<VaNilLa[m]>
what's the "canonical" way for setting it up? The manual is fairly outdated, and the Git Hub issues regarding it are a year old now.
<simpson>
VaNilLa[m]: Hydra is not an easy package to administer. You will likely want to be running on NixOS.
<wrl>
is hydra the best option for self-hosted nix-based CI?
<VaNilLa[m]>
Well it certainly isn't easy given that the first page of the manual is outdated.
<simpson>
wrl: Not really, but people seem to think that just running plain Nix commands is not an option.
<VaNilLa[m]>
I have the same question as wrl: I want self-hosted CI, and I'd like to try nix with it
<simpson>
VaNilLa[m]: Make your stuff build with Nix first, *then* figure out Hydra.
<VaNilLa[m]>
I tried hydra largely because it also provides a web interface, and that's one of the things I'd like.
<simpson>
Exactly, apparently just running plain Nix commands isn't an option for you, for some reason. https://nixos.wiki/wiki/Hydra explains how to Hydra on NixOS.
<VaNilLa[m]>
Alright, guess I'll do that. I'm mostly just experimenting with it
<simpson>
And what will you do once you've got a Hydra set up? You'll need to give it some Nix expressions and a place to pull them from.
<VaNilLa[m]>
right, which goes back to your comment of getting my stuff to build with nix
emptyfl4sk has quit [Ping timeout: 258 seconds]
<hpfr[m]>
How should I set up the Emacs daemon with home-manager?
wfranzini has quit [Remote host closed the connection]
wfranzini has joined #nixos
<VaNilLa[m]>
Out of curiosity though, I still would like to know why hydra is only provided in the NixOS channel and conflicts with nix-build
<clever>
VaNilLa[m]: after you ran nix-build, did you `ls -l result/bin/` ?
wildtrees[m] has joined #nixos
<VaNilLa[m]>
I didn't run nix build. I merely tried to install the hydra package. Following this issue https://github.com/NixOS/hydra/issues/532 I ended up installing nix-build, with no sign of hydra. Finally, I added the NixOS-19.03 channel and tried to install hydra, which conflicted with nix-build.
<{^_^}>
hydra#532 (by gilligan, 1 year ago, open): Link for installling hydra is outdated
<clever>
VaNilLa[m]: what is the exact command you ran?
<Ashy>
i've been looking at getting some personal projects to build with nix and that has been quite helpful
<VaNilLa[m]>
Ashy: thanks!
<VaNilLa[m]>
Clever: I ran nix-env -i hydra (no matches). As per the issue, I ran nix-env -iA hydra (installed nix-build). Then I added the nixos 19.03 channel via nix-channel --add, and finally ran nix-env -i hydra (package conflict in file Hydra.pm)
<clever>
VaNilLa[m]: `nix-env -iA hydra` shouldnt install nix-build, what does `nix-env -q` report?
noudle has quit []
<VaNilLa[m]>
let me fire up my VM.
<jackdk>
I am packaging a brother printer driver for my printer, because it's not in nixpkgs and is similar to existing ones. Before I go full copypasta on this, anyone know if there's work afoot to unify the brother driver packaging (e.g., a function over URL, printer name, etc?)
<VaNilLa[m]>
`nix-env -q` reports only nix-build. I gave you the wrong command: I ran `nix-env -iA nixpkgs.hydra`
<clever>
VaNilLa[m]: can you pastebin the entire output of `nix-env -q` ?
<VaNilLa[m]>
clever, when I say only nix-build, I'm not kidding. That's the __only__ entry.
<{^_^}>
[nixpkgs] @utdemir opened pull request #68178 → shunit2: init at 2019-08-10 → https://git.io/fjjW1
<VaNilLa[m]>
it's a fresh install, and nix is not the "original" package manager
<clever>
VaNilLa[m]: there is no package called nix-build, can you pastebin the contents of ~/.nix-profile/manifest.nix ?
<VaNilLa[m]>
this will be difficult considering I am in a VM without x-org. Give me a second
<clever>
VaNilLa[m]: can you enable ssh, then just ssh into it?
<VaNilLa[m]>
`pkgs.lightdm-gtk-greeter`, and that doesn't exist. What's the "correct" way?
<buckley310>
i think lightdm is the default login manager if you just turn on xserver.enable, right?
<VaNilLa[m]>
Is it? and the gtk greeter is the default greeter/
georges has joined #nixos
<buckley310>
yeah i think so
<VaNilLa[m]>
I'll try that. But hypothetically, should I have wanted a different greeter, is setting lightdm.package = pkgs.some-greeter-package the right way?
<simpson>
Yeah, but overriding .package in config is always a bit of a gamble. It's generally not what you'll want to do.
<VaNilLa[m]>
uhoh, I didn't quite mean that specific line
<VaNilLa[m]>
lightdm's package has an option called "greeter.package"
<VaNilLa[m]>
which I assume is a way to set the greeter?
<simpson>
It's how to choose the Nix package that will be called by the greeter service configuration.
noogie has joined #nixos
<simpson>
For most of the Xorg stuff, the way to choose which of the various flavors of similar compatible package (DM, WM, DE, etc.) is to simply `.enable = true;` it.
<VaNilLa[m]>
interesting.
<VaNilLa[m]>
thank you!
kvda has joined #nixos
wfranzini has quit [Remote host closed the connection]
<CMCDragonkai>
When attempting to build a container, I'm getting a kvm error on one of my machines (but not on another machine with the same nixos configuration)
<switchy>
on one of my machines, I use `nix-shell -p meta-pkg1 python-env` to get a build environment for some particular software (where meta-pkg1 and python-env are defined in my config.nix) -- how do I get the store paths for these to copy to another machine with nix-copy-closure?
mmlb has joined #nixos
aither has quit [Ping timeout: 244 seconds]
snajpa has quit [Ping timeout: 264 seconds]
<mmlb>
hey all, I'm having trouble installing and booting on a GPT bios HPE ml350p gen 8 machine. I've tried a bunch of things to no avail, but just had this last thought, re `boot.loader.grub.device`
<ivan>
mmlb: do you mean UEFI boot or MBR boot of a GPT disk?
<mmlb>
I've got `ef02` type
<mmlb>
I've been setting `boot.load.grub.device = "/dev/disk/by-id/wwn-....";` should that be `wwn-...-part1` maybe?
snajpa has joined #nixos
<mmlb>
ivan: BIOS boot
<ivan>
you don't have to put the EF02 partition in the nixos configuration, grub detects it
<ivan>
boot.load.grub.device should be set to the whole disk device
<ivan>
how does the boot fail?
<mmlb>
doesn't boot, skips hdd and goes off and tries pxe which is next in line
<ivan>
also I think I've had HP's RAID stuff mangle my grub before
<mmlb>
nope no raid, I set the raid to HBA and that wasn't working either before. I'm currently attempting to boot from usb2sata converter
Supersonic112 has joined #nixos
Supersonic has quit [Disconnected by services]
Supersonic112 is now known as Supersonic
<ivan>
nixos-rebuild output is good and grub doesn't complain?
<mmlb>
everything looks great
<ivan>
maybe your BIOS is just fubar
jtojnar has quit [Ping timeout: 245 seconds]
<mmlb>
yeah could be, I'm going to try with just msdos
<mmlb>
I have not ever experience this situation of BIOS not booting from GPT, though I've heard of it
jtojnar has joined #nixos
domogled has quit [Remote host closed the connection]
<gyroninja>
I don't need that for my droplet (nor do I get that problem with low entropy)
palo has quit [Ping timeout: 258 seconds]
palo1 is now known as palo
<averell>
do you have anything that generates network or disk i/o on boot? cause that solves it automatically.
<gyroninja>
I'm not sure what counts as that but it's a pretty normal web server with FDE
growpotk- has quit [Ping timeout: 258 seconds]
o1lo01ol1o has quit [Remote host closed the connection]
<ivan>
averell: thanks
<ivan>
I think ssh startup was waiting for crng init, but I also have wireguard
<cocreature>
How do I debug “error: cannot auto-call a function that has an argument without a default value ('mkDerivation')”? Is there any way to get some backtrace of what exactly is going wrong?
<savanni>
cocreature: Did you get this when running nix-shell or nix-build?
<cocreature>
savanni: nix-shell
<savanni>
k. You can get a backtrace with --show-trace.
<cocreature>
That doesn’t seem to work for this error for some reason
<savanni>
Usually, the problem is that one of your functions (with me, it's frequently the shell.nix function), has a required parameter that I didn't provide a value to.
<cocreature>
oh actually I’m just stupid, I was in the wrong directory …
<savanni>
lol
<savanni>
:)
<cocreature>
thanks for the help :)
<cocreature>
That’s what I get for trying to fix things before having coffee
<savanni>
Anyway, I have my own question. I'm wanting to create version of git that uses a particular configuration file that I've defined. First question: should I be using mkDerivation with git as a buildInput, or should I be using buildEnv with git as one of the paths?
<savanni>
Like, my strategy is going to be to write the configuration files out to the store, and then probably use wrapProgram to set XDG_CONFIG_HOME to the configuration file... but I'm actually starting with just trying to understand what kind of thing I'm building to make this work.
<elvishjerricco>
Is there a way to access a nix store that's been mounted read only? I'm booted from a usb stick and want to `nix copy` from my hard drive, but I want the hard drive mounted readonly to make sure I don't mess with anything
<elvishjerricco>
I get `error: opening lock file '/mnt/nix/var/nix/db/big-lock': Read-only file system`
<dminuoso>
elvishjerricco: What do you mean by "access a nix store" exactly?
<elvishjerricco>
dminuoso: Mainly just `nix copy` from it without having to mount it rw
<elvishjerricco>
the nix store on a non-root disk
<dminuoso>
elvishjerricco: Curious, why does one need rw access to copy from the store?
ixxie has joined #nixos
<elvishjerricco>
Dunno. Tries to acquire some lock file
marusich has quit [Remote host closed the connection]
Fare has quit [Ping timeout: 252 seconds]
judson_ has quit [Remote host closed the connection]
o1lo01ol1o has quit [Ping timeout: 268 seconds]
judson_ has joined #nixos
Ariakenom has joined #nixos
dansho has quit [Ping timeout: 246 seconds]
<worldofpeace>
savanni: hey
<worldofpeace>
I think you can you use symlinkJoin with git in the paths
<worldofpeace>
and in a let in use one of the trivial builders like writeTextFile for your configuration file, and wrap the programs in postBuild in symlinkJoin
domogled has joined #nixos
endformationage has quit [Quit: WeeChat 2.5]
zupo has joined #nixos
wfranzini has joined #nixos
DariusTh` has joined #nixos
<{^_^}>
[nixpkgs] @jb55 opened pull request #68187 → grub: enable copyKernels by default → https://git.io/fjjBG
mexisme has joined #nixos
teej has joined #nixos
infinee has quit [Ping timeout: 268 seconds]
zupo_ has joined #nixos
<Enzime>
hyperfekt: gyroninja: does sound work for you guys in Minecraft? are you guys using ALSA or Pulse?
<{^_^}>
[nixpkgs] @avdv opened pull request #68189 → wtf: No longer use vendored dependencies → https://git.io/fjjBA
<srhb>
It definitely needs a clearer release note at least. :D
<srhb>
Actually, I also think it's a bad idea. Isn't it pretty common with .xsession-based logins?
<srhb>
But maybe we can make that clearer anyway instead of relying on that session.
<worldofpeace>
I don't think in other distros xterm is enabled as a session default. always seemed weird to me and someone can always enable it.
<worldofpeace>
srhb: what DE do you use? (maybe wm)
Tucky has joined #nixos
<srhb>
xmonad.
abbiya has joined #nixos
schjetne has joined #nixos
<srhb>
And that's entirely possible. My main concern is that I think most people who use a home-manager-based xsession start up will experience silentish breakage with this PR. Maybe expanding the release note to explain what's going on and who's affected will suffice..
<worldofpeace>
I'm not familiar with how people use home-manager based xsessions. Do you mean the "breakage" you just experienced?
<srhb>
Yes. :)
<worldofpeace>
Was it with the gtk greeter?
<srhb>
Without the scare quotes though xD
<srhb>
Yeah :)
infinee has quit [Ping timeout: 268 seconds]
<srhb>
Maybe we should just throw a huge warning if lightdm is enabled but no sessions are.
asheshambasta has joined #nixos
<srhb>
I literally had an empty list, which I didn't think about because I never open that menu.
<worldofpeace>
Ok, so I'm not sure where the code is or if it's there. But the greeter should discard sessions that don't exist anymore in the sessions list.
<worldofpeace>
Your list was empty even when using xmonad?
<srhb>
I'm not using xmonad as desktopManager (the NixOS option)
<srhb>
I was relying on xterm being there by default.
<worldofpeace>
ooooh, Can you paste your actual config?
<srhb>
Erm.. I can probably cook up a minimal example, but can't paste it all. The gist of it is that lightdm is enabled, no desktopManagers were, and I was relying on ~/.xsession taking over from there
<srhb>
Which is sufficient to reproduce the issue.
<worldofpeace>
Yeah you can paste a minimal example :D I just need to be able to reproduce it.
<srhb>
Sure. :)
m0rphism has joined #nixos
vonfry has quit [Quit: WeeChat 2.5]
<Ariakenom>
noob here. trying to make a package. C compilatin fails because it needs #include <GL/glut>. I tried adding to freeglut to buildinputs but same issue in a freeglut file.
<srhb>
Basically I'm opposed to the idea that you can end up with no valid sessions after the upgrade, especially because the failure is pretty silent.
<boxdot>
Ashy: Ah, this one I saw. But it looks like it is NixOS related. To be clear, I am just using nixpkgs.
<boxdot>
Basically, what I try to understand is the fact that if there is no configuration nix-build decides based on the system (linux or macos) whether to activate sandbox or not, and it is not clear for me where this switch is done.
<srhb>
boxdot: You will probably just need to create that file then. If you're on Linux, sandbox is likely already on, but you can check with nix show-config
<srhb>
boxdot: In the Nix source code :)
Zer0xp has quit [Ping timeout: 244 seconds]
<boxdot>
srhb: good point. Found it in src/libstore/globals.hh:228.
<srhb>
boxdot: There's also a hard switch that sets the default to smEnabled for Linux builds, and Disabled for others.
<srhb>
boxdot: but nix show-config is your friend.
<worldofpeace>
No DE or WM is enabled and xserver is enabled we enable xterm
<worldofpeace>
And it stays default false.
veske2 has joined #nixos
chloekek has joined #nixos
knupfer has joined #nixos
abbiya_ has joined #nixos
<{^_^}>
[nixpkgs] @bzizou opened pull request #68191 → bayescan: init at 2.1 → https://git.io/fjj0D
<srhb>
worldofpeace: That sounds like a reasonable fix to me. As long as we don't end up with none. :-)
janneke_ is now known as janneke
abbiya_ has quit [Read error: Connection reset by peer]
noudle has quit []
asymptotically has joined #nixos
abbiya_ has joined #nixos
<worldofpeace>
srhb: right, and in the end it would be no-op for users like you who rely on it. We could use stateVersion, but I think it's kinda silly. And a warning if no DE or WM might be annoying.
<srhb>
worldofpeace: Yeah, I completely agree. I mean, even a release note that warned "if you didn't have any enabled before, take note that you will need to enable one now" would be a cheap "fix"
abbiya has quit [Read error: Connection reset by peer]
abbiya_ is now known as abbiya
boxdot1 has joined #nixos
boxdot has quit [Read error: Connection reset by peer]
domogled has quit [Remote host closed the connection]
domogled has joined #nixos
noonien has quit [Quit: Connection closed for inactivity]
<Taneb>
I've got something using "pkgs.writeScript". Is there a way I can add a derivation to its runtime environment easily?
<worldofpeace>
Taneb: maybe writeTextDir ( I think that's name) could do that. and if runtime environment is directories the scripts are in. Or perhaps wrappers
<Taneb>
Wrappers was what I was thinking but I don't quite know how to do this
<{^_^}>
[nixpkgs] @worldofpeace merged pull request #68144 → Switch my qt packages to qt-specific mkDerivation → https://git.io/fjjY6
<{^_^}>
[nixpkgs] @worldofpeace pushed 3 commits to master: https://git.io/fjjE1
<{^_^}>
[nixpkgs] @adisbladis opened pull request #68195 → Drop unsupported go version 1.11 → https://git.io/fjjEM
knupfer has quit [Ping timeout: 264 seconds]
ng0 has joined #nixos
<{^_^}>
[nixpkgs] @worldofpeace opened pull request #68196 → rl-1909: make services.gnome3 links sensible → https://git.io/fjjE5
abbiya has quit [Quit: abbiya]
<{^_^}>
[nixpkgs] @worldofpeace merged pull request #67280 → fortune: 2.6.2 -> 2.10.0, use official tarball → https://git.io/fjNOS
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/fjjEb
boxdot1 has quit [Quit: Leaving.]
sb0 has quit [Quit: Leaving]
qubasa has quit [Ping timeout: 245 seconds]
philr has joined #nixos
qubasa has joined #nixos
<blitzclone_>
infinisil: regarding the node2nix issue: I'm using nodePackages.node2nix. This gives me 1.6.0. There seems to be 1.7.0, but I'm not sure where it comes from
o1lo01ol1o has quit [Remote host closed the connection]
<infinisil>
blitzclone_: Are you not using the generate.sh script?
<blitzclone_>
yes, I am
<blitzclone_>
it uses nodePackages.node2nix as dependency
__monty__ has joined #nixos
vonfry has joined #nixos
shabius has quit [Remote host closed the connection]
veske2 has quit [Read error: Connection reset by peer]
shabius has joined #nixos
<vonfry>
Does someone use parallels desktop to run nixos on macos? How to run the new version of prl-tools on 19.03? I try to update the scripts in nixpkgs for prl-tools, and now it can be built but run failed.
<colemickens>
worldofpeace: hey, just wanted to say thank you - you've been reviewing many of my PRs lately and giving advice/fixes and I appreciate that :) Cheers
<worldofpeace>
colemickens: thanks, and thank you too 😃 My idea is to make a culture around real helpfulness without any harshness that open source communities can have. So something pretty tender we can all share in and capitalize off of. I'm glad it has reached you ✨
o1lo01ol1o has quit [Remote host closed the connection]
<eyJhb>
If I run e.g. Plex in a NixOS container, is there any simple way of opening the ports, like if it did not run in a container, I could use 'services.plex.openFirewall'?
dasj19 has quit [Read error: Connection reset by peer]
<etu>
eyJhb: you can use the regular firewall thingys, but if it's web stuff I would consider having ngnix as a proxy in front and have a hostname to send it to the right container
<manveru>
eyJhb: i think that's still done with firewall.extraCommands
<eyJhb>
I would have hoped `containers.<name>.forwardPorts` did it, but that seems like host->guest. But it is more of, could the openFirewall be reused?
<adisbladis>
As a nice bonus you'll get letsencrypt tls
<eyJhb>
That
<eyJhb>
Looks
<eyJhb>
Awesome
<manveru>
hehe
<eyJhb>
But need to test how well plex.tv reacts to that... :p So the standard is, that host -> container communcation works fine? Or? Because shouldn't the ports still be forwarded from container to host?
<manveru>
afaik you still need the NAT
ixxie has quit [Ping timeout: 246 seconds]
<eyJhb>
manveru: the answer specifies, that you can use "forwardPorts", but it just says, "from host to container", so it seems one way, and in the wrong direction
<eyJhb>
BUt it might just be worded badly
Zer0xp has quit [Ping timeout: 244 seconds]
<manveru>
well, the container has a different IP, if you know the IP you could put that into nginx config as well and it'd work
chr15 has quit [Ping timeout: 250 seconds]
<manveru>
if you want to make the container appear as a normal port on localhost, you need the forwarding
<manveru>
at least that's my understanding
noudle has joined #nixos
<manveru>
the config of adisbladis doesn't look like it's using containers
<eyJhb>
Guess I have to try when I get the time
<infinisil>
blitzclone_: Ah damn, the generate script doesn't pin nixpkgs
<infinisil>
Meaning it uses whatever nixpkgs you're using on your system, not the one in the nixpkgs you're editing
<elvishjerricco>
Can you boot a `nesting.clone` config from grub?
<elvishjerricco>
Rather than having multiple different nixos installs, I can just have my main config have an alternate version with the minor tweaks I need. Namely the stuff for a VFIO VM on my only GPU. Currently the second system is bare bones so my file servers and stuff go down when I want to game on Windows.
<infinisil>
Neat
ixxie has joined #nixos
<blitzclone_>
infinisil: but even in nixpkgs I only get 1.6.0. There is 1.7.0. If I install this manually, I get changes in pkgs/development/node-packages/node-env.nix. but that actually works
<blitzclone_>
I'm confused ;)
<infinisil>
Huh
jtojnar has quit [Remote host closed the connection]
<blitzclone_>
ah, the generate script defaults to nodejs 6
<blitzclone_>
which is probably not helpful
jtojnar has joined #nixos
<blitzclone_>
I'm such a n00b with this node.js stuff... sorry for the noise
<worldofpeace>
infinisil: didn't realize there were directives from peterix. https://minecraft.gamepedia.com/Petr_Mr%C3%A1zek working on the launcher for linux so pretty much the prefect person :)
<worldofpeace>
infinisil: Right. I usually have to verify people's credentials somehow before I can take their word for it. Good nonetheless because I don't think there'd be a good way to resolve the issue without a correspondence from someone internally.
bnh^ has quit []
iyzsong has quit [Remote host closed the connection]
<infinisil>
Yeah.. no fixed url for the .deb, no fixed url for the icon, possibly can't distribute the item by embedding into nixpkgs either..
civodul has quit [Read error: Connection reset by peer]
<savanni>
worldofpeace: why symlinkJoin instead of buildEnv? buildEnv has basically no documentation, so are we intended to not be using it? I'd encountered a comment I think on the discourse, indicating that it was commonly used in adding configurations to derivations, but the examples I found in nixpkgs weren't quite clear.
<worldofpeace>
savanni: fuzzy search for wrapper.nix for examples maybe. some of which I've authored so It's just how I do it. Usually I find `buildEnv` is more for joining multiple things, I'm not sure if they're really different.
<worldofpeace>
It seems that symlinkJoin is defined in trivial-builders, and it looks pretty simple. buildEnv looks a bit more complex. but the comments say they pretty much do the same thing.
<worldofpeace>
so typical duplication maybe.
mounty has quit [Read error: Connection reset by peer]
pingveno has quit [Ping timeout: 245 seconds]
mounty has joined #nixos
veske3 has quit [Ping timeout: 268 seconds]
<worldofpeace>
oh and you can control buildEnv more
pingveno has joined #nixos
barrett has joined #nixos
<hc>
win 32
Ariakenom has quit [Ping timeout: 258 seconds]
Ariakenom has joined #nixos
<wedens>
back to my cache on ssd question, is it possible/feasible to copy current generation closure to ssd and symlink currernt generation to this copied closure?
<infinisil>
wedens: Nix probably doesn't care where your symlinks in /nix/var/nix/profiles go, so yeah that might wokr
<alexarice[m]>
infinisil: The link is all about installing fonts system wide, is there no way to supply a font to one program without access to the system configurationL
<alexarice[m]>
I'm just trying to get my emacs to not depend on my system configuration at all. Perhaps there is an emacs setting on where to look for fonts
<manveru>
but the normal convention for nixpkgs would be, i think, to add a `programs.<name>.enable` option that puts the right font in your `fonts.fonts`
dasj19 has joined #nixos
<manveru>
uh... yeah, good luck with that...
<exarkun>
apparently I just failed at reading stack docs last night...
<manveru>
i've had my fair share of font issues with emacs recently :P
emptyfl4sk has quit [Ping timeout: 246 seconds]
<alexarice[m]>
one day I'll get a unicode font to work
<multun>
Is there anything anybody can do to fix the unstable channel :/ ? It hasn't been updated for 4 days and the current active revision has some major issues
<worldofpeace>
frequently, because how I'm getting how the nixos option interface is. I'm making pretty dummy options just so it looks pretty.
<infinisil>
worldofpeace: Hm I feel like the ones in programs.* are often very simple things for adding support for programs, whereas services.* usually have a bunch of additional options for configuring it or so
<infinisil>
Just based on that programs.* sounds better for this, but yeah the distinction isn't very clear in this case
<worldofpeace>
infinisil: don't look at any of the gnome modules then because the services are just one option :P. in particular here, the only think we have to do for it too be used as a program is all the config for the service and just dropping it in systemPackages.
<worldofpeace>
I have a hard time with the `program` interface because I feel like it really just boils down to interactive applications IMHO that require more setup
<infinisil>
Yeah it's used like that currently, but the term "program" really doesn't mean that specifically
<infinisil>
Maybe the current programs.* should be applications.*
Guillaum has quit [Ping timeout: 245 seconds]
selfsymmetric-pa has joined #nixos
<worldofpeace>
infinisil: I'd really like to do an RFC on option restructuring. Like a nice redesign of the interface and that's documented.
<infinisil>
I feel like there's a fundamental problem about how options can't be assigned a single category
chloekek has joined #nixos
<infinisil>
I wonder if NixOS options could be based on a tag-like system
<worldofpeace>
Well one thing is that certain programs need this special setup for like dbus, systemd, etc. This is nice, but I really expect for when a program is just going to be globally installed to just work. not sure if that's a popular or pure opinion though
<infinisil>
Ah you mean like, add it to environment.systemPackages, and it would somehow pull in the necessary nixos configuration automatically?
<{^_^}>
[nixpkgs] @FRidh pushed commit from @r-ryantm to master « python37Packages.loguru: 0.3.1 -> 0.3.2 »: https://git.io/fjjaV
<infinisil>
I was gonna say "Then go through each entry in systemPackages and include the modules if they exist"
<infinisil>
But that will give infinite recursion because of how the module system works
<worldofpeace>
I think I discussed in the past playing with stuff that packaged could define what dbus interfaces they need with passthru. But I think in the end it'd be a feature i'd like in nix if that sounds crazy
<worldofpeace>
infinisil: currently packages installed to systemPackages do get added to the dbus config automatic but we could do much better for like everything else
<infinisil>
marius851000[m]: In nix code?
<marius851000[m]>
infinisil: yes (that's a fixed hash derivation)
Fare has quit [Ping timeout: 252 seconds]
<marius851000[m]>
so I can do somthing like home.path."a/path.png" = pkgs.fetchurl {...}
<infinisil>
marius851000[m]: You want to know this even though it breaks purity?
<marius851000[m]>
It break purity ?
<infinisil>
marius851000[m]: Ahh I misread your question
<infinisil>
marius851000[m]: To get the output path of a derivation you can just look at the outPath attribute
<clever>
> let drv = stdenv.mkDerivation { name = "foo"; }; in "${drv}"
cartwright has quit [Remote host closed the connection]
Guillaum has joined #nixos
cartwright has joined #nixos
<worldofpeace>
I'm not sure why, but it just seems best. In the end a minor detail.
asymptotically has quit [Quit: Leaving]
drakonis has quit [Read error: Connection reset by peer]
pie_ has joined #nixos
iyzsong has quit [Ping timeout: 258 seconds]
v88m has joined #nixos
<exarkun>
dang, annoying, stack wants to rebuild Haskell stripe-core any time I change the derivation for a native library that a *different* Haskell library depends on...
<jonreeve[m]>
Hey if I'm trying to factor out some of my config into modules, is there a way I can get access to the variables from the parent config?
<eraserhd>
jonreeve[m]: do you mean options? Or do you mean let bindings in the caller?
<jonreeve[m]>
Let bindings
<clever>
jonreeve[m]: not possible, but you can define your own options, and then just set them like any other option
<tejok[m]>
hey there, when trying to upgrade I'm getting `The fork "dmenu2" is not maintained by upstream anymore. Please use the
<tejok[m]>
original "dmenu" instead.`, then when I try to switch to `dmenu` I get an error telling me the package is broken... is there anybody here that did face similar issue? no more dmenu ☹️ ?
<alexarice[m]>
manveru: fonts.fonts seems to work fine for my emacs, anyway thanks for the help, I at least have some unicode now, even if it looks horrific
<jonreeve[m]>
Oh nice. How can I do that? It's only just stuff like my email address that I'm putting in a lot of configs
Okinan has joined #nixos
<clever>
jonreeve[m]: this is the config for my vim, and it also defines a programs.vim.fat option, that controls if my vim config should include some heavy deps
<clever>
jonreeve[m]: i can then just programs.vim.fat = false; to trim the fat
<manveru>
alexarice[m]: yeah, i think it must be some issue with spacemacs that only sets a single font at startup
<manveru>
alexarice[m]: anw, good it works :)
<jonreeve[m]>
Awesome, thanks!
waleee-cl has joined #nixos
<{^_^}>
[nixpkgs] @jtojnar pushed 49 commits to staging-next: https://git.io/fjjwW
<zajko>
my .config/nixpkgs/config.nix contains "{ allowUnfree = true; allowBroken = true; }" ... how can I parse this? like, can I parse this, see if allowUnfree is there, if not, add it there and get an ouput?
<zajko>
I need this from script
<zajko>
something like jq command if it would be json
<clever>
$ nix-instantiate ~/.nixpkgs/config.nix --eval -A allowUnfree
<clever>
true
<zajko>
and how to add into the atribute set if ti's not here
<zajko>
I would not like to do echo >> as that's ugly
<clever>
zajko: you would need to edit it manually, theres not really an automated way to do it yet
<zajko>
just checking if allowUnfree exists, if not add it and return the whole expression
<zajko>
yeah, need it for scripting
<clever>
but you can just ignore config.nix entirely
<zajko>
what I want to do is to setup unfree for users
<zajko>
so I will put it in bashrc
civodul has quit [Quit: ERC (IRC client for Emacs 26.3)]
<infinisil>
If you are configuring the desktop from scratch, then can't you assume that config.nix is empty initially?
<clever>
zajko: you could just modify your nix exprs to ignore the users config.nix file
<zajko>
clever: you mean system wide? how so?
<clever>
zajko: in the nix file that is loading unfree software, you can just set allowUnfree = true;
<zajko>
oh, they want to use nix-env -iA
wfranzini has quit [Remote host closed the connection]
<clever>
with import <nixpkgs> { config.allowUnfree = true; };
<zajko>
from nixpkgs directly, for stuff like spotify or opera
<clever>
and you cant just have thst installed by default for them, via configuration.nix?
<zajko>
yeah, I want them to have the freenode to install it only if they want to
<zajko>
but I guess it's pretty complicated :/
<clever>
one other option
<clever>
$ echo $NIXPKGS_CONFIG
<clever>
/etc/nix/nixpkgs-config.nix
<clever>
if this file exists, then nixpkgs will entirely ignore config.nix in $HOME
<infinisil>
zajko: As I said, if you set up the computer, config.nix is probably gonna be empty, you can just write to config.nix a new string like { allowUnfree = true; }
<clever>
but thats just going to cause more confusion? when config.nix doesnt do anything
<zajko>
infinisil: you are probably right, I will probably create some sort of lock file or input a special string :/
<zajko>
kinda ugly but it will do
<infinisil>
Huh what lock file/special string?
<zajko>
like at bashrc if "mystringisthere" do nothing, if not, create allowUnfree
<clever>
zajko: how about a simple if file doesnt exist, create it with this string
<infinisil>
I'm saying just do `echo '{ allowUnfree = true; }' > ~/.config/nixpkgs/config.nix`
<zajko>
ah right, so that file is not there from skel?
<clever>
fresheyeball: run this, on a path that is failing to copy, what signatures are on it?
phreedom has quit [Ping timeout: 260 seconds]
krgn has joined #nixos
<fresheyeball>
clever: the path-info one?
<clever>
fresheyeball: yeah
<boxdot>
Has anyone tried ccache for local development with nix? I am just curious. I have a working setup, but obviously ccache does not cache that much since all flags contain ever changing hashes. :)
<fresheyeball>
error: path '/nix/store/4nlyyzbl1r1vr4jiqlnl134w5xpznh34-binutils-wrapper-2.30' is not valid
<clever>
fresheyeball: on a path that is failing to copy
<krgn>
hi. I was wondering how I could set the default gcc package globally. there are some hints out there, and I tried replacing stdenv in packageOverrides with the output of stdenvAdapters.overrideCC but that didn't work
<clever>
fresheyeball: find a path on one machine, and not another, preferably also not in the upstream cache (like `ls -l /run/current-system`) and then run `nix-store -r /nix/store/thatpath` on the other machine
<clever>
fresheyeball: or you have a 2nd nix.conf in $HOME, or you used mkForce
<fresheyeball>
I was using `imports = [ ../shared/nix.nix ]`
<fresheyeball>
to compose my nix configs together
<fresheyeball>
let me try inlining it
<clever>
fresheyeball: compare /etc/nix/nix.conf to `nix show-config | grep substituters`
<clever>
fresheyeball: do they agree?
<fresheyeball>
wait actually
colemickens has joined #nixos
worldofpeace has joined #nixos
<worldofpeace>
always colemickens :)
<colemickens>
To be clear, right now, I do not have the IWD module enabled (networking.wireless.iwd.enable), nor do I have your proposed change of adding the wantedBy to the PR.
<colemickens>
And yet, it works, so I'm not sure I understand why the wantedBy should be copied into the NM module and activated when `iwd` backend is used for NM.
<colemickens>
worldofpeace: ping, mind if we sync real quick instead of more back/forth on the PR? I can add a summary reply for the thread, but this might be easier real quick.
<fresheyeball>
how is hie in there?
<fresheyeball>
I don't even have that in my nix code anymore
Guillaum has quit [Ping timeout: 245 seconds]
<colemickens>
(I do agree with the change to pre-create the /var/lib/iwd directory though)
<clever>
fresheyeball: do you have a nix.conf in $HOME?
<colemickens>
(s/precreate directory/use the tempfiles rule/g)
<fresheyeball>
clever: no
<fresheyeball>
ok
<boxdot>
I see some dev dependencies as part of the closure (e.g. openssl-dev). Did I propagate them in a wrong way?
<fresheyeball>
nix.conf shows dunlap and no hie
<fresheyeball>
but show-config does not
<fresheyeball>
show-config is showing wrong stuff
<clever>
fresheyeball: `find ~ -name nix.conf`
<clever>
fresheyeball: or just: ls -lh ~/.config/nix/nix.conf
<worldofpeace>
colemickens: So, not sure if you saw the second half of my latest reply. But I did look over the code in networkmanager and cfg.packages adds the package to all the options iwd module already does. Reading some wiki articles etc. most distros want iwd installed global so I don't think we need to assert iwd not being enabled. My suggestion to duplicate what the module does is now null after figuring that out.
<fresheyeball>
ah!
<fresheyeball>
there is one there
<clever>
fresheyeball: so you lied :P
<worldofpeace>
colemickens: so instead we need to enable iwd when the backend is iwd
<fresheyeball>
clever: I looked to shallow
<fresheyeball>
just at ~
<fresheyeball>
that nix.conf must a relic from years ago
<clever>
2019-08-23 01:13:55< ryantm> "We do not recommend using ~/.config/nix/nix.conf unless you are a power user. It's simpler to have only one global config file."
<worldofpeace>
colemickens: sorry if I did cause any confusion. I did look over the past PRs and it seems people think this is somehow like how you can't use networking.wireless.enable and networking.networkmanager. Are there other combinations we need to guard against?
<colemickens>
worldofpeace: I am indeed confused. :) I am running this configuration now, I don't enable IWD, I just have the current PR, and NM seems to be starting IWD.
Guillaum has joined #nixos
<colemickens>
I don't have IWD enabled, I don't have the wantedBy in the NM module, and yet it works, so I'm confused why we need to chain enable IWD.
alp has joined #nixos
<colemickens>
I think that's why the original author added the specific assert to prevent the networking.wireless.iwd.enable from being enabled at the same time as making NM backend=IWD, as that would cause IWD to start and then NM to try to start it again and fail? I don't know, I haven't exactly tried it.
<worldofpeace>
I ran that and it worked fine. I think it actually relies on the service running.
<fresheyeball>
clever: its been a while
<worldofpeace>
At runtime the issue I ran into was /var/lib/iwd not exists which is what the tempfile rule should create. So I think your system just has it there.
<colemickens>
for example, that's my config, I had to disable wireless.iwd.enable to get the NM/IWD scenario to work.
<colemickens>
And I don't have your wantedBy addition to the module. And I've rebuilt/reloaded this config and restarted, so it's not some fluke of IWD just happening to be running.
<clever>
fresheyeball: the private key just goes into the default dir, run `ssh-keygen` as root, and do all defaults
<colemickens>
So I'm confused why we need to chain-enable wireless.iwd.enable or copy the wantsBy... it seems like NM is somehow doing the right thing without either of those.
<worldofpeace>
colemickens: The wantedBy thing is in the actual iwd module. looking at the service it has wants network.target
<worldofpeace>
#47217 added the wantedBy. didn't get lots of review, maybe disputable.
<colemickens>
I don't dispute that it's necessary with IWD by itself...
<colemickens>
Let me restate - let's ignore the temp files thing, I think we would either need to copy it, or chain enable wireless.iwd.enable in order to fix the temp files portion.
<colemickens>
However, on my machine, `systemctl status iwd` shows it as running, even though I have the IWD module disabled.
schjetne has joined #nixos
<colemickens>
Which makes me think that we do NOT need to copy the wantsBy, NOR enable the iwd module when using NM+IWD.
<joepie91>
!tofu
<joepie91>
,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.
<worldofpeace>
Ok I see what you're saying.
<colemickens>
BUT, I also think being able to manually enable IWD and NM (w/ IWD backend) makes conceptually more sense.
asheshambasta has quit [Ping timeout: 245 seconds]
<colemickens>
BUT2, I assume the assert was added for a reason.
<colemickens>
How about this -- I'll do some testing and see what happens if I manually enable the IWD module along with the PR. In my opinion, that would be conceptually cleaner anyway.
<worldofpeace>
"However, on my machine, `systemctl status iwd` shows it as running, even though I have the IWD module disabled." and this is with the networkmanager PR? Anything in cfg.packages is added to systemd.packages in that module.
<colemickens>
(or put another way, make the PR chain enable wireless.iwd.enable, and see what happens when something other than NM is managing IWD. Maybe it will all work out)
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
<worldofpeace>
And the iwd module also add it to systemd.packages. These module implement the sames things in identical ways.
<fresheyeball>
clever: so I generated a key on the workstation with `sudo ssh-keygen -t ed25519`
<fresheyeball>
accepted all defaults
<colemickens>
worldofpeace: well, almost
<colemickens>
worldofpeace: the IWD module explicitly adds it to wantedBy.
<fresheyeball>
and added the public key on the build server under nix.sshServe.keys
<colemickens>
worldofpeace: whereas my PR merely tells NM to use IWD and doesn't add the wantedBy.
<worldofpeace>
colemickens: right, I'm not sure if that's even needed in iwd.nix
<colemickens>
oh
<fresheyeball>
error: cannot connect to '192.168.1.10'
<fresheyeball>
error: cannot connect to 'isaac@192.168.1.10'
<fresheyeball>
on the workstation when running nix-build
justanotheruser has quit [Quit: WeeChat 2.4]
<clever>
fresheyeball: your not using the nix-ssh user
<fresheyeball>
clever: I don't understand
nschoe has quit [Quit: Program. Terminated.]
<clever>
2019-09-06 13:50:33 < fresheyeball> error: cannot connect to 'isaac@192.168.1.10'
<clever>
fresheyeball: nix adds the keys to a user called nix-ssh, not isaac
<fresheyeball>
clever: I also tried it without specifiying the user
<fresheyeball>
error: cannot connect to '192.168.1.10'
<fresheyeball>
I made this key as root
<fresheyeball>
perhaps I need to generate the key as nix-ssh?
o1lo01ol1o has quit [Remote host closed the connection]
* colemickens
is realizing he isn't quite sure how the IWD module works anyway :S
justanotheruser has joined #nixos
<colemickens>
Is uh, systemd.packages = [ pkgs.iwd ]; automatically making the service be seen to systemd?
<clever>
fresheyeball: no, you want nix-ssh@192.168.1.10
<colemickens>
It must be, maybe that's where my confusion is all based on :S
<fresheyeball>
error: cannot connect to 'nix-ssh@192.168.1.10'
<clever>
fresheyeball: can root ssh to that user&ip?
<fresheyeball>
no
<clever>
fresheyeball: double-check /etc/ssh/authorized_keys.d/ on the remote machine, is ~/.ssh/id_rsa.pub authorized on nix-ssh?
<fresheyeball>
there is only one file in there
<worldofpeace>
colemickens: yeah if you look at what cfg.packages (which iwd is in in networkmanager module) adds the package to: systemPackages, dbus.packages, systemd.packages, udev. The iwd.nix does the exact thing so they'll work identically minus the wantedBy and tempfiles rule. (minus udev rules because iwd doesn't have them)
<fresheyeball>
err sorry
<fresheyeball>
was looking at the wrong machine
<fresheyeball>
ok on the build machine
<fresheyeball>
there is a /etc/ssh/authorized_keys.d/nix-ssh
<fresheyeball>
and the contents looks good
<worldofpeace>
colemickens: all that config is correct os there's no reason to even leverage the networkmanager module when there's iwd.nix. it duplicating it will actual make maintainership harder
<worldofpeace>
(leverage meaning adding it to cfg.packages) the main backend config is needed
<fresheyeball>
clever: however the workstation has no such file
<clever>
fresheyeball: its created when you enable sshServe
<fresheyeball>
right
<fresheyeball>
but my workstation has no nix-ssh user
<fresheyeball>
the build server does though
FRidh2 has joined #nixos
<clever>
fresheyeball: check the journal on the machine with nix-serve, does it say anything in the sshd logs, when you try to ssh into nix-serve?
<clever>
nix-ssh*
<fresheyeball>
clever: how can I check the journal?
<clever>
fresheyeball: journalctl -f -u sshd
<colemickens>
worldofpeace: I think I more or less understand what you're saying, but I don't think I'm articulating my concern very well. It will be easier for me to try out a couple options and report back on what works.
<colemickens>
Ideally, I'd prefer to just chain... networkmanager.wireless.backend="iwd" triggers wireless.iwd.enable=true and it all just works. That sounds ideal, I just have a concern that it might not be that simple. But I can test and report back.
<colemickens>
(I think it sounds like that's what you want too. Remove any possible duplication of IWD settings/pkgs in the NM module, separate the concerns.)
<worldofpeace>
Was you that's exactly what I mean :D
<worldofpeace>
colemickens: I think that's good to check. don't take my word for it.
<worldofpeace>
* yse that's exactly
<colemickens>
If it weren't for the existing assert, I wouldn't think anything of it.
<colemickens>
Might just be one of those things I have to prove to myself or struggle through for the last piece to click ;)
<{^_^}>
[nixpkgs] @FRidh pushed commit from @doronbehar to master « tinc: 1.0.35 -> 1.0.36 »: https://git.io/fjjKE
<fresheyeball>
/nix/store/ja42xx9pfrv6g1qd8pm9743vjbygs2km-kassir <- on workstation
<clever>
fresheyeball: seperately, do `nix-instantiate something.nix -A something` on your project, on both machines, then do `nix-copy-closure --from 192.168.2.10 /nix/store/something.drv` to get both versions of the drv on the same machine, then use nix-diff to compare the 2
<fresheyeball>
/nix/store/891q5c3xk5ipdpx89zvwah381vnq2m9f-kassir <- on build box
<{^_^}>
[nixpkgs] @FRidh merged pull request #68194 → Drop unsupported go version 1.10 → https://git.io/fjjEK
<{^_^}>
[nixpkgs] @FRidh pushed commit from @vladdy to master « dep: 0.5.1 -> 0.5.4 »: https://git.io/fjjKV
<worldofpeace>
colemickens: the assert probably existed because when they built their configuration, it duplicated systemd units in the system path breaking it. But if they'd just enabled iwd module it would have been exactly the same.
phreedom_ has quit [Remote host closed the connection]
<DariusTh`>
So I presume I set up cachix at some point in the past
<__monty__>
Hmm, how do I install ffmpeg-full with libfdk_aac? I see that I need to pass `fdkaacExtLib=true` but I also have to pass `fdk_aac`, which defaults to `null`.
phreedom has joined #nixos
<clever>
fresheyeball: cleanSource will already ignore .git by default
tokudan has joined #nixos
<clever>
2019-08-23 01:13:55< ryantm> "We do not recommend using ~/.config/nix/nix.conf unless you are a power user. It's simpler to have only one global config file."
<clever>
DariusTh`: does that file exist?
<fresheyeball>
clever: I see that, just saying I am doing it explicilty"
<{^_^}>
[nixpkgs] @FRidh pushed commit from @dtzWill to staging « cups: 2.2.11 -> 2.2.12 (security!) »: https://git.io/fjjKA
<DariusTh`>
clever: in my nix store?
<clever>
DariusTh`: in your home dir
<clever>
2019-08-23 01:13:55< ryantm> "We do not recommend using ~/.config/nix/nix.conf unless you are a power user. It's simpler to have only one global config file."
<Acou_Bass>
hey everyone, im looking to switch over my mini server to Nixos, curious to know if anyone's ran a Piwigo instance on nixos and how that setup goes?
Izorkin has joined #nixos
<fresheyeball>
clever: I don't see how I would use that
pikajude0 is now known as pikajude
pikajude has joined #nixos
pikajude has quit [Changing host]
Izorkin has quit [Client Quit]
<clever>
fresheyeball: that will return a string with a sha1 in it, which you then just add to your drv
<fresheyeball>
clever: ok
pikajude has quit [Client Quit]
Izorkin has joined #nixos
pikajude has joined #nixos
pikajude has joined #nixos
pikajude has quit [Changing host]
pikajude has quit [Client Quit]
<fresheyeball>
clever: commitIdFromGitRepo returns a string?
pikajude has joined #nixos
<clever>
fresheyeball: yes
<pikajude>
haven't been on IRC in over a year btw, hi everyone
<Acou_Bass>
welcome ;D
Guest64138 is now known as LnL
ixxie has joined #nixos
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
evanjs| has quit [Ping timeout: 244 seconds]
<fresheyeball>
clever: that worked perfectly
<fresheyeball>
clever++
<{^_^}>
clever's karma got increased to 195
<fresheyeball>
infinisil++
<{^_^}>
infinisil's karma got increased to 138
fusion809 has quit [Remote host closed the connection]
emptyfl4sk has joined #nixos
knupfer has joined #nixos
knupfer has quit [Remote host closed the connection]
<Ericson2314>
Anyone know what controls the mount setting for /run/user/xxxx ?
<{^_^}>
[nixpkgs] @Infinisil merged pull request #67996 → lib.cleanSourceWith: Allow name to be set, optional filter, doc → https://git.io/fjpNz
<{^_^}>
[nixpkgs] @LnL7 opened pull request #68223 → rustracerd: fix darwin build → https://git.io/fjjXT
<aminechikhaoui>
Any systemd gurus around :) ? I have a server @work that started showing the following when I deploy/run a simple command such as systemctl
<aminechikhaoui>
`System has not been booted with systemd as init system (PID 1). Can't operate.`
<aminechikhaoui>
first time I see this kind of error which is really weird
<ivan>
aminechikhaoui: did systemd crash?
sigmundv has quit [Ping timeout: 244 seconds]
<multun>
Init crashing results in a kernel panic.
<multun>
So no
<aminechikhaoui>
yeah don't see anything in the logs that show any crash
<{^_^}>
[nixpkgs] @erictapen opened pull request #68224 → scribusUnstable: use harfbuzzFull instead of harfbuzz → https://git.io/fjjXt
<aminechikhaoui>
and I'm sshing to the server so sshd for instance is running
<aminechikhaoui>
seems like services are running
<aminechikhaoui>
just some kind of issues in the way systemd checks itself ? :D
ixxie has joined #nixos
<ivan>
aminechikhaoui: systemd source says it checks for /run/systemd/system/
<aminechikhaoui>
right strace shows it couldn't find that
<{^_^}>
[nixpkgs] @vcunat pushed to master « ghc on aarch64: try to work around #66277 »: https://git.io/fjjXq
<ivan>
check your journal for things touching /run or trying to delete everything?
<DariusTh`>
builder for '/nix/store/2scpxw8yic4f46qa0bz8msz7w5b3srlx-mesa-noglu-18.3.4.drv' failed with exit code 1
<{^_^}>
[nixpkgs] @simonchatts opened pull request #68226 → pythonPackages.ncclient: enable Python 3 support → https://git.io/fjjXg
boxscape has quit [Ping timeout: 258 seconds]
lord| has quit [Read error: Connection reset by peer]
<infinisil>
So if somebody wants to potentially have their PR in the next release, now's the time to mention it :)
lord| has joined #nixos
chr15 has joined #nixos
iqubic` has joined #nixos
iqubic` has quit [Remote host closed the connection]
evanjs| has joined #nixos
iqubic has joined #nixos
wfranzini has quit [Remote host closed the connection]
wfranzini has joined #nixos
ilikeheaps has joined #nixos
drakonis has joined #nixos
<{^_^}>
[nixpkgs] @volth opened pull request #68227 → network-interfaces.nix: escape '.' in interface names passed to sysctl → https://git.io/fjjXM
ozychhi has quit [Quit: WeeChat 2.5]
<codedmart>
What is the best way to set vim as nvim. Ie I installed nvim and have it working. I added an environment.shellAlias but when I git commit it expects vim.
DariusTh` has quit [Ping timeout: 246 seconds]
<codedmart>
Rather then change the git config there are other programs that rely on vim. What is the best way to global have vim run nvim?
<waleee-cl>
codedmart: you have to set the EDITOR environment variable
<codedmart>
in bash?
<waleee-cl>
yes
<waleee-cl>
... or did you ask if you can set it in configuration.nix?
<codedmart>
I was thinking in configuration.nix.
<evanjs>
man configuration.nix example is { EDITOR = "nvim"; VISUAL = "nvim"; }
<evanjs>
in environment.variables
<waleee-cl>
environment.variables.EDITOR = "nvim"
<evanjs>
But I just use home-manager :P
<waleee-cl>
ah you found it
<evanjs>
Which has programs.neovim.vimAlias and programs.neovim.viAlias
<codedmart>
Ah right thanks!
<waleee-cl>
which I guess doesn't quite cut if for git
<emily>
yeah, aliases don't help you with external programs
<evanjs>
welp. Mine seems to work
<evanjs>
e.g. help.txt Nvim
<evanjs>
inside a git commit -v .
<evanjs>
Perhaps the wording of the option is misleading
<evanjs>
description: Symlink `vim` to `nvim` binary.
<waleee-cl>
hm, isn't there some options for vim where you can set which vim-package it should use?
<evanjs>
For NixOS, I'm not certain, outside of overrides. For home-manager: programs.neovim.package and programs.vim.package.
<codedmart>
hmm... I set EDITOR = "nvim" but when I echo I still see vim?
<codedmart>
Is something else overriding it possibly?
xkapastel has joined #nixos
selfsymmetric-pa has quit [Remote host closed the connection]
<evanjs>
might need to log out and back in. I can never remember
knupfer has joined #nixos
<codedmart>
I did that :(
<{^_^}>
[nixpkgs] @FRidh pushed 26 commits to staging-next: https://git.io/fjjXh
<waleee-cl>
do you use zsh or some other shell as your interactive shell? I don't remember if environment.variable.___ will set it in non-bash shells
<codedmart>
I just use bash
WhittlesJr has joined #nixos
<waleee-cl>
... try the "nuclear" option and reboot and see if it still persists?
<emptyfl4sk>
I've just been looking into nixops and how to use it in a work environment where any developer could provision/deploy -- what's the currently preferred way of handling this? having a nixops repo and using `-s localstate.nix` and committing, using krops, or what?
<{^_^}>
[nixpkgs] @FRidh pushed commit from @jonringer to master « pythonPackages.elpy: drop package »: https://git.io/fjj1m
orivej has quit [Ping timeout: 264 seconds]
dansteeves has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
dansteeves has joined #nixos
DariusTh` has joined #nixos
<pbb>
,locate include efi.h
<pbb>
,locate bin bash
<WhittlesJr>
emptyfl4sk: Our network + modules are all in a git repo, and each developer's machine can clone the repo and create a local deployment targeting the whole network. The trick lies in ensuring that everyone uses the same channel. Only way I know of to deal with that is documentation, but maybe someone here has a better idea.
<{^_^}>
Found in packages: bash, bash_5, bashInteractive, bashInteractive_5
<{^_^}>
[nixpkgs] @alyssais pushed commit from @sikmir to master « alpine: fix links »: https://git.io/fjj1W
FRidh2 has quit [Quit: Konversation terminated!]
<emptyfl4sk>
WhittlesJr: same channel as in nixpkgs channel? I was thinking of just using one specific commit... But what do you do about the nixops state?
dansteeves has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<WhittlesJr>
emptyfl4sk: Yeah, nixpkgs channel. Like you said, you can also use a specific commit. Specifically what we do is this: we have a branch at whatever commit of nixpkgs we want, which may include PRs that aren't yet merged into master. Then we create the deployment like this:nixops create <path to network definition>.nix -d <name of deployment> -I
<WhittlesJr>
nixpkgs=https://<private github address>/<organization>/nixpkgs/tarball/<branch name>Each machine can do this. I've never encountered a need to share the "NixOps state." As long as the network definition stays in sync and everyone's on the same channel, it works just fine. I usually add the `-k` flag to `nixops deploy` to clean up machine names
<WhittlesJr>
that may have gotten deleted or renamed in the network definition.
<numkem>
Hello, I'm planning on switching to NixOS soon (already running it on a server) and was wondering if there was a way to use the packages for KDE from unstable instead. Some packages expose an option for doing just that.
tokudan has joined #nixos
jgeerds has quit [Ping timeout: 244 seconds]
civodul has quit [Quit: ERC (IRC client for Emacs 26.3)]
<numkem>
alexarice[m]:I'm gonna pull some packages from unstable since I'd like to get a recent version. Or maybe I should just use unstable. I'm really not sure what would be best
<alexarice[m]>
numkem: it is easy to switch between stable and unstable channels so you could always try unstable and if you don't like it you could switch to stable
<alexarice[m]>
but it is also possible to get some packages from stable and some from unstable
<numkem>
so far on my server I've done the trick to add the unstable channel as the "unstable" variable in the config. I haven't really played much with channel appart from that