<ryantm>
worldofpeace: Yep, much better! Thanks to blaggacao for pushing on it.
Fare has joined #nixos
<Henson>
if I have two nix installation commands going at the same time (say a nixos-rebuild build and a nix-env --upgrade) I get warnings about the mysql database being busy. I assume since they're only warnings that nothing bad will happen because of it. I won't somehow corrupt by Nix package database by having multiple Nix commands working on the store at the same time?
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/JTMkc
<siraben>
clever: if there's no gcc for z80, how do I specify an alternate cross-compiler?
<sorear>
hmm, i tried 20.09 and I don't have a graphic environment with that (systemctl status reports that .sddm-greeter-wrapped segfaulted)
<clever>
siraben: is it clang based?
<la-jesystani>
does someone have an example of a nix flake that outputs one package, and an overlay for that package? im having trouble figuring out how to do it myself
<la-jesystani>
i have: https://github.com/techieAgnostic/vim/blob/main/flake.nix so far, however when trying to add the overlay to my system configuration i get an error about the .overlay attribute in that link not being an overlay
<clever>
siraben: hmmm, not sure how that would work, try just feeding it in as the gccFun fed into stdenv, look near where i modified gccFun in my pr
<clever>
siraben: Ericson2314 may know more, you can also look to see if llvm has z80 support, then use clang instead maybe?
<colemickens>
Do people just not use flake check, or does no one have multi-arch packages in a flake or am I missing something obvious?
<colemickens>
it seems like something in release-lib.nix should help with filtering a package list against a supportedplatform list but I can't figure it out
<julm>
bqv: AFAIU I can't bump #93659 yet because d5d543d2636 from #100031 is only in staging not master. I've checked with: git branch -a --contains d5d543d2636
<lukego>
Hey is it straightforward to use an x86 machine to build NixOS images to run on a RaspberryPi Zero W? Or even just to use Nix to build Raspberian ubuntu images? I'm a little lost on whether this is normal or exotic
<{^_^}>
Mic92/nixpkgs-review#142 (by SuperSandro2000, 19 hours ago, open): Add platform run on to report
vonfry has joined #nixos
<makefu>
lukego: you mean doing a cross-compile of the complete image? this is possible but more of the `exotic` side. samueldr created https://github.com/samueldr/cross-system/ to help with this
<makefu>
the repo essentially contains fixes for the cross-compiling issues currently in nixpkgs
<lukego>
I suppose that I'm looking for the least exotic option that will give me a RaspberryPi image with some custom software and configs setup.
<lukego>
and whether that should be based on Nix or some other RPi image generator that so far all look like nasty hacks
hnOsmium0001 has quit [Quit: Connection closed for inactivity]
tldr32 has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @teto merged pull request #101265 → wrap neovim but without its configuration file → https://git.io/JTuzt
tldr32 has joined #nixos
<{^_^}>
[nixpkgs] @teto pushed to master « neovimUtils: neovim utilities to handle more usecases »: https://git.io/JTMMD
<makefu>
lukego: if you have some time on your hands to build the world(tm) then you can definitly check out cross-system
<lukego>
ok I'll fire up a build of that in the background while I noodle around for other options, thanks
<makefu>
cheers!
<lukego>
can I expect e.g. wifi to work out of the box? or does ubuntu package a bunch of weird firmware etc that nixos doesn't?
townrb has joined #nixos
<lukego>
my use case is having the RPi stream from the HQ camera to Wifi with gstreamer
<{^_^}>
[nixpkgs] @kode54 opened pull request #102010 → vscode: Fix "Save as Admin" #49643 → https://git.io/JTMMN
townrb has left #nixos [#nixos]
<colemickens>
bqv: I just want a function to be able to filter the packages I have packaged by system
sife has joined #nixos
<makefu>
lukego: do you have a single raspi with this dedicated use case or are you planning to build more than one?
<lukego>
more than one
<makefu>
understood
<lukego>
hence my sinking feeling as I started e.g. cloning the latest git repo of gstreamer onto the rpi and its unreliable sd card
<colemickens>
bqv: otherwise right now when I run nix flake check, since I just have the same pacakges emitted for all system-tuples, I just get an error that my aarch64-linxu packages can't be built on x86_64-linux, etc.
<bqv>
colemickens: I get around that by filtering based on meta.platforms and ${system}
<{^_^}>
[nixpkgs] @teto pushed commit from @braunse to master « CastXML: refactor to avoid python2 dependency (#101996) »: https://git.io/JTMDi
vonfry has left #nixos ["ERC (IRC client for Emacs 27.1)"]
<lukego>
ok thanks. I don't need any GUI.
<lukego>
(though I note that the instructions in the README don't offer any choice about whether to build a bsae or GUI image? I'll dig a bit)
<{^_^}>
[nixos-homepage] @github-actions[bot] pushed commit from GitHub Actions to master « Update flake.lock and blogs.xml [ci skip] »: https://git.io/JTMDh
<{^_^}>
[nixpkgs] @peterhoeg merged pull request #101994 → ffmpeg-full: patch for libsrt deprecated API → https://git.io/JTM0b
<{^_^}>
[nixpkgs] @peterhoeg pushed to master « ffmpeg-full: patch for libsrt deprecated API »: https://git.io/JTMyU
<CRTified[m]>
lukego: If you do not need anything special, I can help you. I've got a basic cross-compiling running but struggle with deviceTree overlays
<lukego>
I'm not sure what qualifies as special... basically I want to be compatible with Raspberry Pi Zero W (ARMv6?) and use the HQ camera and Wifi and have a bleeding edge Gstreamer. weird or normal?
<Yaniel>
I tried setting crossSystem.config = "riscv32-none-elf" in shell.nix but that just says "unknown emulation for platform"
amfl_ has joined #nixos
<bqv>
colemickens: flake.nix:534, then e.g. pkgs/dev/dotnet/azurefn.nix:21
<bqv>
(guess I misspoke, I don't quite use platforms)
<bqv>
Knew it was one of the meta ones
amfl has quit [Ping timeout: 260 seconds]
amfl_ is now known as amfl
dutchie has quit [Ping timeout: 240 seconds]
dutchie_ has joined #nixos
<{^_^}>
[nixpkgs] @rnhmjoj pushed 3 commits to release-20.09: https://git.io/JTMSh
dutchie_ is now known as dutchie
GlennS has joined #nixos
boegel has joined #nixos
<{^_^}>
[nixpkgs] @pstn opened pull request #102015 → nixos/murmur: add murmur group, don't run as nogroup → https://git.io/JTMHz
zakame has quit [Remote host closed the connection]
zakame has joined #nixos
<sorear>
Yaniel: i got (import <nixpkgs> { crossSystem = lib.systems.examples.riscv64; }).buildPackages.gcc to work
malook has quit [Remote host closed the connection]
malook has joined #nixos
malook has quit [Remote host closed the connection]
zupo has joined #nixos
maridonkers[m] has joined #nixos
<ocharles>
Does anyone know why this produces an empty file? nix-build -E 'import <nix/fetchurl.nix> { name = "m4-1.4.18.tar.xz"; url = "ftp://ftp.gnu.org/gnu/m4/m4-1.4.18.tar.xz"; sha256 = "0mdqa9w1p6cmli6976v4wi0sw9r4p5prkj7lzfd1877wk11c9c73"; }'
<ocharles>
The result is /nix/store/fni1i01khaambp44gr99r7461y1hn7hw-m4-1.4.18.tar.xz, but file /nix/store/fni1i01khaambp44gr99r7461y1hn7hw-m4-1.4.18.tar.xz = empty
<lukego>
makefu: fyi the cross compile build only took about 20 minutes on a mid-range ryzen box
<{^_^}>
[nixpkgs] @humancalico opened pull request #102016 → vimPlugins.srcery-vim: init at 2020-10-26 → https://git.io/JTM53
<Yaniel>
thanks sorear
<Yaniel>
I'll try that when I get back to my nixos box
moredhel_ has joined #nixos
hyper_ch4 has joined #nixos
<nschoe>
Oh boy... maybe I switched too soon to 20.09. I guess hydra did not have time to build binaries, right? It's rebuilding everything ^^
<lukego>
Hey is there an easy way to make NixOS use an NFS root file system mounted via WIFI? That's the first thing I want to customize on my RaspberryPi to avoid having to flash a new SD card all the time
<eyJhb>
If I nix-shell https://...../archive.tar.gz , that has a default.nix, but I want to run shell.nix, how would I go about that?
<{^_^}>
[nix] @edolstra pushed to 2.3-maintenance « Bump version »: https://git.io/JTDGf
anderslu1dstedt has joined #nixos
gustavderdrache has joined #nixos
peterhoeg has joined #nixos
<patagonicus>
I'm currently setting up automatic backups using borg, but I have multiple jobs that go to the same repo. I'd like them to run sequentially and I tried adding systemd.services.borgbackup-job-foo.after = ["borgbackup-job-bar.service"];, but they are still running in parallel. Any way I can model that without having to rely on just spacing out the
<patagonicus>
startAt times enough?
Heirlung- has joined #nixos
Heirlung has quit [Ping timeout: 240 seconds]
Heirlung- is now known as Heirlung
anderslundstedt has quit [Ping timeout: 272 seconds]
<{^_^}>
[nixpkgs] @ehmry opened pull request #102048 → elf-dissector: init at unstable-2020-11-14 → https://git.io/JTDZI
<patagonicus>
Ah, actually adding "wants" as well and only having the foo job on a timer could work.
Rusty1 has joined #nixos
anderslu1dstedt has quit [Ping timeout: 265 seconds]
<patagonicus>
Nope, still stars both at the same time, making one fail because it can't acquire the lock. :(
fendse has joined #nixos
<{^_^}>
[nixpkgs] @siraben opened pull request #102049 → cr16: add cross-compile support → https://git.io/JTDZr
<patagonicus>
Ah. The module doesn't set Type=oneshot, so systemd considers the backup job up (done) as soon as it's being started and doesn't wait for completion. That sounds like a bug in nixpkgs.
<kkoekoe>
But I found that myfun (which builds a derivation) has `a` as an extra inputDrv when used from nix
<hexa->
nbathum: i'm currently busy at work, will take a look later
<kkoekoe>
gchristensen: Because is called from a script that creates `a` temporary stored at /tmp/<hash>-name. Then myfun fetches /tmp/<hash>-name from the path of `a`
<kkoekoe>
This is done bc nix cannot create `a` by itself
<kkoekoe>
(`a` is the result of a benchmark that runs in a remote cluster, and nix doesn't have the ssh keys to fetch it)
<{^_^}>
[nixpkgs] @onsails opened pull request #102064 → cargo-tauri-bundler: init at 0.9.3 → https://git.io/JTDgI
<kkoekoe>
I found that I can remove the context from the input a in myfun, then the derivation from the script and the derivation from inside nix are the same, but I need the inputDrv to point to a, as I want nix to build the dependency if needed. So I'm looking for a way to manually add a derivation in the inputDrv set
jonringer has joined #nixos
<utsl>
How do people usually deploy builds from Hydra? Just run Nixops or whatever from a Hydra jobset, or some other method?
<eyJhb>
If I nix-shell https://...../archive.tar.gz , that has a default.nix, how do I say it should run my shell.nix instead?
<noonien>
i'm having a problem that i'm not sure is nixos-related, but can probably be fixed for other users as well.
kkoekoe has quit [Remote host closed the connection]
<LambdaDuck>
V: Thanks. But when I try it, it says "chmod: cannot access 'subdir': No such file or directory". The directory does exist
<{^_^}>
[nixpkgs] @zakame opened pull request #102071 → perlPackages.NetAsyncWebSocket: init at 0.13 → https://git.io/JTDaA
<noonien>
i connect to the internet via a PPPoE connection, which has a MTU of 1492. the problem is that, because of this, when i run nixos-containers with private connections over a bridge, the bridge connection has a MTU of 1500, because of this, the tcp MSS is set to a value that is too big for packets to arrive back without being dropped
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « python37Packages.py_stringmatching: 0.4.1 -> 0.4.2 »: https://git.io/JTDVv
<noonien>
this happens on my server, which runs nixos, however, at home, where i have a laptop connected to a router that also has the same kind of pppoe connection, when i do `ping -n -s 1470 -M do google.com`, i get `ping: local error: message too long, mtu=1480`, in the nixos-container the command just hangs
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « python37Packages.bids-validator: 1.5.6 -> 1.5.7 »: https://git.io/JTDVz
<noonien>
from what i can tell, people in docker land just set a lower mtu on their bridges, but this seems like hack, perhaps the realm problem is that ICMP packets with the "message too long" don't arrive back in the container
zupo has joined #nixos
<pikajude>
are there docs living anywhere for using a flake dependency in a nixops deployment?
<cransom>
noonien: you can research 'mss clamping' to set in iptables that will force tcp connections to advertise a max mss that you set.
<noonien>
interesting, thanks for pointing that out. I wonde if that's what my router does
<noonien>
i'm guessing i can check this by doing tcpdump on both my laptop and a remote host, and check what MSS i send vs what MSS the target receives, right?
tobeportable has quit [Ping timeout: 264 seconds]
ManiacOfMadness has joined #nixos
<cransom>
yes.
mzcore_ has joined #nixos
<noonien>
thanks!
ckauhaus has quit [Quit: WeeChat 2.7.1]
andreas303 has joined #nixos
timstott has joined #nixos
<Taneb>
How can I specify a path to nixpkgs for nixos-rebuild
<patagonicus>
So, I have one disk with /boot and / and related things, but I also have a second disk that has LVM on LUKS and I'd like to automatically unlock LUKS, activate LVM and mount one of the logical volumes at boot (the key is stored on /). Is it enough to set the encrypted options under filesystems.foo and it'll just handle the LVM transparently? If so,
<patagonicus>
how would that work for multiple logical volumes, would I have to copy the encryption settings for each?
<patagonicus>
Hmm, looking at the implementation, my guess is a) you'll only want to specify an encrypted luks PV once and b) it shouldn't matter which filesystem you use for it. But now I'm curious how you'd make that work if the volume group spans multiple physical volumes, I think you'd have to unlock them manually.
<ncteeeest>
hi
<vandenoever>
just upgraded to 20.09 and seeing that different packages use different qt versions
<vandenoever>
but kmail will not install because is marked as brokebreeze-qt5-5.18.5n
nschoe has quit [Ping timeout: 268 seconds]
dbmikus has joined #nixos
timstott has quit [Ping timeout: 268 seconds]
solene has quit [Remote host closed the connection]
solene has joined #nixos
lorenzleutgeb has joined #nixos
andreas303 has quit [Ping timeout: 240 seconds]
erasmas has quit [Read error: Connection reset by peer]
therivercass has joined #nixos
reirob has joined #nixos
<reirob>
Hello everybody
<therivercass>
heya, trying to use appimage-run on a package but getting "libpulse.so.0: cannot open shared object file: No such file or directory". anyone know how I can make that available?
sss2 has quit [Quit: Leaving]
<lorenzleutgeb>
Hi! I have used overrideAttrs to modify z3 (a package in Nixpkgs) to produce an additional output called "java". In my devShell I'd like to use files from that output, so I'd like to get its location inside /nix. I have achieved getting the path with `nix show-derivation $(nix-store --query --deriver $(readlink -f $(which z3))) | jq -r
<reirob>
Just upgraded from 20.03 to 20.09. First, of all, many thanks to all the contributors, I am very grateful. I ran in few problems, and I thought maybe I share with you:
shibboleth has joined #nixos
<{^_^}>
[nixpkgs] @Taneb opened pull request #102076 → Set DynamicUser=true for hoogle → https://git.io/JTDPD
<reirob>
(1) When enabling the new hardware.video.hidpi.enable, then rebuild fails with an error
<reirob>
builder for '/nix/store/sri3bv1h184z0pqww63vranlbzq67p7j-extra-utils.drv' failed with exit code 1
<reirob>
Try 'cp --help' for more information.
<reirob>
cp: missing destination file operand after '/nix/store/ba4p71msbb4bvcjc8isxn9w7d5m21qpb-extra-utils/share/consolefonts/font.psf'
<jasom>
Can I have nix-build ignore my user configuration in ~/.config/nixpkgs?
<reirob>
Maybe somebody can look into this hardware.video.hidpi.enable?
<reirob>
(2) The upgrade took over 3 hours. A lot of stuff was compiled (ThunderBird, LibreOffice, VirtualBox). This seems to be different from 20.03. Will I have to wait so long for every update?
ddellacosta has joined #nixos
ransom has quit [Ping timeout: 268 seconds]
<reirob>
That's all for today. Will join maybe tomorrow again. Have a nice evening everybody
<jonringer>
I might just have us backport plasma 5.20 and gnome 3.38
<jonringer>
we are having a lot of issues with trying to "polish" those Dm's around system 246, and versions of the DM which weren't developed against it
<jonringer>
I don't really like the idea of scripts. As usually partitioning is very specific to a persons use case
<jonringer>
and generally we don't want to be making assumptions for them.
nschoe has joined #nixos
<jdnixx[m]>
<jonringer "and generally we don't want to b"> isn't that the entire point of a convenient installer though lol
<jonringer>
however, using something like calamares, give some "smart defaults" but the ability to change as necessary
<jdnixx[m]>
and I made mine with prompts asking for certain things to make sure
<supersandro2000>
give 5 standard partition setups and let everything else be done by the people itself
<supersandro2000>
they probably know what they want anyway
<jdnixx[m]>
<supersandro2000 "give 5 standard partition setups"> yeah there u go^
<jdnixx[m]>
well if it helps for context, I hadn't really used linux-based anything at all prior to March of this year
<jdnixx[m]>
I kinda consider myself still a fresh "convert" a little
<jdnixx[m]>
and it was pretty disappointing when I was doing my first nixos install
<supersandro2000>
and I always just use boot + ext4 /
<supersandro2000>
jdnixx[m]: are you already into aliases and functions?
<hexa->
fwiw: I came from arch and partitioning is something I've mostly done manually
<jdnixx[m]>
the partitioning was a huge bitch, I didn't know how to use parted (only ever stuck with graphical installers/gparted/KDE partition manager) and its syntax absolutely sucks tbh.
<jdnixx[m]>
<supersandro2000 "jdnixx: are you already into ali"> yep
<hexa->
so I honestly lack motivation to get into the installer use case
CyberManifest has joined #nixos
<jdnixx[m]>
just working my way up
<hexa->
at least I would be unlikely to use it
vandenoever has quit [Quit: Konversation terminated!]
emmanuel_erc has quit [Ping timeout: 260 seconds]
<jonringer>
the main pain point i have is just "this is not aligned for performance"
<jonringer>
with no clear way to get it to move over a few bits
<jonringer>
It's like... I have 1TB, you can take 4k in either direction. I don't care
<gchristensen>
parted will auto-align if you use %'s
sangoma has quit [Quit: WeeChat 2.9]
tput has joined #nixos
<hexa->
jonringer: even tools like fdisk and gdisk automatically align these days
<supersandro2000>
jdnixx[m]: little challenge: replace --force in git push with --force-with-lease in a git function. That was fun doing correct
<gchristensen>
parted won't auto-align if you use mb/gb/sectors/etc
<hexa->
with ssds you probably should align for erase block size though
<vika_nezrimaya>
When setting nixpkgs.pkgs to `import <nixpkgs> { overlays = [overlayWithPythonPackages]; }`, it looks like some NixOS modules continue to somehow pull packages out of a different Nixpkgs, since Python packages don't seem to work in uwsgi. Fully reproducible example in gitlab:vikanezrimaya/nix-flake/nixops - look at the containers.pyindieblog, when I try to make a container using a very old Nixpkgs version to make things work without
<vika_nezrimaya>
fiddling with legacy software that I need before I finish my rewrites
asheshambasta has quit [Ping timeout: 264 seconds]
tobeportable has joined #nixos
<vika_nezrimaya>
Is there a way to sidestep that? e.g. by setting `containers.pyindieblog.path` to `(<nixpkgs>.lib.nixosSystem { modules = [<config>]; })`?
davidv7 has joined #nixos
<vika_nezrimaya>
oh wait I forgot to mention that the file the error in is nixops/logical.nix
domogled has quit [Ping timeout: 260 seconds]
ManiacOfMadness has quit [Ping timeout: 268 seconds]
ransom has joined #nixos
domogled has joined #nixos
mallox has joined #nixos
justan0theruser has quit [Ping timeout: 264 seconds]
ris has joined #nixos
hlavaty has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @lorenzleutgeb opened pull request #102086 → z3: Add output for Java bindings (*.jar and *.so) → https://git.io/JTDH1
<vika_nezrimaya>
ok turns out you can't set path to a derivation
<vika_nezrimaya>
nah it's just me
<vika_nezrimaya>
i'm dumb
<vika_nezrimaya>
ok it seems to work
<vika_nezrimaya>
i wonder if it will be properly included in the system closure
CyberManifest has quit [Quit: Leaving...]
lorenzleutgeb has quit [Remote host closed the connection]
<qyliss>
I'm sure I've asked this before, but can I pretty-print a .drv?
alp_ has joined #nixos
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos
<exarkun>
nix show-derivation?
<cole-h>
^
<qyliss>
ah yes thank you!
<exarkun>
I can't build anything anymore. "error: stroull". Help?
<cole-h>
qyliss: Looks like it outputs in JSON, so you can use jq to parse it if you need.
<tilpner>
exarkun: Do you have build machines registered?
<exarkun>
yes
<exarkun>
hmm
<exarkun>
but only for armv7
<tilpner>
Check the number formatting
<exarkun>
oh no there's multiple architectures registered, my mistake
<exarkun>
The number formatting?
<tilpner>
Somewhere in the middle
<exarkun>
I have them configured with buildMachines in configuration.nix
<exarkun>
oh there's a /etc/nix/machines too
<tilpner>
Yes, I meant /etc/nix/machines
<tilpner>
stoull is used to parse something in the middle of each line of that file
<tilpner>
(Or for fetchers, but your gist didn't suggest anything was being fetched)
<exarkun>
ah. okay I deleted /etc/nix/machines and stuff builds now.
<exarkun>
thanks!
<tilpner>
No, I just meant to narrow it down
<tilpner>
We can probably fix your machines file
<sephii>
I'm using a custom keyboard layout with services.xserver.extraLayouts.symbolsFile, which works fine, but I'd like to test some changes without having to `nixos-rebuild switch` every time (because it then compiles X every time). Is there a way to do that? I tried copying the contents of my layout in a file and use `xkbcomp mylayout $DISPLAY` but it didn't do anything.
<exarkun>
I'm not sure I need that configuration anymore at all. The file has been the same for months, maybe years. The thing that maybe has changed is upgrading nixpkgs.
<tilpner>
I suspect your inlined ssh key is confusing the tokenisation
kreyren has quit [Remote host closed the connection]
<exarkun>
tilpner: Yes I have no idea if it has been used recently. I haven't _tried_ using it recently. :)
kreyren has joined #nixos
<halfbit>
paho-mqtt-cpp has a cmake find_package for paho-mqtt-c, which is in the nativeBuildInputs of paho-mqtt-cpp mkDerivation, but it can't find it
<halfbit>
I'm really unsure of why
<halfbit>
building it using nix-build -A paho-mqtt-cpp packages.nix with the above gisted files
<tilpner>
wnklmnn: When did you install it? Before or after updating to 20.09? Does your users channel also point to 20.09?
katyucha has joined #nixos
<witchof0x20[m]>
I'm having an issue using nat to allow a container with private networking to access the internet. I have nat = { enable = true; internalInterfaces = ["ve-container"]; externalInterface = "eno1";}
<witchof0x20[m]>
firewall.service is failing to start: with the error: `firewall-start[2777]: iptables v1.8.5 (legacy): unknown option "--set-mark"`
<wnklmnn>
I was running 20.03 previously. I changed the nixos channel to point to 20.09 ran a nixos-reuild switch restarted the system. logged back in and ran nix-env --upgrade.
<cole-h>
jonringer: Oof, inf recursion due to that change...
<tilpner>
wnklmnn: Does "nix-channel --list" (as your normal user) mention 03 or 09?
Fare has quit [Ping timeout: 264 seconds]
<wnklmnn>
the command does not return any channels. so I assume it is using the system nixos channel
<{^_^}>
[nix-mode] @matthewbauer pushed to master « Skip descriptions in nix completions »: https://git.io/JTDdv
selfsymmetric-mu has quit [Ping timeout: 240 seconds]
<tilpner>
If you install firefox via environment.systemPackages (or users.users.wnklmnn.packages), does it fail too?
<tilpner>
(Don't forget to verify you're starting the correct one. Use "which", or temporarily uninstall the nix-env version)
<wnklmnn>
tilpner I'll give that a try.
<cole-h>
I think you need to `fc-cache -f` to regenerate your fontconfig cache?
<tilpner>
wnklmnn: Try that ^ first, it's quicker
saschagrunert has quit [Remote host closed the connection]
<wnklmnn>
installing it via users.packages fixes the issue.
<tilpner>
halfbit: Blind guess: are you lacking pkgconfig?
<tilpner>
(nativeBuildInputs = [ pkgconfig ];)
wnklmnn has quit [Remote host closed the connection]
<witchof0x20[m]>
Ah, solved that step of the issue. Looks like NAT requires the `ipt_mark` module to be loaded and I'm using the hardened profile so it wasn't loaded
<supersandro2000>
is there a good guide for builders option in nix.conf? I am ahving trouble finding the syntax of it
<wnklmnn>
updating the font-cache does not fix the issue though.
<Taneb>
I just updated to latest unstable and now steam and minecraft aren't launching :(
alp_ has quit [Ping timeout: 264 seconds]
<tilpner>
Time to write VM tests you can execute before upgrading next time /s
<theduke>
I'm currenlty working on updating the `nvidia-docker` package and could use some guidance. nvidia-container-cli expects a ldconfig cache file to be present and loads it manually. Currently there is a pretty ugly path to the C code that works around this, but there were some upstream changes and the patch is pretty invasive anyway. Is it reasonable to create a ldconfig cache in the installPhase?
<theduke>
Currently the patch just changes the code to create it at runtime in a temp path.
<bqv>
jonringer: that's the thing, i'm not talking about one package in particular, i've found at least 5 just today, and i feel like the vast majority of those in the list are not in fact broken
<jonringer>
hmm, now that I think about it, I think I only marked two haskell applications broken
<jonringer>
I'm just concerned that I may have "over-marked" certain items as broken
<bqv>
i feel like there should be a more effective system for this, like perhaps marking it broken only for a certain hackage version
<bqv>
from what i see from peti's stream, they only get marked unbroken if their "brokenness" breaks some important package
<{^_^}>
[nixpkgs] @luc65r opened pull request #102099 → lib.licenses: add sata license → https://git.io/JTDx0
<jonringer>
I skipped marking the kernel modules broken because I was concerned about erroneously marking some things broken.
ransom has quit [Ping timeout: 264 seconds]
<bqv>
yeah, i'm not sure there's a trivial way to achieve that in haskellPackages, so it's probably best to discuss it with peti or similar
wnklmnn has joined #nixos
<bqv>
.tell peti perhaps check #nixos logs for this discussion
<jonringer>
I added some helpers like `kernelOlder`
<bqv>
(bot?)
<jonringer>
however, I think haskell modules has some file of all the broken package names
<jonringer>
not sure if you even can do conditional broken marking
wnklmnn has quit [Client Quit]
<jonringer>
for haskell, it might be nice to have something like `ghcOlder` or `ghcAtLeast`, so the
<bqv>
it does, yeah. there is also pkgs.haskell.lib.markUnbroken, which is the usual way around it, but the fact that it's just a list of names means it'd be a pretty significant change to add more metadata
<jonringer>
yea....
alp_ has joined #nixos
<halfbit>
how does cmake get run by nix
<halfbit>
I mean I have mkDerivation with src, but no where in there do I specify something like mkdir build && cd build && cmake .. kind of thing
<energizer>
i'm missing something; https://bpa.st/DPTTQ makes the output a directory rather than a script
<tobiasBora>
Hello, I'd like to know, a friend of mine has an old computer with a NVIDIA Corporation C61 [GeForce 7025 / nForce 630a]. She was running ubuntu 14.04 without any issue, but now it's not maintained I tried to upgrade to ubuntu 20.04, but unfortunately the nvidia driver is not available anymore on ubuntu (at least it's not automatically detected by ubuntu in the proprietary driver list). The problem
<tobiasBora>
is that the system is not stable with the open source driver. Any idea if there is a chance to have a working driver on NixOs? I don't even know how to check if the card is supported on a given driver. Here is the output of lshw and lspci: http://paste.debian.net/1169128
<halfbit>
ok now I get some bogus thing about /sbin/ldconfig not found when building paho-mqtt-c
<jonringer>
you need prior to that, >=4.18 is broken
<jonringer>
so i think you're stuck with _4_4
<tilpner>
That is what I said :)
<jonringer>
oh, yea.... I'm tired
costrouc has quit [Quit: costrouc]
<tobiasBora>
tilpner: oh interesting... Thanks a lot, I'll try (just need to find a way to install NixOs remotely, from a 32bits host). Latest NixOS works with this kernel? And is there a way to list all nvidia drivers that work for a given card?
<tilpner>
There is a linuxPackages_4_4 on 20.09, yes. Is the target machine 32-bit as well?
<tilpner>
There might be a way, but I don't know of it
<exfalso>
hi, I'm writing an IFD where my derivation is producing a .nix file. I'd like this .nix file to refer to a *global /nix/store* path inside a string, however this doesn't work as the string context of said string is empty. If I do ''${/nix/store/bla}'' then it's trying to copy the path into the store again. Any ideas? Is there an explicit way to attach a string context to a string?
<{^_^}>
[nixpkgs] @jonringer pushed to master « libguestfs: inherit hydraPlatforms from appliance »: https://git.io/JTDjC
<{^_^}>
[nixpkgs] @NeQuissimus pushed 7 commits to release-20.09: https://git.io/JTDjl
alp_ has quit [Ping timeout: 264 seconds]
<exfalso>
to be more precise, currently the derivation is producing a .nix file that contains "''/nix/store/bla''", but when evaluating this .nix file through IFD it doesn't see the path as the sandbox doesn't expose it
<{^_^}>
[nixpkgs] @NeQuissimus pushed 5 commits to release-20.03: https://git.io/JTDjz
la-jesystani has joined #nixos
<la-jesystani>
howdy, im really stumped with this. im trynig to use an inputs overlay in my nixos flake, as shown in the tweag article, but i keep getting an error that its not of the type nixpkgs overlay, even though as far as I can tell it is?. heres the file where im defining the flakes overlay, and in that same repo is the file with the overlay itself (curried with a single argument)
<alienpirate5>
warning: dumping very large path (> 256 MiB); this may run out of memory
<alienpirate5>
what does this mean?
<jonringer>
most of the time, that means you're refering to a local path e.g. "./."
<alienpirate5>
I am referring to a local path
<jonringer>
nix has to "nixify" paths, it will add that path to the nix store. I believe it has to archive the directory into a format that can be hashed, then hashes it
<exfalso>
alienpirate5: if the local path contains big files (for example build folders) then all of those files will be copied into /nix/store
<alienpirate5>
I see
<notgne2>
because it's given to eachDefaultSystem
<alienpirate5>
is that why I removed 137GB of garbage
<jonringer>
yea
<alienpirate5>
got it
<notgne2>
which the flake schema disallows, `overlay` and `overlays` should not be per-system as you can use `pkgs.system`
lsix has quit [Ping timeout: 268 seconds]
<alienpirate5>
now I am having a problem with ruby / bundix. when I try to run my ruby app from its location in the nix store, I get an error about bundler not being able to find gems
<jonringer>
nix isn't great to iterate on for development, since it can't share previous build results
<exfalso>
alienpirate5: use buildins.filterSource, or pkgs.lib.cleanSourceWith to filter stuff from your source path
<notgne2>
la-jesystani: tl;dr put `overlay =` in a separate attrset to the one given to `eachDefaultSystem`, merge this attribute set with the result of `eachDefaultSystem` instead
<jonringer>
generally you want to use nix to create a dev environment. However, keeping around a default.nix might be nice for CI or quick installation
<alienpirate5>
<jonringer "nix isn't great to iterate on fo"> I'm just trying to get the package working so I can deploy the app
<exfalso>
this is also important in order to not get your derivations constantly invalidating
<jonringer>
oh yea, for packaging, obviously nix is a great choice
<tobiasBora>
tilpner: cool! So not really, the problem is that I installed by mistake a 32bit xubuntu. Do you think it's possible to install by SSH nixos 64 bits from a 32 bit system?
<{^_^}>
[nixos-search] @turboMaCk pushed 3 commits to turboMack/refactor-route-structure: https://git.io/JTyeo
<alienpirate5>
I am unable to find any details on the error I am getting, probably because bundix is very obscure
<alienpirate5>
is there a nixos-ruby channel?
<qyliss>
alienpirate5: there is, by that name
<alienpirate5>
ok
<alienpirate5>
matrix says it hasn't seen a message since May but I don't think that's right
<qyliss>
it's not, there was activity there a couple of days ago
ransom has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<qyliss>
maybe that's the last time a matrix user was in there or something
luc65r has quit [Quit: WeeChat 2.9]
<tilpner>
tobiasBora: I don't think you can kexec 32->64, but I haven't tried it
<tilpner>
tobiasBora: Can you instruct the other side to boot a specific ISO?
<la-jesystani>
notgne2: that did it! thank you!
moredhel_ has quit [Quit: Connection closed for inactivity]
mallox has quit [Quit: WeeChat 2.9]
<la-jesystani>
notgne2: im having a smaller problem now, where i need to pass a list of built vim plugins to my overlay, and to do that i need a pkgset to access vimUtils from, and from what I can tell i need a system to inherit in order to do that
<alienpirate5>
this is very strange. my app works great in a nix-shell, but when i build the package and try to run the launcher script, it fails
ifthenelse has quit [Quit: see ya!]
<alienpirate5>
actually the launcher script fails in general
ifthenelse has joined #nixos
<la-jesystani>
alienpirate5: are you trying in a pure nix shell?
<notgne2>
la-jesystani: if I'm understanding right, you could instead pass it a function that either takes `pkgs` or `pkgs.system` and then returns the plugins
acarrico has quit [Ping timeout: 264 seconds]
<alienpirate5>
let me try that
ransom has joined #nixos
<{^_^}>
[nixos-search] @turboMaCk pushed to turboMack/refactor-route-structure « simplify further »: https://git.io/JTyv6
jabster28 has joined #nixos
proofofkeags_ has joined #nixos
vesper has joined #nixos
<{^_^}>
[nixpkgs] @kalekseev opened pull request #102103 → ssm-session-manager-plugin: 1.1.61.0 -> 1.2.7.0, add macos support → https://git.io/JTyvS
<notgne2>
la-jesystani: what I would do personally is just put your `buildVimPlugin` definition inside the overlay
katyucha has quit [Quit: Lost terminal]
<jonringer>
just realized we don't have a cmake section in the nixpkgs manual :(
kirc has joined #nixos
<halfbit>
jonringer: thanks!
<tobiasBora>
tilpner: not really, but I can try to change the bios to boot it automatically... But I'm affraid this remote install is quite risky: if anything fail, it can easily break the boot process :-\
<halfbit>
yes, there isn't a good guide on how to package up a cmake project that I found
<notgne2>
but you could also change `customPlugins = {` to `customPlugins = pkgs: {` and call it from inside of overlay.nix
<{^_^}>
#102104 (by jonringer, 15 seconds ago, open): Add section about using CMake in the nixpkgs manual
proofofkeags_ has quit [Remote host closed the connection]
proofofkeags_ has joined #nixos
<tilpner>
tobiasBora: You can probably install Nix on Xubuntu and build nixos-install, then do a normal-ish install. It will deviate slightly from the manual, and definitely won't be risk-free
<woffs>
Hi. Last merge from staging-20.09 (including pango: 1.45.3 -> 1.45.5) into release-20.09 did apparently not include some big packages (libreoffice, ...) in cache.nixos.org. Is there something broken in the channel management?
werner291 has quit [Remote host closed the connection]
samhza has quit [Remote host closed the connection]
proofofkeags_ has quit [Remote host closed the connection]
sandcat is now known as jess-o-lantern
<jonringer>
woffs: is it marked "big-parallel", maybe it's still queued?
knupfer1 has joined #nixos
proofofkeags_ has joined #nixos
<la-jesystani>
oh i see now, thanks notgne2
<tilpner>
tobiasBora: You don't stand a realistic chance at debugging driver issues without remote assistance. At least remote eyes and reboot are needed if you get any freezes
<worldofpeace>
gchristensen: thx, all done I think
<jonringer>
a channel will only block if a required "constituent" fails. Libreoffice is not one of those. As long as the build is in a terminal state, it will move forward
<jonringer>
libreoffic is pretty painful to build locally
<gchristensen>
jonringer: there was an error in the configuration of hydra and the machine where hydra thought it was big-parallel but the machine wasn't, that should be fixed now
<gchristensen>
someone with restart-jobs access to hydra should be able to restart it
<nbathum>
hexa-++
<{^_^}>
hexa-'s karma got increased to 8
<nbathum>
thanks for reviewing
alp_ has joined #nixos
<jonringer>
eventually those should be queued again, as most of the big-parallel packages have huge dependency closures
<woffs>
I see. Thanks for restarting, I think, many people just upgrading to 20.09 are heavily wondering why libreoffice (and Thunderbird and Virtualbox were mentioned, too) locally
<tobiasBora>
tilpner: I can have eyes, but my friend has very basic computer knowledge... I may give it a try if I've the time, otherwise I'll wait a bit and do it in real time. Thanks a lot for the help!
<tpw_rules>
very basic computer knowledge and you're trying to install nix for him?
rnea has quit [Ping timeout: 260 seconds]
kirc has quit [Read error: Connection reset by peer]
<energizer>
it says i havent specified the root partition, but i have. also it wants me to specify grub devices, but grub is disabled. what's going on? https://bpa.st/VG2VQ
<jonringer>
if what they do most of the is email and browse the web, then really any distro will do fine
<jonringer>
I had my dad go to ubuntu in like 2014, and just told him to run this update script and enter his password once every so often
<tpw_rules>
i've tried that for relatives a couple times and then next time i see them they have a new computer...
<energizer>
unattended-upgrades ftw
<jonringer>
Well, he was able to use it up until 2018, and then was having 32bit issues with browsers. And then moved over to back to windows, as that's what he was used to from work
<{^_^}>
#101132 (by deviant, 1 week ago, open): iosevka-bin: add support for variants
<woffs>
A funny side effect: csvkit is suffering from python-modules/agate-sql marked as broken although csvkit overrides agate-sql's version and is not affected from its brokenness
justanotheruser has quit [Ping timeout: 240 seconds]
<woffs>
It is even already in the cache
<halfbit>
can I get a nix-shell in a sandbox?
<jonringer>
halfbit: nix-shell --pure
<halfbit>
thank you
la-jesystani has quit [Quit: Lost terminal]
tilpner has left #nixos [#nixos]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<jasom>
I see: /nix/store/ic0pky86h7q0xjh8w3pv7lxiyjfpf926-icu4c-67.1-dev/bin/uconv but when I do a nix-env -iA nixos.icu.dev I get no uconv in my enviroment bin directory?
neiluj has joined #nixos
<jonringer>
woffs: hmm, really agatesql should do a release with the most recent sqlalchemy
<jonringer>
they haven't done a release since July
neiluj has quit [Client Quit]
aaabbbbbbbbbb has quit [Read error: Connection reset by peer]
<jonringer>
actually, there's been no commits to it since their last release
<{^_^}>
[nixpkgs] @MetaDark opened pull request #102109 → git-review: build with python3 → https://git.io/JTykn
Siilwyn has joined #nixos
<mananamenos>
hi, my `sudo nix-channel --list` is nixos https://nixos.org/channels/nixos-unstable. Why doesnt `sudo nixos-rebuild switch --upgrade` upgrade to 20.09? I still see 21.03pre249162.1dc37370c48 (Okapi) after
<nicolas[m]>
Because the "new" unstable is 21.03 > 20.09
domogled has quit [Quit: domogled]
domogled has joined #nixos
<mananamenos>
nicolas[m], embarrasing i thought i was looking at 20.03..