<simpson>
juliusdeane: I don't see anything obviously wrong in the libtelebot derivation.
<juliusdeane>
simpson: yeah, I think the issue is somewhere in the gf-hegel (the binary linked with libtelebot) derivation (https://pastebin.com/F0hacATd)
<simpson>
juliusdeane: I can't actually debug that src.
supersandro2000 has quit [Disconnected by services]
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
morr has quit [Quit: WeeChat 2.8]
morr has joined #nixos
<{^_^}>
[nixpkgs] @github-actions[bot] pushed 38 commits to staging-next: https://git.io/JLRdB
<{^_^}>
[nixpkgs] @github-actions[bot] pushed 39 commits to staging: https://git.io/JLRdR
Darkmatter66_ has quit [Ping timeout: 272 seconds]
<simpson>
juliusdeane: If you remove the custom buildPhase and installPhase, then things almost work as-is, including building the echobot example. The main problem, and part of your build problem, is that libtelebot.pc isn't being configured.
<simpson>
The ./Build/ directory was just to do the same sort of thing that Nix does with /build/, so we don't need it.
<{^_^}>
[nixpkgs] @lheckemann opened pull request #107181 → linux: fix generate-config's handling of "no-choice" options → https://git.io/JLRF6
lovesegfault has quit [Quit: WeeChat 3.0]
ddellacosta has joined #nixos
LnL- has joined #nixos
LnL- has joined #nixos
LnL- has quit [Changing host]
LnL has quit [Ping timeout: 268 seconds]
<{^_^}>
[nixpkgs] @abathur opened pull request #107182 → bashup-events: init at it's-complicated → https://git.io/JLRN6
<Raito_Bezarius>
is it just import <nixpkgs> { crossSystem = …; } ?
<Raito_Bezarius>
because it's more like I need the binutils on host arch which targets riscv32 arch
<Raito_Bezarius>
and not the binutils on riscv32 arch which targets riscv32 arch
<qyliss>
Yeah
<Raito_Bezarius>
so I believe it won't really work
<Raito_Bezarius>
as binutils is broken on riscv32 for example
<Raito_Bezarius>
but riscv32-none-objcopy just works fine on x86_64 for example
<Raito_Bezarius>
and I have a derivation that depends on the toolchain
camsbury has quit [Ping timeout: 260 seconds]
berberman_ has joined #nixos
h0m1 has quit [Ping timeout: 260 seconds]
b42 has quit [Ping timeout: 260 seconds]
glowpelt has quit [Ping timeout: 260 seconds]
omnigoat has quit [Ping timeout: 260 seconds]
d10n-work has quit [Ping timeout: 260 seconds]
Ken[m] has quit [Ping timeout: 260 seconds]
redcedar[m] has quit [Ping timeout: 260 seconds]
grin[m] has quit [Ping timeout: 260 seconds]
KaboomShebang[m] has quit [Ping timeout: 260 seconds]
ctp has quit [Ping timeout: 260 seconds]
emily has quit [Ping timeout: 260 seconds]
aswanson_ has quit [Ping timeout: 260 seconds]
LnL- has joined #nixos
LnL- has joined #nixos
LnL- has quit [Changing host]
Ultrasauce_ has quit [Read error: Connection reset by peer]
Ultrasauce has joined #nixos
berberman has quit [Ping timeout: 260 seconds]
mitsuhiko has quit [Ping timeout: 260 seconds]
lopsided98_ has quit [Ping timeout: 260 seconds]
lblasc[m] has quit [Ping timeout: 260 seconds]
hax404 has quit [Ping timeout: 260 seconds]
avalenn has quit [Ping timeout: 260 seconds]
glowpelt has joined #nixos
Ken[m] has joined #nixos
redcedar[m] has joined #nixos
grin[m] has joined #nixos
KaboomShebang[m] has joined #nixos
ctp has joined #nixos
LnL has quit [Ping timeout: 256 seconds]
lopsided98 has joined #nixos
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
mitsuhiko has joined #nixos
h0m1 has joined #nixos
<Raito_Bezarius>
hm
b42 has joined #nixos
<Raito_Bezarius>
stdenv might be what I want
emily has joined #nixos
aswanson_ has joined #nixos
d10n-work has joined #nixos
omnigoat has joined #nixos
avalenn has joined #nixos
lblasc[m] has joined #nixos
hax404 has joined #nixos
<qyliss>
Raito_Bezarius: I've never done this, but my best guess from the manual is that you want (import <nixpkgs> { crossSystem = ... }).pkgsHostTarget.binutils
<qyliss>
Raito_Bezarius: although if you're trying to use this in a derivation, then yeah you just need to use stdenv
supersandro2000 has joined #nixos
<Raito_Bezarius>
did the job, indeed I believe, thanks qyliss !
<Raito_Bezarius>
qyliss++
<{^_^}>
qyliss's karma got increased to 0o151
<Raito_Bezarius>
depsBuildBuild = [ stdenv.cc ]; on the hostNixpkgs.stdenv.mkDerivatoin
<Raito_Bezarius>
then nativeBuildInputs in the mkShell
<th3ta>
is there a reason why cmake would only detect c++14 and c++20, but not c++14? i'm having different results depending if gcc, clang, or llvm is used
<th3ta>
not C++17*
<siraben>
Is there a Nix overlay for GHC 8.12?
LnL has joined #nixos
LnL has joined #nixos
LnL has quit [Changing host]
tobeportable has quit [Quit: Leaving]
xcmw has joined #nixos
th3ta has quit [Remote host closed the connection]
xcmw has quit [Client Quit]
<siraben>
GHC 9*
<ZaraChimera>
I am looking to install NixOS on a Hyper-V VM. In addition to the steps listed in the manual for installing as a VirtualBox Guest, is there anything else I need to configure to get things working under Hyper-V? The most recent info I could find on Hyper-V was from 2015.
h0m1 has quit [Ping timeout: 260 seconds]
h0m1 has joined #nixos
myName_isMethos has quit [Quit: ERC (IRC client for Emacs 27.1)]
<andi->
the hyper-v drivers should all be mainline in all our kernels so it should just work
<ZaraChimera>
andi-: Okay, so just following the recommended sizes from the VB section and then running the install as normal should work?
<ZaraChimera>
Seems I have to use the CLI initially, the GNOME live download does boot but hangs and never shows the desktop.
<matthewcroughan>
nhs: >matthewcroughan: why don't you just do a fresh nixos install once on one rasp pi and then use the inverse of your dd command to create an iso from it
<matthewcroughan>
Because that's impure, anti-nix, imperative and unreproducible. That is why I can't do that. Because what would be the point?
<matthewcroughan>
I'm looking for ways to generate system images. Not create golden images like a record company.
<matthewcroughan>
nix-build subprocesses qemu doesn't it? Given that it can do that, you should be able to generate a system image that you can flash, and I believe you can with nixos-generators.
orivej has quit [Ping timeout: 264 seconds]
<jophish>
Is there anyone here with a Mac who could check that a change compiles, I don't have the hardware to test
<Ke>
if I want to autologin sway in a normal logind/freedesktop idiomatic way, do autologin options under X11 work for me
<Ke>
ie. does the substring x11 in variable name mean it's only for Xorg sessions
<Ke>
or can I just skip display manager without losing anything relevant
<Ke>
well ok xserver, not x11
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<patagonicus>
Ke: From what I understand it really is only for xserver. One solution I've heard about, but haven't tried yet, is setting up autologin for tty1 and adding an if [[ $(tty) = /dev/tty1 ]]; then exec scay; fi to your .bash_profile
veleiro has quit [Read error: Connection reset by peer]
veleiro has joined #nixos
CMCDragonkai1 has quit [Remote host closed the connection]
jonringer has quit [Ping timeout: 258 seconds]
CMCDragonkai1 has joined #nixos
rajivr has joined #nixos
CMCDragonkai2 has joined #nixos
<patagonicus>
Ke: There's an older pull request for Phosh, where people argue that it shouldn't be merged until there's a nice NixOS module way of configuring wayland based desktop environments. So the problem seems to be entirely that no one has bothered to write something for it yet.
growpotkin has quit [Ping timeout: 260 seconds]
respawn_ has joined #nixos
turlando has joined #nixos
<DigitalKiwi>
i saw a github issue a few days ago about how xserver settings might still apply even when you think they shouldn't since there are things that you think are wayland but are really both
<pinpox>
mmonit: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked, interpreter /lib64/ld-linux-x86-64.so.2, for GNU/Linux 2.6.18, BuildID[sha1]=0cc9b581f6cc6c0f98d7b0ea0eb76258ad97ff42, not stripped
<simpson>
`file` to check the interpreter? Or `ldd`, but note that GNU ldd (and other ldds) usually have to *run* the binary to find out, which is dangerous.
<pinpox>
Should I package it? (note, the licence is unfree I guess )
<simpson>
Meh. On one hand, I don't encourage packaging unfree software. OTOH, writing a derivation is going to be simpler and easier to understand in a few months when you've totally forgotten that you made a local patchelf'd copy of that one binary.
<pinpox>
simpson: Ok, thanks for the info. I'll read up on how to do this
<simpson>
(On the gripping hand, then, I'd recommend moving to Prometheus~)
<pinpox>
simpson: Prometheus and netdata seemed like an overkill, but I'll conside them. I really like the minimalistic approach of monit and mmonit
<pinpox>
btw. the packaged monit (not m/monit) works great
<{^_^}>
[nixops] @tewfik-ghariani opened pull request #1424 → upgrading install-nix-action to fix build → https://git.io/JLEYu
<simpson>
pinpox: Minimalism is relative. I think Prometheus is quite minimal; rather than a dedicated monitoring binary, it's possible for individual server processes to open up an extra port and export Prometheus-compatible scrape data.
<pinpox>
simpson: will look into it. How is the data gathered on the servers? My main use case is to have my 5 machines run something to periodically check if everything is in order and get a simple notification like "hard drive nearly full" or "website abc.com is down" via email
<pinpox>
It's more of a personal/small setup with a few websites and some webapps like bitwarden_rs
cosimone has joined #nixos
meh` has quit [Ping timeout: 264 seconds]
<ollie[m]>
Another endorsement for Prometheus here, it's excellent
<simpson>
pinpox: A central daemon (Prometheus itself) collects data by HTTP scraping each monitored service's exported port. Another daemon, AlertManager, scrapes Prometheus and considers whether or not to be in a state of alert.
<pinpox>
hm i see.
<simpson>
AlertManager can email you, yes. It can also do many other things. And the whole affair scales down to the size of a homelab.
<pinpox>
I guess I'll try to package it for myself anyway. Seems like I could learn something on the way for nixos
m0rphism has joined #nixos
<aristid>
lol i thought i had waited out the great firefox extension nuking, but turns out i had just been one reboot away from it
<sphalerite>
ehmry: especially having a valid (not self-signed) cert via LE :)
<pinpox>
aristid: I had trouble after updating too. Had my extensions all configured in home-manager and suddenly everything was gone
<pinpox>
aristid: If you need a fix, I found a workaround that works for me
* sphalerite
is also a satisfied prometheus user
<aristid>
pinpox: shouldn't the latest master have it fixed by now?
<pinpox>
aristid: Specifically using "package = pkgs.firefox-bin" seems to solve or work around the issue
<aristid>
pinpox: ah just switch to firefox-bin? yeah well let me try just upgrading the nixos-unstable channel first
<pinpox>
aristid: yep
<pinpox>
let me know, if the workaround is no longer necessary. I don't have any problems with the -bin version though, not sure if it's maybe older?
<aristid>
fortunately i don't use many extensions, still annoying to have to reconfigure them
fintohaps[m] has joined #nixos
<aristid>
pinpox: isn't the -bin version just the binary released by mozilla with a wrapper / maybe just some patchelfing?
<pinpox>
aristid: yeah me too. But just logging in and not having a password-manager and an add blocker is terrible :D
<aristid>
also i heard it will now sync the extension nuking to all my other computers (not all of them use nixos)
<pinpox>
seems like it, bot the -bin and non-bin versions in nixpkgs are v83
<pinpox>
aristid: it did
<pinpox>
If you have a firefox account set up with sync for extensions at least
<aristid>
i do.
veleiro has quit [Read error: Connection reset by peer]
<pinpox>
Bonus: you can't even manually install extensions because it complains "you can't have a mix of manually installed and administered extensions"
veleiro has joined #nixos
<pinpox>
so basically, you're screwed
<aristid>
pinpox: well that'll be a good way of verifying that it's actually fixed
<pinpox>
Yeah, but just using -bin worked for me and I was lazy so it seemed fine to leave it like this
<pinpox>
guess I'll just revert it back to the normal package in some time
<aristid>
pinpox: ok i verified, unstable channel of nixos is fine now.
<aristid>
but i don't know what the disadvantage of using firefox-bin is, i guess you may as well stick to it if it works for you
<Yaniel>
fwiw, I copied my firefox profile over from my old machine and it looked like the extensions were gone
<Yaniel>
but they were only gone from the UI, still listed in extensions
<Yaniel>
so I just toggled them all once and they came back
<Yaniel>
with settings intact
andi89gi has joined #nixos
meh` has joined #nixos
<{^_^}>
[rfcs] @FRidh opened pull request #83 → [RFC 0083] common interface package sets → https://git.io/JLEsd
chiefgoat has joined #nixos
<ehmry>
sphalerite: oh I don't care about that stuff
veleiro has quit [Read error: Connection reset by peer]
veleiro has joined #nixos
<ehmry>
I should publish the mumble configuration anyway
__monty__ has joined #nixos
andi89gi has quit [Ping timeout: 260 seconds]
asheshambasta has quit [Ping timeout: 260 seconds]
cosimone has quit [Remote host closed the connection]
veleiro has quit [Read error: Connection reset by peer]
veleiro has joined #nixos
Alexey49 has joined #nixos
veleiro has quit [Read error: Connection reset by peer]
veleiro has joined #nixos
vidbina has joined #nixos
<Alexey49>
Hi. Sorry for noob question, but how to enable broadcom_sta on nixos 20.09? All googled advices seem to be outdated. I have a fresh install of nixos 20.09, networkmanager is enabled, my user in group networkmanager, and i've added linuxPackages.broadcom_sta to environment.systemPackages, I've rebooted system, but still does not have wifi.
<Alexey49>
clever: Thank for your response. I've added boot.extraModulePackages = [ config.boot.kernelPackages.broadcom_sta ];, removed broadcom_sta from environment.systemPackages, did nixos-rebuild switch. Still no wifi, should I reboot?
<clever>
Alexey49: rebooting may help, some stuff only loads the first time the card is detected
veleiro has quit [Read error: Connection reset by peer]
Alexey49 has quit [Remote host closed the connection]
veleiro has joined #nixos
vldn has joined #nixos
fuzzypixelz has joined #nixos
Alexey12 has joined #nixos
<nature>
Is someone familiar with netsurf ? I have a problem with the gtk icons displaying as squared boxes instead of actual icons
<nature>
etu: how would I go about adding them, I tried starting a shell with `paper-icon-theme` for instance and then starting netsurf, but it wouldn't work
<clever>
this confirms that the module was called "wl"
<etu>
nature: I usually pull in: gnome3.adwaita-icon-theme
<clever>
03:00.0 Network controller [0280]: Broadcom Inc. and subsidiaries BCM43142 802.11b/g/n [14e4:4365] (rev 01)
<clever>
Alexey12: yep, there it is
<etu>
nature: I think icons may have a cache somewhere, so I'm not sure just pulling it in in a nix shell is enough... I know that before i put it as one of my system packages I often had boxes in gtk applications.
<clever>
Alexey12: what does `lsmod | grep wl` say?
andi89gi has quit [Remote host closed the connection]
veleiro has quit [Read error: Connection reset by peer]
<clever>
yep, missing there
veleiro has joined #nixos
<clever>
Alexey12: what if you do `rmmod bcma wl ; modprobe wl` ?
<nature>
Okok etu, let me try adding it to my system packages
<nature>
It worked! Thanks etu :)
SomeoneSerge has joined #nixos
<etu>
nature: No worries! :)
Alexey12 has quit [Ping timeout: 245 seconds]
Alexey12 has joined #nixos
<{^_^}>
[nixpkgs] @Profpatsch opened pull request #107197 → Tree sitter fixes → https://git.io/JLEEf
<Alexey12>
clever: sorry, does not see, that I was disconnected from chat. After `rmmod bcma wl ; modprobe wl` wifi works, thank you! Do you know how to persist this settings?
veleiro has quit [Read error: Connection reset by peer]
cosimone has quit [Remote host closed the connection]
cosimone has joined #nixos
tejing has joined #nixos
andi89gi has joined #nixos
<andi89gi>
Yaniel: ok. Well, I have installed home-manager. Should I list up every program on home.nix?
veleiro has quit [Read error: Connection reset by peer]
veleiro has joined #nixos
<{^_^}>
[nixpkgs] @mcbeth opened pull request #107199 → qrcodegen: init at 1.6.0 → https://git.io/JLEXA
SomeoneSerge has quit [Ping timeout: 256 seconds]
veleiro has quit [Read error: Connection reset by peer]
veleiro has joined #nixos
<{^_^}>
[nixpkgs] @peterstuart opened pull request #107200 → nodePackages.firebase-tools: init at 9.0.1 → https://git.io/JLE12
andreas303 has quit [Remote host closed the connection]
davidv7 has quit [Ping timeout: 240 seconds]
andreas303 has joined #nixos
Alexey12 has joined #nixos
strikerlulu has joined #nixos
<lukego>
How do I get nix-shell to setup my LD_LIBRARY_PATH? I thought it would automatically fill in the buildInputs but doesn't seem to...?
<tpw_rules>
in theory it doesn't need to because all your programs are patched to find their libraries in the nix store
veleiro has quit [Read error: Connection reset by peer]
veleiro has joined #nixos
<tpw_rules>
however sometimes that fails in practice because programs dynamically calculate library names and stuff so you can manually set it by adding an attribute in your shell.nix that's like LD_LIBRARY_PATH = pkgs.lib.makeLibraryPath [ pkgs.mesa_drivers pkgs.ocl-icd ];
<lukego>
Thanks. I remember now at some point writing a Nix expr that generates a script with the appropriate calls. I'm in this case running a binary that expects to be able to dlopen() some libraries.
<tpw_rules>
yeah a wrapper script is another option too
veleiro has quit [Read error: Connection reset by peer]
veleiro has joined #nixos
jmeredith has joined #nixos
mallox has joined #nixos
<andi89gi>
makefu: Do u use LinuxKernel 5.10?
domogled has quit [Remote host closed the connection]
domogled has joined #nixos
andi89gi has quit [Remote host closed the connection]
<lukego>
Hey is there a good Git/Nix workflow for gradually/asynchronously upstreaming a new package? Like, I already have a Nix source tree with my various derivations, but one of them really belongs in nixpkgs. Is there an easy way to PR it there without making a second copy of the source code in git?
<adisbladis>
lukego: A lot of us are using local nixpkgs checkouts for that very reason
Fare has joined #nixos
<adisbladis>
What are you trying to avoid?
<adisbladis>
Avoiding a second copy of what exactly?
andi89gi has joined #nixos
<lukego>
Well so adding a new package I guess I have three options: 1) fork nixpkgs. 2) overlay repo. 3) derivation in my own repo. The only one of these that I can PR to nixpkgs upstream would be (1) but that makes it a bit tricky for me to do my own dev e.g. might be required to rebase the branch and so on. But if I do (2) or (3) then it seems like I _also_ need to do (1) separately (= code dup) for upstreaming purposes.
<lukego>
Maybe I'm missing something obvious
<lukego>
In the past I've started off doing (1) of packaging a program up and sending it to nixpkgs upstream but then, alas, finding that too cumbersome and making a separate package for my own use instead.
<qyliss>
lukego: It sounds like the best thing you can do is to try to get more comfortable with rebasing
<simpson>
What does your "own dev" entail? It's possible to use nixpkgs expressions on upstream source trees, depending on the exact package.
<a64>
seems that firmware-linux-nonfree does not contain rx6800's firmware. overrided with latest linux-firmware now amdgpu loads without error but i am still having a black screen
<Yaniel>
does the textmode tty at least work?
<Yaniel>
and which mesa version
<DigitalKiwi>
lukego: i have lots of different nixpkgs directories
<DigitalKiwi>
so i can "stay" on a branch i'm using when i do nixpkgs-rebuilds and work on ones that are for PRs
cole-h has joined #nixos
<a64>
nomodesetting can get me to tty
<lukego>
okay but "my own dev" here is building a larger system using Nix that is pushed to github etc and contains its own derivations too. basically what I'm lacking is a smooth way to "export" my own derivations to nixpkgs when it makes sense to do so.
<DigitalKiwi>
and then i git push/pulll between them and i add remotes
<a64>
mesa 20.2.4
<DigitalKiwi>
i use ghq https://github.com/x-motemen/ghq to clone repos so they're organized by owner ~/projects/github.com/Kiwi/nixpkgs ~/projects/github.com/NixOS/nixpkgs
<Yaniel>
you might want 20.3.1
<lukego>
DigitalKiwi: okay but I'm building an application to collaborate on with other people and I need to publish it including packages that haven't landed upstream in nixpkgs yet.
<Yaniel>
it's on nixpkgs master already I think
<lukego>
DigitalKiwi: So to do the same as you I'd need to be pushing a separate branch for each copy and updating the sha256 etc when they reference each other.
<DigitalKiwi>
i nixos-rebuild off of my personal fork/branch that might be ahead of NixOS/nixpkgs
<DigitalKiwi>
also what qyliss you'll have to rebase eventually to do the PR anyway
<simpson>
Rather than using upstream/channel nixpkgs, you can have your projects directly pin a nixpkgs version. That way, you can send PRs upstream to nixpkgs, and there's still a way to atomically switch each individual package from your bundled expression to nixpkgs.
<DigitalKiwi>
i frequently pull in specific packages from different channels/git checkouts when they break on nixos-unstable
<qyliss>
That way it's very easy to modify it if you need to, and you can easily update it by just git fetch nixpkgs && git merge -Xsubtree=nixpkgs nixpkgs/nixos-unstable or whatever
<Yaniel>
a64: ah sorry, it's in staging atm (see #105817)
<lukego>
qyliss: oh so you embed the whole nixpkgs tree in your repo and use git-subtree to pick off the changes and upstream them?
<a64>
i saw fedora users reporting mesa 20.2.4 working so it might not be the problem?
<lukego>
maybe git-subtree is the answer... to be able to have my own main fork of nixpkgs with all my changes and to be able to semi-automatically pick off each change into a topic branch that can be PR'd and rebased without f'ing up the main copy that I'm working with (where I don't want to be scrambling git history)
<qyliss>
lukego: yeah, it's really nice and easy
<lukego>
So thinking aloud... I could have one fork of nixpkgs that I push to Github and includes all of my changes. This is never rebased and it represents my own development. Then for each change that needs to go upstream I could use git-subtree to spin it off into a topic branch with a revised history and send that off to nixpkgs upstream.
<qyliss>
lukego: yeah, although it's probably easier to use cherry-pick to do that than subtree
<simpson>
lukego: Fare has been trying something similar recently, and I can emphatically recommend *not* doing this unless you have a deep need to embed your packages deep into nixpkgs.
<simpson>
It is so much easier to just have your own nix-support/ folder and write new expressions locally for each project which needs new expressions.
<lukego>
simpson: So what's a better alternative though? Thus far I've always started off doing "upstream first" but finding too much friction and just keeping all my code private
<simpson>
lukego: Why not upstream *incrementally*? Send changes upstream in PRs *without* removing the changes from your private codebase. When PRs get merged, bump your project's pinned nixpkgs and *then* drop the private changes.
veleiro has quit [Read error: Connection reset by peer]
veleiro has joined #nixos
civodul has joined #nixos
<andi89gi>
sphalerite: Should we talk about how we can start building TMW pkg for NixOS?
<lukego>
but I don't know how long the upstreaming will take and I'll be stuck duplicating any change into two places throughout that process (and testing each copy spearately, etc)
<simpson>
lukego: No; while your upstream PRs are pending, you don't have to test integration with new nixpkgs, just with old pinned nixpkgs.
<simpson>
You never actually have to have integration working with more than one version of nixpkgs at a time, ever.
<lukego>
Sure but e.g. if the thing I've packaged is updated while the PR is still pending
<lukego>
I'll need to bump both the version I'm using and the version I'm upstreaming and make sure that both work
lgcl has quit [Quit: ERC (IRC client for Emacs 27.1)]
<simpson>
If you're the cause of that package update, then maybe the package isn't ready for upstream nixpkgs. Otherwise, just sigh and bump the PR *or* ignore the update entirely.'
<simpson>
(Does this actually happen with non-pathological packages? I only see it happening with unfree games.)
<sphalerite>
andi89gi: sure. Want to do voice chat of some description for it?
<simpson>
(Oh, also stuff like Discord clients and youtube-dl. But even then, they rarely release multiple times per day.)
<lukego>
Thing is that when the source package updates I'll want to adopt the new version quickly into my application -- and if I end up sending an older version that this upstream to nixpkgs then it defeats the purpose because the upstream code won't be usable for me anywya
<Yaniel>
mesa gets PRs on .0 releases but they are rarely merged until they get bumped to .1
medvid has quit [Quit: WeeChat 2.3]
<Yaniel>
but I think that one is mostly by policy?
<simpson>
Yaniel: This should tell you something about the quality of upstream's realeases, too~
medvid has joined #nixos
<simpson>
Yaniel: *releases. Just like when I used to contribute to Mesa, I bet that the first draft's not perfect.
<andi89gi>
sphalerite: Well, yeah - I have also questions about home-manager :D Where can we do a voice chat?
<lukego>
I feel like I must be missing something major here though or else lots of other people would be having the same problem as me.
<simpson>
lukego: Modify your desires, I suppose. Why do you even want Nix expressions to go upstream to nixpkgs, in those cases? What *are* these packages, exactly?
<a64>
ok i think it's just xf86-video-amdgpu being too old (19.1.0)
sss2 has joined #nixos
<lukego>
simpson: So in this case it's GToolkit (https://gtoolkit.com/) which is a Smalltalk based dev environment that's quite new and a massive pain in the ass to install by yourself (distributed as binaries, source build requires a whole raft of Rust stuff, Smalltalk code includes /usr etc references that patchelf can't see.)
<simpson>
lukego: Fun times. Probably isn't ready for nixpkgs yet. Do you have other folks that wanted to use common derivations?
veleiro has quit [Read error: Connection reset by peer]
veleiro has joined #nixos
<lukego>
I've had the same issue with the Pharo Smalltalk VM itself too which was already in nixpkgs but too cumbersome for me to take over maintenance off for similar reasons, ended up just maintaining my own private derivation instead (and original maintainer moved on leaving it orphan)
<Yaniel>
yeah sounds like something for a NUR
<simpson>
Maybe that's how things should be.
domogled1 has joined #nixos
<simpson>
NUR's a possibility. Flakes are another possibility.
domogled has quit [Read error: Connection reset by peer]
domogled1 is now known as domogled
<Yaniel>
and once it's stable enough you can always try getting the already existing derivation into nixpkgs
domogled has quit [Read error: Connection reset by peer]
domogled has joined #nixos
<lukego>
but okay I'll modify my desires and just maintain my own derivations and leave nixpkgs to other people. Seems worth popping up and looking for a way to upstream like this every year or two anyway.
veleiro has quit [Read error: Connection reset by peer]
veleiro has joined #nixos
<simpson>
There's nothing wrong with putting stuff in nixpkgs. The problems are the ones you're experiencing, around integration.
<Yaniel>
that sounds like a reasonable approach
<simpson>
nixpkgs works best for stuff that is sufficiently common that, by the time you learn that you need e.g. a security patch, there's already been a merged PR and Hydra is building updated cache artifacts.
<Yaniel>
maybe make your custom derivations public so others interested can use them too / help out
<lukego>
I had imagined that nixpkgs would be the place where you do make your derivations public for other people to share and build on
noudle has quit []
<lukego>
I suspect that nixpkgs contribution workflow is geared towards people who are living on the bleeding edge and don't need to integrate their pre-upstreamed packages into other stable ecosystems. So for example it would work fine for me to upstream an application that I'm casually using but not one that I'm building a system around.
<lukego>
and maybe that's fine because that's still plenty of people to move nixpkgs forward. Just feels weird not to upstream stuff that I'm sure other people will want sooner or later and end up duplicating my work on.
andi89gi has quit [Remote host closed the connection]
<Yaniel>
don't duplicate your work if you don't need to?
<Yaniel>
push code changes upstream, update nixpkgs whenever there are releases
<Yaniel>
and you can override the package locally if you want to use code changes before they hit nixpkgs
<lukego>
(maybe the fact that I'm having so much trouble explaining my situation is also a sign that I'm in a small minority within the nixpkgs universe and should accept that the workflow is rightly optimized for other people :))
<simpson>
In mobile-nixos, there's already two PRs for a phone I have; I probably will need to make a third PR which incorporates work from both of them. CLV.
andi89gi has joined #nixos
<simpson>
lukego: As a contributor, I don't live on the bleeding edge. I have configured channels for my system, and a git clone of nixpkgs for when I want to send stuff upstream.
<simpson>
But also, in the Typhon repository I linked earlier, I have a third nixpkgs which is pinned by reference, and only changes when I desire something from a newer nixpkgs. Only at that time do I have to do integration work.
zakame has quit [Ping timeout: 256 seconds]
domogled1 has joined #nixos
domogled has quit [Ping timeout: 260 seconds]
domogled1 is now known as domogled
<sss2>
hi all, how to use fileSystems.<name>.neededForBoot properly ?
<sss2>
i can't find wjere fs is mounted during initramfs init
<sss2>
some shady hint found in fileSystems.<name>.encrypted.keyFile description, but i can't get it work
<viric>
sphalerite: ahhh that's your module, not ready in nixos
veleiro has quit [Read error: Connection reset by peer]
<sphalerite>
sss2: it should be mounted under the path where it will be in the final system as well, but under /mnt-root
<sphalerite>
viric: yeah
veleiro has joined #nixos
<viric>
sphalerite: I like it
<viric>
does nix automatically sign with that?
<sphalerite>
yeah, it'll sign any paths built on that machine with it
<sss2>
sphalerite, i will check again, but it not worked last time, probably i made a typo somewhere
<viric>
I didn't know nix would automatically sign
<sphalerite>
viric: man nix.conf describes the option
fendor_ has joined #nixos
<viric>
thank you
<viric>
would nix-copy compress?
<lukego>
On reflection maybe maintaining my stuff as a nixpkgs overlay makes the most sense. Then it's my own code, but it's also easy for other people to try out and depend on, and it is will be straightforward to upstream into nixpkgs whenever the time is right (e.g. package becomes more stable or more popular, etc)
<{^_^}>
[nixpkgs] @vcunat pushed 8 commits to release-20.09: https://git.io/JLuvE
strikerlulu has joined #nixos
<viric>
sphalerite: I'll adopt that signing & keys
lgcl has quit [Client Quit]
andi89gi has quit [Remote host closed the connection]
lgcl has joined #nixos
<patagonicus>
sss2: I'm pretty sure the neededForBoot bit happens after / has been mounted. I'm not sure if NixOS natively supports having the key outside of the initramfs, but on a filesystem. You could probably work around it with preLVMCommands and postMountCommands or something.
lgcl has quit [Quit: ERC (IRC client for Emacs 27.1)]
<patagonicus>
nrb: Are you using the NixOS usb installer? It already has nix installed, you don't need to download the installer.
lgcl has joined #nixos
<simpson>
What's the HTML say? I get text, using `less` instead of `sh` to see what I got.
lgcl has quit [Client Quit]
<nrb>
I have an older usb stick, tried the 'quick' command from that
<DigitalKiwi>
simpson++
<{^_^}>
simpson's karma got increased to 44
<DigitalKiwi>
that's what i wanted to know too
<nrb>
downloads starting <!DOCTYPE html>
<sss2>
patagonicus, i see init checking for specified filesystem, and after this waiting for file to be visible, but this wait is ending with error
<patagonicus>
sss2: Yeah, that makes sense, because it won't try mounting the filesystem the file is on, so the file will never appear. At least that's my understanding of the LUKS integration in the NixOS initrd, I only read parts of it a while ago.
<nrb>
using older NiXOS usb iso and running: curl -L https://nixos.org/nix/install | sh # pulls down html starting <!DOCTYPE html> instead of plain text
<patagonicus>
nrb: Coud it be that it's your ISP doing weird things? Try replacing sh with less and look at more of the page, for example the <title> tag
<sss2>
patagonicus, is it possible to use some correct way to wait for key device to appear instead of while loop polling ?
rb2k has joined #nixos
lgcl has joined #nixos
<sss2>
patagonicus, i mean if i will implement some workaround via pre/post commands
spinus has joined #nixos
<nrb>
less gives same text starting <!DOCTYPE html>
<patagonicus>
nrb: What's after the <!DOCTYPE html>?
strikerlulu_ has joined #nixos
<sss2>
hmm, one ugly hack come in mind
<patagonicus>
sss2: Ooh, I just found boot.initrd.luks.devices.<name>.preOpenCommands that's exactly what you need. Stick something in there that waits for the key device and mounts it, then use postOpenCommands to unmount it.
lgcl has quit [Client Quit]
<sss2>
yes, but how to weait for device properly ?
<patagonicus>
nrb: Can you do `curl -L https://nixos.org/nix/install | curl -F 'sprunge=<-' http://sprunge.us`? That will upload the output of curl to sprunge.us and give you a link that you could share with us so we can look at the HTML.
<sss2>
also, i see default timeout is 10 seconds, can i make it longer or betterinfinite somehow ?
<Fare>
simpson, lukego: one of the purposes of my multi-inheritance extension system is precisely to enable more "decentralized" development of nix packages, that can seamlessly move between extensions to nixpkgs to nixpkgs itself.
lgcl has joined #nixos
<patagonicus>
sss2: At least for waitDevice it seems to be hardcoded to 20 tries with a 1 second sleep. For LUKS it does seem to be hardcoded at 10 seconds.
lgcl has quit [Client Quit]
<patagonicus>
sss2: You could open a feature request for that, doesn't seem unreasonable.
<patagonicus>
nrb: It says file not found - are you *sure* you're using https://nixos.org/nix/install and don't have a typo in there? Or maybe replaced an l with an I or something?
maxfield has joined #nixos
mallox has quit [Ping timeout: 260 seconds]
<nrb>
sure
<DigitalKiwi>
isn't this what flakes is for
<patagonicus>
Then I have no idea why the server would claim that the file is not found for you. :/
maxfield has quit [Remote host closed the connection]
<nrb>
that is it, the redirect not rendering text...
lgcl has quit [Quit: ERC (IRC client for Emacs 27.1)]
<nrb>
note same from another IP (OVH VPS completely unrelated network I sshed into )
<lukego>
Fare: more on this multi-inheritance business somewhere?
mkaito has quit [Quit: WeeChat 2.9]
<hexagoxel>
Can somebody who has built nix please share the doc/manual/builtins.json file? (instead of me trying to get nix to build from source, which somehow runs into errors no matter what I try..)
<{^_^}>
nix#4381 (by nrbray, 4 minutes ago, open): boot from NiXOS usb iso and running: curl -L https://nixos.org/nix/install | sh # pulls down html starting <!DOCTYPE html> instead of plain text
mallox has quit [Ping timeout: 265 seconds]
<victorbjelkholm>
nrb: also, could you paste the output of running `curl -v -L https://nixos.org/nix/install`? There is an added -v flag in there
<nrb>
no copy paste all manually typed on a virgin box, curl to the redirected URL works, to the URL in the documentation it renders the html instead of plain text
<nrb>
no copy paste all manually typed on a virgin box, curl to the redirected URL works, to the URL in the documentation it renders the html instead of plain text
<simpson>
Huh. Yeah, the -v flag will help a lot, then.
<victorbjelkholm>
nrb: if you're still trying to use sprunge with the curl -v command, you'll need `2>&1 |` instead of just `|` as -v doesn't output to stdout
<victorbjelkholm>
nrb: no need. I'm not a maintainer of anything nix, but in the future you should probably wait to create an issue until it's confirmed it's one by at least someone else, just to avoid adding more notifications to the poor fellows who get notifications from that repo
<{^_^}>
[nixpkgs] @ehmry pushed commit from @r-ryantm to master « picard: 2.5.2 -> 2.5.4 »: https://git.io/JLu3S
neiluj has joined #nixos
andi89gi has quit [Remote host closed the connection]
strikerlulu has joined #nixos
<victorbjelkholm>
I'm trying to setup my wifi connection, and got the following example from the docs: `networking.wireless.networks.Wifi_name.pskRaw = "pskRaw generated";` however, my network name is all numbers, and seems nix doesn't like that, giving me a "unexpected float, expected '.' or '='" error, any ideas on how to work around?
andi89gi has joined #nixos
<patagonicus>
victorbjelkholm: networking.wireless.networks."123456".pskRaw should work
<victorbjelkholm>
patagonicus: ah, of course. It did the trick, thanks a lot
<{^_^}>
[nixpkgs] @veprbl opened pull request #107204 → texlive.combine: set allowSubstitutes = true → https://git.io/JLusN
strikerlulu has quit [Remote host closed the connection]
cosimone has quit [Remote host closed the connection]
<b00n>
I have to settle: "(chromium.override { enableVaapi = true; })" in the configuration.nix, but I don't know how to do this right.
<b00n>
Tried "packageOverrides = pkgs: with pkgs; { chromium.override.enableVaapi = true; };" and hundreds of variants, but I don't get the right syntax. Does somebody give me a helping hand, please?
<notgne2[m]>
b00n: you can probably just do `nixpkgs.config.enableVaapi = true;`
ddellacosta has quit [Ping timeout: 260 seconds]
SomeoneSerge has joined #nixos
mallox has quit [Quit: WeeChat 3.0]
<b00n>
notgne2[m]: oh boy, this worked instantly. Thank you sooo much!
Alexey12 has quit [Ping timeout: 245 seconds]
Fare has joined #nixos
b00n has quit [Quit: leaving]
ajhpics has joined #nixos
<ajhpics>
Hey all, I am trying to get nix on my m1 Mac, could somebody please point me to a guide, the usual steps sadly didn't work.
andi89gi has quit [Remote host closed the connection]
ajhpics has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @github-actions[bot] pushed 26 commits to staging-next: https://git.io/JLuVu
<{^_^}>
[nixpkgs] @github-actions[bot] pushed 27 commits to staging: https://git.io/JLuVg
<infinisil>
Cadey: NixOS doesn't manage your state/disks
alexrelis[m] has joined #nixos
<alexrelis[m]>
Is this an official channel?
<infinisil>
alexrelis[m]: Yup :2
<infinisil>
:)*
<alexrelis[m]>
I'm using Debian and I am looking for something different. My instincts was to try Fedora, but I'm looking to see if NixOS is right for me.
<alexrelis[m]>
A big frustration of mine is that installing software in more traditional distros can sometimes be a pain. Freezing packages only really works with mature programs but completely falls apart for things like Matrix clients which are constantly ever-changing at this point in time. Third party repositories easily break systems if not maintained, and flatpaks/snaps/AppImages just don't feel native and there not every program
<alexrelis[m]>
is packaged for it. I'm left pretty stressed as to what method I should install for each package.
veleiro has quit [Ping timeout: 264 seconds]
SomeoneSerge has quit [Ping timeout: 256 seconds]
ddellacosta has quit [Ping timeout: 268 seconds]
lgcl has joined #nixos
<{^_^}>
[nixpkgs] @Lassulus opened pull request #107212 → treewide: use auto diskSize for make-disk-image → https://git.io/JLuK8
lgcl has quit [Client Quit]
<shapr>
Is there an IRC channel where I can ask questions about getting started with haskell in NixOS?
lgcl has joined #nixos
lgcl has quit [Remote host closed the connection]
<shapr>
I'm reading Gabriel Gonzalez haskell-nix intro and I don't want to use nix-env to install tools.
<{^_^}>
[nixpkgs] @Ericson2314 opened pull request #107214 → WIP: lib: Clean up how linux and gcc config is specified → https://git.io/JLuKA
<lassulus>
#haskell.nix ?
<DigitalKiwi>
i think that's for the iohk one isn't it
<lassulus>
ah, could be, I just looked at the log list
gustavderdrache has joined #nixos
lgcl has joined #nixos
veleiro has joined #nixos
lgcl has quit [Remote host closed the connection]
<shapr>
lassulus: good suggestion, but I got frustrated trying to get started with haskell.nix and am trying to start with stock nix now.
lgcl has joined #nixos
<shapr>
I think my problem is that I lack enough basic nix understanding to jump directly into haskell.nix
<Vq>
shapr: I have a small nix-shell nix file with ghc, cabal-install and cabal2nix. Then I generate a project specific shell.nix file with cabal2nix and use that instead.
<shapr>
Vq: do you have step by step instructions written up somewhere?
<Vq>
shapr: Some tweaks may be required so I'd say it's still necessary to know some nix basics.
<Vq>
shapr: I don't
<shapr>
so I could probably use "nix-shell -p ghc cabal-install cabal2nix" and then "cabal init"?
<shapr>
not sure what to do after that though.
veleiro has quit [Ping timeout: 260 seconds]
joesventek has quit [Quit: Quit]
berberman_ has joined #nixos
noudle has joined #nixos
berberman has quit [Ping timeout: 260 seconds]
<Vq>
shapr: cabal2nix . --shell > shell.nix
dstzd has joined #nixos
* shapr
adds that to the almost blank tutorial
<Vq>
shapr: Then exit the ghc-etc-shell and run nix-shell in the project directory.
Darkmatter66 has quit [Ping timeout: 260 seconds]
dstzd has quit [Client Quit]
<{^_^}>
[nixpkgs] @Ericson2314 pushed 2 commits to fix-buildRustCrateTests: https://git.io/JLuPW
<Vq>
shapr: I'm certainly no expert on the matter but I've had surprisingly few problems building my Haskell projects with cabal2nix and nix.
lgcl has joined #nixos
<shapr>
I'm still a n00b, so I shall pester you with questions now :-)
<shapr>
At this point I do normal Haskell dev with cabal?
<Vq>
I haven't figured that part out yet.
lgcl has quit [Client Quit]
FRidh has quit [Quit: Konversation terminated!]
<Vq>
I've been adding dependencies to both the cabal file and the nix-file by hand but that may very well not be the most efficient way.
lgcl has joined #nixos
<shapr>
How do add a dependency to the nix-file after it's added to the cabal file?
<Vq>
I suppose you could either re-run cabal2nix on it or manually add it to the executableHaskellDepends attribute in the nix-file.
<shapr>
I'll try the re-run cause that's automated
<ascj09>
Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed100 26.6M 100 26.6M 0 0 9829k 0 0:00:02 0:00:02 --:--:-- 9825kNote: a multi-user installation is possible. See https://nixos.org/nix/manual/#sect-multi-user-installationperforming a single-user
<ascj09>
installation of Nix...copying Nix to /nix/store............................................./var/folders/5x/nnnv57y96p560w4gsg7s8g940000gn/T/nix-binary-tarball-unpack.XXXXXXXXXX.Ej4CeXLf/unpack/nix-2.3.10-x86_64-darwin/install: line 174: /nix/store/8c98r6zlwn2d40qm7jnnrr2rdlqviszr-nix-2.3.10/bin/nix-store: Bad CPU type in
<ascj09>
executable/var/folders/5x/nnnv57y96p560w4gsg7s8g940000gn/T/nix-binary-tarball-unpack.XXXXXXXXXX.Ej4CeXLf/unpack/nix-2.3.10-x86_64-darwin/install: unable to register valid paths
<shapr>
I prefer declarative cause it's easier to reuse.
<shapr>
I'm against nix-env use (for myself at least).
blueberrypie has quit [Quit: leaving]
blueberrypie has joined #nixos
<shapr>
for commands I rarely use, I do: nix-shell -p unzip
<shapr>
though I do wish I knew how to get my familiar zsh and ~/.zshrc in a nix-shell
joesventek has joined #nixos
blueberrypie has quit [Client Quit]
<Vq>
I'm the same way. I love the isolation of running things in their own nix-shells. I also haven't cracked the zsh in nix-shell nut yet. :)
blueberrypie has joined #nixos
joesventek has quit [Client Quit]
meh` has quit [Ping timeout: 240 seconds]
growpotkin has joined #nixos
joesventek has joined #nixos
rb2k has joined #nixos
joesventek has quit [Client Quit]
malook has joined #nixos
<ascj09>
Specific issue after running with deamon flag: ---- sudo execution ------------------------------------------------------------I am executing: $ sudo /nix/store/8c98r6zlwn2d40qm7jnnrr2rdlqviszr-nix-2.3.10/bin/nix-store --load-dbto load data for the first time in to the Nix Databasesudo: unable to execute
<ascj09>
/nix/store/8c98r6zlwn2d40qm7jnnrr2rdlqviszr-nix-2.3.10/bin/nix-store: Bad CPU type in executable---- oh no! --------------------------------------------------------------------Jeeze, something went wrong. If you can take all the output and openan issue, we'd love to fix the problem so nobody else has this issue.:(We'd love to help if you need it.If
<ascj09>
you can, open an issue at https://github.com/nixos/nix/issuesOr feel free to contact the team, - on IRC #nixos on irc.freenode.net - on twitter @nixos_org
<rb2k>
The "c" is important. Having grown up in "Deutsche Mark" Germany, all the "should we call you "Euro" now?" jokes semi backfired.
strikerlulu has joined #nixos
ddellacosta has joined #nixos
<rb2k>
Either way, I met flokli when he visited Facebook and he gave me the confidence (and liquor) to try switching our datacenter macs to use nix to install 3rd party packages. We actually use nix to build and package the resulting store directories as RPMs (for now).
<rb2k>
and that naturally also makes its way onto the developer laptops (custom prefix)
<rb2k>
so nix on darwin just got a few 10k more unsuspecting users :)
<Ke>
how does nix on macos work these days?
<rb2k>
surprisingly well :)
<rb2k>
we had dozens of homebrew packages (also deployed as RPMs on macs) and migrated all of them over
<DigitalKiwi>
error: while evaluating the attribute 'constituents' of the derivation 'nixpkgs-20.09pre999999.c695c6d' at /var/lib/ofborg/checkout/repo/38dca4e3aa6bca43ea96d2fcc04e8229/mr-est/packet-spot-eval-1/pkgs/build-support/trivial-builders.nix:7:7:
<pinpox>
What is the best way to define some own packages to be able to install them in my configuration.nix? I'm using flakes for my config, can I just create a /packages subdirectory and put them there?
<pinpox>
E.g. this derivation I'm building for mmonit https://termbin.com/9rsp where would I put it in my configuration repository (github.com/pinpox/nixos) ?
<pushqrdx>
i need to use this patched version of meson
<pushqrdx>
but i don't wanna upgrade
maxfield has joined #nixos
<pushqrdx>
i wonder if i reassigned the depsHostHostPropagated in an overlay if that would do the trick
terlars has joined #nixos
cosimone has quit [Quit: cosimone]
cosimone has joined #nixos
marusich has joined #nixos
<terlars>
Does anyone have experience with packaging a meson project with subproject with nix? Is this solved somehow, or is this uncharted territory? It seems the subproject will be pulled down at build time which of course doesn’t work since network traffic is not allowed.
spinus has quit [Quit: Connection closed]
<simpson>
terlars: You will probably have to package the subproject on its own, and then patch the superproject to not pull its own copy.
<terlars>
simpson: thank you, that was what I had thought about as well, unfortunately, I am not that familiar with meson, but I guess I just have to read up on it a bit to figure out where I can drop in these dependencies for them to be found
nature has quit [Ping timeout: 240 seconds]
lgcl has quit [Ping timeout: 260 seconds]
<{^_^}>
[nixpkgs] @viric pushed to master « frogatto: update to latest master and fix build »: https://git.io/JLux0
<pushqrdx>
apparently just by creating a meson overylay that removes the propagated gcc dependency, i triggered a near system wide rebuild of pretty much every shell i go into
<sheeldotme>
Does anyone know how to get grub to boot with a tmpfs root? I get this error: "grub-install: error: failed to get canonical path of `tmpfs'."