<blibberblob>
hi all, is anyone using zoom (video chat) on nixos? They have a linux version but I can't seem to find an expression for it
bebarker has joined #nixos
<colemickens>
zoom-us
<blibberblob>
colemickens: aha, thanks! does it work well?
<colemickens>
blibberblob: I'm fortunately in a position to refuse to join Zoom meetings which I plan to do for the forseeable future.
Brandon__ has joined #nixos
rardiol has joined #nixos
<blibberblob>
no video chat at all, or are you using some other program?
bbarker has quit [Ping timeout: 250 seconds]
m0rphism has quit [Ping timeout: 256 seconds]
<colemickens>
I use Jitsi Meet (meet.jit.si, totally free no signups) and have had an impeccable experience... other than swallowing the "theres no native app" pill.
<colemickens>
I've used it with technical and non-technical folks with very good success.
<colemickens>
There are Android/iOS apps that work well. The browser app is fine.
<blibberblob>
meh, no native app isn't such a big deal as long as the web app works
<blibberblob>
that's cool, I'll give it a look
<colemickens>
cool. :) good luck!
<blibberblob>
although I've been having issues getting nixos to work with my webcam...
<blibberblob>
but that's only via the `cheese` app so I'm not sure if it's specific to that
<blibberblob>
(haven't tried other apps yet)
<colemickens>
interesting, I can't say I've had nixos-specific trouble but I do have some fairly serious problems with v4l2 and most apps choosing sub-par defaults for certain webcams.
bebarker has quit [Ping timeout: 256 seconds]
<blibberblob>
I was getting a bunch of errors about buffers being dropped and it had a super low framerate
<colemickens>
Oh, yeah, maybe try meet.jit.si (you can test by yourself). The WebRTC stacks in the browsers seem... well, idk, I think they're fairly well tested.
<blibberblob>
I'm not sure if that's the kind of thing that's configured at the app level or the kernel level
<blibberblob>
oh sweet, yeah I'll give that a try
<colemickens>
Oh, weird. I know some certain older, non-USB3 webcams that do high resolutions have usb bandwidth trouble too. I think v4l2 tries to quirk around it, but it was something else I'd discovered.
<colemickens>
Never could get a LifeCam Studio one to work right through certain (powered) USB hubs, etc.
<blibberblob>
Out of curiosity, what webcam are you using now?
<colemickens>
I splurged and bought a Brio in December.
<blibberblob>
ah I see
<colemickens>
Can't find them anywhere now though.
<blibberblob>
I'm on an old logitech C200
<blibberblob>
also my desktop is pretty old; I'm not even sure if I have usb3
LysergicDreams has quit [Ping timeout: 250 seconds]
<colemickens>
not to say those are issues for you, you might be fine.
<blibberblob>
well the webcam doesn't seem to be working in my browser
<colemickens>
Oof.
<blibberblob>
Meh, not in a super rush
<blibberblob>
I have a working webcam on my laptop...
<colemickens>
Next steps (for me) would be grabbing v4l2-utils (or whatever it's called in nixpkgs) and then poking around.
<colemickens>
FWIW, last time I fought with "it's not showing up" I had the wrong usb cable plugged in :P
<blibberblob>
what is v412 by the way?
<peelz>
video 4 linux
<cole-h>
2
<blibberblob>
I'm assuming the reason it's not working in the browser is related to the super low framerate when using cheese
<blibberblob>
ahhh it's an l not a 1 haha
<colemickens>
uh, video4linux2 or something. It's the webcam infra in linux
<colemickens>
Not sure. ANother note - the way that Firefox and Chrome enumerate webcams are different, so maybe try both.
<blibberblob>
good call
<colemickens>
(At least, when you use v4l2loopback, the settings you choose affect whether it will work in Chromium. Usually Firefox is less picky. And usually these don't affect "normal" webcams.) Just sorta dumping anything I think might help you later on.
<blibberblob>
I'm not seeing much related to v4l2 in nixpkgs... there's v4l2loopback. Do you know what package provides v4l2-utils?
<Kyndig>
oh man; I borked something up. I did a nix-garbage-collect -d and now it looks like a corrupt database of some kind (help please):
<Kyndig>
$ nix-env -q error: executing SQLite statement 'pragma synchronous = normal': file is not a database (in '/home/foo/nix/var/nix/db/db.sqlite')
<Kyndig>
$ sqlite3 /home/foo/nix/var/nix/db/db.sqlite 'pragma integrity_check' Error: file is not a database
<Kyndig>
I didn't even realize nix used a db; thought it was all managed through symlinks and what's actually on the filesystem. Hoping this is just a caching database for speed and can be reconstructed? I could try a nix-store repair, but thought I should ask first, to not mess it up more...
<Kyndig>
so I started reinstalling a package I already had installed
<peelz>
first Q/A
<Kyndig>
and it's building
<Kyndig>
everything
<Kyndig>
like maybe reindexing?
<Kyndig>
should've mentioned i'm single user compiled environment
<Kyndig>
but you could see from the nix path
<Kyndig>
in my home
<peelz>
oh you have it set up in your home? o.O
selfsymmetric-mu has left #nixos ["gone to the land of dead hiccups and extinguished light bulbs"]
<peelz>
usually the nix store is /nix/store. It's important to use that path for cached binaries... otherwise you end up having to recompile everything
<peelz>
but yeah I would assume moving the corrupted file allows nix to rebuild it, so you're probably fine
<Kyndig>
yeah; i had to compile everything
<Kyndig>
i don't own the server
<peelz>
ah
<peelz>
gotcha
<Kyndig>
the FAQ has 2 similar issues; but not exactly the same; seems my db was completely corrupt
<peelz>
did you try inspecting its contents?
<Kyndig>
seems like it's fixing itself now as it rebuilds the packages
<peelz>
or checking the size?
<Kyndig>
naw; I figured this pretty much means a corrupt db: $ sqlite3 /home/foo/nix/var/nix/db/db.sqlite 'pragma integrity_check' Error: file is not a database
<peelz>
yeah but you might want to know what happened
<peelz>
so that it doesn't happen again or so you can report it
<Kyndig>
oh; probably this crappy NFS which I have to use
Pwnna has joined #nixos
<Kyndig>
i doubt it's a nix issue
<peelz>
uhhh hmm
<Kyndig>
our admins have all kinds of crappy NFS settings
<Kyndig>
caching
<Kyndig>
it's a pain
<peelz>
>_<
romildo has joined #nixos
romildo has quit [Client Quit]
<Pwnna>
so i have a python app that calls pyside2. I'm using propagatedBuildInputs = [ (python37.withPackages ( ps: [ps.pyside2 ps.pyside2-tools ps.shiboken2] )) qt5.full ]
romildo has joined #nixos
<Pwnna>
and result/bin/whatever seems to launch the Qt app
<Kyndig>
you wouldn't believe some of the code at this place I work, we have sleep commands everywhere to work around the NFS caching latency
<Pwnna>
but if i use nix-shell, I get This application failed to start because no Qt platform plugin could be initialized. Reinstalling the application may fix this problem.
rogue_koder has quit [Remote host closed the connection]
<Pwnna>
i read some stuff about wrapQtApp, but i don't understand hwo to make it work with a python package
rogue_koder has joined #nixos
<Kyndig>
at least I have the build dir pointing to /tmp though (not NFS)
<peelz>
Kyndig: oh god...
captjakk has quit [Remote host closed the connection]
<Pwnna>
so I'm using python37.withPackages and python37Packages.buildpythonApplication.
<Pwnna>
I think I'm doing this wrong but not certain
<monokrome>
seems like I just needed to add a udev rule
rogue_koder has quit [Ping timeout: 256 seconds]
captjakk has quit [Ping timeout: 265 seconds]
rogue_koder has joined #nixos
jumper149 has quit [Quit: WeeChat 2.7]
<romildo>
If I was going to write a package update script that would work for many unrelated packages (from github, for instance) where in the tree would it reside?
<peelz>
Pwnna: I've never had to package a qt app before, but afaik qt derivations need to be called with `libsForQt5.callPackage` instead of the regular `callPackage`
<Pwnna>
well i'm not calling callPackage at all?
<Pwnna>
it's just shell.nix
<Pwnna>
i could move the basedeps.nix file's content into shell.nix just fine
<peelz>
Pwnna: yeah I know, but you might wanna look into that
<Pwnna>
and there would be no callpackage anywhere?
<Pwnna>
but how does callPackage fit into shell.nix? I thought callPackage is only used for the repository
anirrudh has quit [Quit: Leaving.]
andi- has joined #nixos
<peelz>
Pwnna: I'm not quite sure... maybe if you use `pkgs.mkShell { buildInputs = [ libsForQt5.callPackage ./your-pkg.nix ]; }` ?
<peelz>
* ^ as your shell.nix
<Pwnna>
i mean i could
<Pwnna>
but it doesn't look like i should _always_ put my-pkg.nix in a separate file?
<peelz>
well you could have them in the same file
<Pwnna>
how do i call another function in the same file?
<Pwnna>
callPackage (what do i put here?)
anirrudh has joined #nixos
<peelz>
callPackage takes a path and an argset
<peelz>
soo idk
<peelz>
try passing it a derivation instead of a path, it probably works too
<peelz>
> pkgs.callPackage pkgs.hello {}
<{^_^}>
cannot import '/nix/store/7hz2ms6mxdbfwj8bl4fcd171lly4l9wx-hello-2.10', since path '/nix/store/1g5j94c1jb7dm1qd0wdrxbax2a16qba7-hello-2.10.drv' is not valid, at /var/lib/nixbot/nixpkgs/master/repo/lib...
<energizer>
Pwnna: usually my shell.nix is `{ pkgs ? import <nixpkgs> {} }: let foo = pkgs.callPackage ./default.nix {}; in pkgs.mkShell { buildInputs = [foo]}`
<Pwnna>
huh that's an interesting pattern
<peelz>
Yeah, that's how I always do it. I've never tried putting everything in one file
<energizer>
i'm not an expert but it has been working for me
<Pwnna>
i guess the weirdness that i'm seeing is that nix-shell doesn't work (qt won't start and crashes) while nix-build's output binaryt works perfectly
<Pwnna>
in fact since development mode is active, result/bin/mycmd even works with the current development source code
<Pwnna>
so there's some sort of voodoo magic that i don't understand
<Pwnna>
I feel uncomfortable at this amount of magic as it shows that i really have no idea what is going on
<peelz>
I think qt need some sort of wrapping... try it with `libsForQt5.callPackage`
<Pwnna>
which seems like a bad position to be in
<Pwnna>
i guess i have to read up about that function
<peelz>
Pwnna, looking at <nixpkgs/top-level/all-packages.nix>, it seems like libForQt5.callPackage allows you to refer to qt5 libraries through your package's argset
<peelz>
nothing special going on as far as I can tell
<peelz>
Pwnna: that shell script might seem daunting but it's pretty simple. If you look at the bottom, you'll see this: fixupOutputHooks+=(wrapQtAppsHook)
<Pwnna>
hmm thanks i'll take a look
<peelz>
this is the "entry point" that allows it to run on your derivation's fixup phase
<peelz>
yw
nwspk has quit [Quit: nwspk]
aw has quit [Quit: Quitting.]
spacefrogg has quit [Quit: Gone.]
aw has joined #nixos
nwspk has joined #nixos
spacefrogg has joined #nixos
infinee has joined #nixos
<peelz>
Pwnna: oh one more thing, if you wanna see what happens during the build process, you can run the phases one by one in a nix-shell
<peelz>
Pwnna: nix-shell -E 'with import <nixpkgs> {}; callPackage ./. {}', then you can use the following bash functions: unpackPhase, configurePhase, buildPhase, installPhase, fixupPhase, etc.
<Pwnna>
or maybe it does, just via nix shell
<Pwnna>
ok i'm just bad at reading lol
<Pwnna>
it's getting late
<peelz>
:D
felixfoertsch has quit [Ping timeout: 272 seconds]
felixfoertsch23 is now known as felixfoertsch
<peelz>
I usually run the shell with `--pure` to make sure it doesn't get tainted by my global packages
<peelz>
oh and `type unpackPhase` to see the source of a phase
iyzsong- has joined #nixos
iyzsong has quit [Ping timeout: 272 seconds]
<{^_^}>
[mobile-nixos] @samueldr pushed 21 commits to master: https://git.io/JvAZf
<{^_^}>
[mobile-nixos] @samueldr merged pull request #122 → doc: Add options listing and other fixes → https://git.io/JvAmo
tomberek has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @orivej opened pull request #84543 → gst_all_1.gst-plugins-good: enable jack plugin → https://git.io/JvAZr
<peelz>
> vimHugeX.meta.longDescription
<{^_^}>
attribute 'longDescription' missing, at (string):293:1
<energizer>
cole-h: "The most popular clone of the VI editor" revers to VIM
<{^_^}>
[nixpkgs] @mkg20001 opened pull request #84544 → pythonPackages.pam: init at 1.8.4 → https://git.io/JvAZo
<cole-h>
Yes, I know
<cole-h>
I was hoping it would have a more useful description
reallymemorable has quit [Quit: reallymemorable]
evax has quit [Ping timeout: 268 seconds]
numbskull has joined #nixos
<numbskull>
What is the right attribute to use to depend on glibc headers, as glibc folder in the store does not contain the headers, but glibc-dev has it.
<samueldr>
glibc.dev, or am I misunderstanding?
anirrudh has joined #nixos
anirrudh has quit [Client Quit]
<hoplaahei>
So I managed to remove the files I want from firefox-unwrapped and copy them to the $out of a new package my-firefox-unwrapped. Now how can I deal with the wrapper script?
anirrudh has joined #nixos
anirrudh has quit [Client Quit]
<numbskull>
I tried nix-env -i glibc.dev, that gives error: selector 'glibc.dev' matches no derivations
<hoplaahei>
The wrapper script depends on the unmodified firefox-unwrapped.
<samueldr>
-i without -A searches through all derivations by name, so it won't find one with that name, as you saw, it has "glibc-dev" in its name
<samueldr>
-i with -A uses the attribute name, instead
<{^_^}>
[nixpkgs] @mkg20001 opened pull request #84545 → timezonemap: init at 0.4.5 → https://git.io/JvAZ7
kleisli has quit [Ping timeout: 250 seconds]
<{^_^}>
[nixpkgs] @AndersonTorres opened pull request #84546 → rs: start at 20200313 → https://git.io/JvAne
blibberblob has joined #nixos
morgrimm has joined #nixos
anirrudh has joined #nixos
<blibberblob>
Hi all, I'm seeing an issue with slack on nixos where it's unable to send notifications. The error I'm seeing in the logs is "(slack:18209): libnotify-WARNING **: 22:25:07.504: Failed to connect to proxy". Has anyone seen this before?
<blibberblob>
Notifications do work from my browser, btw
<codygman__>
I have `bson` and `mongo` in my overrides because `bson` and `mongo` are broken. Since `mongo` depends on `bson`, the `mongo` override still breaks because apparently each override item doesn't take the overrides next to it in account? How can I get the behavior where only the `bson` I define in an override is used in any case including in fellow
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « python27Packages.numpy-stl: 2.10.1 -> 2.11.0 »: https://git.io/JvAnP
<codygman__>
Put more clearly: How do I ensure not only my direct dependency on `bson` uses my override but that **all packages** which depend on`bson` also use my overriden version.
<jtojnar>
blibberblob some browers create their own notifications, whereas libnotify will use your desktop environment's mechanism. What DE do you use? does notify-send work for you?
<charukiewicz>
Hello. I'm running du -ah on my Haskell dev server's /nix/store to look for it's largest items, and it looks like there's like 3 versions of GHC 8.6.5, one of 8.8.2, one of 8.8.3. They're taking up a lot of SSD space. Is there some way I can clean this up? Running nix-collect-garbage doesn't remove them for some reason
evertedsphere has joined #nixos
<srhb>
charukiewicz: ghcWithPackages is very amenable to store optimizations where identical files are hardlinked together
<evertedsphere>
is there a way i can get, e.g. nixops to give me the "new-style" build output that i get from e.g. nix build (as opposed to nix-build)
<srhb>
charukiewicz: Try running nix-store --optimise and see how much you reclaim
<evertedsphere>
is there like some kind of hidden nixops -2 flag
<srhb>
charukiewicz: If that suffices for you, you can turn it on automatically with auto-optimise-store, see man nix.conf
<charukiewicz>
srhb: okay, I read about that but haven't tried that. I'll give it a shot. Thanks.
<srhb>
charukiewicz: It'll take a long time the first time you run it, but auto-optimise-store is quite fast from then on
<evertedsphere>
i've put my / on a zfs pool (i think that's the term) on my laptop's (quite small) ssd and it seems to be getting ~1.9x compression which is nice
<evertedsphere>
i think that's enough for me to stop worrying about my nix-store filling up
<charukiewicz>
srhb: okay, I'll run it manually and probably add that to my configuration.nix
<srhb>
evertedsphere: I'm at 2.00x exactly, which is nice. :)
<srhb>
charukiewicz: Sounds like a plan :)
<evertedsphere>
oh neat a nixos-on-zfs user! do you have any sort of docs on your setup
<evertedsphere>
i've managed to crib together something that boots and runs but that's about it
<srhb>
evertedsphere: Hmm, no, there's nothing fancy going on. ZFS on luks. I have quite a lot of datasets because I want different parameters for a lot of stuff.
<srhb>
evertedsphere: eg. steam has its own, because I don't want snapshots of games.
<evertedsphere>
right, i have yet to do anything fancy with datasets
<evertedsphere>
wait you snapshot your store? or..?
<srhb>
Not my store either, but my homedir modulo ~/Downloads and ~/.steam
<evertedsphere>
right, i see
<srhb>
Being able to go back 15 minutes in time to recover that thing I accidentally deleted is glorious.
<evertedsphere>
i have /home on a separate hdd which is snapshotted
<evertedsphere>
yes that's what got me to do it haha
<srhb>
Everything's the same pool here.
<srhb>
My personal laptop has two nvme disks, but my work one which is the only one I use currently only has one, so no mirror, sadly. :(
<evertedsphere>
i've long thought about just putting ~/Downloads on a tmpfs like i saw someone do
<srhb>
I want mine around for a bit longer, but I don't want it snapshotted.
<evertedsphere>
i'm getting a new disk which comes with one nvme but has space for one more nvme and one "big" hard disk
<evertedsphere>
*new laptop
<evertedsphere>
err. is that a 2.5" bay that i'm calling "big". i swear i know a lot about computers
<evertedsphere>
so in principle i could have two ssds + a spinning disk (or even three if i managed to find the money) at some point
<evertedsphere>
or i could set up backup to a small raspi server over my local network like i've always wanted to :)
fusion809 has joined #nixos
<srhb>
I don't have huge space requirements, it's really only /nix/store for me...
<leonardp>
srhb: do you backup your zfs pools somehow?
<srhb>
leonardp: Totally (no, but I really really shoul)
<leonardp>
haha, same here
<srhb>
I need some off-site solution but storage is so expensive ._.
<leonardp>
thats why i was asking :D
<evertedsphere>
i discovered `znapzend` poring through infinisil's setup
<leonardp>
true
<evertedsphere>
would love to set that up sometime
<srhb>
Yeah, znapzend is great. I have used it, for testing.
<srhb>
I should just pay up, really.
<infinisil>
The longer I use znapzend, the less I'm a fan of it
<leonardp>
why is that?
<evertedsphere>
idk i'm somewhat uncomfortable about sending my data to someone, however much i encrypt it /shrug
<infinisil>
It has a bunch of issues and isn't really being maintained anymore
<srhb>
It has sharp edges, fundamentally what I like is the incrementality..
<cole-h>
infinisil: Sounds like someone should write an alternative...
<infinisil>
E.g. one issue I'm currently having is that even an incremental send is like 20GB, and with my slow internet that takes a while to upload
<infinisil>
znapzend doesn't support resuming interrupted sends
<leonardp>
oh shi-
<evertedsphere>
can't you just rsync zfs snapshots or something
<evertedsphere>
are they un-diffable binary files
<infinisil>
zfs can diff between snapshots in almost zero time (compared to rsync)
<infinisil>
And zfs also supports interrupting sends and resuming them
<infinisil>
znapzend just doesn't make use of that
<infinisil>
I've been meaning to look into syncoid as an alternative
<evertedsphere>
huh. is there anything wrong with just using zfs's native send feature + a cronjob or something then
iqubic has quit [Remote host closed the connection]
<{^_^}>
#72060 (by lopsided98, 23 weeks ago, merged): sanoid: add package, NixOS module and test
<evertedsphere>
iow how do you add/remove/work with cachix caches
<infinisil>
evertedsphere: That's true. I add caches with `cachix use -d system/config <name>`
<evertedsphere>
and that overwrites the file i linked?
<evertedsphere>
hm.
<infinisil>
Yup
<evertedsphere>
so the -d switch makes cachix think the config dir is like /etc/nixos/cachix or whatever
<evertedsphere>
great. i'm honestly quite enjoying this sort of declarative setup
<infinisil>
Yeah, by default it uses /etc/nixos
<evertedsphere>
always felt like a shame to configure a nixos system with imperative commands and then have to trawl through, like, my old .zsh_history to see what the hell i did to set up my previous install the way i did
<evertedsphere>
are there any plans for a declarative alternative to nix-channel?
<evertedsphere>
are they planned to include the option to pin the entire system to a commit hash the way we currently do with nixos-rebuild -I or NIX_PATH=foo nixops deploy in my case
<evertedsphere>
instead of imperatively updating channels in a way that isn't tracked in the configuration
<srhb>
evertedsphere: flakes is most likely to make that viable
<srhb>
Every solution right now sort of involves hoop-jumping.
<evertedsphere>
mhm.
<evertedsphere>
are flakes usable rn or is the sort of stuff we've all separately stolen from clever^W^W^W hacked into our configs still better to use on a daily driver
eoli3n__ has joined #nixos
<infinisil>
Flakes are usable, but I'm not sure what the status of it is in relation to NixOS configs
<evertedsphere>
i mean the idea of your configuration.nix being a flake that i saw on the rfc
<infinisil>
(which is very WIP still, I wouldn't recommend using it (yet))
* cole-h
is still waiting for the rename to Nixus
<evertedsphere>
does that use nixops under the hood or just raw nix-copy-closure etc
<cole-h>
Not nixops, no
<infinisil>
The raw thing
rogue_koder has quit [Quit: Konversation terminated!]
<evertedsphere>
i'm torn between wanting to continue using nixops so i can Know How to Use It for when people ask me about it during interviews and just learning to use the raw stuff
morgrimm has joined #nixos
* infinisil
isn't a fan of nixops
<infinisil>
But there's been some work on improving it recently
<cole-h>
Using raw copy-closure will familiarize yourself with the underlying mechanisms
<evertedsphere>
i don't know enough about *waves hands* nix/nixos/nixops to have an informed opinion yet but it does seem like a "useful thing to know"
<cole-h>
Maybe you can make your own nixops, or nixus, or the other many Nix-deployment tools out there
<srhb>
And probably cause you to roll-your-own-nixops, like the rest of us did.
<evertedsphere>
i guess there's nothing wrong with learning both
<srhb>
:-)
<cole-h>
down the line
<cole-h>
:P
<srhb>
There must be at least like 20 copy-closure nixopsy things out there by now.
<infinisil>
I think with my nixoses/nixus I have an edge, because it allows multi-host abstractions
<srhb>
infinisil: [to e documented] :)
<srhb>
If only I could type...
<evertedsphere>
lol your auto-rollback would've saved me a couple days ago when i was modularising my config and accidentally forgot to include my ./services.nix and disabled sshd
<infinisil>
Hehe
<evertedsphere>
what timezone is most of this chat in?
<infinisil>
Yeah the autorollback has been very nice
<evertedsphere>
i feel like it's significantly less NA-heavy than most irc channels
<evertedsphere>
lots of nix in the eu
scribbler has joined #nixos
<cole-h>
PST (or PDT, I can never remember which it currently is) here
<infinisil>
Yeah definitely
<infinisil>
> PST
<{^_^}>
undefined variable 'PST' at (string):293:1
<evertedsphere>
rude
scribbler has quit [Remote host closed the connection]
<evertedsphere>
i'm +0530/IST
<cole-h>
(maybe we should -> #-chat again lol)
morgrimm has quit [Ping timeout: 265 seconds]
* evertedsphere
should get back to writing haskell instead of blaspheming in channels for untyped things
<evertedsphere>
is there a good guide anywhere to the way nixos handles nix?
<evertedsphere>
i mean the way pkgs/config etc are automagically passed around
<evertedsphere>
how imports = [...] works, all that
<evertedsphere>
i guess the "nixos module system" is the right term perhaps
<srhb>
evertedsphere: The best way is probably to read the actual module system code.
<evertedsphere>
right
<evertedsphere>
oh and on that topic is there any autocomplete at all in any editor for nixos stuff
<srhb>
evertedsphere: Or you could scrap the duck typing, which is most of it really, and try implementing extensible fixpoints like that yourself. Good exercise.
<djanatyn>
i'm trying to install the steam package. it looks like the runtime URLs were updated in nixpkgs:master 3 days ago in runtime-generated.nix. but repo.steampowered.com is down, and the mirrors defined in build-support/fetchurl/mirrors.nix don't have the latest files
<djanatyn>
i tried adding this mirror to pkgs/build-support/fetchurl/mirrors.nix in a local checkout, and to install the steam package: nix-env -iA nixpkgs.steam -I nixpkgs=/home/djanatyn/repos/nixpkgs --arg config '{allowUnfree = true;}'
<djanatyn>
however, it doesn't look like the changes i made to the mirror list are being applied
<djanatyn>
how can I update pkgs/build-support/fetchurl/mirrors.nix in nixpkgs for a nix-env steam install using a local checkout?
<djanatyn>
alternatively, is this an appropriate reason to open an issue with more details?
cartwright has quit [Remote host closed the connection]
cartwright has joined #nixos
<Amit[m]>
On a fresh install in a Xen VM, I'm failing to restart the systemd-udev-trigger service, with an odd error message:"Failed to write 'add' to '/sys/devices/virtual/input/input5/uevent': Cannot allocate memory"
<Amit[m]>
Does this seem familiar to anyone? I haven't found much useful from looking up the error message
<Amit[m]>
(I'm able to restart the same service perfectly on my laptop & desktop, which are also running NixOS)
<Amit[m]>
a seffault in dhcpcd seems bad but also not obviously related?
<clever>
Amit[m]: yeah, that is weird
thc202 has joined #nixos
linarcx has joined #nixos
mallox has joined #nixos
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
rauno has joined #nixos
<Amit[m]>
booting now without DHCP (static IP instead), that dhcpd segfault is no longer there (unsurprisingly), but the uevent message still are there every time I try to restart the systemd-udev-trigger service
LysergicDreams has quit [Ping timeout: 256 seconds]
<clever>
Amit[m]: what if you run `nix-store --verify --check-contents` ?
<Amit[m]>
clever: checking. and to clarify, it runs on boot successfully, it's just when I restart it (e.g. on a nixos-rebuild switch, or manually) that it fails
<{^_^}>
[nixpkgs] @invokesus opened pull request #84559 → perlPackages.RPM2: init at 1.4 → https://git.io/JvAlq
<Hi-Angel>
Ah, okay, they're alphabetically sorted, I scrolled forward to git. Though the site doesn't show list of files :c
domenkozar[m] has quit [Ping timeout: 246 seconds]
morgrimm has quit [Ping timeout: 265 seconds]
domenkozar[m] has joined #nixos
<pjt_014>
Hi-Angel: prot tip: search ^git to find results that start with git, not just contain it
rotaerk has quit [Ping timeout: 246 seconds]
init_6[m] has quit [Ping timeout: 246 seconds]
thefloweringash has quit [Ping timeout: 246 seconds]
rotaerk_ has joined #nixos
<Hi-Angel>
Oh, thanks, that's cool, I've never seen an online site for package view with regex support
<Hi-Angel>
s/an online site/a website
chagra has quit [Quit: WeeChat 2.7.1]
dermetfan has joined #nixos
apeyroux` has left #nixos ["ERC (IRC client for Emacs 26.3)"]
init_6[m] has joined #nixos
thefloweringash has joined #nixos
nikola2 has joined #nixos
hax404_ has joined #nixos
<pjt_014>
the 'nix search' command does regex too, If you didn't know.
das-g[m]1 has joined #nixos
das-g[m] has quit [Ping timeout: 246 seconds]
hax404 has quit [Remote host closed the connection]
nikola1 has quit [Ping timeout: 246 seconds]
<Hi-Angel>
pjt_014, yeah, command line utilties typically support regexes, I'm wondering about websites functional
<Hi-Angel>
It's like unexpected functional, I think some obvious pop up on the website may help to new people
cr4y1 has quit [Remote host closed the connection]
<tyrion-mx>
Hola, `nixos-rebuild switch` is not picking up changes in my configuration.nix ... this never happened to me before. Somebody has any hints on how to proceed?
cr4y1 has joined #nixos
<tyrion-mx>
It only shows "building Nix... \ building the system configuration..." and then exits
civodul has joined #nixos
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Guest51734 is now known as palo
palo has quit [Quit: WeeChat 2.7.1]
zupo has joined #nixos
palo has joined #nixos
opthomasprime has quit [Remote host closed the connection]
<energizer>
what is the difference between `nix cat-store` and `cat`?
opthomasprime has joined #nixos
<LnL>
a nix store isn't necessarily /nix
<clever>
energizer: `nix cat-store` can work even if you `--store s3://bucket` or `--store ssh://host`
<Hi-Angel>
Hey, I'm not on Nix OS, could please someone tell under what directory did your "git" package put a script `diff-highlight`? Background for the question: I'm arguing with someone who says that the best way to find the file is by using `find /usr`, and I'm trying to find an example where this script is not under this dir. I think on Nix OS it may not be there.
<Hi-Angel>
clever, what `dirname` and `realpath` in your example are? :)
<hyper_ch>
hmmm, something has changed in samba since 20.03 git
<clever>
Hi-Angel: realpath is a program to turn a symlink into an absolute path, dirname is a program to give the dir that a file is in
<hyper_ch>
I got lots of aborts and locks now
thibm has joined #nixos
linarcx has joined #nixos
<boothead>
Hey folks - what's the best way to get and xmonad and gnome session set up? I have windowManager.default = "xmonad" and gnome enabled but desktopManager.default = "gnome" doesn't seem to work...
linarcx has quit [Client Quit]
<Hi-Angel>
clever, okay, thanks. FYI, this doensn't work on Archlinux and Ubuntu, because there're other paths. On Arch it's `/usr/share/git/diff-highlight/diff-highlight` and Ubuntu it's `/usr/share/doc/git/contrib/diff-highlight`
<energizer>
boothead: the value should be a package not a string
<clever>
Hi-Angel: this will search, relative to where the git binary is
<boothead>
energizer - ah that makes sense. What should I put in there? I want gnome running with all services, but xmonad as the wm...
<Hi-Angel>
clever, heh, yeah, that probably gonna work :)
<energizer>
boothead: i have `services.xserver.displayManager.sddm.enable = true;`
<energizer>
you can do something similar with windowManager and desktopmanager
andydarcyjewell2 has joined #nixos
<andydarcyjewell2>
Hey #nixos
<boothead>
I had lightdm as the displayManager, when I switched over to gdm, i didn't seem to have a gnome+xmonad session option. Does sddm have that?
<andydarcyjewell2>
I'm new to Nix, just installed it in order to get Factor (language) working. It's installed, but the GUI isn't working - complains about missing GTK this and that
<energizer>
boothead: i think you can set all of those independently
<andydarcyjewell2>
When I installed Factor from source on stock Debian, I had to install several "dev" packages to get it to work, but I can't work out how to do that under Nix.
<energizer>
boothead: like `services.xserver.windowManager.xmonad.enable = true;`
<{^_^}>
[nixpkgs] @dguibert opened pull request #84573 → darcs: compile with ghc 8.8 → https://git.io/JvAB0
cybrian has joined #nixos
<boothead>
energizer, yes I think you're correct - just wondering what the right way is to have a gnome + xmonad set up (I already have desktopManager.gnome3.enable and xmonad.enable)
<boothead>
I get a gnome desktop and in the login screen I don't have the expected gnome+xmonad option for the session
o1lo01ol1o has joined #nixos
opthomasprime has left #nixos [#nixos]
<energizer>
boothead: i do my window manager config with home-manager rather than configuration.nix so i haven't tried that myself. perhaps the services.xserver.desktopManager.gnome3.flashback.customSessions.*.wmCommand option is relevant (man configuration.nix)
<boothead>
Ah - been meaning to look at home manager... Do you have a link to your set up public (or anything else you could point me at)?
aswanson has quit [Quit: WeeChat 2.7.1]
<{^_^}>
[nixpkgs] @filalex77 opened pull request #84574 → meilisearch: init at 0.9.0 → https://git.io/JvABr
<energizer>
mine isn't public but a lot of people have public ones
<evertedsphere>
what's with store paths like this?
<evertedsphere>
checking contents of '/nix/store/5vvdi8ni626jp111hw2zsar6a42xbrh4-wm92mnm0fz045xrhjm715ck4vi926ij5-ixx9wbs9sr72jxlx72h9wkh0gkbc322s-17h9mdgqxb5hb9razx8m9i1hip0hc20a-source'
<evertedsphere>
i had to actually once use builtins.path or something like that to shorten the name so as to not hit the filename length limit
<evertedsphere>
i think its like <hash>-(<hash-(<hash>-(<hash>-name))) or something?
<evertedsphere>
it might be from incorrectly trying to copy clever's nixpkgs=/run/current-system/nixpkgs + system.extraBuilderCms = "ln -sv ${pkgs.path} /run/current-system/nixpkgs" trick
<evertedsphere>
extraBuilderCmds*
<evertedsphere>
but i had to give pkgs.path a shorter name using builtins.path
<Athas>
I'm having some trouble with 'pass' on my NixOS. It can encrypt passwords, but not decrypt them, as it fails with 'gpg: public key decryption failed: No pinentry'. I can manually decrypt the passwords with 'gpg --decrypt --pinentry-mode=loopback'.
<Athas>
It looks like it's unable to ask me for my password, unless I pass that --pinentry-mode. Has anyone here seen anything like this before?
<evertedsphere>
perhaps you need `pinentry` installed?
<Athas>
I have 'pinentry' installed.
<evertedsphere>
ah, okay
rauno has quit [Remote host closed the connection]
<andydarcyjewell2>
Maybe I should re-phrase my question: when a program tries to dynamically bind to a library, it needs the C header file to know what endpoints are exported. How do you handle that with Nix?
ottidmes has joined #nixos
psy3497_ has joined #nixos
<psy3497_>
I'm trying to build a Haskell project with the latest revision of nixos-19.09, it has base at 4.12, but for some reason the `hslogger` fails to compile
<andydarcyjewell2>
If so, surely it wouldn't throw a load of GTK errors? Yes I read a bit about that, but it didn't seem to apply. New to Nix, so still trying to work it out.
mallox has quit [Quit: WeeChat 2.7.1]
<psy3497_>
Perhaps there is something more compilcated inside my build where a different hslogger is being referenced I wonder? I admit I did not try to build hslogger in isolation
<{^_^}>
[nixpkgs] @alyssais pushed to master « gitAndTools.git-cinnabar: init at 0.5.4 »: https://git.io/JvARG
<srhb>
psy3497_: Seems likely.
<psy3497_>
I will first try that, thanks
<thibm>
andydarcyjewell2: for instance if I do nix-shell -p glib, then $NIX_CFLAGS_COMPILE contains "-isystem /…glib-2.60.7-dev/include"
<srhb>
psy3497_: Great :)
<srhb>
psy3497_: (I checked like this in a checkout on that commit: nix-build -E 'with import <nixpkgs> { config = {}; overlays = []; }; haskellPackages.hslogger')
<srhb>
Er, where <nixpkgs> is a checkout of that commit, I should say
fendor has joined #nixos
<psy3497_>
srhb: I was using ghc882 for the build
<psy3497_>
it looks like ghc865 is the default one in nixos-19.09
<srhb>
Ah, that would explain.
<psy3497_>
I am not supposed to use ghc882 for this LTS?
<srhb>
psy3497_: For nixpkgs haskellPackages, you're supposed to use the version in haskellPackages itself, unless you want to forgo the binary cache.
<srhb>
psy3497_: As in haskellPackages.ghcWithPackages
<andydarcyjewell2>
thibm: thanks, but it's not compiling in that sense; Factor (the package) is dynamically binding to the libraries, like for instance Python's ctypes module does
<psy3497_>
srhb: I see. I still wonder why the error on base's version while it does seem to be inside the bounds
palo has joined #nixos
<psy3497_>
I found in configuration-ghc-8.8.x.nix:
<psy3497_>
# Disable GHC 8.8.x core libraries.
<psy3497_>
base = null;
<{^_^}>
[nixpkgs] @doronbehar opened pull request #84577 → nodePackages.clean-css: rename to clean-css-cli → https://git.io/JvAR4
<srhb>
psy3497_: Base is bundled with ghc along with some other packages
<psy3497_>
Ah
<psy3497_>
I see, so actually it does not matter that I saw base_4.12 in haskellPackages.nix. I will see what base version is bundled with ghc882
<andydarcyjewell2>
thibm: When Factor loads modules (written in factor) they can call out to arbitrary libraries, and that's where it's failing. For instance, to run its wiki app, it needs to bind to sqlite, but is failing to find it, despite it being installed. I think the problem with GTK may be similar.
<psy3497_>
srhb: Cool I found it, ghc88x bumped base to 4.13
<psy3497_>
srhb: does the `base` dependency ignores jailbreak? I tried to see if hslogger would build if I jailbreak it. But it kept complaining about the dependency version
apeyroux` has joined #nixos
<thibm>
andydarcyjewell2: I see. It seems that this cannot really run seamlessly with Nix… You cannot find arbitrary libraries like that. There's just mechanism to find them when building derivations
<thibm>
Neither the C headers of the library itself
<andydarcyjewell2>
How does Nix support other languages then?
<srhb>
psy3497_: I wouldn't have expected it to ignore the jailbreak. You were jailbreaking hslogger, right?
<psy3497_>
Yes
<psy3497_>
srhb: Yes
<psy3497_>
srhb: it even wrote it in the output
<srhb>
psy3497_: You can try running cabal2nix --jailbreak by hand on a cabal file
<thibm>
andydarcyjewell2: what's usually done is generating an environnement with everything you need in scope (compiler, libraries, etc), then build the project.
<srhb>
psy3497_: Maybe the error was actually coming from some other dependency?
<andydarcyjewell2>
Python, perl, php.. even ones not stating in "p" do this kind of thing
<psy3497_>
srhb: During the build it says: "Run jailbreak-cabal to lift version restrictions on build inputs."
<srhb>
psy3497_: Otherwise I don't know, I don't see how it could ignore it if the bounds are removed correctly.
<psy3497_>
Configuring hslogger-1.2.12...
<psy3497_>
CallStack (from HasCallStack):y
<psy3497_>
die', called at libraries/Cabal/Cabal/Distribution/Simple/Configure.hs:1022:20 in Cabal-3.0.1.0:Distribution.Simple.Configure
<psy3497_>
srhb: and then more lines that I will abbreviate
<srhb>
psy3497_: You probably need to inspect the output of cabal2nix --jailbreak, as said
<thibm>
andydarcyjewell2: if I want to build a python project that uses ffmpeg, I need to create an environment with python and ffmpeg. Nix helps you do that, and normally sets all environment variables such that your tooling find the appropriate stuff (starting with $PATH)
<psy3497_>
srhb: I see, thanks for the help though! You helped me to make some progress
<thibm>
andydarcyjewell2: so I think it's not easy to use an IDE that assumes it can find libs at runtime. But there may be a way to tell your IDE where to find the libs you need to build your project
apeyroux` has quit [Remote host closed the connection]
<andydarcyjewell2>
thibm: is there something I can read that discusses how to approach that?
<thibm>
Hum… Is it possible to build Factor projects from command line (usually easier to understand what's going on). Is there a Nix abstraction for this? (Like buildGoModule for go, for instance)
<andydarcyjewell2>
thibm: Factor is it's own command-line
xelxebar has quit [Remote host closed the connection]
<psy3497_>
srhb: it looks like setting the small_base flag should make it compile
lordcirth has quit [Remote host closed the connection]
kleisli has quit [Remote host closed the connection]
<srhb>
psy3497_: You're welcome :)
<thibm>
andydarcyjewell2: maybe try mkShell (https://nixos.org/nixpkgs/manual/#sec-pkgs-mkShell). You need to find out how factor looks for its libraries. Maybe there's a environment variable, if you're lucky
kleisli has joined #nixos
<andydarcyjewell2>
thibm: ok, will look at that, but it doesn't seem particularly sustainable, as each time I try to use a new facility in Factor, I'll have to manually hack in the new libraries it needs
xelxebar has joined #nixos
chagra has joined #nixos
balsoft_ has joined #nixos
<andydarcyjewell2>
thibm: I'll look at mkShell too then, thank you
balsoft_ has quit [Client Quit]
pbogdan has joined #nixos
<sshow>
I want to get the two first elements of a list. Slice, splice, shift? builtins.tail was the closest I could find :P
<sshow>
I have come to find that as a source of confusion. Three different docs that covers each other without overlapping
<srhb>
Yes, everyone is confused by that, I think. :)
<srhb>
The division makes perfect sense once you understand why it's divided like that, but by then it's too late. :P
<sshow>
I *do* know, however, that I could contribute myself ;)
<srk>
> lib.take 2 [ 1 ]
<{^_^}>
[ <CODE> ]
<srk>
mkey
o1lo01ol1o has joined #nixos
mounty has joined #nixos
<sshow>
srhb++
<{^_^}>
srhb's karma got increased to 95
<sshow>
thibm: ++
<sshow>
thibm++
<{^_^}>
thibm's karma got increased to 1
chloekek has joined #nixos
<srhb>
> seq
<{^_^}>
<LAMBDA>
<srhb>
oops
<evertedsphere>
pls to not kill {^_^}
<thibm>
srhb: used Nix for years, I still do not understand the builtins.* vs map/toString/… things
o1lo01ol1o has quit [Ping timeout: 240 seconds]
captn3m0 has quit [Ping timeout: 240 seconds]
captn3m0 has joined #nixos
<{^_^}>
[nixpkgs] @vbgl pushed to master « jackline: use OCaml 4.08 »: https://git.io/JvA0o
jbrock has quit [Ping timeout: 240 seconds]
<srhb>
thibm: builtins are primitives provided directly by Nix. library functions are implemented on top of those (and are sometimes just compatibility aliases(
<thibm>
I know for builtins/lib. But map, toString and other are accessible outside of builtins.*
<evertedsphere>
would it be unwise to enable dedup on a zfs pool (or is the term dataset?) containing only /nix/store
<evertedsphere>
i think it might work like a sort of automatic nix-store --optimise
<srhb>
evertedsphere: Yes. always
<srhb>
evertedsphere: You can stop the sentence by "would it be wise to enable dedup on a zfs" to which the answer is no, never.
<evertedsphere>
hah
<srhb>
It's a very short distance from opt-in data-loss.
<evertedsphere>
wow, really, TIL
<evertedsphere>
does it overzealously "dedup" and just delete files for no reason
<hyper_ch>
dedup.... ufff.... :)
<hyper_ch>
dedup works on block size, not files
<evertedsphere>
i've just been thinking about what cool zfs stuff i could be making use of on my home machine apart from just compression and encryption (i went with zfs encryption instead of luks)
<evertedsphere>
or things i could tweak
<srhb>
evertedsphere: You're probably best off googling why it's a bad idea, I only remember "no, seriously, don't do it" :)
<evertedsphere>
right.
<evertedsphere>
for one i think i read somewhere that you want to make fine-grained datasets and tweak settings on each
<evertedsphere>
like a dataset for things like big videos/movies with a larger recordsize parameter (iirc?)
<Valodim[m]>
the main argument against dedup I remember was an a very high increase in required memory
knupfer has quit [Quit: knupfer]
knupfer1 has joined #nixos
knupfer1 has quit [Client Quit]
knupfer has joined #nixos
<evertedsphere>
yeah isn't it like 5GB/TB up from the usual 1GB/TB number
<hyper_ch>
evertedsphere: there's some good zfs videos
<srhb>
evertedsphere: By the way, for automatic nix-store --optimise, there's auto-optimise-store, which I heartily recommend, in case you missed it earlier.
<srhb>
evertedsphere: That will efficiently dedup on registration of all new store paths.
<zimbatm>
bqv[m]: that happens when deleting branches :)
<bqv[m]>
ah
<bqv[m]>
i still sometimes do the old-fashioned push origin :<branch>
<thibm>
srhb: what's the cost of enabling auto-optimise-store?
morgrimm has quit [Ping timeout: 250 seconds]
<srhb>
thibm: Each file of the new store path is hashed, its hash checked for existense in /nix/store/.links, if it exists, it's hardlinked to that.
<srhb>
thibm: So you trade some (minimal, in my experience) time for space savings.
<srhb>
thibm: For some workloads (say, haskell development work) the space savings are quite huge, because the derivations tend to produce many identical files.
<thibm>
srhb: OK, seems a good solution
<evertedsphere>
most of my nixos use is indeed haskell so i'll give it a try
<srhb>
evertedsphere: Note that you'll have to do a one-time nix-store --optimise too, to get existing paths hardlinked.
<thibm>
I'm interested because I build several GB derivations (each) on a machine, with multiple outputs, and some of these outputs have a lots of file in common. And I don't have much disk space.
boothead has quit [Remote host closed the connection]
<bqv[m]>
i originally didn't have optimize on locally cause i thought it would be expensive on my home pc. turned it on later and the initial one took half an hour
<thibm>
Is it possible to enable it on a current installation? I guess first store registration will be long
<bqv[m]>
add nix.autoOptimizeStore = true;
laduke has quit [Ping timeout: 265 seconds]
<bqv[m]>
see also: nix.gc, nix.optimize
<thibm>
Yes, I was just wondered if changing the config would work. Hope so
<srhb>
thibm: auto optimise will not traverse the whole store iirc, you need to manually run nix-store --optimise for that.
<thibm>
(It's not of a Nixos. It's on a old CentOS machine with a custom Nix path because I'm no root. It's terrible.)
<thibm>
srhb: OK, thanks
<srhb>
thibm: Yeah, man nix.conf
<thibm>
Yeah I saw. But no mention of turning it on/off
mallox has joined #nixos
<srhb>
It's fine. :)
<bqv[m]>
auto-optimise-store = true
<bqv[m]>
add that as a line in nix.conf
<{^_^}>
[nixpkgs] @peti pushed 2 commits to haskell-updates: https://git.io/JvAuM
<thibm>
bvdw: yes I saw. Just saying this is not on NixOS. Thanks anyway*
<thibm>
bqv[m] (sorry)
<srhb>
THe only thing I don't know is if a custom store location might be an issue, but I haven't heard of that being the case.
<{^_^}>
[nixpkgs] @AndersonTorres pushed 2 commits to master: https://git.io/JvAou
<bqv[m]>
exarkun: 👍️
<exarkun>
wonder what I'm doing wrong, then.
<gchristensen>
given your overlay starts with: final: super: ... are you getting packages from final or super?
<bqv[m]>
have you ordered the overlays in the right direction?
<exarkun>
I probably need to create a smaller example to be able to answer those questions... I can't hold all of these pieces in my head at once.
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
orivej has quit [Ping timeout: 256 seconds]
<yorick>
adisbladis: I'm trying to poetry2nix, but some thing it generates doesn't find setuptools_scm_git_archive
turion has joined #nixos
<yorick>
is that a known thing?
<{^_^}>
[nixpkgs] @doronbehar opened pull request #84599 → nodejs: use python3 if possible → https://git.io/JvAoy
<gchristensen>
yorick: adisbladis is afk at the moment, probably should open an issue / direct your question to other people :)
<yorick>
okay :)
troydm has quit [Quit: What is Hope? That all of your wishes and all of your dreams come true? To turn back time because things were not supposed to happen like that (C) Rau Le Creuset]
aveltras has quit [Quit: Connection closed for inactivity]
<exarkun>
okay, I guess maybe I understand. I guess I thought maybe I understood before, too. I don't really remember what I was thinking when I decided to put these overlays in the old, broken order instead of the new, working order. Maybe I didn't think about it at all since apparently it worked either way before and only some other change recently made elsewhere made it necessary to get the order right...
<{^_^}>
[nixpkgs] @invokesus opened pull request #84600 → perlPackages.FileMap and dep → https://git.io/JvAKz
<exarkun>
I do remember why some of this code uses super instead of final, though, and I even left copious comments explaining it
<{^_^}>
[nix] @Ninlives opened pull request #3477 → `nix run` using $SHELL as default command → https://git.io/JvAKw
<gchristensen>
in my example, self and super are replaced with final and super
<thibm>
Is it recommended to use mkEnableOption or mkOption for new services? I remember some discussion about reducing number of functions to help newcommers read files
<thibm>
s/services/modules
<bqv[m]>
i would use mkEnableOption. the former can be converted th the latter easily en-masse. the reverse is far less trivial
mauli has quit [Ping timeout: 260 seconds]
<thibm>
True. Was wondering if any compromise was found. If not, you're right. Thanks
vika_nezrimaya has joined #nixos
jakobrs has joined #nixos
<jakobrs>
How does the live usb image get an IP address if it's not running dhcpcd?
<morgrimm>
Quick question folks - is there a best practice for what phase stuff should occur? For example, installing binaries in $out I assume in the postInstall phase, etc.
<morgrimm>
Right now I'm moving my binaries in the build phase, which I assume probably isn't great
<evils>
if nothing else is installing stuff, i think installPhase is the right place to do that
<thibm>
morgrimm: postInstall or installPhase?
<thibm>
``Phases'' in nixpkgs' manual explains each phase role. But you could do what you want. As you said, they are best practices
<morgrimm>
I see some derivations moving completions in the postInstall, or flat out moving binaries in postInstall
<morgrimm>
My package source doesn't need to be built, just pulled and installed, so...
<teto>
morgrimm: in which case you can also use a plain `runCommmand`
chagra has quit [Quit: WeeChat 2.7.1]
<Orbstheorem>
morgrimm: if your package is just copying stuff you may want to consider copyToNixStore
<Orbstheorem>
Or plain runCommand as teto mentioned ^^
<morgrimm>
copyToNixStore? that's one I haven't run into yet
<Orbstheorem>
> copyToNixStore
<{^_^}>
undefined variable 'copyToNixStore' at (string):296:1
<thibm>
If you don't need build, you can skip buildPhase by setting dontBuild = true, and keep unpackPhase+installPhase (+ others)
<Orbstheorem>
> copyToStore
<{^_^}>
undefined variable 'copyToStore' at (string):296:1
* Orbstheorem
is pretty certain something of the sort exists x)
<thibm>
because it's usually useless. ${./path/to/files} works
<morgrimm>
I'm also fetching the src from GitHub, so I don't know if copyToStore would work
<morgrimm>
copyPathToStore*
<morgrimm>
runCommand likely would though
<thibm>
you can also just use fetchFromGitHub
<thibm>
This will put the files in the store
<morgrimm>
That's what I'm using now, yeah
<thibm>
no need for src = fetchFromGitHub {} + cp $src $out
<thibm>
OK, perfect
<morgrimm>
My derivation is just fetching from GH, and then I'm using postInstall to wrap the main executable script
<morgrimm>
and put it in $out;bin
<morgrimm>
$out/bin*
mauli has joined #nixos
<morgrimm>
because I don't really need any of the other stuff it comes with. I'm trying to decide how best to actually structure this. The way it is now, $src won't get gcd
<morgrimm>
Because the lib dir it needs to run needs to exist somewhere - I wonder if I'd be better served putting it explicitly in $out as well
jakobrs has quit [Quit: WeeChat 2.7.1]
<morgrimm>
instead of leaving it in the store
dermetfan has quit [Ping timeout: 272 seconds]
iyzsong- has quit [Read error: Connection reset by peer]
<Orbstheorem>
When you cp stuff inside the nix store, is there anything deduplicating content using hardlinks?
amosbird_ is now known as amosbird
<bdju>
how do I see the list of generations that grub shows me but from my booted environment?
<morgrimm>
Orbstheorum: Not explicitly, no. I'm just fetching from GH, then running a postInstall to install shell completions and a wrapped binary into $out
<ottidmes>
bdju: easier to just check /boot/grub/kernels
<bdju>
okay I guess I did say grub but I'm thinking now that I'm not actually using grub, rather systemd-boot or something
<Orbstheorem>
morgrimm: I didn't mean in your specific case, I meant in general ^^
<DamienCassou>
I've tried to install Slack on NixOS. But when clicking on the login button, Emacs is opened instead of Firefox. It's apparently related to KDE's KIO but I don't know how to configure that
<thibm>
Orbstheorem: yes and no, see `nix-store --optimise' or Nix `auto-optimise-store' option
<est31>
bdju: for me it's inside ls /boot/EFI/nixos/
<est31>
also see files in /boot/loader/entries/
<bdju>
est31: thank you! that does show what I w anted
<bdju>
I've been bounching between "latest" and 4.19 kernels because new stuff is very broken and I wanted to see what the version of the newer kernel was
LysergicDreams has joined #nixos
<bdju>
5.5.13 was what I last used it seems
<bdju>
my dock kept disconnecting for several seconds causing sound and inputs issues. it was pretty maddening
<ottidmes>
bdju: one way to see them regardless of boot loader would be to check /nix/var/nix/profiles any system-*-link in there has /kernel pointing to their version of the kernel used
hlavaty has joined #nixos
<bdju>
ottidmes: thanks, that seems to sort of work, but it's hard to check the kernel for multiple gens at once as far as I can tell. you have to get down to a specific one with your ls -l to see the kernel
<ij>
is it possible to evaluate the full set of config attributes for a nixos config definition?
<ottidmes>
bdju: just because I could: find -L /nix/var/nix/profiles -maxdepth 2 -path '*/system-*-link/kernel' -exec realpath {} + | uniq | cut -d- -f3- | cut -d/ -f1
<bqv[m]>
finally realised the reason my gpg-agent has been broken for weeks is that i have it enabled with nixos AND home-manager
<ottidmes>
ij: what's the use case?
<ij>
ottidmes, I'm modifying the nixos live iso creation config to suit my needs and I've imported something that changes autoLogin
<ottidmes>
ij: is nixos-option not nicely suited for that?
<ij>
I should be auto-logged in as with my default config, but I suspect some import is changing that. If I knew what the output is, I could go through the nixpkgs/nixos sources and find what.
<ij>
I haven't used it – I'll go see what it does.
<ij>
it's probably the right tool
LysergicDreams has quit [Ping timeout: 260 seconds]
bqv has quit [Quit: WeeChat 2.7.1]
LysergicDreams has joined #nixos
<ij>
the examples from the manual aren't working :(
<morgrimm>
What's the proper way to specify a GH src as the src for a derivation? I'm specifying it as part of the attrset, but I think I'm doing it wrong
<morgrimm>
Since I expect it to unpack that src into the workdir for the phases
<morgrimm>
But I can only refer to it using $src, so it's definitely not doing what I think it is
<morgrimm>
I guess it's putting that into a separate store path, and then just running the builder in another
zeenk has joined #nixos
<bdju>
ottidmes: Wow, that is neat. Thank you for that. Maybe I'll save it as an alias.
<ottidmes>
morgrimm: it should put it in /nix/store and then unpack it to the build directory, if you want to know what youre doing wrong, share your derivation with a paste
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<morgrimm>
ottidmes: The problem with copying parts directly into $out is it relies on the structure in its repo (afaik), and blindly copying files into $out will end up polluting ~/.nix-profile if it's installed in the env
ayuce has joined #nixos
_e has quit [Quit: WeeChat 2.7.1]
logzet has joined #nixos
linarcx has quit [Read error: Connection reset by peer]
johnjay has quit [Ping timeout: 256 seconds]
<chr0n1c>
You are invited to take part in an important scientific lecture and chat with Dr Gamme Martin @ irc.butt.es #GAMME
<plato[m]>
A haskell package needs the pkg-config files of the "brotli" package. They seem to be in the "dev" output of the package. How do I tell the hackage derivation that?
<morgrimm>
ottidmes: How so? Also, as an aside - what did you change to have it correctly unpack into the working directory?
<ottidmes>
morgrimm: nothing at all, it does so automatically for stdenv.mkDerivation
revtintin has quit [Ping timeout: 240 seconds]
<ottidmes>
morgrimm: give me a moment, almost done fixing it
lordcirth has quit [Remote host closed the connection]
lordcirth has joined #nixos
fenedor has joined #nixos
mauli has quit [Ping timeout: 265 seconds]
linarcx has joined #nixos
<ldlework>
b42: will you submit a pr for it?
fendor has quit [Ping timeout: 264 seconds]
johnjay has joined #nixos
justanotheruser has quit [Ping timeout: 265 seconds]
<thibm>
Is there anything special about putting glibc in a derivation dependency?
<thibm>
I need to add `-I${glib.dev}/include/glib-2.0 -I${glib.out}/lib/glib-2.0/include' for the file to compile, otherwise it does not find headers. Is this normal? (Maybe the source file is just broken)
turion has quit [Quit: Leaving.]
<{^_^}>
[nixpkgs] @mmilata opened pull request #84612 → pokerth: use mkDerivation from Qt → https://git.io/JvA16
<{^_^}>
#84612 (by mmilata, 50 seconds ago, open): pokerth: use mkDerivation from Qt
mauli has joined #nixos
<ldlework>
b42: you fricken champ
copy` has left #nixos ["Kicked by @appservice-irc:matrix.org : Idle for 30+ days"]
<thibm>
To sum up: mypackage -> vips -> glib (which is in vips' propagatedBuildInputs. But building mypackage leads to error: `glib.h: No such file or directory' in vips-8.9.1-dev/include/vips/vips.h
contrun[m] has left #nixos ["Kicked by @appservice-irc:matrix.org : Idle for 30+ days"]
mauli has quit [Ping timeout: 264 seconds]
Victor[m] has left #nixos ["Kicked by @appservice-irc:matrix.org : Idle for 30+ days"]
justanotheruser has joined #nixos
codyopel has left #nixos ["Kicked by @appservice-irc:matrix.org : Idle for 30+ days"]
anirrudh has joined #nixos
anirrudh has quit [Client Quit]
<morgrimm>
ottidmes: That works! That's not an approach I thought of
<thibm>
Nevermind, adding the flags through pkg-config do the trick (pkg-config --cflags -- glib-2.0). $NIX_CFLAGS_COMPILE is not sufficient here. Sorry for the noise
<morgrimm>
No, that's how I'm getting around it, but I'm putting that in the derivation - I'd rather use that in the cli if I can
<morgrimm>
ottidmes: That's exactly what I was looking for, yeah. Thank you :D
edomfjp[m] has quit [Quit: Idle for 30+ days]
<bqv[m]>
that's what i meant lol..
<bqv[m]>
ok
<bqv[m]>
guess that's my fault for not being specific enough
<ottidmes>
I actually have it as part of a nicer shell function so I can pass it any file and any arguments
Hi-Angel has quit [Remote host closed the connection]
<ottidmes>
bqv[m]: well, just going with what you showed, is what many do to workaround it, which I agree with morgrimm, I do not like, the file should not be modified
ecou[m] has quit [Quit: Idle for 30+ days]
maljub01[m] has quit [Quit: Idle for 30+ days]
grahamc[m] has quit [Quit: Idle for 30+ days]
aquarial has quit [Quit: Idle for 30+ days]
RainDroppz[m] has quit [Quit: Idle for 30+ days]
ZamaneKz[m] has left #nixos ["Kicked by @appservice-irc:matrix.org : Idle for 30+ days"]
VaNilLa[m] has left #nixos ["Kicked by @appservice-irc:matrix.org : Idle for 30+ days"]
<bqv[m]>
yeah but i meant it as part of the command line invocation
<bqv[m]>
don't mind me i'm just salty cause my ISP is being a ****
<ottidmes>
were you the one with Orange problems?
rogue_koder has joined #nixos
<bqv[m]>
no i'm the one that's been dropping packets all day thanks to Sky
mauli has joined #nixos
<ottidmes>
hopefully its a one time thing (or uncommon), so far the only problems I've had with my ISP were it breaking due to maintanance or some infrastructure problem, or things that were simply fixed by restarting the router
anirrudh has joined #nixos
anirrudh has quit [Client Quit]
hax404_ is now known as hax404
<bqv[m]>
just as i get the super speed boost of switching from tinc to wireguard as well
cybrian has quit [Ping timeout: 264 seconds]
wildtrees has joined #nixos
mauli has quit [Quit: WeeChat 2.7.1]
rogue_koder has quit [Ping timeout: 240 seconds]
rogue_koder has joined #nixos
erasmas has joined #nixos
LysergicDreams has quit [Ping timeout: 265 seconds]
<{^_^}>
[nixpkgs] @teto opened pull request #84615 → vimPlugins.vim-dasht: init at 2019-09-29 → https://git.io/JvAyU
noudle has quit []
chloekek has quit [Quit: WeeChat 2.7]
LysergicDreams has joined #nixos
Guest41574 has quit [Remote host closed the connection]
revtintin has quit [Ping timeout: 265 seconds]
wildtrees has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @teto opened pull request #84619 → gitAndTools.gitbatch: init at 2019-12-19 → https://git.io/JvAy9
<srid>
My default.nix requires .git to be present. Is there an alternative CLI where the derivation would be built *with* the .git directory? I'm not sure if `builtins.fetchgit` can be used in CLI?
<{^_^}>
[nixpkgs] @grahamc merged pull request #84620 → nixos: kafka test: fix building for other arches → https://git.io/JvAyF
hyper_ch has quit [Client Quit]
linarcx has joined #nixos
pbogdan has quit [Remote host closed the connection]
snicket has joined #nixos
noudle has joined #nixos
pbogdan has joined #nixos
hyper_ch has joined #nixos
<bsima>
what is the nix equivalent of "python3-dev" on other systems?
<clever>
bsima: `nix-shell -p python3`
<jjwatt[m]>
bsima: If you mean a package built from git HEAD there isn't one, but you can customize one. pythonPackages38 is in there, too, which is pretty dang new!
<bsima>
oh it comes with the headers already?
<jjwatt[m]>
ohh you mean `-dev` for the development libs
<bsima>
ya
<jjwatt[m]>
Yep, what clever said
linarcx has quit [Client Quit]
reallymemorable has joined #nixos
<bsima>
ok thank you
linarcx has joined #nixos
growpotkin has joined #nixos
lsix has quit [Quit: WeeChat 2.7.1]
linarcx has quit [Client Quit]
<vika_nezrimaya>
So, what's the state of nixpart? Doesn't look like nixpkgs has support for it. It is in a PR that is only accessible via API and gives HTTP 500 on attempt to read it in the browser
<vika_nezrimaya>
And I'm really interested in it since I've got a use-case where NixOS starts to really shine
linarcx has joined #nixos
<vika_nezrimaya>
fully stateless laptop that could be wiped and reimaged at any time
linarcx has quit [Client Quit]
<jjwatt[m]>
`python3` so like `nix-shell -p python3 gcc zlib somecoollib gnumake`, etc. just off the top of my head and warning: I'm new to this too
<jjwatt[m]>
bsima: but if you're going to build something, you'll need to do it in the shell and likely make your own shell with any extra needed C stuff. When packages are built, you specify which dependencies they need, it's possible that just `python3` wouldn't have everything you needed, but you can probably also get around that without writing a nix file by passing the packages you want in the shell to `nix-shell -p` along with
<vika_nezrimaya>
If you're going through a border or caught by the police, you just delete all the data, then insert a flash drive that provisions your system back
<clever>
vika_nezrimaya: i think gchristensen has a setup, where it will use `zfs rollback` to revert / back to empty on every boot
<clever>
vika_nezrimaya: so only /nix can persist data
<clever>
vika_nezrimaya: basically, you make an install media with justdoit included, boot it on something, type justdoit into the shell, and it does it
linarcx has joined #nixos
<vika_nezrimaya>
ok, great, prior art is always great! :3
morgrimm has quit [Ping timeout: 256 seconds]
<vika_nezrimaya>
I'm looking for a more customizable approach
<jjwatt[m]>
I noticed some neat settings in configuration.nix that could probably help with something like that the other day, too: to automatically grow partitions
NieDzejkob has quit [Ping timeout: 264 seconds]
<bqv[m]>
what in the world...
<vika_nezrimaya>
that could, e.g. even maybe detect a Windows install and maneuvre around it (by shrinking Windows partition to a set size and creating a bunch of partitions)
<bqv[m]>
is zfs actually more common than btrfs these days?
linarcx has quit [Client Quit]
bqv has joined #nixos
<vika_nezrimaya>
bqv: I've heard some people use ZFS in production setups
<vika_nezrimaya>
not so much about btrfs
cole-h has joined #nixos
<vika_nezrimaya>
is it really better or is it just butter?
<vika_nezrimaya>
(pun intended)
<clever>
vika_nezrimaya: the nixos mac builders all use darwin in qemu on nixos+zfs
<Yaniel>
well you could consider SailfishOS "in production"
<clever>
vika_nezrimaya: every time the darwin guest reboots, nixos will do a zfs rollback of it
linarcx has joined #nixos
<Yaniel>
and that used btrfs back when it was still cool
<bqv[m]>
honestly i've never seen a reason to switch
<vika_nezrimaya>
darwin guest?
<Yaniel>
and by cool I mean unstable as hell
<clever>
vika_nezrimaya: darwin under qemu, on nixos
<bqv[m]>
zfs was still in it's diapers when btrfs was going strong
<vika_nezrimaya>
You could install MacOS in QEMU? isn't that technically kinda illegal
<vika_nezrimaya>
and on ZFS?
<clever>
vika_nezrimaya: its only legal if you do it on apple hardware
<vika_nezrimaya>
what is this magic, do y'all have a Nix file for this setup?
<clever>
vika_nezrimaya: so, nixos, on apple hardware, with a zfs root, and then darwin in qemu, with zfs volumes for the disk image
<niso>
how would i run `something = pkgs.nixosTest ./mytest.nix` interactively? reimplement nixosTest to override makeTests driver? or is there a better way?
<gchristensen>
LnL: thanks for merging that PR. did it backport to stable?
<vika_nezrimaya>
Looks like nix flake info --json has an additional key "outputs" but it's broken, it doesn't show beyond the first level - e.g. if I have nixosConfigurations.maria it only shows nixosConfigurations: {}
<gchristensen>
(woot, 1,000 people in the channel)
<hyper_ch>
hmmm, the new ethernet standard 802.3ck is going to be really fast
<Unode>
hi all, a basic nix question. If I include/import a function into a nix expression (i.e. first line) but don't use it at all, does this actually cause the function to be executed?
<jared-w>
bqv[m]: Do you happen to know if wireguard can be used as a client to connect to openvpn? Kinda curious because I have an openvpn config currently to use my work's VPN but it would be interesting to see if I'd even get any performance boost at all from wireguard
<bqv>
Unode: it may be partially evaluated i think
<bqv>
jared-w: totally incompatible, afaik
<Unode>
I'm asking because while trying to fix mnemosyne using mkDerivationWith, simply adding it to the import line is causing error: anonymous function at /var/nixpkgs-channels/nixos-19.09/pkgs/games/mnemosyne/default.nix:1:1 called without required argument 'mkDerivationWith', at /var/nixpkgs-channels/nixos-19.09/lib/customisation.nix:69:12
<jared-w>
darn. That's what I thought
<jared-w>
bqv: What'd you change in your flake config setup vs nrdxp's? Just curious
<Unode>
I'm a little confused about what is going on here. I see the function used in qutebrowser without errors, but can't seem to understand why it causes this error.
* jared-w
is debating bikeshedding his nixos configs into flakes at some point
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « munin: 2.0.51 -> 2.0.59 »: https://git.io/JvAdN
<bqv>
jared-w: i mean the only downside to flakes at the moment is that not everything supports it, but nixpkgs and home-manager(ish) do, as well as a few other tidbits
<blibberblob>
Hi all, I'm seeing an issue with slack on nixos where it's unable to send notifications. The error I'm seeing in the logs is "(slack:18209): libnotify-WARNING **: 22:25:07.504: Failed to connect to proxy". Has anyone seen this before? By the way, I've tried running `notify-send hi hello` and it also doesn't work (nothing happens)
<evertedsphere>
bqv: wait your repo looks h*cking amazing
<evertedsphere>
god i wish i had some more time this week
<bqv>
for my next trick i'd like to find a neater way to declare dependencies between machines, like ssh keys
<bqv>
lol
<evertedsphere>
it's soooo ... well-thought-out
<jared-w>
bqv: nice. Yeah I'm really hopeful for flakes. It would be great to start pulling out all of the various bits of nixpkgs into their own separated bits. Especially the various language nix integration ecosystems.
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « nss_wrapper: 1.1.10 -> 1.1.11 »: https://git.io/JvAFe
<evertedsphere>
do you use disnixos/disnix etc at all?
<evertedsphere>
i see a commit message mentioning it
<Athas>
I'm writing a derivation where the checkPhase needs to make network connections. It looks like Nix is providing an environment where even github.com cannot be resolved. Is there a way to enable network access for the checkPhase?
linarcx has joined #nixos
<jared-w>
evertedsphere: check out nxrdxp's repo as well. Lots of their configuration comes from there and it'll be useful to see the differences.
<evertedsphere>
i think there's a typo
<evertedsphere>
ah figured it out, it's nrdxp
linarcx has quit [Client Quit]
<jared-w>
Whoops. Yeah. I can never remember the exact spelling. (apparently not even when staring at it...)
<bqv>
evertedsphere: i'm considering it
<bqv>
but i can't wrap my head around it
<bqv>
yeah, the first third or so commits of my repo are nrdxp's, and the template branch is all but untouched
<bqv>
but afair that template will no longer even build
<evertedsphere>
i'm trying to get a taste of all the nixos devops tools (and *taps head* devops in general) by testing them out on my localhost
<bqv>
:)
drozdziak1 has quit [Ping timeout: 256 seconds]
<evertedsphere>
i've become quite happy with nixops so far even if all i do is just use nixops deploy with a little envrc to pin NIX_PATH
<evertedsphere>
i could just use nix-copy-closure + switch-to-configuration or whatever instead and i think i will, to get a bit more comfortable with lower-level nix stuff
<evertedsphere>
imagining something like a disnixos-based qubes os minus the tor stuff
<evertedsphere>
(as (iirc) taktoa[c] would put it)
<{^_^}>
[nixpkgs] @andir merged pull request #84584 → firefox: 74.0.1 -> 75.0 and related updates → https://git.io/JvA2u
<jared-w>
Nixops irrationally annoys me but I need to sit down and fully dogfood it before I really pass judgment on it. Using terraform for more than toy examples will help too.
<gchristensen>
nixops 2.0 comes with some major changes on that front
<jared-w>
Really wish we had something a weeee bit more usable than ovirt at work. The API is there, but nothing interfaces naturally with it, so it ends up being super annoying to integrate it with any other tooling.
<bqv>
gchristensen: quick nix question, can i access the build dir in some way?
<jared-w>
gchristensen: nice! Is there any documentation or writing about it? This is the first I've heard of a v2
<gchristensen>
bqv: say more?
<bqv>
i've built a package, is the build dir still around? if not, can i rebuild it and keep the build dir?
<gchristensen>
you can only keep it if it failed
<gchristensen>
do you need it for a specific reason?
shafox has quit [Remote host closed the connection]
<thibm>
bqv: use --keep-failed, see man page of nix-build
cr4y1 has quit [Read error: Connection reset by peer]
<ixxie>
hello folks, trying to write a script that create a NixOS machine on Hetzner and I noticed the graphical ISO doesn't run sshd by default. Is the minimal ISO any different?
<thibm>
There may be a better way now… keepBuildTree was written in 2014 and never touched since
<niso>
ixxie: you might want to checkout nixops
<clever>
ixxie: they both stop ssh from running on boot by default
rogue_koder has joined #nixos
<thibm>
ixxie: you can turn on sshd in your own ISO
lsix1 has joined #nixos
<ixxie>
niso: nixops only helps once NixOS is installed; in anycase I am planning to go for morph in this project.
<ixxie>
good to know clever
<niso>
ixxie: nixops does have a hetzner backend
<clever>
ixxie: nixops has a script to auto-install on hetzner
<ixxie>
thibm: yeah... now I just need to see if I can upload custom isos to hetzner
<niso>
(for root servers)
<bqv>
thibm: that's incredibly useful to know, thanks
<thibm>
ixxie: found it. Back to that time nixops was broken, so I did it that way: boot to vkvm rescue mode (Debian), from there you can install any distro as usual. It's like if you had the machine in front of you
<thibm>
(it took me a while to understand that vkvm was free of charge. I don't speak german)
<ixxie>
hmm it seems nixops only supports the dedicated hardware on Hetzner :(
<thibm>
as usual: no, because you're not on the NixOS installation media live. So I installed Nix, installed the NixOS installation tooling (nixos-install & friends) and then it's as usual
<{^_^}>
nixops#1189 (by cleverca22, 33 weeks ago, open): plan for supporting custom partition layouts and custom FS's on any backend
<thibm>
hu?
<clever>
thibm: if you install just nix, then you can use phase 3 here, to copy a pre-built nixos to /mnt/nix/store on a remote machine
<clever>
assuming things have already been partitioned and installed
<clever>
mounted&
<ixxie>
I have a script that used clever's kexec trickery to do this on a Hetzner VM... but I can't seem to get this to work reliably...
<clever>
ixxie: kexec is also somewhat optional, since you could also use the nix copy trick in the above issue
VShell is now known as Shell
<clever>
ixxie: you could also package the kernel+initrd from it, into a normal /boot with grub, and then just boot that as a disk image
<clever>
depends on what the cloud provider offers
<thibm>
clever: what I did: 1. clean up 2. install dependencies with apt-get (zfs!) 3. zfs+partitions 4. mount 5. install nix 6. add nixbld users & stuff 7. setup channel 8. install nixos-install & friends 9. nixos-generate-config + config + nixos-install
<ixxie>
clever: interesting... but the approach in the issue seems simpler for a scripted install
zupo has quit [Ping timeout: 260 seconds]
<obadz>
(any nixers happily (or not) using btrfs btw?)
<{^_^}>
[nixpkgs] @rnhmjoj opened pull request #84645 → [20.03] haskellPackages.pandoc-crossref: downgrade to latest working → https://git.io/JvAbm
<lordcirth>
obadz, do you have a more specific question about btrfs?
<clever>
thibm: if your doing zfs, it would probably be simpler to kexec into nixos
<thibm>
clever: indeed you can dump an installation instead of letting nixos-install do it
<clever>
thibm: but for any FS the distro already supports, nix + nix copy would likely do
<obadz>
lordcirth: I've been playing around a bit and thinking about flipping some disks, but it seems a bit scary
<thibm>
clever: may be. I tried your kexec at that time (October 2018). Either I did not manage to make it work, or I had an issue with ZFS
o1lo01ol1o has joined #nixos
zupo has joined #nixos
<lordcirth>
obadz, btrfs works ok, as long as you stay away from the experimental features, and don't fill your disk. But I would recommend ZFS instead
<thibm>
I remember I really tried 10 ways to install it on the damn machine :D
revtintin has quit [Quit: WeeChat 1.9.1]
<thibm>
with nixos-infect and others. Maybe ZFS was not a good idea for the first install here…
blibberblob_ has joined #nixos
<obadz>
lordcirth: I struggle to understand ZFS tbh. Looks like the volume manager wants to take over the world.
<thibm>
But you're right, copy the thing is a good way to go, mostly universal
<lordcirth>
obadz, being originally ported in from outside Linux, it does some things a bit differently. But it's the best filesystem there is.
<obadz>
lordcirth: Does it do the cool heterogenous raid1/raid10 trick that btrfs does?
<thibm>
clever: finally I don't agree. What's the point of copying a prebuilt nixos if you have Nix? Just build it on the server (except if it's a 100MB RAM VSP)
<thibm>
VPS*
* obadz
going AFK but will read replies when I'm back
<lordcirth>
obadz, no, that kind of flexibility is the main thing btrfs has over ZFS
<clever>
thibm: more for when your using nixops, you can `nixops deploy --build-only`, then copy that product up
pbogdan has quit [Remote host closed the connection]
<zeta_0>
hello everyone, i want to setup a sendmail instance on my nixos machine, so that i can use: darcs send, to push patches, anyways, when searching online, i can't seem to find a good link to the nix documentation with instructions on how to do this, could you guys paste a link with the documentation(i'm sure it's buried in the nix doc somewhere)
iqubic has quit [Ping timeout: 260 seconds]
cybrian has joined #nixos
<{^_^}>
[nixpkgs] @matthuszagh opened pull request #84654 → fparser: init at 4.5.1 → https://git.io/JvAN9
<morgrimm>
zeto_0: Seems like you could explore the mail service in the nixpkgs repo, or potentially ssmtp (also in the nixos modules) as another option
<zeta_0>
darcs has some good instructions on setting up msmpt, but i'm not sure how to do it the nix way
<kraem>
i've been playing around with the firewall settings lately - is there an obvious reason why iptables aren't flushed when the system is rebuilt and the firewall.service is restarted?
spinus has quit [Ping timeout: 265 seconds]
c0c0 has joined #nixos
spinus has joined #nixos
<spinus>
kraem: not sure what you mean by flushed, but when I rebuild the system and apply new config, new firewall rules are applied
Darkmatter66_ has joined #nixos
kiloreux has joined #nixos
<kiloreux>
I am installing fontconfig and trying to write to update the cache in installation. But it's failing with failed to write cache. How can I go about install fonts? (Nixpkgs manual page for fonts didn't help)
Darkmatter66 has quit [Ping timeout: 264 seconds]
<zeta_0>
which option do i need to configure in order to be able to send email using: darcs send ? there's no enable option with default features, also, am i going to need to install external nix packages as well? like sendmail and msmtp? https://nixos.org/nixos/options.html#networking.defaultmailserver
<zeta_0>
dam, i didn't think that it would be this difficult to setup a sendmail instance, i thought there would be a option to enable, where this functionality would work out of the box
<simpson>
Really? Email is super-hard to administer. It's a place where it's possible to be an expert in only one of the three languages, two of which are required to actually perform an email.
<simpson>
The jargon has jargon. MTA and MUA are distinct problems.
<gchristensen>
back when I was doing more contracting work I had a clause saying anything having to do with email services cost extra
teto has quit [Quit: WeeChat 2.7.1]
<MichaelRaskin>
Not only outgoing?
<gchristensen>
nope
<gchristensen>
incoming you have to deal with spam
<simpson>
zeta_0: It seems like msmtp is a sendmail replacement; their documentation has examples for configuration. https://marlam.de/msmtp/msmtp.html#Examples msmtp is packaged, so that you might merely have to add it to your darcs environment. I'm not sure, though, as I don't use darcs or msmtp.
<kraem>
spinus: the nixos fw rules actually are deleted, and reapplied with the new rules. old rules that are entered manually, or by other services are not flushed though so the iptables get stateful.
o1lo01ol1o has quit [Ping timeout: 265 seconds]
<thibm>
bqv: builtins.stringLength
<thibm>
length is for lists
andydarcyjewell2 has joined #nixos
<thibm>
No you won't :p
<bqv>
will remember in the future
<{^_^}>
[nix] @domenkozar closed pull request #3457 → downloader: when retry but can't resume, retry the whole file → https://git.io/Jv71F
<bqv>
no, i won't
<bqv>
but i appreciate the thought :p
<thibm>
:D
<cole-h>
lovesegfault: Yeah, should just need parens wrapping the first override/overrideAttrs
lsix has joined #nixos
<kleisli>
is it intended that the man pages for gcc are not installed by default? ie `nix run nixpkgs.gcc -c man gcc` fails?
<bqv>
works for me
<thibm>
bqv: A few days ago, I was reading the OpenBSD's signify paper, where it's written "I still can't use ln without reading the man page every time". So true for me too
<bqv>
haha
drozdziak1 has joined #nixos
<gchristensen>
"I still can't use cp without reading the man page every time"
<MichaelRaskin>
I think it is about nix run not caring enought about man pages
<gchristensen>
I just don't get what is so hard about ln :|
<bqv>
i had trouble with the ordering, originally
<bqv>
or was that another utility...
<thibm>
Honestly? I have no idea
<MichaelRaskin>
Who remembers which program had among its known issues «has more flags than ls»?
<thibm>
Maybe I use cp 10× more often
<bqv>
find was a frustrating one, until i got the message "PATHS MUST PRECEDE EXPRESSION" burned into my memory
<gchristensen>
aye heh
<kleisli>
MichaelRaskin: it doesn't work in nix-shell either. in fact i don't know how to get the gcc man pages at all
<thibm>
bqv: so true
<spinus>
kraem: I see what you mean, I guess this is implemented that way to avoid this moment when you are either, opened to the world or cut off
<zeta_0>
thibm: cool, so how did you setup msmtp using nix? could you link a paste bin?
<clever>
,ifd srid
<{^_^}>
srid: import-from-derivation (IFD) is when you evaluate nix from a derivation result, for example `import (pkgs.writeText "n" "1 + 1")` will evaluate to 2. This is sometimes problematic because it requires evaluating some, building some, and then evaluating the build result. It has been described as "such a nice footgun."
<zeta_0>
sorry i took so long to respond i had to run an errand
<clever>
srid: simplest, is to use builtins.readFile to read the file in the derivation
<zeta_0>
thibm: ok, the instructions say i need to configure it in the ~/.msmtprc file, i wasn't sure if there was some nix option that's used to write to this file
<{^_^}>
[nixops] @grahamc merged pull request #1256 → Use Poetry & Poetry2nix for environment and plugin management → https://git.io/JviO4
zeenk has quit [Read error: Connection reset by peer]
zeenk has joined #nixos
<niso>
zeta_0: there is if you decide to configure msmtp with home-manager
thibm_ has joined #nixos
thibm has quit [Ping timeout: 256 seconds]
<{^_^}>
[nixops] @grahamc opened pull request #1277 → Docs: Update authoring.rst after a few more implementations. → https://git.io/JvAhf
cybrian has joined #nixos
<xfix>
updated my computers to 20.03, no issues so far :) (i know it's a beta, but i'm impatient, and the current blocker issues seem to be non-issues for my usecase)
<morgrimm>
zeta_0: You can just define that configuration file as a home-manager file
<srid>
In pkgs.runCommand, what makes it unambigously clear that a shell env variable ($foo or ${foo}) is to be filled in by Nix or the runtime shell env?
<lovesegfault>
Is there an analogue of pythonWithPackages to PHP? like phpWithExtensions or something
<morgrimm>
srid: As far as I know, ${foo} is interpolation syntax for Nix, and $foo will delegate to the environment consuming the string
<gchristensen>
is that an in-place transition from ext4?
<lordcirth>
obadz, a migration? no, only btrfs
cybrian has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @doronbehar closed pull request #84610 → notmuch: set emacs off by default → https://git.io/JvAP0
<lordcirth>
gchristensen, yeah, you can migrate ext4 to btrfs online, it's pretty cool
<gchristensen>
wow!
<obadz>
gchristensen: yes, and reversible too. it's awesome.
<gchristensen>
wat
<clever>
lordcirth: online, damn!!
<lordcirth>
Reversible to orginal state*
<cole-h>
Online and reversible?!
<gchristensen>
obadz: I'm a big ZFS fanboi these days if you have questions :P
<obadz>
gchristensen: it creates a fake sparse file over all the ext4 metadata and a snapshot of the drive post migration, which means all new writes happen in previously unused space and can be ignored.
<lordcirth>
Once you've written new data to btrfs, you can only revert to the original, you won't have your new data
lsix has quit [Quit: WeeChat 2.7.1]
<cole-h>
Right, that's to be expected.
<cole-h>
That's pretty neat-o
<niso>
zeta_0: yea, you'll need programs.msmtp enabled, however it will take it's configuration from accounts.email
<obadz>
gchristensen: I'm leaning more towards btrfs atm but might hit you up. Is there a way to encrypt subvolums with different keys in ZFS?
<bqv>
then again, i haven't used ext4 extensively for over half a decade
<gchristensen>
obadz: you bet
<obadz>
gchristensen: about the keys?
<gchristensen>
yep
<obadz>
hmmm I'm changing my mind rapidly then :)
<xfix>
the way conversion back is pretty cool too, during conversion from ext4 to btrfs, a snapshot is created which prevents old ext4 data from being modified
<obadz>
so you can have zfs volume over LUKS over zfs volume manager?
<xfix>
conversion back works*
iqubic` has quit [Quit: ERC (IRC client for Emacs 28.0.50)]
<obadz>
xfix: yes with a fake sparse file for the metadata :)
<atemu12[m]>
obadz: You could but...why?
iqubic has joined #nixos
<obadz>
atemu12[m]: different users have different keys?
<cole-h>
gchristensen: Do you use native ZFS encryption, or LUKS/something else?
<gchristensen>
obadz: yeah so you can use LUKS if you want, or native ZFS encryption, and each dataset can have a different key
<gchristensen>
(they're called datasets to zfs, not volumes)
<obadz>
hmmm ok, I'm going to have to read all about ZFS again then
<gchristensen>
obadz: I have a single partition on my disk and it is LUKS, and then ZFS on top of that, with no LUKS on top of that. if I were to erase and go again, I'd probably just go for ZFS native encryption instead of LUKS.
<xfix>
i kinda would like to migrate an ext4 partition to btrfs myself, but i'm kinda worried things may go wrong with LUKS encryption...
<obadz>
xfix: LUKS is underneath all that, I don't think it should mattter?
<clever>
gchristensen: of note, zfs cant encrypt the dataset names, so if you have a tank/uber-secrets, somebody is going to have questions for you when they notice it :P
<atemu12[m]>
obadz: The openzfs GH's wiki, the Archwiki, the zfs and zpool manpages and, of course, the NixOS wiki page
<xfix>
yeah, but bootloaders are complicated
<xfix>
i'm not concerned about the operating system after I boot, i'm worried about process before boot
<kraem>
i had luks and zfs on top on that. last install i made on my home server i went with native zfs and it's super :)
<gchristensen>
obadz: yeah, notably ZFS encryption doesn't provide plausible deniability and what-not. there is some metadata around like, names of your datasets and a few other things
<clever>
xfix: i choose to use a either a fat32 or ext4 /boot, with zfs on luks for /
<obadz>
gchristensen: I'm fine with that. Can't be worse than ecryptfs..
<xfix>
yes, i have a separate /boot partition too
<srid>
It appears -E is not really supported in nix-env, only in nix-build. Is this true?
<atemu12[m]>
clever: Evil maid attacks are very hard to protect against but in this case you could use secure boot
<atemu12[m]>
Not perfect either since it's black box proprietary firmware but should be very hard to get around for your average hacker
<atemu12[m]>
gchristensen: Sure
<clever>
atemu12[m]: measured boot in the TPM also helps, smaller firmware blob involved in the critical thing
justanotheruser has quit [Ping timeout: 256 seconds]
<atemu12[m]>
clever: Yeah but you don't know whether the TPM actually verified the boot, a hacker could've just disabled it and put in their own bootloader
<atemu12[m]>
clever: TPM can be handled in firmware but are also often actual hardware chips
<qyliss>
atemu12[m]: you do if the computer proves to you it was measured, a la Heads
<CptCaptain>
What's the best way to go with Nvidia Optimus right now?
<{^_^}>
[nixpkgs] @jonringer opened pull request #84666 → palp: correct metadata for repology → https://git.io/JvAjB
<gchristensen>
clever: would it be interesting if the metadata from installation time were made available to the nixos configuration on a packet machine?
<clever>
gchristensen: for normal nixos, or nixops?
<simpson>
CptCaptain: hardware.nvidia.optimus_prime, looks like; it changed for 20.03, I guess.
<atemu12[m]>
qyliss: Indeed lol but that does look like something that could protect against Evil maid attacks reasonably well
<clever>
gchristensen: simplest option i can see is to just encode all of that metadata as json (or hnix it into nix), and then have a nix file in /etc/nixos/ that sets something like hardware.packet.metata = builtins.parseJSON (builtins.readFile ./metadata.json);
<clever>
gchristensen: and nixops can do similar, via a file in /tmp/
<gchristensen>
exactly my thought!
<gchristensen>
do you think it'd be interesting?
<CptCaptain>
simpson: optimus_prime is old and prime is new?
<CptCaptain>
And the wiki article talks about the unreleased NixOS 20.03?
<clever>
gchristensen: yeah, and then you can use stuff like ram size to adjust limits of programs, and cpu core count to adjust concurrency
<gchristensen>
cool
<gchristensen>
clever: should it be embedded in the data nixops-packet puts in the statefile?
<{^_^}>
[nixpkgs] @jonringer opened pull request #84669 → python3Packages.py3to2: fix metadata for repology → https://git.io/JvAjM
<lovesegfault>
How can I call nix build on my nixpkgs checkout without it using my overlays?
<clever>
gchristensen: probably
<simpson>
CptCaptain: Yes, that's what I'm gathering. I don't use this stuff; I'm reading the same articles you are.
<clever>
gchristensen: can the metadata be obtained at a later time, via the api?
<gchristensen>
yeah, but I think it can only be done from the machine itsnelf
<atemu12[m]>
<gchristensen "I'm pretty sure every consumer c"> Firmware TPMs are also allowed unfortunately and, firmware being firmware, there has already been an attack on Intel's fTPM that could leak the private keys. You still need to watch out for hardware TPM support when buying a new machine
<gchristensen>
sure
<atemu12[m]>
gchristensen: Are the hydra job status error messages documented anywhere?
<lovesegfault>
What's the right way to manually patch the interpreters?
m0rphism has quit [Quit: WeeChat 2.7.1]
<clever>
lovesegfault: patchShebangs, it uses the binary already in $PATH
<lovesegfault>
clever: but fixupPhase seems to not call that? (but it does patch interpreters)
<jjwatt[m]>
Does anyone know if it would be preferable to pull and build from git in a Nix package in the case that the PyPI source distribution does not include tests or should we prefer using PyPI in that case and skip tests? A whole lot of python packages seem to go the last route.
fusion809 has quit [Remote host closed the connection]
<clever>
lovesegfault: yeah, anything in the buildInputs (and some parts of the stdenv) can add themselves to phases
Bambus has left #nixos ["Tschüss"]
drozdziak1 has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @globin pushed to structured-attrs « treewide: add C{,XX}FLAGS to env attrs »: https://git.io/Jvxea
m0rphism has joined #nixos
zupo has joined #nixos
johnw` has joined #nixos
<{^_^}>
[nixops] @grahamc merged pull request #1277 → Docs: Update authoring.rst after a few more implementations. → https://git.io/JvAhf
<{^_^}>
[nixops] @grahamc pushed to master « Docs: Update authoring.rst after a few more implementations. (#1277) »: https://git.io/JvxeX
<{^_^}>
[nixpkgs] @zimbatm pushed commit from @yvesf to master « qcad: init at 3.24.2.1 (#81209) »: https://git.io/Jvxe1
fendor has quit [Read error: Connection reset by peer]
zupo has quit [Remote host closed the connection]
zupo has joined #nixos
<ij>
is there any reason why sudo shouldn't be present in systemd service file ExecStart environment?
<AmandaC>
... why would you use sudo in an systemd start line. You can set what user the job will run as with `User`
<gchristensen>
I'd like to flip that question around: why should it be?
<gustavderdrache>
gchristensen: you try running veruca_salt.exe without sudo
<cole-h>
worldofpeace: I'm currently trying to package a gnome app, but am getting "JS ERROR: ImportError: No JS module 'main' found in search path" when I try to run the binary. Could you take a look at my derivation? https://gist.github.com/cole-h/9d65b27609a3d2930d48d8e858d423a6
<obadz>
atemu12[m]: thx
<gchristensen>
O.o
<cole-h>
(when you have a moment; ping because I remember you're on the gnome team)
<jjwatt[m]>
Usually, you specify to what you want to step down your privs with systemd
<cole-h>
worldofpeace: Oh wow :D Very cool, thanks.
<AmandaC>
I assumed it was to run as a lower user, not becausse they were trying to elevate
<obadz>
gchristensen / atemu12[m]: "Once a device is added to a VDEV, it cannot be removed." ⇒ how do you live with this limitation?
<jjwatt[m]>
AmandaC: ah yeah. true.
phreedom has quit [Remote host closed the connection]
<atemu12[m]>
obadz: I don't remove devices ;)
<worldofpeace>
cole-h: It needs someone to pick it up though. maybe you can re-PR it 😸
<gchristensen>
obadz: how often do you need to do that?
<gchristensen>
(though device removal does exist now, with some limitations)
<obadz>
gchristensen: I mean, can I downgrade from raid1 to 'single' until my new drive arrives or what not?
<cole-h>
<3 worldofpeace If I end up using it in earnest, I probably will. Thanks!
<{^_^}>
worldofpeace's karma got increased to 98
<cransom>
you can go with a drive missing, that's fine.
<cole-h>
Wow this is a beautiful epub reader holy moly
<gchristensen>
obadz: your raidz1 stays a raid1, but is "degraded" until you get your extra disk
<cransom>
replacing a drive doesn't mean you remove something from the pool
bqv has quit [Ping timeout: 260 seconds]
<atemu12[m]>
obadz: You can always detach drives from mirrors
phreedom has joined #nixos
<obadz>
ok maybe the limitation isn't formulated precisely then
<obadz>
"When creating file VDEVs [..] the image files must be preallocated, and not sparse files or thin provisioned" ⇒ another one that's slightly annoying but again not a common usecase
<ottidmes>
cole-h: I just do $(nix-build ...)/bin/<name> stuff
<zeta_0>
do they need to be in their own seperate strings?
<cole-h>
zeta_0: Not at the end of the `from`, `user`, etc lines, at the end of the extraConfig block (last double quote).
zarel has quit [Ping timeout: 240 seconds]
<cole-h>
Darn, the one time `fish` lets me down. "Command substitutions not allowed"
<ottidmes>
zeta_0: what editor do you use? I use vscode and its Nix highlighter makes it almost impossible to make syntax errors, they can highlighted immediately
aveltras has quit [Quit: Connection closed for inactivity]
knupfer has quit [Ping timeout: 260 seconds]
kiloreux_ has joined #nixos
<ottidmes>
cole-h: how about just making a function/binary/script that does this for you
<gchristensen>
^ much better idea
<cole-h>
Yeah, fair.
<gchristensen>
move the smarts in to a nix build
<zeta_0>
zeta_0: i'm using emacs, it has bare bones support for nix, it underlines syntax errors with a little message, but that's it
<zeta_0>
ottidmes: ^
<energizer>
does vscode do something more than that?
<zeta_0>
whoops accidentily typed my own nick name
zarel has joined #nixos
<niso>
zeta_0: i thought you wanted to configure msmtp with nix?
<ottidmes>
energizer: nope, just highlighting of the errors, but they are hard to miss and so far have been able to prevent me making any syntax errors
<{^_^}>
[nixpkgs] @mmilata opened pull request #84673 → Use qt5's mkDerivation in packages that otherwise crash → https://git.io/Jvxfe
kiloreux has quit [Ping timeout: 256 seconds]
<energizer>
i see people have been working on nix lsp servers and tree-sitter parsers, are those somehow gonna provide more semantic information to the ide?
tv has joined #nixos
<zeta_0>
niso: i am that piece of code is a component of the programs option in home.nix: https://dpaste.org/GY1H
<zeta_0>
but the syntax errors are slowing me down, it's this one option that's messing me up
<lovesegfault>
What do I do if my installCheckPhase errors with this: /nix/store/izwfky32fcnz4gg9vj6yl35x0x4x2g4v-stdenv-linux/setup: ./arcanist/bin/arc: /usr/bin/env: bad interpreter: No such file or directory
<andi->
You probably didn't patch the shebangs before
<niso>
zeta_0: you want to configure your account with something like this: https://dpaste.org/pUra
<lovesegfault>
andi-: I disabled the fixupPhase because it was patching php all wrong
<andi->
lovesegfault: -v to all wrong
<lovesegfault>
andi-: I need php.withExtensions ... and it was patching to just php
<ottidmes>
zeta_0: you really need to take the time to learn Nix, cause I immediate see syntactical errors with that paste
<flokli>
lovesegfault: I'd just override php to a php with extensions
<flokli>
and pass that into buildInputs
<flokli>
and still let the rest take over of shebang patching
* lovesegfault
tried
<ottidmes>
zeta_0: 1) 3 open braces, 2 close braces 2) a string contained in braces, that is no valid attrset or argument destructering 3) Multiline strings are best done with '' rather than "
<lovesegfault>
* tries
<zeta_0>
ottidmes: agreed, i've been real busy with other stuff, i need to eventually set some time to learn nix, i read the nix pills 2 months ago, but that's about it
<evils>
would a PR that adds a test that requires a custom kernel be acceptable? (this is for rasdaemon)
<zeta_0>
ottidmes: what? the syntax errors are gone, so i figured that code is correct?
<flokli>
lovesegfault: PR incoming
<zeta_0>
ottidmes: i didn't paste bin my entire home.nix, only part of it
LysergicDreams has quit [Ping timeout: 265 seconds]
cr4y1_ has quit [Ping timeout: 258 seconds]
LysergicDreams has joined #nixos
endocrimes has quit [Ping timeout: 265 seconds]
<huevo5050>
hi, how I could install external software that uses nix as build system?
<huevo5050>
for instance. https://github.com/urbit/urbit. I can git clone and install with nix-env -i. But how can i install it from my configuration.nix?
endocrimes has joined #nixos
<huevo5050>
have sense? I tried to make a derivation but nix commands fails into nixos-rebuild environment
tv has quit [Ping timeout: 256 seconds]
<zeta_0>
huevo5050: in your configuration.nix you can use: environment.systemPackages = []; , to install the nix pkg you need
<huevo5050>
its an external nix pkg, not in nixpkgs repositories
<notgne2>
in this case it should be `environment.systemPackages = [ (import (builtins.fetchGit { url = "https://github.com/urbit/urbit.git"; rev = "2bfcf20bf7523b353cf99ba5f93dae5873f9c0db"; })).urbit ];` I think
<huevo5050>
notgne2 i get problems because Makefile in that project is using nix-build underneath, and looks like nixos-rebuild environment doesnt like run nix into (recursive nix)
cybrian has joined #nixos
<notgne2>
I don't think that should use the makefile unless I'm missing something
<notgne2>
actually, it looks like it uses LFS stuff which I think breaks builtins.fetchGit
tv has joined #nixos
<thequux[m]>
Another option is to just check out the tree somewhere on your filesystem and simply import it by path
pbogdan has quit [Ping timeout: 256 seconds]
selfsymmetric-mu has joined #nixos
<selfsymmetric-mu>
I'd like to install `dhall-flycheck.nix` from here, but it's not called `default.nix` so I can't just import a fetchFromGitHub expression.
<selfsymmetric-mu>
Also (secondarily) I tried to install `dhall.el` with `melpaBuild`, but it doesn't like the fact that it's called `dhall.el`. I guess it's not a real package.
cosimone has joined #nixos
<notgne2>
I actually can't find how to make fetchGit (or pkgs.fetchgit either) work with LFS at all
<huevo5050>
notgne2 i read that i could leverage postfetch
reallymemorable has quit [Quit: reallymemorable]
<selfsymmetric-mu>
Yeah, I don't know how to fetch an arbitrary file…
cybrian has quit [Ping timeout: 264 seconds]
<selfsymmetric-mu>
How can I reference the signature of a function like `fetchFromGitHub`? Maybe using the REPL?
reallymemorable has joined #nixos
<selfsymmetric-mu>
If I just write `pkgs.fetchFromGitHub` I get a mess like { __functionArgs = { ... }; __functor = «lambda @ /nix/store/irv1wqx914wfsn35pybjf6zh1xixs69y-nixpkgs-20.03beta1155.29eddfc36d7/nixpkgs/lib/trivial.nix:318:19»; override = { ... }; }
<niso>
notgne2: couldn't you just build it with nix-build nix/release.nix -A linux64?
<selfsymmetric-mu>
Oh using `:p <expr>` helps here somewhat.
reallymemorable has quit [Client Quit]
<selfsymmetric-mu>
How do I override `default.nix` in `fetchFromGitHub`?
<notgne2>
niso: oh, I must've missed something, because `(import "${builtins.fetchTarball { url = "https://github.com/urbit/urbit/archive/master.tar.gz"; sha256 = "1zbq094gmwj62a0arph2hmchw03hy4j4xchmd03bx2syhbi9bxjk"; }}/nix/release.nix").linux64.urbit` seems to work actually
<notgne2>
huevo5050: ^
snicket has quit [Ping timeout: 240 seconds]
<energizer>
i haven't been using gnome-keyring, i'm wondering when it's useful?
<niso>
nice
srid has quit [Quit: Connection closed for inactivity]
cosimone has quit [Quit: Quit.]
mbrgm_ has joined #nixos
<Amit[m]>
Does anyone have experience installing nixos specifically on a prgmr.com vps? The recommended configuration from their wiki almost works except the xen-kbdfront kernel module causes issues with systemd-udev-trigger restarting. I'm now strating from scratch using the xen-domU base in the repository with... very slow success so far and wondering if anyone is ahead of me
mbrgm has quit [Ping timeout: 256 seconds]
mbrgm_ is now known as mbrgm
cosimone has joined #nixos
<niso>
selfsymmetric-mu: i think you are looking for something like this:
<selfsymmetric-mu>
How can I query my current configuration from `nix-repl`? Steam is telling me to add `hardware.opengl.driSupport32Bit = true;` even though I already have that setting in my configuration.nix.
<andi->
> (<nixos/config>).config.time.timeZone
<{^_^}>
file 'nixos/config' was not found in the Nix search path (add it using $NIX_PATH or -I), at (string):296:2