first approximation, no more paths will be added to the cache once the channel is bumped.
morgrimm has joined #nixos
<peelz>
srhb: do you know if there's a way to get home-manager to to find shell completion files from packages installed in `home.packages`? This is my current solution (host-wide): https://termbin.com/0dkj ... thought it won't work if I deploy my home-manager config on a non-NixOS machine
evertedsphere has quit [Remote host closed the connection]
<srhb>
peelz: It should work out of the box iirc
<peelz>
srhb: huh? well it's not finding them unless I have zsh enabled globally
<srhb>
What you're doing is unnecessary. Something else is up if it's not working.
<srhb>
Or at least, I'm pretty sure that's the case.
<peelz>
srhb: the zsh "programs" module relies on `environment.pathsToLink = [ "/share/zsh" ];`, which makes shell completion scripts available in `/run/current-system/sw/share/zsh/site-functions`
FRidh2 has joined #nixos
rauno has joined #nixos
FRidh has quit [Ping timeout: 258 seconds]
<srhb>
peelz: And I believe the same should be true for ~/.nix-profile
<mlatus[m]>
what does the experimental-feature ca-references do?
cfricke has joined #nixos
<peelz>
srhb: oh right... uhh
morgrimm has quit [Ping timeout: 265 seconds]
phreedom has quit [Ping timeout: 240 seconds]
<peelz>
srhb: I guess I have to rely on the host to make system autocompletion things available (like systemd and stuff)... hmm
logzet has joined #nixos
<srhb>
peelz: I'm pretty sure I have completions come in from both /run/current-system and ~/.nix-profile without doing anything really special.
peelz has quit [Remote host closed the connection]
<sevanspowell>
I'm wanting to do something maybe a little unusual: get the packages/haskellPackages specified in an overlay.
<sevanspowell>
I've thought about creating an empty package set/haskellPackage set and applying the overlay to it.
<sevanspowell>
This was easy enough to achieve for the package set, but more difficult for the hackage package set, as I use the "override" function in the specification of my overlay.
<sevanspowell>
Does anyone have any ideas for getting the attribute names out of a haskellPackages overlay?
<srhb>
sevanspowell: Out of an overlay? Aside from the overlay collecting them itself via naming the attrset that's eventually merged in, I don't think there's a way.
<sevanspowell>
This is so I can build just the projects specified in the overlay.
revtintin has joined #nixos
<srhb>
sevanspowell: (as in: let packages = ...; in packages // { specialOverlayPackages = packages }
fusion809 has joined #nixos
<sevanspowell>
Thanks srhb, I think I'll just do that.
mexisme has quit [Ping timeout: 246 seconds]
max34 has joined #nixos
iqubic` has joined #nixos
iqubic` has quit [Remote host closed the connection]
<srhb>
sevanspowell: It's a bit simple/stupid, but I think it's the cleanest way to do it from within the overlay like that. :)
<Profpatsch>
selfsymmetric-mu: use the overlay.nix
LysergicDreams has quit [Ping timeout: 250 seconds]
<colemickens>
worldofpeace: were you actually able to test the webrtc change btw? It builds but I don't have gnome around, nor am I familar with xdg portal stuffs.
spinus has quit [Ping timeout: 256 seconds]
<worldofpeace>
what is alt worldofpeace_ doing here?
<worldofpeace>
like hun, you can't sis with us
<worldofpeace>
it's not you're moment
<{^_^}>
[nixpkgs] @euank opened pull request #84693 → k3s: fix build for aarch64 → https://git.io/JvxGQ
<srhb>
worldofpeace: Harsh!
<worldofpeace>
Cole Mickens: care to show pr link?
<worldofpeace>
(I look at soo many, it's hard)
<worldofpeace>
srhb: it was supposed to be my moment :D
<worldofpeace>
but really though, I'm not even running that irc client
<{^_^}>
#84233 (by colemickens, 4 days ago, open): firefox: add pipewire patch for wayland users
<colemickens>
IT seems like the xdg-desktop-portal-wlr PR has some usage instructions, and I pkg it already in my overlay, so maybe I can give it a shot with sway.
<worldofpeace>
does anyone know how I can check who is identified as who and such? I need to find out whyworldofpeace_ is here
<worldofpeace>
Cole Mickens: I tried building it and it actually failed weirdly
<pjt_014>
slightly ugly but the output is pretty and that's what matters :))
<Emantor>
Lots of no such file or directory D:
<leonardp>
pjt_014: i get a lot of "no such file or directory"
<pjt_014>
is realpath telling you that?
<Emantor>
indeed.
est31 has quit [Remote host closed the connection]
<pjt_014>
does it do the same thing if you replace it with readlink?
est31 has joined #nixos
<pjt_014>
readlink seems to not care about errors as much
<andydarcyjewell>
Hi, I'm reading throught the nix pills, but can't seem to run the repl. It's a fresh install, but ~/.nix-profile/etc/profile.d/nix.sh won't run, and obviously I need to use this to progress through the "Pills".
rauno has quit [Remote host closed the connection]
drvirgilio has quit [Remote host closed the connection]
<xfix>
interesting, the newest April Fools xkcd comic used Nix for deployment purposes (using dockerTools to create a Docker image)
doublej472 has quit [Ping timeout: 246 seconds]
doublej472 has joined #nixos
drvirgilio has joined #nixos
o1lo01ol1o has joined #nixos
arjen-jonathan has joined #nixos
magnetophon has joined #nixos
max34 has quit [Remote host closed the connection]
o1lo01ol1o has quit [Ping timeout: 240 seconds]
<Valodim[m]>
nice :)
__monty__ has joined #nixos
<asheshambasta>
https://github.com/lspitzner/brittany/blob/master/release.nix - if I want to override compiler and I use import "${src}/release.nix" { compiler = "ghc865" }; and nix reports the attribute "ghc865" is missing what am I doing wrong?
<typetetris>
Where can I look, wether a package was already built by hydra on release-20.03?
<asheshambasta>
and what does the attribute-missing error mean?
<xfix>
but I guess that's mostly due to automatically generated `node-packages.nix`
o1lo01ol1o has joined #nixos
<asheshambasta>
from reading the release.nix derivation, I can see that it is an anon. function that accepts a compiler attribute as an optional string which has a default value of "ghc822"
<thibm>
I noticed something strange: command-not-found.pl hardcodes the channel as `nixos' in the output messages. The database is configurable and defaults to root's nixos channel database
<thibm>
But: I've found a program (lutris) which is in the database, but not available in this channel
<wedens[m]>
I'm using lorri and I'm trying to define a shell function in shellHook in shell.nix. I've added `eval "$shellHook"` to .envrc but function is not defined. If I put `echo` in the hook, it prints
fendor has joined #nixos
<wedens[m]>
it probably uses different shell after .envrc execution without propagating definitions. is there a workaround?
<betaboon>
manveru: ouh thanks. gonna look at that. looked at https://github.com/cloudwatt/nix-container-images and arion and some of the other issues already. the story doesnt seem quite clear atm
gulplante has joined #nixos
pbogdan has quit [Remote host closed the connection]
jjakob has joined #nixos
<betaboon>
manveru: looks interesting but doesnt reuse nixos-service-modules :(
<roberth>
betaboon: I think `arion build` does what you want
pbogdan has joined #nixos
<roberth>
you might want to inspect the generated compose file too, to make sure you have all the settings, `arion cat` or `arion config`
<betaboon>
roberth: my goal is to build docker images that use the nixos-services and use those images with terraform. does that sound possible with what you're suggesting ?
pjt_014 has quit [Ping timeout: 256 seconds]
<roberth>
betaboon: possibly. It depends on which privileges you have on the docker host
<{^_^}>
[nixpkgs] @Infinisil merged pull request #84129 → Support removing python from zfs/grub closure → https://git.io/JvF0K
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<roberth>
integrating that with arion would be really nice for your local testing
never_released has quit [Quit: ZNC 1.7.5 - https://znc.in]
<betaboon>
roberth: yeah thats what i have been looking at nix-container-images. seems most promising atm.
cybrian has quit [Ping timeout: 256 seconds]
never_released has joined #nixos
<betaboon>
but if i can use that for building the docker images i dont see the benefit of using arion in my situation, as i am using terraform to create the containers
<roberth>
betaboon: you get quicker builds during image development when you set `useHostStore = true`
<niso>
assuming i'm deploying 2 nodes (nixops). one of them is a DNS server for the other. How would i wait on the second node till the first ones DNS server is ready?
<niso>
by writing a systemd service?
<roberth>
betaboon: and you don't have to fork arion until you want to upstream it; just do `imports = [ ./nixos-s6.nix]` or similar at the service level
chagra_ has quit [Quit: WeeChat 2.7.1]
<betaboon>
roberth: thanks for the pointers. i will look into it :)
morgrimm has quit [Ping timeout: 256 seconds]
<manveru>
niso: make sure it retries until the dns is up? also helps when you reboot that server or it fails for some other reason :)
<ZoomZoomZoom>
Hi. Anyone using Plasma here? Can't access smb:// from Dolphin. Installed dolphin-plugins and kdeApplications.kio-extras but it didn't help. "Unable to create io-slave. klauncher said: Unknown protocol 'smb'""
o1lo01ol1o has joined #nixos
<hyper_ch>
why not mount it locally?
<ZoomZoomZoom>
hyper_ch, I could mount it with systemd of course, but the problem at hand is Dolphin not cooperating. It has the ability, it should work.
<hyper_ch>
mount it with systemd so that it only mounts when you access the folder
<hyper_ch>
ZoomZoomZoom: well, that example mounts the smb shares when you enter the folder... so pretty convenient :) but I can't think of "non-permanent" samba shares
<{^_^}>
[nixpkgs] @nh2 merged pull request #84382 → openblas: Add `singleThreaded` option. Off by default. → https://git.io/JvNEg
mlatus has quit [Remote host closed the connection]
reallymemorable has quit [Quit: reallymemorable]
reallymemorable has joined #nixos
<{^_^}>
[nixpkgs] @filalex77 opened pull request #84715 → monolith: init at 2.2.2 → https://git.io/JvxEW
<ZoomZoomZoom>
hyper_ch, You may be right, but let's say I'm at the stage 2 already.
<hyper_ch>
well, good luck :)
knupfer has quit [Quit: knupfer]
knupfer has joined #nixos
<hyper_ch>
oh... also I need to check what smb changes were in the last 2 months or so on nixos... upgrading to latest unstable causes issues at the other office with windows clients
<asheshambasta>
if I understand correctly, release.nix is a function that accepts 2 params, a pkgs set, and a compiler text. It has defaults in either case. I'd like to be able to supply it with a pkgs using channel 18.09 but with a new key added under haskell.packages.
<ZoomZoomZoom>
There's a bunch of stuff with '? false' there, I'm not sure how to set them to true. They are not in nixos options list
kleisli has quit [Remote host closed the connection]
kiloreux has quit [Remote host closed the connection]
<sphalerite>
ZoomZoomZoom: the override function allows replacing arguments, so you could e.g. use `pkgs.qemu.override { smbdSupport = true; }` (if you have `pkgs` in your scope)
<ikwildrpepper>
that's where the error comes from, haskell is not expected
<ZoomZoomZoom>
sphalerite, Ah, thanks. Will try
romildo has joined #nixos
<asheshambasta>
ikwildrepepper: that is still not clear to me. My intention there is to add the ghc865 key to the imported pkgs under its haskell.packages set.
<bqv[m]>
from what i've seen of nix, camelcase is preferred in almost every scenario
<bqv[m]>
besides store path basenames, which are dash delimited
<ikwildrpepper>
asheshambasta: I don't know much about haskell in nixpkgs, so don't know how best to add ghc865. It doesn't seem to be in the nixpkgs that the brittany project uses in nixpkgs
<ikwildrpepper>
is it in a newer nixpkgs? if so, you could use a different nixpkgs and pass that into the release.nix function
<{^_^}>
[nixpkgs] @NeQuissimus pushed 2 commits to release-20.03: https://git.io/JvxEN
<{^_^}>
[nixpkgs] @peterhoeg opened pull request #84716 → g15daemon: init at 1.9.5.3 → https://git.io/JvxEx
<asheshambasta>
ikwildrpepper: yes, I've tried passing a newer nixpkgs and while that works, the build still fails since some tests fail due to newer dependencies in the newer nixpkgs.
cr4y1_ has quit [Ping timeout: 258 seconds]
<asheshambasta>
the idea here was to just use the nixpkgs used by brittany but with a newer ghc version being available to it.
<{^_^}>
[nixpkgs] @bennofs opened pull request #84717 → nixos/release-combined.nix: test hibernate only on x86_64 → https://git.io/Jvxue
<ikwildrpepper>
asheshambasta: again, no haskell expert here, but wouldn't that just give you the same dependency issue?
<ikwildrpepper>
(most likely)
<asheshambasta>
not sure, my understanding is that only the compiler version is newer and the rest of the dependencies are the same
cosimone has quit [Remote host closed the connection]
cosimone has joined #nixos
gustavderdrache has joined #nixos
<niso>
bqv[m]: yw, note: writing tests for that is more work than expected (due to things like node-startup-order)
<{^_^}>
[patchelf] @pespin opened pull request #198 → Update doc to describe multiple filename args allowed → https://git.io/JvxuK
<asymmetric>
are channels configured in configuration.nix?
<gchristensen>
no
<mallox>
hi, how can I add a custom file to a package using override? thanks!
<mallox>
(i'm thinking about distribution/policies.js to deploy a custumized firefox)
<spinus>
asymmetric: no, but you can bind configuration.nix to specific nixpkgs version
<mallox>
asymmetric: have a look at the option system.stateVersion = "..."
<gchristensen>
mallox: that doesn't do it either
<gchristensen>
,stateversion
<{^_^}>
Changing stateVersion doesn't upgrade anything and can only break your setup at best. To actually upgrade NixOS see https://nixos.org/nixos/manual/#sec-upgrading. If you need to change stateVersion for some odd reason regardless, Ctrl-F for "stateVersion" in https://nixos.org/nixos/manual/release-notes.html to see things that need to be manually migrated with the new value
romildo has quit [Quit: Leaving]
never_released has quit [Quit: ZNC 1.7.5 - https://znc.in]
never_released has joined #nixos
o1lo01ol1o has joined #nixos
<mallox>
gchristensen: you're right, thanks for the clarification
<mojjo>
I've built this '*2nix' tool for purescript: https://github.com/thought2/spago2nix-ree My main motivation was that dependency resolution is done by nix - so that only actual project sources have to be rebuilt if something changes. Once one of its dependencies is built it's in the nix store and you never have to build it again. It works! But there's
<mojjo>
one huge problem:The actual evaluation phase is very long. For a project with about 10 dependencies one rebuild (without any code change) already takes one minute. No good :)It's really just evaluation, the dependencies don't get rebuilt or so. Tbh, I thought that this would not happen because of the lazy evaluation and 'caching' of derivations in
<ottidmes>
mojjo: a lot of caching in Nix is based on files, importing the same file will make use of the cache of subsequent imports, you might want to implement your own caching for the packages if there is a lot of shared dependencies, another thing is that it will still have to produce the derivation to see if anything changed and needs to be rebuilt, so if producing the derivation is costly, it will still
<ottidmes>
take time, regardless of its outputs already being in the nix store
vykook has joined #nixos
anirrudh has quit [Ping timeout: 250 seconds]
jco has joined #nixos
<ZoomZoomZoom>
sphalerite, So I've managed to pass the override with nixpkgs.overlays, and the system rebuilt qemu, it launches with the appropriate smb options, but libvirt still gives me the same error. Does it use another binary for qemu? How can I check it?
romildo has joined #nixos
<ZoomZoomZoom>
To clarify, launching qemu-kvm with '-net user,smb=/share' works, but the same option added to libvirt xml results in error "could not find /usr/sbin/smbd", which was specifically overriden for qemu..
<ij>
is it possible to set up network managre connections without making NM connect? I want to set up a bunch of them in a script, so it can connect to any if I'm around one.
<ij>
I'm not sure which channel I could ask
reallymemorable has quit [Quit: reallymemorable]
kleisli has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @Mic92 pushed to acpi-call « linuxPackages.acpi-call: switch to nix-community fork »: https://git.io/JvxzQ
<{^_^}>
[nixpkgs] @grahamc merged pull request #84717 → nixos/release-combined.nix: test hibernate only on x86_64 → https://git.io/Jvxue
<{^_^}>
[nixpkgs] @grahamc pushed 2 commits to release-20.03: https://git.io/Jvxzx
<srk>
ij: all the configs are just text files so yes
vykook has quit [Ping timeout: 256 seconds]
jco has quit [Quit: WeeChat 2.7]
kleisli has quit [Remote host closed the connection]
domogled has joined #nixos
kleisli has joined #nixos
<ZoomZoomZoom>
I'm not really sure what's libvirt using, but it says "do not use "''${qemu_kvm}/bin/qemu-kvm""
<jonge[m]>
hey there, i use `pkgs.recurseIntoAttrs` on my attrsets to make `nix-instantiate` and `nix-build` recurse into them. but now i realized that hydra doesn't like that. hydra emits evaluation errors like `in job ‘initrds.recurseForDerivations’: unsupported value: true`. is that something that i need to filter out before letting hydra look at it?
vykook has joined #nixos
gulplante has quit [Quit: WeeChat 2.7.1]
<{^_^}>
[nixpkgs] @jtojnar merged pull request #84458 → librsvg: fix build on darwin → https://git.io/JvNdM
<{^_^}>
[nixpkgs] @jtojnar pushed commit from @marsam to master « librsvg: fix build on darwin »: https://git.io/JvxgI
<ZoomZoomZoom>
sphalerite, Huh, I've checked my config and libvirt.qemuPackage was set to kvm, not qemu_kvm.
mexisme has joined #nixos
<ZoomZoomZoom>
thanks a lot, now it launches without errors
kleisli has quit [Remote host closed the connection]
kleisli has joined #nixos
<srk>
ij: there's #nm channel
<ij>
I was thinking it was probably going to be empty, because gnome has their own IRC servers... boy, I was wrong
<{^_^}>
[nixpkgs] @Mic92 pushed to acpi-call « linuxPackages.acpi-call: switch to nix-community fork »: https://git.io/Jvxgx
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
mexisme has quit [Ping timeout: 265 seconds]
vykook has quit [Ping timeout: 256 seconds]
mananamenos has joined #nixos
kleisli has quit [Remote host closed the connection]
kleisli has joined #nixos
vykook has joined #nixos
cosimone has quit [Quit: Terminated!]
LysergicDreams has quit [Ping timeout: 256 seconds]
<{^_^}>
[nixpkgs] @talyz opened pull request #84721 → php.buildEnv: Don't inherit dev from the original php → https://git.io/Jvx23
LysergicDreams has joined #nixos
gulplante has quit [Quit: WeeChat 2.7.1]
kleisli has quit [Remote host closed the connection]
kleisli has joined #nixos
<timokau[m]>
What are the differences between building aarch64 packages with a remote builder and `--option system aarch64-linux` and building them on the builder directly?
<timokau[m]>
Apparently there are differences, because for the remote case I'm getting "exec format error" while the non-remote case works
anirrudh has joined #nixos
kleisli has quit [Remote host closed the connection]
vykook has quit [Ping timeout: 258 seconds]
<timokau[m]>
samueldr: I'm assuming you probably know a lot about this ^
kleisli has joined #nixos
<dhess>
gchristensen: ping
zupo has joined #nixos
<clever>
timokau[m]: id recomend against using `--option system`, that will break a lot of things
morgrimm has quit [Ping timeout: 240 seconds]
<clever>
timokau[m]: if you use `--option system`, then your lying to nix that the LOCAL machine is aarch64, which causes builtins.currentSystem to repeat that lie, and nix will try to run aarch64 binaries (causing your issue)
vykook has joined #nixos
<clever>
timokau[m]: if you use `--argstr system aarch64-linux`, then your telling nixpkgs to do an aarch64 build, and leaving the local arch intact
<benny>
so I don't know how to continue here, because they build fine, if you have the necessary resources
<benny>
that information is from https://github.com/NixOS/nixpkgs/issues/83080 where it might be a good idea to offer graalvm8 as a graalvm8-bin, which is how it used to be before a big refactor in January 2020
<contrun[m]>
srk: nixos-20.09pre220217.39247f8d04c unstable. I changed my kernelPackages config, it still does not work. Also in `nix repl '<nixpkgs/nixos>'`, we I type system, the same error echoed.
<adrielus>
so, im trying to include home-manager in configuration.nixand im doing literaly the same thing as a few configs i found on github
anirrudh has quit [Quit: Leaving.]
cjpbirkbeck has joined #nixos
<adrielus>
(sorry for weird formating, idk why it didnt pick up my newline)
<evanjs>
gah, does _anyone_ currently have `gh` working on nixos/nixpkgs?
<evanjs>
I know there's nixpkgs-review, that's what I'm currently using, but gh just seems to not work for pr checkouts period on nixpkgs, and it prints garbage after its output
<adrielus>
and try importing it like that, but I get this error:
<adrielus>
```
<adrielus>
```building Nix...building the system configuration...error: attribute 'submoduleWith' missing, at /nix/store/pl11m7hr5d6qhfgkrcama94w9n5r1i7k-home-manager-src/nixos/default.nix:11:14(use '--show-trace' to show detailed location information)```
vykook has quit [Ping timeout: 264 seconds]
<zeta_0>
yesterday i configured smsmpt in home.nix but when i try to test it out it's throwing errors, am i missing something? https://dpaste.org/iSeU
<adrielus>
anyone has any idea what's causing that error?
<srk>
contrun[m]: quite recent, I would try bisecting if it worked before
erasmas has joined #nixos
<srk>
adrielus: looks like incompatible home-manager vs nixpkgs
anirrudh has joined #nixos
vykook has joined #nixos
<selfsymmetric-mu>
zeta_0: Is `/var/mail/zeta` your intended maildir, and if so has it been initialized?
<niso>
zeta_0: yesterday we concluded that you can't use protonmale with smtp because you don't wait to pay for the bridge, then switched to gmail, where you got an authentication error...
<niso>
*protonmail
<srhb>
Anyone else experience steam failing to launch (nothing happens...) since the last 20.03 bump?
<adrielus>
@srk and how can I fix it?
mallox has quit [Ping timeout: 256 seconds]
<srk>
adrielus: well by either using more recent nixpkgs or downgrading home-manager
<srk>
adrielus: iirc home-manager has some compatibility notes in README
<adrielus>
ok, im very new to nixos (just installed it yesterday), whats the best way to switch to the latest nixpkgs?
<srk>
adrielus: you can easily switch to unstable which is a tested channel
<srhb>
adrielus: for what it's worth, I believe rycee maintains some release branches of home-manager that matches stable NixOS releases
<srhb>
adrielus: So there's a release-19.09 one that's intended to be used with NixOS 19.09
<kraem>
I'm trying to do `gopass show -c` in a script which is invoked by a pop up terminal started by i3. It doesn't have access to the clipboard it seems though. Rofi, and all other processes spawned by i3 have access to the clipboard though. Any pointers?
<srhb>
adrielus: (that is, definitely not master)
<srhb>
adrielus: Which may be what niv is giving you, unless you specifically requested something else.
hyper_ch has joined #nixos
mananamenos has quit [Remote host closed the connection]
<srhb>
adrielus: So maybe, if you're on NixOS 19.09, you just need to point niv at the right branch/ref and update, so that NixOS matches home-manager? :)
<{^_^}>
[nixpkgs] @Mic92 merged pull request #83858 → perlPackages.starman: add shortenPerlShebang on darwin → https://git.io/Jv5cm
<zeta_0>
selfsymmetric-mu niso : the people in the #darcs channel had me run these commands to try and troubleshoot msmtp: https://dpaste.org/V9tn
<srhb>
zeta_0: sendmail-esque setups usually need some sort of NixOS setup, because they all depend on root in intricate ways
<srhb>
(ie. they try to deliver to some mail spool in /var/something and expect a daemon to pick it up from there)
<jakobrs>
Does nixos-unstable-small wait for initrd to be built before advancing?
<srhb>
There are rootless senders, but I don't remember which
<visl>
I just installed nixos-20.03 from the iso yesterday and i have this config (plus more or less the default config from the install) to install sway: https://pastebin.com/U0nTu7k7 when i run sway -V --debug it hangs at Preparing Wayland server initialization before shutting my computer down. 1. does it shut my computer down? 2. how do i fix it?
<raboof>
I have a nixpkgs branch in which I updated a package to the latest upstream snapshot. This was not trivial (the upstream build system changed), and not ready to be merged (there are known problems with the snapshot). Would it be helpful to submit this as a 'draft pr' so when the project releases later this can be found and used as a starting point, or would that be considered noise?
<visl>
1. Why* does it shut my computer down
<srhb>
jakobrs: If it's any of these, or a transient dependency of any of these.
<zeta_0>
srhb: i'll i have is smsmtp configured in home.nix, is there some other functionality that i need to setup? https://dpaste.org/9wDH
non_freethink has joined #nixos
<srhb>
zeta_0: And the mail binary you're seeing is from that?
fendor_ has quit [Read error: Connection reset by peer]
<niso>
zeta_0: thus it's actually resulting in a worse error message?
kleisli has joined #nixos
<jakobrs>
srhb: My kernel setup completely trivial, except for the fact that I've got linuxPackages_latest from nixos-unstable-small
reallymemorable has quit [Quit: reallymemorable]
<zeta_0>
niso: i copied my gmail password directly from firefox's emails and preferences section, so i know that the password is correct, so i'm not sure what else it could be, and ya, that message isn't very clear on why it cannot send mail
<srhb>
jakobrs: Then you better share the initrd drv for analysis. Bearing in mind that _trivial_ isn't relevant, only _different_. :)
<srhb>
Along with the nixpkgs rev you're expecting it to be cached for.
<niso>
zeta_0: did you try to connect to gmail with another client (as recommended yesterday?)
<niso>
zeta_0: since another user mentioned that there might be some gmail flag
<{^_^}>
[nixpkgs] @primeos pushed 5 commits to release-19.09: https://git.io/JvxrD
jakobrs has quit [Quit: WeeChat 2.7.1]
knupfer has quit [Remote host closed the connection]
knupfer has joined #nixos
<non_freethink>
Random thought from Nix n00b - consider removing the "More .." buttons that all link to the manual. (https://nixos.org/learn.html) It makes things cluttered and confusing
turion has quit [Ping timeout: 246 seconds]
<selfsymmetric-mu>
I just want to extend my gpg timeout so I don't keep having to unlock my email daemon's key so often…
kleisli has quit [Remote host closed the connection]
LysergicDreams has quit [Ping timeout: 260 seconds]
kleisli has joined #nixos
LysergicDreams has joined #nixos
<zeta_0>
niso: yes, i was able to sign into my gmail and access my email through thunderbird, so now we know that my gmail's username and password is correct
<niso>
bqv[m]: because legacy (fun stuff like e.g. order-matters) -> which is why nftables werde merged
<bqv[m]>
and yet, because i'm using nixos's networking.firewall, i can't use nftables :)
<bqv[m]>
fun times
mananamenos has quit [Remote host closed the connection]
OrangeTetra has quit [Remote host closed the connection]
jbrock has joined #nixos
OrangeTetra has joined #nixos
vykook has joined #nixos
OrangeTetra has quit [Remote host closed the connection]
rogue_koder_ has quit [Ping timeout: 256 seconds]
rogue_koder has joined #nixos
OrangeTetra has joined #nixos
OrangeTetra has quit [Remote host closed the connection]
OrangeTetra has joined #nixos
OrangeTetra has quit [Max SendQ exceeded]
turion has joined #nixos
turion has quit [Client Quit]
OrangeTetra has joined #nixos
adrielus has quit [Remote host closed the connection]
OrangeTetra has quit [Max SendQ exceeded]
philr has quit [Ping timeout: 256 seconds]
OrangeTetra has joined #nixos
<dsx>
iptables was wizardry for me for a long time too. But after few years I finally read man. It is not wizardry anymore, just shitty ui for a firewall. Nonetheless, it does what I want it to do, so no hard feelings.
rogue_koder has quit [Ping timeout: 265 seconds]
<theduke[m]>
anyone know at which version Grub got luks2 support?
OrangeTetra has quit [Remote host closed the connection]
vykook has quit [Ping timeout: 250 seconds]
<{^_^}>
[nixpkgs] @teto opened pull request #84736 → broadcom_sta: fix build on 5.6 → https://git.io/JvxKd
<zeta_0>
what is a good utility to cut and past a part of the screen as a picture, and can pictures be paste binned?
rogue_koder has joined #nixos
OrangeTetra has quit [Remote host closed the connection]
<bqv[m]>
internet --- zeta --- delta . Zeta has two IPs, both publicly reachable, and is wireguard connected to delta (10.0.0.3). I'd like the internet to be able to use zeta's second IP to talk to delta, but also to be able to talk directly to delta at 10.0.0.3
OrangeTetra has joined #nixos
<dsx>
zeta_0: scrot can screenshot selection. As for upload, there were some scripts built around it AFAIR.
romildo has quit [Quit: Leaving]
<evertedsphere>
anyone here do rust gui stuff? i'm having some trouble with wgpu-native, which is a dependency of the gui library i'm using
<manveru>
zeta_0: fireshot
<evertedsphere>
(and i think it's nixos-related)
<manveru>
zeta_0: i mean flameshot
OrangeTetra has quit [Remote host closed the connection]
<dsx>
bqv[m]: DNAT for internet ←→ zeta ←→ delta datapath. zeta ←→ delta traffic shouldn't be affected
non_freethink is now known as freethink
OrangeTetra has joined #nixos
<cole-h>
evertedsphere: What's the issue? Not an expert, but maybe if you just ask, somebody can help.
<bqv[m]>
just DNAT?
vykook has joined #nixos
<bqv[m]>
i thought I'd need to forward packets back too
<dsx>
bqv[m]: I guess you're looking for last example
oranginal has joined #nixos
<zeta_0>
thanks, that flameshot looks tool seems to be just what i need for my use case
<bqv[m]>
dsx: something like that, yeah, but i've been fiddling with similar for hours and getting nowhere...
<{^_^}>
[nixpkgs] @disassembler pushed 3 commits to release-20.03: https://git.io/Jvx6y
<bqv[m]>
i'm sure this is simpler than i'm making it
wavirc22_ has joined #nixos
OrangeTetra has quit [Ping timeout: 256 seconds]
<evertedsphere>
cole-h: http://0x0.st/iSqU.txt it's quite inscrutable, looking at the source it fails to "request an adapter"
<dsx>
bqv[m]: if you want Internet to have tcp with delta — that's all you need. However, other iptables rules on both zeta and delta may affect that. What I suggest is you start with a completely clean firewall with default ACCEPT and see if NAT works.
<cole-h>
evertedsphere: Are you running nouveau?
wavirc22 has quit [Ping timeout: 256 seconds]
<evertedsphere>
i think i'm missing some sort of graphics drivers. from my experience using Iced (the gui library) on Arch i needed to do some sort of Vulkan fiddling
<evertedsphere>
i'm running on nvidia.prime.sync if that's what you mean
<dsx>
bqv[m]: otherwise it's a hell to debug, trust me.
OrangeTetra has joined #nixos
<evertedsphere>
had to blacklist i915 because it kept softlocking my computer for some reason
<cole-h>
Ah, can't help then. I just know that there is no Vulkan support in Nouveau (open-source Nvidia drivers) yet, which is what could also cause that error (I ran into it myself)
<evertedsphere>
oh no.
<evertedsphere>
can i somehow run the code on a vm or something?
oranginal has quit [Ping timeout: 256 seconds]
<evertedsphere>
yeah it might be that, i remember needing to install Vulkan drivers to make this library work on Arch
<dsx>
bqv[m]: if for some reason you can't just open everything to the world, you could always try two vms running on your localhost
OrangeTetra has quit [Remote host closed the connection]
vykook has quit [Ping timeout: 256 seconds]
<evertedsphere>
i guess i could also just use rust's wasm backend *taps head*
<cole-h>
evertedsphere: You might need `vulkan-headers`
OrangeTetra has joined #nixos
<bqv[m]>
that's true
<evertedsphere>
i'm not using rustup, how do i install non-native rustc targets into a nix-shell?
OrangeTetra has quit [Client Quit]
<evertedsphere>
i just use nix-shell -p rustc cargo (well i usea shell.nix but you get the idea)
<dsx>
bqv[m]: another thing to keep in mind is MTU mismatch. That's another potential source of trouble, but before you travel there, see if simplest case possible works.
<ij>
if I want to change the image I'm burning onto a usb that I'm going to boot, would it make more sense to make it a ext4 rather than keep making ISOs all the time?
vykook has joined #nixos
freethink has joined #nixos
freethink has quit [Changing host]
averell has quit [Quit: .]
<dsx>
bqv[m]: oh, another common mistake: forgetting to enable ip_forward :)
<{^_^}>
[nixpkgs] @jjwatt opened pull request #84739 → dasm: init at 2.20.13 → https://git.io/Jvxim
<bqv[m]>
checked that one :p
vykook has quit [Ping timeout: 256 seconds]
LysergicDreams has quit [Ping timeout: 264 seconds]
<evelyn>
I am a bit confused - I've customised a package, and I can do something like `nix-build -I ./pkgs -A nixpkgs.$pkg`, but I can't work out how to install the result into my nix-profile in ~/.nix-profile. I get the output in ./result, but I can't see an obvious option in `nix-build`
azdle has joined #nixos
<zeta_0>
maybe it's because i have the wrong port number setup in my home.nix, what do you guys think?
<evelyn>
part of the problem is that there are two 'rival' portable versions of it, the one that nixpkgs has does not work with darwin from high sierra onwards (I... think) and the other one does, but it has lots and lots of patches that add things the mac C library is missing, and I have not auditied this code and cannot vouch for it, which is why I'm reluctant to submit the changes
<evelyn>
and from Catalina onwards they put it in the base system, to replace their ancient emacs (it is a tiny emacs-like editor)
<evelyn>
I will submit the work but I suspect I am the only one using it
<evelyn>
and I don't want to be responsible for any mayhem that results mainly
o1lo01ol1o has quit [Remote host closed the connection]
linarcx has joined #nixos
<evertedsphere>
"TypeError: el is null" yup, really happy about the decision to compile rust to wasm
<asymmetric>
is anyone else seeing SSL errors on yarn install on 20.03?
<to8>
Hello! I'm trying to package hponcfg a tool to setup iLO on HP Proliant servers. The software is only delivered as .deb or .rpm but this might be a good starting point to learn how to package anyways... Once unpacked I have a binary and a .so lib. The binary doesn't find the lib unless it's part of the runtimeDependencies but I cannot figure out how to add it as a "relative" path (to the build package). The gist says more than thousand words:
<evertedsphere>
so since i can't run this app on my system because of nouveau not supporting vulkan... can i run it in a vm and get around the problem?
kleisli has quit [Remote host closed the connection]
<evertedsphere>
do any virtualisation options available on nixos provide virtual graphics drivers that support vulkan
<Brandon__>
what's the current recommendation on composing multiple nix-shell scripts? My usual method is to do: `nix-shell first.nix`, then in that shell do `nix-shell second.nix`.
kleisli has joined #nixos
kleisli has quit [Remote host closed the connection]
<drakonis>
imports maybe?
<evils>
or the horrible solution of having `shellHook = "nix-shell second.nix";`? xD
linarcx has quit [Quit: WeeChat 2.8]
<Brandon__>
evils ... hmm, i like it, will have to try it. but yes, not idea heh
<evils>
which i suppose has the advantage of keeping second.nix easily usable on its own
<Brandon__>
drakonis, will imports execute the shellHook?
<Brandon__>
evils - yes, good point! and you could even swap out `second.nix` easily enough with a different nix expression
to88 has joined #nixos
<drakonis>
do you have multiple shellhooks?
<Brandon__>
drakonis, unfortunately, sometimes i do
to8 has quit [Ping timeout: 250 seconds]
<drakonis>
i see
<Brandon__>
i get that all bets are off at that point since each shellHook is about as mutative as you can get
asheshambasta has quit [Ping timeout: 265 seconds]
<srk>
Amit[m]: ^^ it is released automatically when tests are passing
<Yaniel>
in other words, "when it's ready"
<drakonis>
as soon as there's nothing preventing it from passing
anirrudh has joined #nixos
lsix has joined #nixos
<Amit[m]>
Aha. and is that the entire repository needs to pass? Or each package individually?
anirrudh has quit [Client Quit]
<Amit[m]>
(I'm trying to understand what I need to look out for to switch back to nixpkgs-unstable for a particular package that is updated and merged but doesn't seem to be in the released version)
shafox has quit [Remote host closed the connection]
knupfer has joined #nixos
knupfer has quit [Client Quit]
<drakonis>
no mass rebuilds i think?
knupfer has joined #nixos
<drakonis>
unless necessary
hax404 has quit [Quit: WeeChat 2.7.1]
hax404 has joined #nixos
<drakonis>
it advances to a specific commit and tries to build everything from it?
<Amit[m]>
Right. I'm trying to find a recently passing build in Hydra but can't (I'm sure I'm looking in the wrong place or reading the tables incorrectly). E.g. the nixpkgs-unsable channel (https://releases.nixos.org/nixpkgs/nixpkgs-20.09pre218717.05f0934825c) says it's based on a release from 03-28, but builds that recent seem all have packages that failed
<drakonis>
so you might occasionally find packages that arent available in the cache
<Amit[m]>
For example, https://hydra.nixos.org/eval/1578270 seems to be the build corresponding to what's currently in nixpkgs-unsable (based on the date as well as the git commit hash), and it has many failing jobs still
cybrian has quit [Ping timeout: 250 seconds]
<Amit[m]>
drakonis: right, though looking at my unpacked channel, it looks like it doesn't have up to date expressions (at least for a particular package that was merged... a week ago?)
<drakonis>
have you tried upgrading?
domogled has quit [Ping timeout: 265 seconds]
<drakonis>
the current commit dates to 4 days ago
<Amit[m]>
huh... as in `nix-channel --update`? (this is for packages installed in my user profile, not the my main configuration, which tracks nixos-19.09)
<Amit[m]>
I did just do that a few times. drakonis how are you seeing what the current commit is?
o1lo01ol_ has quit [Remote host closed the connection]
<drakonis>
it points out which commit the channel is currently at
o1lo01ol1o has joined #nixos
anirrudh has joined #nixos
zupo has joined #nixos
<Amit[m]>
Oh i see that, thank you. Odd, I must have done something incorrect locally
selfsymmetric-mu has joined #nixos
<selfsymmetric-mu>
Profpatsch: I am seeing you EVERYWHERE on the internet lately.
o1lo01ol1o has quit [Ping timeout: 256 seconds]
<selfsymmetric-mu>
Thanks for updating the README for `dhall-flycheck` btw. It works for me now. :)
<fusion809>
Anyone know of a way to get the pulseaudio module of i3pystatus to work on NixOS? Due to the non-FHS compliance of NixOS it gives the error, "pulseaudio: Fatal Error - OSError(libpulse.so.0: cannot open shared object file: No such file or directory)". I've just assumed there's no way to fix it, but I thought I'd ask in case there is one.
<selfsymmetric-mu>
Profpatsch++
<{^_^}>
Profpatsch's karma got increased to 14
<Yaniel>
fusion809: whatever i3pystatus is using to access that should be fixed to not hardcode paths
<fusion809>
So I should report this as a bug, Yaniel?
<fusion809>
At NixOS/nixpkgs
<Yaniel>
and then it needs to be wrapped to set the library path
<fusion809>
Thanks Yaniel, I'll file a bug report.
<cizra>
cole-h: I'm already happily installing a small selection of unstable packages in my /etc/nixos/configuration.nix. I hoped to do the same in nix-shell -p.
<cizra>
cole-h: ... exactly how the faq describes.
<cole-h>
Do you have the unstable channel setup, i.e. it shows in `nix-channel --list`?
<cole-h>
If so, I think you should just need `nix-shell '<unstable>' -p insert-package-here`
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Darkmatter66_ has quit [Ping timeout: 256 seconds]
reallymemorable has quit [Quit: reallymemorable]
thonkpod has quit [Quit: WeeChat 2.4]
bqu has joined #nixos
Darkmatter66 has quit [Client Quit]
<bqu>
Christ I screwed up
Darkmatter66 has joined #nixos
<bqu>
Locked myself out my server again
<bqu>
I hate iptables :D
<to8>
At least do hast a proof that iptables is working.
<bqu>
Ya
o1lo01ol1o has joined #nixos
vika_nezrimaya has joined #nixos
<bqu>
The frustrating thing is I can't even use a backdoor in scenarios like this, cause all traffic to that machine is now null routed
zupo has joined #nixos
<bqu>
I just need to implement auto rollback :/
mallox has quit [Quit: WeeChat 2.8]
zupo has quit [Client Quit]
zupo has joined #nixos
opthomasprime has joined #nixos
opthomasprime has quit [Remote host closed the connection]
domogled has joined #nixos
<vika_nezrimaya>
NixOS geekiness is when the first place you go to learn how to configure something isn't Nixpkgs/NixOS documentation, it's the Nixpkgs repository itself
apeyroux` has joined #nixos
reallymemorable has joined #nixos
captjakk has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @zupo opened pull request #84748 → Automatically restart netdata on failures → https://git.io/JvxDd
<mekeor>
vika_nezrimaya: that shouldn't be nixos-geekiness. that just sounds like bad documentation which is bad also in case of nix(os)
captjakk has joined #nixos
<srk>
bqu: NetworkManager has checkpoint/restore feature :)
leotaku has quit [Ping timeout: 256 seconds]
<vika_nezrimaya>
Well sometimes documentation is not enough
cosimone has quit [Remote host closed the connection]
<bqu>
I don't use networkmanager :p
<vika_nezrimaya>
sometimes I need to actually see it with my own eyes to understand it
<srk>
yeah, just saying
<bqu>
Ty :)
cosimone has joined #nixos
<srk>
something like that should be doable easily with NixOSes rollbacks as well
<jared-w>
Most of my reasons for needing to constantly grep through the nixpkgs repo is because none of the documentation gives you any indication of how things are put together and how you override stuff. Things are all hunky dory until you need to not do the default; then its source code time :p
shibboleth has joined #nixos
<vika_nezrimaya>
gchristensen: yep, the trust bootstrapping
<vika_nezrimaya>
I could generate keys, e.g. on install time and use known SSH keys on the custom install media
<vika_nezrimaya>
Maybe I could have a CA keyset on the install media
<vika_nezrimaya>
this looks fine unless someone steals the install USB
<vika_nezrimaya>
a YubiKey sounds nice
<lordcirth>
vika_nezrimaya, a secret that the CA recognizes, not the CA key itself
<vika_nezrimaya>
lordcirth: that's better
<gchristensen>
usually you have a way to trust that a machine was safely installed to, and use that to get its host key signed
<vika_nezrimaya>
hash of install media?
cybrian has joined #nixos
<gchristensen>
aws provides the metadata api and identity documents for example
<vika_nezrimaya>
I could hash the USB drive contents and then send the hash with the CA signing request
<gchristensen>
how often are you going to do this?
<lordcirth>
Secrets are better, as a hash can lie
<colemickens>
seems easy to spoof the hash
apeyroux` has quit [Ping timeout: 258 seconds]
<vika_nezrimaya>
gchristensen: As often as I'll be installing NixOS
<colemickens>
would it be stupid to pre-ship the host keys encrypted and decrypt them manually on first boot (that presumably you control timing of)?
<energizer>
why do packages take the whole config as an argument instead of just taking their own config?
Naptra has quit [Remote host closed the connection]
<gchristensen>
vika_nezrimaya: is that once a year, month, week, day, 10 a day, 1,000 a day...?
<vika_nezrimaya>
gchristensen: Last time I installed NixOS on something was... this winter! I got an RPi 4 and proudly flashed NixOS onto the SD card
* colemickens
just realized what he said out loud >_<
<gchristensen>
because unless you're in the like, >=1 a day, I'm not sure I'd call SSH CA's worth the effort :P
<energizer>
for example why does home-assistant use `config.services.home-assistant` instead of just taking those values as argument, with defaults set by `override` ?
mgdm has joined #nixos
<vika_nezrimaya>
colemickens: decrypt on first boot with what?
<vika_nezrimaya>
this only protects keys in transit and only if the encryption key is transmitted separately
<mgdm>
Hello. Is there a way in configuration.nix to create a storage pool for libvirtd? I've just done virtualisation.libvirtd.enable = true; but I can't see where I'd configure the storage
<colemickens>
vika_nezrimaya: idk, I was thinking you'd ssh to it and decrypt (which uhm yeah)
<energizer>
this seems like the whole config mutable global state and i'm puzzled why it's there
<vika_nezrimaya>
ssh to a machine which can't present a valid host key and could've been compromised as soon as I yanked the flash drive out
anirrudh has quit [Quit: Leaving.]
<vika_nezrimaya>
and went somewhere else
<energizer>
config is*
evertedsphere has quit [Ping timeout: 265 seconds]
<colemickens>
vika_nezrimaya: well yeah, if you TOFU, what's the point of the whole exercise :S
<notgne2>
betaboon: I caught the logs from earlier, Continix actually does work with existing NixOS service modules (but as of now you can only have 1 ongoing service, and any number of oneshots)
<ZaraChimera>
Following the curl -L command to install nixpkgs it has me run a . /home/user/.nix-profile etc command. I am running zsh as the default shell, and I appear to be missing something. I found a search result that mentioned sourcing the profile to .zshenv, but didn't mention the specific command for that.
<colemickens>
but if you can't trust the machine after you walk away, you can't really trust it anyway, no? Unless you've got some sort of other sophisticated FDE that would presumably suffer similar problems.
sam_w has joined #nixos
<vika_nezrimaya>
hmmm
<vika_nezrimaya>
FDE + generating host keys on install
<gchristensen>
+1
<notgne2>
betaboon: modules are added on a whitelist basis though, as lots introduce changes/bloat even when not explicitly enabled
<vika_nezrimaya>
that's a great idea
<lordcirth>
colemickens, even if you trust the machine, you can't trust the network between you/CA and the machine
<vika_nezrimaya>
lordcirth: SSL?
<gchristensen>
lordcirth: then cat the public host key before you walk away
<vika_nezrimaya>
s/SSL/TLS/
<lordcirth>
vika_nezrimaya, TLS, again, requires certs to exist
<vika_nezrimaya>
I have a Let's Encrypt certificate which proves I own and control a domain! :3
<gchristensen>
vika_nezrimaya: I get the impression you're worried abou tthinsg without having a clear ideaabout why you're worried about them, and so we're inventing more and more complicated solutions for something we haven't fleshed out
<vika_nezrimaya>
as long as Let's Encrypt is not compromised which is not probable since they know their security stuff better than a lonely girl who has nothing to do on self-isolation
<colemickens>
FDE just moves the problem of course. But I think there are OSS projects out there meant to help coordination of automatic boots of FDE'd machines. (like a central server that will release keys back to a host for up to 30 secs after they announce they're rebooting, etc). I don't have experience with those myself.
<gchristensen>
oh well if it is for justthat, carry on :P
<sam_w>
hey all, is there an idiomatic way to write arbitrary text to a file as part of nixos configuration? For context: I'm deploying bind and have an entirely static zonefile that I'd like to write to the filesystem - it feels a shame to not be able do it as part of the main config. Apologies if the answer is blindingly obvious
<gchristensen>
I've been thinking about playing with vault for signing host keys
<colemickens>
My other idea - FDE + a initrd Tor hidden service that would allow you to SSH in and unlock FDE. Sorta solves trust by having the hidden service public key be pre-shared and embedding the hidden service private key in initrd. Also haven't had time to build this yet.
<vika_nezrimaya>
sam_w: builtins.writeFile writes a file to Nix store and returns a path to it
<vika_nezrimaya>
then you can symlink it on systemd unit prestart
<vika_nezrimaya>
if you even need to
<colemickens>
gchristensen: I was thinking of your setup and wondered if you were doing that already
<gchristensen>
nah
<sam_w>
vika_nezrimaya: that sounds like exactly what I need, thanks
ratsclub has quit [Ping timeout: 256 seconds]
<vika_nezrimaya>
no problem :3
<vika_nezrimaya>
colemickens: Tor is sometimes blocked
<vika_nezrimaya>
you can inject another SSH host key in the initrd and use SSH
<colemickens>
vika_nezrimaya: for sure, not a perfect solution. So far I have more issues with NAT than Tor though :)
thibm has quit [Ping timeout: 250 seconds]
<vika_nezrimaya>
that's what I did on sakura which had FDE
<vika_nezrimaya>
(almost FDE)
<vika_nezrimaya>
more like everything encrypted except /boot
<colemickens>
mhm, nixos has dropbear support already too; been using it on rpi4 a bit
<vika_nezrimaya>
hm, encrypted / on RPi
<vika_nezrimaya>
never tried it and I'm thinking if I should
<vika_nezrimaya>
an RPi is easy target for stealing
<vika_nezrimaya>
yank the SD card out and boom
<colemickens>
;)
<mgdm>
how well does nixos on rpi4 work? I have one which is just running conbee and home-assistant but on raspbian. I'd need wifi
<vika_nezrimaya>
mgdm: I use NixOS on an RPi 4! WiFi is untested since I use Gigabit Ethernet instead, but other features work well
<vika_nezrimaya>
Oh wow
<vika_nezrimaya>
I just remembered that I didn't update it in a long while
<vika_nezrimaya>
oops
<mgdm>
vika_nezrimaya: that's a good start at least, thanks :)
<colemickens>
mgdm: mine is a unifi/home-assistant appliance so it's wired, but afaik wireless is only an issue with certain setups with high HDMI resolutions and interferrence? Everything else is peachy.
<vika_nezrimaya>
I'm not even sure if I have all the files in place to build the configuration
<mgdm>
colemickens: ah cool. I'm not normally using the hdmi on it so that wouldn't really be a problem
<vika_nezrimaya>
colemickens: HDMI interference is only relevant for certain WiFi channels and it's a hardware issue
<vika_nezrimaya>
so if it works on Raspbian it'll work on NixOS too
<vika_nezrimaya>
probably...
<sam_w>
vika_nezrimaya: uh, builtins.toFile?
<vika_nezrimaya>
builtins.writeFile?
<vika_nezrimaya>
I'm not sure
<vika_nezrimaya>
Nix manual totally has it
<sam_w>
vika_nezrimaya: ok cool, I think it's tofile then. no hits for writeFile :p
<sam_w>
well, one hit - irc logs!
<gchristensen>
pkgs.writeFile
<vika_nezrimaya>
maybe I'm confusing it with nixpkgs.writeShellScriptBin
<vika_nezrimaya>
gchristensen: diff?
<gchristensen>
builtins.toFile can't point to other things in the nix store. it is much more limited
<Brandon__>
Is there a way to access default phases? I'd like to do something like `phases = remove "unpackPhase" phases;`, but the phases on the RHS isn't defined
<vika_nezrimaya>
Brandon__: in this case dontUnpack = true
abathur has quit [Ping timeout: 265 seconds]
ratsclub has joined #nixos
<vika_nezrimaya>
in other cases set ${phase}Phase = "";
mojjo has quit [Ping timeout: 240 seconds]
ratsclub has quit [Remote host closed the connection]
morgrimm has quit [Ping timeout: 265 seconds]
zeta_0 has joined #nixos
niso has joined #nixos
mexisme has joined #nixos
<Brandon__>
vika_nezrimaya, thank you!
<zeta_0>
niso: i final gave up on that mail sender, this is thunderbird's default mailing settings: https://dpaste.org/8Bte
morgrimm has joined #nixos
<zeta_0>
niso: ya, it's too much of a hassle, so maybe i'll try to set it up some other time
<zeta_0>
hey guys, i would like to contribute to nixpkgs haskell infrastructure, but i'm not sure where to start, could you send me a link or something that shows where to get started?
oranginal has joined #nixos
kahiru has quit [Ping timeout: 265 seconds]
oranginal has quit [Client Quit]
<niso>
zeta_0: i see, sorry if i couldn't help you further i don't use gmail
<niso>
zeta_0: or even better: improve the documentation where it where it was lacking
cjpbirkbeck has joined #nixos
blibberblob has joined #nixos
<wip_>
hi! i'm really fresh on nixos, and trying to build syslog-ng with java and java-module support. however this seems to require gradle which in turn requires log4j 1.2.16, where only 1.2.12 is in nix
<wip_>
i think i got 1.2.16 to build but im not able to get this into the build of syslog-ng, gradle is unable to find it and tries to download it from within the sandbox, which fails ofc
<zeta_0>
niso: it's ok, thanks for helping me out, i'll try to setup that mailing sender someday when i have more time
<evils>
i'm having trouble adding an output to a package, somehow fails on find: <second output path> no such file or directory, https://hastebin.com/imeyewasin.nginx
cole-h has quit [Ping timeout: 265 seconds]
fendoer is now known as fendor
<bqv[m]>
fun fact, nixos-rebuild switch --rollback doesn't work by default on flakes
<bqv[m]>
learnt that the hard way
anderslu1dstedt has joined #nixos
<vika_nezrimaya>
OOPS
<vika_nezrimaya>
why
<bqv[m]>
error: path '/nix/store/1gfaw9wwfbgkmriinxs95wnv7vz5ifcr-3kmq245g656x23ssviq4qvhvrx28bxwz-source' is not a flake (because it does not reference a Git repository)
<bqv[m]>
no idea what that's supposed to be about but i'll fix it later
<vika_nezrimaya>
are you maintaining the flakes branch?
<bqv[m]>
for now i'll rely on /nix/var/nix/profiles/**/bin/switch-to-configuration switch
<bqv[m]>
me? no
<bqv[m]>
also, what flakes branch
<vika_nezrimaya>
Nix has a flakes branch and Nixpkgs had it for a while before master had flake.nix added
<bqv[m]>
oh, i guess
<vika_nezrimaya>
tbh my dream is flake support for home manager
<bqv[m]>
that flakes branch is exposed in nixpkgs as pkgs.nixFlakes btw
<vika_nezrimaya>
that's what I have in my nix.package
<vika_nezrimaya>
Flakes <3
<bqv[m]>
<vika_nezrimaya "tbh my dream is flake support fo"> rycee/home-manager/bqv-flakes
<vika_nezrimaya>
Ok now that's interesting
<vika_nezrimaya>
is it keeping up with master or do I have to rebase it myself?
anderslundstedt has quit [Ping timeout: 264 seconds]
<vika_nezrimaya>
'cause I don't know how to rebase lol
<bqv[m]>
rebase it manually
<bqv[m]>
i haven't bothered recently because there's nothing new i need
<bqv[m]>
but the deal was that it would be a community maintained branch
<bqv[m]>
cause the bossman hasn't got time to fiddle with flakes
v0|d has joined #nixos
<vika_nezrimaya>
ohhhhh it locks nixpkgs in flake.lock :c will I be able to override it with something like inputs.home-manager.inputs.nixpkgs = <...>?
<vika_nezrimaya>
bqv[m]: it looks like it only exports the NixOS module
<bqv[m]>
oh, yes, huh
<bqv[m]>
does it even make sense to use anything else, with flakes?
<vika_nezrimaya>
I don't see a reason why it wouldn't
anderslu1dstedt has quit [Ping timeout: 240 seconds]
<vika_nezrimaya>
Flakes aren't NixOS-centric by themselves
<bqv[m]>
i suppose it does, but i'm not sure how much work that would be
<spinus>
do you know if sha256 checksums are supported with `-I nixpkgs=https://blabla` syntax?
<vika_nezrimaya>
Rewriting bin/home-manager to use flakes (probably easy)
<bqv[m]>
yeah, my thinking was that home-manager as a program isn't built with flakes in mind
<vika_nezrimaya>
I guess I'll have to do it myself
<bqv[m]>
exactly
<bqv[m]>
eh, clone it locally and see how far you get i guess, if you can do it without breaking the nixos module, it'd be a good thing to PR in so others can use it
<vika_nezrimaya>
are there any tests for the nixos module?
<spinus>
bqv[m]: one sentence how nix flakes are different from overlays?
<bqv[m]>
no, but just ping me a repo, branch and ref and i can check it against my config
pbogdan has joined #nixos
<spinus>
is my understanding correct it's only a description of a structure for artifacts?
<vika_nezrimaya>
spinus: completely different solutions for completely different problems
<bqv[m]>
declaratively defining the nix relationship between git repositories, and declaratively pinning them to revisions
<bqv[m]>
overlays are something you'd use WITH flakes
<spinus>
bqv[m]: but I can do that with fetchgit
<betaboon>
notgne2: thanks for the headsup regarding continix. i have had some success using nix-container-images today. might look into continix next. evaluating the possibilities of building docker-images from nix-configuration was just one stop on a semi-long way to go tho :) off for today
<vika_nezrimaya>
well yeah but a flake is easier to write and use (and re-use!)
<bqv[m]>
spinus: sure you can, see how easy that is compared to flakes :p
<vika_nezrimaya>
`nix flake` writes your "fetchgit"-like stuff automatically, updates it and all the revisions and hashes
<spinus>
:>
<spinus>
vika_nezrimaya: I see
<bqv[m]>
e.g. what vika mentioned earlier, s/he can pull in the home-manager flake, and override which nixpkgs it uses
<bqv[m]>
can't do that if it's using fetchgit
<vika_nezrimaya>
bqv: it's she/her if you're referring to me :3
terlar has joined #nixos
<vika_nezrimaya>
also there's they/them which is relatively gender-neutral and sounds a lot better compared to slash-based kludges
<kraem>
how do i get bundix to fetch private repos? seahorse pops up even though i'm not using it (guess it's some kind of buildInput) and it won't take my credentials in terminal.
<bqv[m]>
i had a feeling, but chickened out once i got to 'h'
<vika_nezrimaya>
bqv: in case of confusion use they/them
shibboleth has quit [Quit: shibboleth]
<vika_nezrimaya>
I'm ok with correcting people if they accidentally do that to
cole-h has joined #nixos
knupfer has quit [Quit: knupfer]
knupfer1 has joined #nixos
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to release-20.03: https://git.io/JvxH1
<{^_^}>
[nixpkgs] @worldofpeace merged pull request #82295 → nixos/release-notes: mention that dhcpcd stopped giving IPv4 addresses to bridges → https://git.io/JvKoD
<{^_^}>
[nixpkgs] @worldofpeace pushed commit from @erictapen to master « nixos/release-notes: mention that dhcpcd stopped giving IPv4 addresses to bridges by default »: https://git.io/JvxHD
civodul has quit [Quit: ERC (IRC client for Emacs 26.3)]
terlar` has joined #nixos
fendor has quit [Read error: Connection reset by peer]
terlar has quit [Ping timeout: 258 seconds]
InvisibleRasta has joined #nixos
morgrimm has quit [Ping timeout: 256 seconds]
bukkitgerman8608 has joined #nixos
bukkitgerman860 has quit [Read error: Connection reset by peer]
<colemickens>
,tofu
<{^_^}>
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
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
__monty__ has quit [Quit: leaving]
<worldofpeace>
we should probably change that to mention lib.fakeSha256
<terlar`>
Is there a way to do something similar to nix-instantiate but from with Nix? I want to get the derivation path with two potentially different sources to compare if they produce the same thing, it would be neat if there was a function to calculate the hash of a derivation without building it
<vika_nezrimaya>
home-manager sometimes uses builtins.getEnv which always returns an empty string with flakes
<vika_nezrimaya>
I'll need to supplement it
selfsymmetric-mu has left #nixos ["gone to the land of dead hiccups and extinguished light bulbs"]
Diogo has joined #nixos
<vika_nezrimaya>
bqv++
<{^_^}>
bqv's karma got increased to 1
<vika_nezrimaya>
flakes++
orivej has joined #nixos
<vika_nezrimaya>
terlar`: nix repl can compare derivations I think
<vika_nezrimaya>
and with Nix you can do derivation1 == derivation2 and afaik it will not build it
<vika_nezrimaya>
it'll compare all the inputs
<vika_nezrimaya>
Nix is very lazy
<terlar`>
Hmm, I thought so too, but I was surprised, I did derivation1.drvPath and it didn't build that derivation, but it built all the dependencies mentioned within that derivation
projects has quit [Remote host closed the connection]
<terlar`>
I will try the d1 == d2 comparison if it does some more magic
<vika_nezrimaya>
It should! probably
<vika_nezrimaya>
drvPath instantiates a derivation, which references all its dependencies' closures
<vika_nezrimaya>
which results in building all dependencies
<vika_nezrimaya>
d1 == d2 may avoid it
spinus has quit [Ping timeout: 265 seconds]
<terlar`>
But then I can't figure out the hash? Might work for my use case though
<vika_nezrimaya>
why will you need the hash if it's calculated from all inputs?
<terlar`>
I was considering to run nix-instantiate within nix, but unfortunately this project have very complicated nix files that require a lot of injected state from the nix-build
<vika_nezrimaya>
if recursively comparing inputs equals true, then derivations are effectively equal as far as Nix knows
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/JvxQ1
<vika_nezrimaya>
(barring indeterminism in build process like $RANDOM in buildPhase)
<dhess>
Has anyone here run this Hydra migration?
<terlar`>
I thought it would be nice to see why they are different, hence the drv hash and you could diff them to see why
morgrimm has joined #nixos
<terlar`>
My use case is checking if the target derivation has changed from the fork point in git, so I check out the code and run the same derivation with that as src and then with the local src to compare if they changed
jbrock has quit [Quit: jbrock]
jbrock has joined #nixos
<terlar`>
Comparing inputs is enough for my case
<bqv[m]>
vika_nezrimaya: oh nice, it worked?
eoli3n__ has quit [Ping timeout: 258 seconds]
<bqv[m]>
sorry, i'm having a minor dishwasher malfunction
<vika_nezrimaya>
Looks like it worked! :3
maddo has quit [Quit: See ya]
<bqv[m]>
nice! how dire were the modifications?
<vika_nezrimaya>
bqv: nix-store --repair-path /nix/store/*-dishwasher might help
<vika_nezrimaya>
bqv[m]: fairly small
<vika_nezrimaya>
looks like maintainers of the flake branch did a great job :3
<vika_nezrimaya>
but this isn't rebased on top of master yet
<vika_nezrimaya>
bqv: currently build tools aren't supporting flakes though... I built it manually
InvisibleRasta has quit [Quit: WeeChat 2.3]
abathur has joined #nixos
morgrimm has quit [Ping timeout: 250 seconds]
<bqv[m]>
hahaha
<bqv[m]>
well, join the small band of flake branch maintainers if you'd like :p i doubt rebasing will pose any issues
<vika_nezrimaya>
bqv: rebasing went completely smooth
chloekek has quit [Ping timeout: 256 seconds]
<vika_nezrimaya>
I'm starting to like this! :3
Jonathan72 has joined #nixos
abathur has quit [Ping timeout: 240 seconds]
<bqv[m]>
nice :)
<Brandon__>
I've got a nix expression that, when it invokes another nix expression via import (itself being called by nix-shell), fails to build the other nix expression (the error is a bit mysterious). However, when I run nix-shell on the called nix expression directly, there are no problems.
<Brandon__>
Is there some reason that the behavior would be different? i would expect the nix store to be used
<vika_nezrimaya>
Brandon__: hm... That looks like a problem that COULD be solvable by flakes in the future
<vika_nezrimaya>
your inputs differ somehow
<vika_nezrimaya>
I'm not sure why and how
spinus has joined #nixos
<Jonathan72>
I am building a D application with SDL but it is failing to link (at run-time or compile-time). I thought I got it solved, with some help, with a nix-shell with `LD_LIBRARY_PATH = "${pkgs.SDL2}/lib";` but it is not working.
<{^_^}>
commercialhaskell/stack#2536 (by stulli, 3 years ago, closed): stack build: "ConnectionFailure getAddrInfo: does not exist (Name or service not known)"
<vika_nezrimaya>
sandboxing?
<vika_nezrimaya>
try disabling sandbox in nix options
<vika_nezrimaya>
if it doesn't fail, this means that your build tries to access the network somehow(?)
<Brandon__>
it is definitely trying to access the network, trying to do a stack build
<Brandon__>
though i'm not entirely sure why
<Jonathan72>
The application is using Dub (the D package manager) which has no special NixOS wrapper, which is why the `LD_LIBRARY_PATH` thing was added.
miscellaneousbus has joined #nixos
<vika_nezrimaya>
Brandon__: Nix build cannot access the network, it's forbidden. It ruins the reproducibility of the build since there are unaccounted non-deterministic inputs
<Jonathan72>
Really?! I am dumb, I figured it out. Sorry.
<{^_^}>
[nixos-homepage] @garbas pushed 0 commits to zimbatm-patch-1: https://git.io/Jvx7i
<Brandon__>
vika_nezrimaya, ah right i see
<{^_^}>
[nixpkgs] @Ma27 merged pull request #84570 → nixos: default nix.maxJobs to auto → https://git.io/JvA4c
<bqv[m]>
vika_nezrimaya: want a hand with the rebasing? looks like that went a bit sideways :D
<vika_nezrimaya>
???!
<vika_nezrimaya>
Well I said
<vika_nezrimaya>
I never rebased stuff so this is my first rebase
<vika_nezrimaya>
no wonder I botched it
gxt has quit [Remote host closed the connection]
<bqv[m]>
hold on, i'll figure out how to repair it
gxt has joined #nixos
<bqv[m]>
(or indeed, if it even has gone wrong
o1lo01ol1o has quit [Ping timeout: 265 seconds]
emilsp has quit [Ping timeout: 619 seconds]
o1lo01ol1o has joined #nixos
<vika_nezrimaya>
bqv[m]: I don't think it has gone wrong, rather, it went too well
<Brandon__>
vika_nezrimaya, just to follow up, if i set buildPhase = '' in the calling expression then it gets past that particular issue. I think the reason it work if I use nix-shell on the *called* expression directly is because nix-shell does not execute a build on the outer derivation
justanotheruser has quit [Ping timeout: 260 seconds]
<Brandon__>
nevemrind, false alarm, '' should have been "" and it caused a separate issue
<bqv[m]>
according to the log, those are the only commits ignoring your merge commit (and i'll assume it was a clean merge so that can be completely skipped)
<bqv[m]>
it's just a good idea to keep these commits tight so when it comes that they get merged, it's easy
rogue_koder_ has joined #nixos
GrimSleepless has joined #nixos
rogue_koder has quit [Ping timeout: 265 seconds]
<bqv[m]>
all good?
<bqv[m]>
(sorry i'm slightly awol, dishwasher is trying to kill me)
<vika_nezrimaya>
bqv: I nuked my local checkout
o1lo01ol1o has quit [Remote host closed the connection]
<vika_nezrimaya>
bqv[m]: diverging branches, 4 ours / 7 theirs, force-push into my fork?
<bqv[m]>
yep, force push
<vika_nezrimaya>
done
<bqv[m]>
much better
<bqv[m]>
wait, why's there a merge conflict still :/
o1lo01ol1o has quit [Ping timeout: 265 seconds]
<bqv[m]>
oh, right. the problem is github is trying to merge
<bqv[m]>
rycee needs to also force push
<bqv[m]>
so that's fine
GrimSleepless has quit [Ping timeout: 264 seconds]
<vika_nezrimaya>
also it's upstream/master not origin/master for me
<vika_nezrimaya>
since rycee's branch is named upstream in my remote list
<bqv[m]>
oh, interesting
<bqv[m]>
i hope github didn't ping all those people who made those commits
<vika_nezrimaya>
ME TOO
<vika_nezrimaya>
>.<
<bqv[m]>
heh
<bqv[m]>
for future reference, assuming you're currently working on a branch, the top 4 commits are the only relevant ones, and you want to rebase onto upstream/master, you would run: git rebase --onto upstream/master HEAD~4
<bqv[m]>
that will take the top 4 commits and replay them (in the same way you just did manually) onto upstream/master's tip
morgrimm has joined #nixos
rogue_koder_ has quit [Remote host closed the connection]
rogue_koder_ has joined #nixos
morgrimm has quit [Ping timeout: 256 seconds]
pbogdan has quit [Remote host closed the connection]
zebrag has quit [Remote host closed the connection]