<17WAA85G0>
I'm running nixos-unstable, freshly updated, and can't seem to get dwarf-fortress-full to run. Getting "SDL initialization failure: No available video device". swaywm, firefox, etc runs fine, any idea how I can start hunting down the issue?
<bkv>
,locate bin update-desktop-database
<{^_^}>
Found in packages: desktop_file_utils
<bkv>
,locate bin gtk-update-icon-cache
<{^_^}>
Found in packages: gtk3-x11, gnome2.gtk
<bkv>
,locate bin glib-compile-schemas
<{^_^}>
Found in packages: glib.dev
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/JfNyB
<17WAA85G0>
kiwiirc: yup, haven't made any progress.
<17WAA85G0>
rather no, I mean. I tried to strace it but couldn't figure out what exactly it's missing
_ris has quit [Ping timeout: 246 seconds]
zebrag has quit [Quit: Konversation terminated!]
zebrag has joined #nixos
superbaloo has joined #nixos
nikita_ has joined #nixos
<{^_^}>
[nixpkgs] @veprbl opened pull request #91264 → texlive: fix build for scheme-infraonly → https://git.io/JfNQ5
nikita` has quit [Ping timeout: 260 seconds]
_ris has joined #nixos
<17WAA85G0>
Got it to run with SDL_VIDEODRIVER=x11. Interesting.
felixfoertsch23 has joined #nixos
felixfoertsch has quit [Ping timeout: 246 seconds]
felixfoertsch23 is now known as felixfoertsch
aw has quit [Quit: Quitting.]
spacefrogg has quit [Quit: Gone.]
voobscout has quit [Ping timeout: 240 seconds]
aw has joined #nixos
spacefrogg has joined #nixos
tsrt^ has quit []
knerten1 has joined #nixos
knerten2 has quit [Ping timeout: 240 seconds]
ltiawabm^ has joined #nixos
icey_ has joined #nixos
sheeldotme has joined #nixos
veleiro has joined #nixos
LnL has quit [Ping timeout: 260 seconds]
lawr3nce has quit [Quit: Leaving]
<veleiro>
what's the best way to replace an attribute in a list? just need to switch out a patch in mesa with a newer version of the patch, in an overlay
LnL has joined #nixos
LnL has quit [Changing host]
LnL has joined #nixos
mtn has quit [Remote host closed the connection]
drakonis has quit [Quit: WeeChat 2.8]
waleee-cl has quit [Quit: Connection closed for inactivity]
<notgne2>
in a package I'm building I have a tarball for one package containing another tarball for another package, this would be easy to deal with as I can just point the src of the other project to "${src}/otherthing.tar.gz" but unfortunately its in a subdirectory, how can I make it cd into that directory before building?
<notgne2>
or alternatively, extract that tarball so I can set src to the directory within it
acarrico has quit [Ping timeout: 240 seconds]
Supersonic112 has joined #nixos
Supersonic has quit [Ping timeout: 240 seconds]
Supersonic112 is now known as Supersonic
_ris has quit [Ping timeout: 246 seconds]
fendor_ has joined #nixos
_ris has joined #nixos
<tnks>
I think I have a handle on a lot of the channels maintained by the Nix community. But can someone explain how `nixos-20.03-darwin` is managed? How do changes from `nixpkgs-unstable` and `nixos-20.03` feed into it?
<DigitalKiwi>
lc_: also for future reference you can check https://logs.nix.samueldr.com/nixos it updates real time and is also useful for finding problems other people had that match yours
<samueldr>
DigitalKiwi: you're too slow :) but better url, date-less
<DigitalKiwi>
i started typing before you :(
<DigitalKiwi>
and yeah had to get the right url! :D
<lc_>
Alright. I'm trying to use coc.nvim in neovim with python. It's not working as I have it setup, because python-language-server from microsoft is broken, and jedi is outdated
<lc_>
how do I manually install a newer version of jedi in my nix-shell
<{^_^}>
[nixpkgs] @Ericson2314 pushed 418 commits to staging: https://git.io/JfNdE
sheeldotme has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
griff_ has joined #nixos
<lc_>
tfw everyone whos' asked similar questions so far got no response ;_;
Graypup_ has joined #nixos
<Graypup_>
hi! I've been setting up nix to manage haskell development environments and it is unclear what the difference is between the channels. I am doing something which I acknowledge is slightly evil: I **for now** want to install tools globally while I learn the Nix language and do better. ghcjs is broken on nixpkgs-unstable, so I want to use nix packages from nixos 20.03 where it works properly
<Graypup_>
can I just add the nixos-20.03 channel as nixpkgs?
<Graypup_>
is that like *egregiously* bad?
<Graypup_>
do I have to do anything odd if it's on a macOS host?
<Graypup_>
reason I'm asking is because there's a separate nixpkgs-20.03-darwin, but not a nixpkgs-20.03 (not darwin), just a nixos-20.03. How does this system work?
Orbstheorem has quit [Ping timeout: 272 seconds]
<Graypup_>
ALSO: how do the different git branches on https://github.com/nixos/nixpkgs work? I see a release-20.03 which is different from nixos-20.03 which seems contradictory?
<lc_>
I think I'm just going to begrudgingly install a debian docker container for my python stuff. Or use an online ide or something
<angerman>
,locate libtoolize
<{^_^}>
Found in packages: libtool, libtool_1_5
<angerman>
,locate glibtoolize
<{^_^}>
Couldn't find in any packages
<MichaelRaskin>
Graypup_: nixos- is snapshot of release- corresponding to the latest succesful-enough Hydra build
thomasjm has joined #nixos
<tnks>
MichaelRaskin: what determines what the definition of "successful enough" is?
codygman has quit [Read error: Connection reset by peer]
codygman has quit [Read error: Connection reset by peer]
codygman has joined #nixos
<jwaksbaum[m]>
I'm running nixos on a raspberry pi 4, and the `/boot` partition fills up. I've set all the `configurationLimit` options but it doesn't seem to getting rid of old profiles even after running `nix-garbage-collect -d` as regular user and root. How do I get it to clean the `/boot` partition?
<pjt_mpt>
Just a note" you should eventually expand /boot by a decent amount (or move it to / so it's not separate). Also, where'd you get the image? I thought /boot had not been separate in the isos for some time now.
griff_ has quit [Quit: griff_]
codygman has quit [Ping timeout: 265 seconds]
codygman has joined #nixos
<jwaksbaum[m]>
Hmm that's a good question, I think I built it myself based on some config somewhere. It worked and I haven't touched it since.
<pjt_mpt>
You built the iso?
orivej has joined #nixos
<jwaksbaum[m]>
How would I change that situation/upgrade to whatever the latest working thing is? I feel like when I made it burning the using the SD image with a separate boot partition was the only thing hat worked
<jwaksbaum[m]>
Yes I think I built the ISO.
<jwaksbaum[m]>
Because I needed to enable SSH and put on my public key, because I had no monitor, so they only way I could talk to it was by plugging it into the router.
<jwaksbaum[m]>
* Because I needed to enable SSH and put on my public key, because I had no monitor, so the only way I could talk to it was by plugging it into the router.
<DigitalKiwi>
that's when you get good at blind typing
<jwaksbaum[m]>
Okay, but is there a reason to totally reflash the pi if I have it working now? Can't I change the configuration.nix or mess with the partitions?
<notgne2>
this one I think you can just dd onto the sd card, no booting an iso and installing
gerschtli has quit [Quit: WeeChat 2.7.1]
<pjt_mpt>
jwaksbaum[m]: it can be fixed without reimiaging, yes
<jwaksbaum[m]>
Cool!
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
<pjt_mpt>
I just had the impression that your install was super out of date
gerschtli has joined #nixos
<notgne2>
yeah I think tweaking hardware-configuration.nix then rebuilding should probably be all it takes
orivej has quit [Ping timeout: 240 seconds]
<notgne2>
though you wont get back the space that /boot already takes up without some more messing around, which is made more difficult by the fact that iirc raspi4 doesn't support kexec
<jwaksbaum[m]>
I think the underlying like partition table is from an old image, yes, but I think it's running off a new version of nixos
<notgne2>
jwaksbaum[m]: I might be completely wrong and this might totally break your system, but you should just be able to remove the `filesystems."/boot"` entry from your hardware-confugration.nix
<jwaksbaum[m]>
notgne2: It needs kexec to resize the main partition while booted from it?
<pjt_mpt>
I think we have a different definition of old--mine was at 18.09 for ages xp
cole-h has quit [Quit: Goodbye]
<notgne2>
jwaksbaum[m]: I don't think you can resize the main partition while it's mounted, so generally I do it by using kexec to boot nixos into ram, then resizing and rebooting
<jwaksbaum[m]>
Oh fascinating, never heard of that before. I can take out the SD and mess with it from another computer
<pjt_mpt>
That's probably the easiest way
<pjt_mpt>
Do make a backup first though
<pjt_mpt>
it can be as simple as:
<pjt_mpt>
dd if=/dev/sdx | gzip > bkup.img.gz
<jwaksbaum[m]>
Sick.
<pjt_mpt>
Might've gotten that a bit wrong but it's close
<pjt_mpt>
also bzip'll be a bit better compression
eoli3n_ has joined #nixos
<pjt_mpt>
remember that backs up the whole thing, all partitions
simba1 has quit [Ping timeout: 260 seconds]
<jwaksbaum[m]>
But what's the end game in terms of boot. Right now I have something like `boot.loader.raspberryPi.enable = true` `boot.loader.raspberryPi.version = 4`
<pjt_mpt>
How big is /boot now?
<jwaksbaum[m]>
Does uboot work now and I can just use `generic-extlinux-compatible`?
<pjt_mpt>
If it fills up it's a pain to deal with
<pjt_mpt>
not sure
<jwaksbaum[m]>
128M
<pjt_mpt>
Hmm..I'm trying to remember what I had mine as.
<jwaksbaum[m]>
I mean I manually deleted some old stuff
<jwaksbaum[m]>
So it's not full anymore
<jwaksbaum[m]>
I just want to keep this from happening again
<pjt_mpt>
How big's the card?
pr06lefs has quit [Ping timeout: 260 seconds]
<jwaksbaum[m]>
I think 32GB
<pjt_mpt>
In that case you can afford sizing boot to like 256M or something for extra cushioning--or more if you want.
<pjt_mpt>
But yeah, setting a configuration limit's a good idea for sure
<pjt_mpt>
I think there's an auto garbage collect too if you haven't done that
<pjt_mpt>
and an auto store optimize option
sheeldotme has joined #nixos
<pjt_mpt>
yeah: nix.gc.{automatic,dates,options}
<jwaksbaum[m]>
Right, but right now it doesn't seem like it's deleting garbage-collected version from `/boot`
<jwaksbaum[m]>
`/boot` doesn't have symlinks, it has actual files.
<samueldr>
jwaksbaum[m]: no u-boot yet
<pjt_mpt>
hmmm
<samueldr>
(for pi 4)
<samueldr>
looked into it early this month
<jwaksbaum[m]>
So what is the current best way to have NixOS on the pi4, slash what even is the current boot situation? Can I get rid of my `/boot` partition?
<jwaksbaum[m]>
But tbh that all seems like beyond the point, I should be able to clean `/boot` using garbage collection, no?
<samueldr>
can't get rid of the boot partition
<samueldr>
it uses the raspberry pi bootloader usual way
user_0x58 has quit [Ping timeout: 244 seconds]
<samueldr>
jwaksbaum[m]: just like on x86_64, nix won't touch anything outside the store
<jwaksbaum[m]>
<samueldr "it uses the raspberry pi bootloa"> don't really know what that means, but that's fine
<samueldr>
pjt_mpt: only with u-boot
<samueldr>
you get into a chicken-and-egg situation where you need to build a generation to run the scripts, but the scripts fail to run because the boot partition is too full
<samueldr>
it is safe to rm the `/boot/old` directory and then rebuil
<samueldr>
rebuild*
pr06lefs has joined #nixos
<jwaksbaum[m]>
right but i manually delete some stuff on `/boot`
<jwaksbaum[m]>
so now there's space
<samueldr>
you would temporarily lose the older generations
<samueldr>
but it's basically the same issue that can happen on x86_64 with a separate boot partition
<pjt_mpt>
Quick question for you: How's the support on the Pi4 so far?
<jwaksbaum[m]>
<pjt_mpt "Quick question for you: How's th"> Like other than this, great.
<samueldr>
biasedly, I'd say terrible because it uses a vendor kernel, not mainline, and doesn't boot using u-boot
<samueldr>
but it works
<jwaksbaum[m]>
Yeah by great I mean it works fine for what I use it for.
<jwaksbaum[m]>
I haven't tried bluetooth, I use it more like a headless server.
<pjt_mpt>
Ah. It's a bit wonky on the 3B. I tend to need a plugin adapter >_<
<jwaksbaum[m]>
samueldr: I'm confused though: Right now I have free space in `/boot`, I have all the `configurationLimits` set, I have run garbage collection as normal user and root, but it's still not cleaning out old stuff in `/boot/old` that's past the configuration limit.
<samueldr>
I am not 100% positive it ever cleans it
<samueldr>
I have little experience with the raspberry pi bootloader
<samueldr>
I value my time too much to support this one-off platform :/
<jwaksbaum[m]>
Okay so you're saying the raspberry pi bootloader on nixos doesn't support this, but like the uboot one does, so once we get support for uboot this will be not an issue.
<jwaksbaum[m]>
<samueldr "I value my time too much to supp"> Gotcha totally fair.
Havvy has joined #nixos
orivej has quit [Ping timeout: 246 seconds]
<samueldr>
the u-boot one also can face a similar issue, when setup on a separate partition
orivej has joined #nixos
<samueldr>
but by default it uses the main big partition so it generally fits, and when it fills everything else fails too :)
palo1 has joined #nixos
<jwaksbaum[m]>
Hmm but on some platforms, I feel like, NixOS can clean `/boot` even on a separate partition. Or is that wrong?
<samueldr>
it does, but I have seen it happen at least with grub, you can get in the same pickle-of-a-situation
<samueldr>
where you need to erase something before you can rebuild into it
<jwaksbaum[m]>
Right, but once you make the space it will clean up after itself.
palo has quit [Ping timeout: 246 seconds]
palo1 is now known as palo
<jwaksbaum[m]>
I've like accepted that if you run out of space you're screwed, but I want it to be able to clean `/boot` assuming there already is some available space, right, by setting `configurationLimit` or something.
<samueldr>
it happens after everything else happened
<samueldr>
so it's "too late", you're full already!
<samueldr>
it never gets to it!
<jwaksbaum[m]>
Right but what if you're not full?
<samueldr>
it should work
<samueldr>
or else... something's to be fixed
<jwaksbaum[m]>
That's what I'm saying, it didn't for me. I manually created space, I redeployed and it worked fine, but there was still old shit in there that wasn't getting deleted.
<jwaksbaum[m]>
That behavior, cleaning old stuff out of boot when its not full, does that work on uboot and grub?
<samueldr>
differently, and yes
phreedom has quit [Remote host closed the connection]
eoli3n_ has quit [Ping timeout: 264 seconds]
<jwaksbaum[m]>
Okay cool. How differently?
<jwaksbaum[m]>
Looking at the code I think those regexes are too restrictive, the files that were left behind didn't match them.
phreedom has joined #nixos
<samueldr>
different scripts, which IIRC work using some trickery where it copies to a nix store
<jwaksbaum[m]>
Btw blowing away `/boot/old` and redeploying worked great, thanks.
<hazel>
how do I run a perl script with libraries? just pulling them from perlPackages.LWP (for example) and then `perl script.pl` does not work
<hazel>
in a nix shell
<jwaksbaum[m]>
samueldr: Right so it is kind of the same, in that they're copied from `/nix/store`, it just correctly cleans it some how.
<gregf>
I'm getting the following error while running nix-build -A netboot nixos/release.nix from the docs a 'aarch64-linux' with features {} is required to build '/nix/store/alsgzjpik8pfxgd34s5bk31fvp16c5bj-append-initrd-secrets.drv', but I am a 'x86_64-linux' with features {benchmark, big-parallel, kvm, nixos-test}
<samueldr>
literally removes everything it doesn't know about
<samueldr>
ah, no, just looked at your link, no regex, same approach as with grub
<samueldr>
checks that the file is one that was copied, if it isn't, removes it
<jwaksbaum[m]>
Right by "correct regex" I meant `*`
<samueldr>
much sturdier
<jwaksbaum[m]>
I think the reason though is the raspberry pi bootloader uses `default-system` and other `default-*` things, so maybe that's why? There must be a reason: clearly the raspberry pi script is based on the others, so why would they change it?
<jwaksbaum[m]>
viric: I think you wrote this, is there a reason why you did it this way instead of `$target/old/*`?
benschza has joined #nixos
jmeredith has quit [Quit: Connection closed for inactivity]
orivej has quit [Ping timeout: 240 seconds]
pr06lefs has quit [Ping timeout: 260 seconds]
orivej has joined #nixos
<jwaksbaum[m]>
Oh no, I think it's because there are others files that aren't being tracked, and those are the one's that aren't being removed. Also, it removes things silently, so maybe it was cleaning up some of the files without me noticing.
thomasjm has quit [Ping timeout: 240 seconds]
lc_ has quit [Ping timeout: 240 seconds]
lc has joined #nixos
orivej has quit [Ping timeout: 240 seconds]
<jwaksbaum[m]>
What's the deal with `system.build.installBootLoader`? I can't seem to set it myself in `configuration.nix`.
orivej has joined #nixos
<srk>
jwaksbaum[m]: it is typically set by bootloaders already, you could possibly mkForce it. uboot works well with rpis (except for pi4 which is still in progress)
abbe has quit [Quit: “Everytime that we are together, it's always estatically palpitating!”]
abbe has joined #nixos
<jwaksbaum[m]>
Right, I'm trying to override it and for some reason `mkForce` isn't doing anything.
<samueldr>
it won't be possible to mkForce as it's not a discrete option
<samueldr>
system.build is an internal-use attrset
<jwaksbaum[m]>
mm. so how should i override it? I can stop using the module that's currently setting it and just copy/paste the rest of the stuff its sets, but is there a better way?
<samueldr>
to me it seems unclear what installBootLoader should be
<samueldr>
ah
<samueldr>
found it
<samueldr>
it's something to run to make a configuration bootable
<jwaksbaum[m]>
sorry, i'll be clear: I'm trying to fix the issue we discussed in the raspberry pi bootloader script. So i want to replace that script with a modified copy and submit a PR once it works.
hazel has quit [Remote host closed the connection]
<quidome[m]>
I'm looking into running root on zfs, encrypted. My preference for encryption today would be to use luks and not native zfs. A few questions:
<quidome[m]>
- uefi machine, so having one esp partition outside of encrypted would be fine, the rest of disk would be luks with zfs inside, that should be working, right?
<quidome[m]>
- does the legacy mount recommendation still stand? Or could we move on? Documentation says legacy mountpoints
fpob has joined #nixos
tpham has quit [Read error: Connection reset by peer]
<srhb>
quidome[m]: Legacy mount still stands
Gaelan has joined #nixos
<srhb>
quidome[m]: To the setup: Yes, that is exactly what I am running
<quidome[m]>
alright, will go for that than. Shame about the legacy mounts. Thanks!
<quidome[m]>
* alright, will go for that then. Shame about the legacy mounts. Thanks!
<srhb>
quidome[m]: I don't feel they're bad at all. The name is just very opinionated and should probably be called "standard mount" :P
<quidome[m]>
:D
<quidome[m]>
well, if each dataset I create should be added to the legacy mounts, thats gonna be a long list
<elvishjerricco>
The Ubuntu installer can do this for you automatically, but it lacks flexibility like the ability to use native encryption or use raidz or mirrors.
<elvishjerricco>
The basic version of it is that you need to set the root dataset to canmount=no, then use `mount -t zfs -o zfsutil tank/root /target-root` in initrd. Then ZFS's packaged systemd services can do the rest in stage 2
yoctocell[m] has quit [Client Quit]
civodul has joined #nixos
<etu>
elvishjerricco: Actually it's quite easy to do encrypted ZFS on ubuntu, one just have to open up the bash script the installer execute and modify the command that creates the zpool. I've seen it been done and it seems to work.
<elvishjerricco>
etu: Yea, but I was mostly pointing out the implementation difficulties for distros
asheshambasta has joined #nixos
<etu>
yeah
<elvishjerricco>
NixOS has yet another issue in that it would have to avoid generating entries in `hardware-configuration.nix` for non-legacy ZFS mounts
yoctocell[m] has joined #nixos
<elvishjerricco>
It would be really nice to be able to set ZFS mountpoints though. For one thing, it'd help decouple system config from file system layout, which is desirable since you don't ever want to rollback to a generation that assumes a different file system layout.
<FRidh>
as it says, the package is not supported on that version. There can be different reasons for that, to find out you need to look into the Nix expression of the package.
fpob has joined #nixos
<{^_^}>
[nixpkgs] @FRidh pushed to master « ruby: use buildPackages.bison_3_5, fixes cross compilation »: https://git.io/JfNpc
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
noudle has joined #nixos
kaliumxyz has joined #nixos
zupo has joined #nixos
user_0x58 has quit [Ping timeout: 240 seconds]
Maxdamantus has quit [Ping timeout: 258 seconds]
Maxdamantus has joined #nixos
o1lo01ol1o has joined #nixos
cosimone has joined #nixos
<{^_^}>
[nixpkgs] @ShamrockLee opened pull request #91270 → chewing-editor: init at 0.1.1 Add package `chewing-editor` → https://git.io/JfNhZ
<{^_^}>
[nixpkgs] @FRidh pushed commit from @DavHau to staging « add autoPatchelfIgnoreMissingDeps to auto-patchelf.sh »: https://git.io/JfAe6
Guest2599 has joined #nixos
Guest2599 has left #nixos [#nixos]
sigmundv_ has joined #nixos
orivej has joined #nixos
KarlJoad has joined #nixos
<KarlJoad>
Hey guys, I am having troubles getting Rust's clippy to run. I have cargo, and can use it. However, when clippy is attempted to be run, I get "error: no default toolchain configured". Any thoughts?
<{^_^}>
[nixpkgs] @misuzu opened pull request #91272 → python3Packages.pip: allow setting reproducible temporary directory via TMPDIR → https://git.io/JfAJK
<dutchie>
how do the `imports = [ ... ];` lines work? Is it "magic" implemented in the configuration builder?
sab7iryudpgf6[m] has joined #nixos
<nwspk>
quick question: how can I build a python package based on pythonFull instead of python?
<nwspk>
using python38Full.pkgs.FOO results in using the "normal" python38. I'm kind of confused
<{^_^}>
[nixpkgs] @Mic92 opened pull request #91273 → home-assistant: add override for pyowm → https://git.io/JfAmM
nixbitcoin has quit [Ping timeout: 240 seconds]
nixbitcoin has joined #nixos
alp has joined #nixos
<elvishjerricco>
dutchie: No, it's just regular Nix code. There's a library function called `evalModules` that's written in pure Nix that implements all the logic for it. `nixos-rebuild` just evaluates `<nixpkgs/nixos>` to build a certain attribute in the resulting Nix value.
<KarlJoad>
Hey guys, I am having troubles getting Rust's clippy to run. I have cargo, and can use it. However, when clippy is attempted to be run, I get "error: no default toolchain configured". Any thoughts?
<KarlJoad>
I also know that clippy is installed and in my environment path because `which cargo-clippy` returns a path in `/run/current-system/sw/bin/cargo-clippy`, which points to the correct location in the nix store.
* DigitalKiwi
wonders if this is one of those things you're supposed to use nix-shell for and not install system wide
kenji has joined #nixos
veleiro` has joined #nixos
<KarlJoad>
DigitalKiwi: I wouldn't be surprised, but I'm not sure how to proceed from here.
<DigitalKiwi>
nix-shell -p clippy
veleiro has quit [Ping timeout: 264 seconds]
<KarlJoad>
Ok, yeah. I'm running through the rustlings course right now, and I am not sure the script/cargo program it uses would use the clippy that a nix-shell would provide.
<KarlJoad>
I'm using the `rustlings watch` command to have it interactively check the work I have done after every file save. It fails on exercises/clippy/clippy1.rs with error: no default toolchain configured
<DigitalKiwi>
also curse you for making me use rust
<DigitalKiwi>
*shakes fist*
<KarlJoad>
But your nix-shell command works and allows clippy1.rs to be evaluated.
matthiaskrgr has quit [Quit: Free ZNC ~ Powered by LunarBNC: https://LunarBNC.net - currently broken?]
<DigitalKiwi>
so problem solved?
matthiaskrgr has joined #nixos
__monty__ has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<KarlJoad>
I guess? I assume I need to make a something.nix file in the root of the project repo to have a "saved" nix-shell environment?
nikivi has joined #nixos
<DigitalKiwi>
i like lorri for that kind of stuff
Neo-- has joined #nixos
<DigitalKiwi>
and just make a shell.nix with those things
<DigitalKiwi>
or you can just do the nix-shell -p ... everytime (or just make the shell.nix don't need to use lorri but it'll make a root so it doesn't get gc)
<DigitalKiwi>
you can make the root yourself too
<DigitalKiwi>
i don't know how
<dutchie>
DigitalKiwi: > complains about using rust > uses lorri :p
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
dramforever has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @edolstra pushed to master « Update nixUnstable and nixFlakes »: https://git.io/JfAWL
<{^_^}>
[nixos-homepage] @edibopp pushed to integrate-nix-dev « Integrate nix.dev tutorials as a flake »: https://git.io/JfAWq
<{^_^}>
[nixos-homepage] @edibopp closed pull request #484 → Integrate nix.dev tutorials as a flake → https://git.io/JfNC3
orivej_ has quit [Read error: Connection reset by peer]
<{^_^}>
[nixos-homepage] @edibopp opened pull request #486 → Integrate nix.dev tutorials as a flake → https://git.io/JfAWG
orivej has joined #nixos
<{^_^}>
[nixos-homepage] @edibopp closed pull request #483 → Enable local link checking and fix a few outdated links → https://git.io/JfNGG
<{^_^}>
[nixos-homepage] @edibopp pushed 2 commits to link-checking: https://git.io/JfAWl
fendor_ is now known as fendor
<meh`>
is there a way to disable tests completely for the python environment? trying to get ceph to build again since the Python package upgrades, cherrypy/cheroot are failing some UNIX socket tests and I haven't been able to nixos-rebuild since then :(
<srk>
for one package or all of them?
<{^_^}>
[nixpkgs] @lsix opened pull request #91278 → pythonPackages.qscintilla-qt5: Do not copy pyqt5 tree → https://git.io/JfAlb
<meh`>
srk, I've been trying for one package but that doesn't seem to work, so I just want a temporary nuke until the issue is fixed in nixpkgs
<meh`>
since I was so dumb I ran a garbage collect after updating the channel and can't rollback
<srk>
meh`: in case of Python you can try overlay + override for buildPythonPackage but there's another catch - it will trigger rebuild of all of the Python packages you use
orivej has quit [Ping timeout: 240 seconds]
ebopp has joined #nixos
<srk>
meh`: so it might be better to doCheck = false only the few failing ones
<meh`>
srk, that's fine at the moment, still better than not being able to do any changes until it's fixed
<{^_^}>
meh`: Overlays look like `self: super: { foo = ...; }`. Use the self argument to get dependencies, super for overriding things and library functions (including callPackage). More info: https://nixos.org/nixpkgs/manual/#sec-overlays-install
<srk>
or maybe try override for buildPythonPackage
<meh`>
srk, that's within an overlay
<srk>
oh, right
<meh`>
well, that's doing something
<meh`>
overriding buildPythonPackage, let's see what happens
<meh`>
mmh, it's till running tests, guess it's not doCheck that's doing that
<srk>
yeah, take a peak at pkgs/development/interpreters/python/mk-python-derivation.nix line 149
<meh`>
srk, thanks, I was looking for the right file :P
* srk
traced it with grep .. pkgs/top-level/python-packages.nix: buildPythonPackage = makeOverridablePythonPackage ( makeOverridable (callPackage ../development/interpreters/python/mk-python-derivation.nix
<meh`>
mmh
<meh`>
even with doInstallCheck = false overridden it still runs the cheroot tests
<srk>
still meh? :D
<srk>
time to look at the package itself
<meh`>
srk, all I see is the checkPhase disabling some stuff, but shouldn't that be irrelevant if
<srk>
hmm, nothing special just a checkPhase
<meh`>
wait, is there also a buildPython3Package?
<meh`>
I'm gonna try forcing an empty installCheckInputs
<meh`>
still runs, I must be doing something dumb
<srk>
don't think so, buildPythonPackage builds for both Python versions iirc
<meh`>
does pipInstallPhase have anything to do with this?
<{^_^}>
[nixpkgs] @GuillaumeDesforges opened pull request #91279 → [WIP] haskell-language-server: init at 0.1.0.0 → https://git.io/JfAan
<srk>
meh`: probably not but I'm not that good with Python packaging, try fiddling with it a bit and maybe someone else responds who knows more about it
<patagonicus>
Hmm, no armv7l related channel. I'm trying to get NixOS booted on a Odroid HC2, but the newest armv7 images I can find (from the Nixos on ARM wiki page) seem to be from 2018. Trying to bootstrap Nix from scratch also proved a bit complicated. Doesn anyone have any pointers for running NixOS on armv7? (I'm also a bit blocked right now because of a hardware defect, but at least I still have a RPi3
<patagonicus>
that can run armv7)
<meh`>
srk, fair, thanks for the help!
<DigitalKiwi>
there's #nixos-aarch64 that it's ok to talk about armv7 in too
<srk>
patagonicus: #nixos-aarch64 is sometimes used for armv7l discussions
<meh`>
srk, altho, I also see a doCheckByDefault, now trying to figure out where to set that and just nuke the entire thing
<srk>
meh`: yeah, I've noticed that as well
<patagonicus>
Ah, makes sense. Thanks, I'll ask there. :)
kleisli has quit [Ping timeout: 256 seconds]
<meh`>
srk, can't figure out where to set it tho
<kiwiirc>
no #nixos-gaming chan? :(
absoluutely has joined #nixos
<DigitalKiwi>
not all nix channels are listed there
<DigitalKiwi>
it's only the largest ones
<DigitalKiwi>
so it's possible there is a #nixos-gaming
<{^_^}>
[nixpkgs] @flokli pushed 2 commits to staging-20.03: https://git.io/JfArT
<{^_^}>
[nixpkgs] @geistesk opened pull request #91281 → pythonFull: add BlueZ support → https://git.io/JfAr3
alp has quit [Ping timeout: 246 seconds]
<teto>
meh`: that was a funky one, I set that for no reason in config.nix, forgot about it and suddenly nix would rebuild world + it would fail. Must have lost one hour there :)
KarlJoad has quit [Quit: Konversation terminated!]
<meh`>
teto, I want your nightmare to happen on my machine as well!
<meh`>
because I already wasted more than hour on this anyway
o1lo01ol1o has joined #nixos
<teto>
meh`: can't you set `doCheck = false;` for the packages yo uare interested in ? you can default the test to false in pkgs/development/interpreters/python/mk-python-derivation.nix but that will rebuild a lot
zupo has joined #nixos
<meh`>
teto, that doesn't seem to work
<meh`>
I don't know what's running these tests anymore
<__monty__>
Can you override the checkPhase for the package?
graf_blutwurst has joined #nixos
<meh`>
__monty__, just tried, still runs tests
<meh`>
I don't know if I'm getting the wrong package or what
<__monty__>
Hmm, override all the phases with noops and verify? : )
<meh`>
__monty__, for Python it would be installCheckPhase, but even that doesn't werk, how do I find the list of all phases?
<meh`>
make-derivation.nix I guess?
<__monty__>
meh`: The package doesn't specify any phases explicitly?
<meh`>
__monty__, it doesn't seem to, it's using buildPythonPackage
<{^_^}>
[nixpkgs] @danieldk opened pull request #91284 → micro: install man page → https://git.io/JfAXN
elkowar has joined #nixos
charmonium has quit [Remote host closed the connection]
knupfer1 has joined #nixos
knupfer has quit [Quit: knupfer]
knupfer1 is now known as knupfer
drdee has quit [Quit: Leaving]
dr-dd has quit [Client Quit]
drdee has joined #nixos
<elkowar>
Hey guys, I'm trying to setup Nix package manager on Arch using the install script (from nixos.org), but it's getting an error (see this pastebin: https://pastebin.com/5QGrA1Le )
orivej_ has joined #nixos
orivej has quit [Ping timeout: 264 seconds]
<elkowar>
any help would be appreciated here,... is the installer broken? or am i messing something up. I had installed nix using the AUR before (for a few seconds xD)
knupfer has quit [Client Quit]
knupfer1 has joined #nixos
<simpson>
elkowar: Somehow a slash vanished from your path; it should look like /nix/store/ily1... and you can check that, if that slash were there, ily14d68xl11cnbbkf9svwnzwsrrnzah-bash-4.4-p23 is in your Nix store.
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<elkowar>
oh, yea that's only a copy pasting issue tho, in the actually error message there / is there ^^
<simpson>
Hm, curious. And the store path exists?
knupfer1 is now known as knupfer
<elkowar>
after attempting to run the installer, /nix/store/ily1* does not exist, no
<dbe>
For nix flakes, should I create my packages using in `packages.x86_64-linux.<pkg>` or using `overlay`? I just noticed that if I use `packages.x86_65-linux.name` then I need to call `nix run .#packages.x86_64-linux.name` - I can't do `nix run .#name`
<elkowar>
when looking at the unpacked tarball the install thing downloads, looking into unpack/nix-2.3.6-x86_64-linux/store contains a bash version that has a different hash (x7fr0bvnwvqvr3zg60di9jxvfwimcw7m-bash-4.4-p23)
<ptotter[m]>
I'm getting "Error: It looks like Stripe.js was not loaded correctly", any ideas?
<ptotter[m]>
used to work before
<buffet>
to add to elkowar's problem: it seems like the bash version in the $nixTarball/store has a different hash than the one it tries to run
<bbigras>
jmgai: Did you ever figure out how to use multicorn on NixOS? If you are the same person from last year's irc log.
<simpson>
elkowar: That's not a problem. Read /install, the install script. It seems that you're getting past l170, but not l177; nix-env seems to be installed and working, somewhat.
<simpson>
buffet: Also on Arch?
<buffet>
i was talking to him earlier, checked on his machine
orivej_ has quit [Quit: No Ping reply in 180 seconds.]
<elkowar>
the /nix/store dir currently contains 38 things, that doesn't sound like it worked well ^^'
<simpson>
elkowar: And yet nix-store from the Nix store did run successfully. So now I'm curious about l175 and what etc/profile.d/nix.sh might have in terms of problematic interactions with your existing shell configuration.
<elkowar>
if you need any more info, tell me where to look ;D I'd love to figure out what went wrong there
<worldofpeace>
aaron: you really kinda called it
<simpson>
elkowar: Well, as you can see from the installer, you're really almost done. What you've gotta do is figure out what $nix/etc/profile.d/nix.sh does to your bash configuration, and then run `$nix/bin/nix-env -i $nix` to get your Nix tools added to your user's environment.
<elkowar>
I'll just go with the AUR nix package for now, and then maybe switch later on if i notice any issues with that ^^ thanks anyways!
<{^_^}>
[nixpkgs] @mweinelt opened pull request #91287 → openttd-jgrpp: init at 0.34.4 → https://git.io/JfAMW
gustavderdrache has quit [Quit: Leaving.]
gustavderdrache has joined #nixos
sbetrium has joined #nixos
<karantan>
I have this default.nix file: https://pastebin.com/E4c79xix and when I do `nix-shell --pure` I expect to have e.g. `php` installed but I don't. If I put those packages (pkgs.tree, pkgs.php, ...) to `environment.systemPackages` (in configuration.nix) I have access to them but only if I do `nix-shell` (so without --pure).
<karantan>
how should I add those packages to default.nix so that I can call them?
<karantan>
hmm maybe I just need to add them next to `pkgs.poetry`
<karantan>
I guess the "nativeBuildInputs" could be for python packages
graf_blutwurst has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @rnhmjoj opened pull request #91288 → Remove old zeromq 3.x packages → https://git.io/JfAM5
<simpson>
karantan: Right, the build inputs to your application's derivation are not automatically listed among the build inputs to your environment's derivation.
orivej has quit [Ping timeout: 256 seconds]
orivej has joined #nixos
elkowar has quit [Remote host closed the connection]
Guest78 has joined #nixos
elibrokeit has joined #nixos
<jluttine[m]>
Can I somehow easily figure out which texlive package in NixOS I should install to get `adjustbox.sty`? I tried `nix-locate adjustbox.sty` but it found nothing. Perhaps Hydra doesn't build larger texlive packages?
jjwatt[m] has joined #nixos
<jjwatt[m]>
jluttine: I seemed to need 'full' for anything significant (e.g., exporting from emacs org-mode, some of the algorithm packages, etc)
acarrico has joined #nixos
rajivr has quit [Quit: Connection closed for inactivity]
<jjwatt[m]>
jluttine: If you really want to stay trim, you can try out tectonic. It's a newer implementation and it detects what you need and downloads them. I've been trying it on some of my documents, and I have been very impressed with most of the output. Usually can't see a difference between it and luatex, pdflatex, etc.
<{^_^}>
[nixpkgs] @gebner opened pull request #91290 → freecad: use python 3.7 → https://git.io/JfAyD
<{^_^}>
[nixpkgs] @ehmry pushed to master « av-98: init at 1.0.2dev »: https://git.io/JfAyA
orivej has quit [Quit: No Ping reply in 180 seconds.]
orivej has joined #nixos
Chiliparrot has joined #nixos
orivej has quit [Quit: No Ping reply in 180 seconds.]
orivej has joined #nixos
<{^_^}>
[nixpkgs] @Ericson2314 merged pull request #85189 → C++ compilers: Be sane with standard libraries → https://git.io/JfeZ0
<{^_^}>
[nixpkgs] @Ericson2314 pushed 3 commits to staging: https://git.io/JfA9k
waleee-cl has joined #nixos
<{^_^}>
[nixpkgs] @jorsn opened pull request #91291 → openlp: init at 2.4.6 → https://git.io/JfA90
griff_ has joined #nixos
orivej_ has joined #nixos
meh` has quit [Ping timeout: 260 seconds]
drdee has quit [Remote host closed the connection]
orivej has quit [Ping timeout: 240 seconds]
griff_ has quit [Client Quit]
pareidolia has joined #nixos
knupfer has quit [Quit: knupfer]
knupfer has joined #nixos
knupfer has quit [Client Quit]
<tnks>
A little guessing the answer to a question I had yesterday. I've read about how changes flow into `nixpkgs-unstable` and after passing some tests get into `nixos-unstable` and eventually through release management these end up in channels like `nixos-20.03`. So is `nixos-20.03-darwin` the same build/test as `nixos-20.03`, but on a MacOS build machine? Or does it get special treatment above and beyond that?
absoluutely has joined #nixos
knupfer has joined #nixos
ok2` has quit [Read error: Connection reset by peer]
<devalot>
What is the recommended source formatting tool for Nix? I started using nix-fmt when I saw home-manager was using it, but it also seems like nixpkgs-fmt is pretty popular.
ok2` has joined #nixos
cfricke has quit [Ping timeout: 246 seconds]
absoluutely has quit [Ping timeout: 258 seconds]
eyenx has quit [Read error: Connection reset by peer]
eyenx- has joined #nixos
Guest78 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<jtojnar>
devalot there is no official recommendation at the moment
Guest78 has joined #nixos
orivej_ has quit [Ping timeout: 260 seconds]
orivej has joined #nixos
<devalot>
jtojnar: Thank you. Do you know if there are plans to eventually require one for PRs? Sort of like how NixOS tests require the Python code to be formatted.
<jluttine[m]>
Jan Tojnar: Thanks!! 👍️
anderslu1dstedt has joined #nixos
<gchristensen>
devalot: it is complicated by having like a million LoC which isn't formatted, nad not wanting to obliterate history
acowley has joined #nixos
<jwaksbaum[m]>
gchristensen: Isn't there a way to have git ignore commits in terms of blame and history?
<gchristensen>
dunno
anderslundstedt has quit [Ping timeout: 258 seconds]
<devalot>
gchristensen: I totally get that. Here's a better question, which formatter, if any, do you use?
<{^_^}>
[nixpkgs] @bouk opened pull request #91292 → mockgen: init at 1.4.3 → https://git.io/JfAQE
<acowley>
Is there a manual page somewhere that describes what nix-shell is?
<acowley>
I'm writing a blog post and wanted to link to such a thing, but can't find one.
<gchristensen>
devalot: I use nixpkgs-fmt, and I turn it off when contributing to nixpkgs
<acowley>
jtojnar: Thank you! I checked the NixOS and nixpkgs manuals, but not the nix one!
alexherbo2 has quit [Ping timeout: 240 seconds]
alexherbo24 is now known as alexherbo2
<jwaksbaum[m]>
gchristensen: Yeah that's all I found, so I guess it wouldn't be worth it yet :(
ckauhaus has joined #nixos
<gchristensen>
I'd _love_ for gh to support it :)
<{^_^}>
[nix] @Ericson2314 opened pull request #3729 → hashed-mirrors: Use parsed derivation output rather than reconstructing it → https://git.io/JfA7e
<{^_^}>
[nixpkgs] @Ericson2314 opened pull request #91293 → treewide: Get rid of `cc.gcc` → https://git.io/JfA7m
<jwaksbaum[m]>
<gchristensen "I'd _love_ for gh to support it "> yeah me too. I hearted the issue, everyone else should too. without that the git feature is almost useless.
anderslu1dstedt has quit [Ping timeout: 260 seconds]
pingiun has joined #nixos
orivej has quit [Ping timeout: 246 seconds]
orivej has joined #nixos
magnetophon has joined #nixos
jmeredith has quit [Quit: Connection closed for inactivity]
<magnetophon>
Can someone recommend me a remote desktop viewing solution? Preferably one where the end that is being viewed, iow my parents PC, doesn't need to do anything.
<magnetophon>
Not really nixos related, but I'm not sure where to properly ask this. Both PC's *are* running nixos, though... :)
<sphalerite>
magnetophon: oh right, doesn't need to do _anything_ at all?
<magnetophon>
sphalerite: if possible
<sphalerite>
magnetophon: this sounds like you might have SSH access? ;)
<magnetophon>
sphalerite: it needs to be extremely simple though. My parents are old and digiphobic
<magnetophon>
sphalerite: thanks. SSH is a partial solution, I often want to show them how do do something.
growpotkin has joined #nixos
<sphalerite>
magnetophon: if you have SSH access and they have an X session open, you can use freerdp-shadow-cli to start an RDP server (also allows remote control) on an existing X session
meh` has joined #nixos
<magnetophon>
sphalerite: Also, I need to look up how to set ssh up. Do I need to open a port in my parents router? How do I make it so the PC's find each other? I know how to use SSH on a local netwerk, also how to set it up with keys and no password, but idk how to do itover the web
<magnetophon>
sphalerite: what are some search terms I need to look up to learn how to set this up?
<sphalerite>
magnetophon: ah right, yeah you'll need to open a port (though if they don't have a fixed IP that's also less useful, and CGNAT can also ruin everything), or use a VPN.
justanotheruser has quit [Ping timeout: 272 seconds]
<magnetophon>
sphalerite: I don't think they have a fixed IP.
<sphalerite>
magnetophon: I'd recommend using a VPN because it doesn't enable direct access.
<magnetophon>
I don't have a VPN yet. would a free one be safe enough for this? We oth do our banking on these PC, for example, so I don't want to put anyone at risk!
<magnetophon>
sphalerite: let me put it another way: do I need to trust the VPN for this, or does their trustworthiness not come into the picture at all, because the tech shields them off?
<sphalerite>
magnetophon: you can build one yourself (if you have a server with a public IP somewhere) using tinc or wireguard, which are slightly more involved options, or use a service like zerotier or… I forgot the name of the other one I had in mind
alp has quit [Ping timeout: 272 seconds]
alp_ has joined #nixos
orivej has quit [Quit: No Ping reply in 180 seconds.]
eoli3n_ has joined #nixos
jb551 is now known as jb55
<sphalerite>
magnetophon: it's not a VPN service as in a glorified proxy, as is often recommended (typically quite misleadingly) for improved privacy, but a VPN as in an actual virtual private network which allows devices to communicate with each other as if they're on the same physical network
<magnetophon>
sphalerite: I have ssh access to my cheap-ass dreamhost webserver, but I don't have root there. Would that work?
orivej has joined #nixos
fpob has left #nixos ["Leaving."]
<sphalerite>
magnetophon: hm, not sure on that one. Depends on whether you can run services I guess.
<sphalerite>
magnetophon: but probably something like zerotier or tailscale is better there.
<sphalerite>
magnetophon: they both have free tiers which would be suitable.
<magnetophon>
sphalerite: I checked: command not found: tinc, command not found: wireguard
* magnetophon
goes loo those up
<magnetophon>
*look
<magnetophon>
sphalerite: thanks, so far!
<sphalerite>
magnetophon: yeah it's unlikely the web server thing will work.
<sphalerite>
magnetophon: you could use a €3/mo hetzner cloud server or something though if you do want to build your own, which also has the advantage of being able to use nixos on it ;)
<sphalerite>
magnetophon: but building your own is rabbit-hole-ifying the problem, if you want to "just get access" I'd recommend a service like zerotier.
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
cinimod`` has quit [Ping timeout: 265 seconds]
<magnetophon>
sphalerite: I've been looking at getting a server like that, and putting nixos on it, but yeah, I know it''l be a massive timesink.
cinimod`` has joined #nixos
<cransom>
tailscale would also probably be an easy to use vpn setup as well
<magnetophon>
cransom: thanks, I'll look into that one too.
<Miyu-saki>
Is there a wrapper function that does the same setup as `nix-shell -p`?
<sphalerite>
magnetophon: anyway, once you have your VPN, you'll have an IP address for your parents' computer which you can access from your computer, and using that you can SSH in, start a remote desktop server, and connect to that.
<sphalerite>
Miyu-saki: a wrapper function in what language? For what?
<{^_^}>
[nixpkgs] @jbaum98 opened pull request #91295 → Rewrite raspberrypi-builder.sh based on extlinux → https://git.io/JfA5N
justanotheruser has joined #nixos
<Miyu-saki>
sphalerite: More context: Global-wide Emacs installation with Org+Jupyter. Jupyter needs Python modules setup, amongst other htings.
drakonis has joined #nixos
<magnetophon>
sphalerite: OK, thanks, sounds doable. Which remote software would you recommend in this scenario? Jitsi's main strenghth is that it fixes the IP issue, right? Oh, and I said I needed a viewer, ut actually, I want/need to interact too!
<Miyu-saki>
Oh right
<Miyu-saki>
I remember one hack that I used to fix this.
<Miyu-saki>
`python3.withPackages (...);`
<Miyu-saki>
sets up the environment
orivej has quit [Quit: No Ping reply in 180 seconds.]
orivej has joined #nixos
<Miyu-saki>
Sadly, that also "pulls in Python"
<sphalerite>
magnetophon: right, jitsi meet is an "out-of-the-box" option in that you don't need to install any special software on either end, you just open the same URL on both ends, have a video chat, and can also share the screen.
<sphalerite>
magnetophon: but yeah no remote control there — the freerdp-based solution does allow remote control though.
noudle has joined #nixos
<sphalerite>
you can also use x11vnc much the same way, but rdp is cooler :p
cinimod`` has left #nixos ["ERC (IRC client for Emacs 26.1)"]
<magnetophon>
sphalerite: I always associate rdp with windows, and thus assue it's not the est thing to use for linux <-> linux connections. I guess that's not true then? Why is rdp cooler?
<magnetophon>
*assume
GiGa has joined #nixos
<Miyu-saki>
sphalerite: This works, but I think is too roundabout. Any ideas? `nix run '(with import <nixpkgs> {}; python3.withPackages (pp: with pp; [jupyter]))' -c jupyter kernelspec list`
<GiGa>
Hello
<GiGa>
I have a new AMD Radeon RX550 graphics card
<GiGa>
Trying to decide what value to set for services.xserver.videoDrivers.
<dutchie>
probably amdgpu
<GiGa>
I can see options of "radeon", "ati_unfree", "amdgpu" and "amdgpu-pro" but I'm not sure how to determine which is the correct one.
<GiGa>
Thanks dutchie p
<GiGa>
*.
<GiGa>
I'll give that a go.
<dutchie>
not sure what the -pro does
abathur has joined #nixos
<GiGa>
I'll do a "nixos-rebuild boot" then powerdown and install the card. Not long until dinner though...
<GiGa>
Thanks for your help
GiGa has quit [Client Quit]
jakobrs has joined #nixos
noudle has quit []
<jakobrs>
Is anybody here using displayManager.startx? I'm trying to figure out how to set it up correctly.
<jakobrs>
I've tried `startx -- -config /nix/store/hash-xserver.conf` but the mouse, the font and all windows are messed up completely.
<jakobrs>
s/mouse/cursor/
cfricke has joined #nixos
<{^_^}>
[nixpkgs] @cawilliamson opened pull request #91296 → Correct typo in onlydev.nix → https://git.io/JfAdE
jb55 has quit [Remote host closed the connection]
<genevino>
jakobrs: i ended up just using home-manager to manage my .xsession file
horek has joined #nixos
<genevino>
jakobrs: that way i have full control over what's started when i type "startx" and that was more or less the only thing that really worked for me.
<genevino>
jakobrs: also has the benefit that i don't need to run X as root.
<jakobrs>
Okay I might look into that.
boxscape has joined #nixos
<jakobrs>
Alternatively it would be neat if I could get sddm to autologin with a command-line parameter.
orivej has quit [Read error: Connection reset by peer]
orivej_ has joined #nixos
<jakobrs>
I noticed that xsession-wrapper isn't in my closure when using startx so I assume that it's a thing that's pulled in by the sddm module?
Guest78 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
horek has quit [Quit: Leaving]
reanimus has quit [Ping timeout: 260 seconds]
_reanimus_ has joined #nixos
<{^_^}>
[nixpkgs] @siriobalmelli opened pull request #91297 → beancount_docverif: init at 1.0.0 → https://git.io/JfAFT
erasmas has quit [Read error: Connection reset by peer]
orivej has quit [Ping timeout: 260 seconds]
orivej_ has joined #nixos
jnbrains has joined #nixos
catharsis_[m] has joined #nixos
meh` has quit [Ping timeout: 265 seconds]
erasmas has joined #nixos
jakobrs has left #nixos ["WeeChat 2.7.1"]
dhess has joined #nixos
<gustavderdrache>
i'm looking into switching to nixos for my personal machine but am having some trouble finding particular pieces of information - for example, does nixos expect to manage which extensions are installed into my gnome-shell session?
<gustavderdrache>
i'm not having much luck finding odds 'n ends like this in the nixos manual
<sphalerite>
gustavderdrache: to answer your concrete question: I don't think so.
<sphalerite>
the documentation is indeed not the most complete.
<gustavderdrache>
ah
<gustavderdrache>
not the end of the world as long as i have some other places to look in case i get stuck
<gustavderdrache>
is there a community preference for here vs. e.g., discourse?
<sphalerite>
gustavderdrache: not really, it's a matter of personal preference. IRC tends to have somewhat lower latency I guess.
<gustavderdrache>
noted
<sphalerite>
I spend a lot more time here than on discourse for example :p
<gustavderdrache>
:)
<gustavderdrache>
in that case, consider this fair warning that i will probably ask some dumb newbie questions in a few days :P
NeoCron has quit [Ping timeout: 264 seconds]
orivej_ has quit [Ping timeout: 260 seconds]
philr_ has quit [Ping timeout: 246 seconds]
orivej has joined #nixos
asheshambasta has quit [Ping timeout: 272 seconds]
<cole-h>
No need to warn, just ask away. Anybody who can help you, will. We were all in your same position at some point ;^)
<multun>
gustavderdrache: there's no such thing as a dumb question, we all started out as newbies!
orivej has quit [Ping timeout: 265 seconds]
<gustavderdrache>
:)
<gustavderdrache>
appreciate the support
<sphalerite>
hyper_ch_: did you ever get the USB-C docking station in initramfs working?
orivej has joined #nixos
<gchristensen>
very proud of this channel right now
user_0x58 has joined #nixos
codygman has quit [Read error: Connection reset by peer]
codygman has joined #nixos
seven-eleven has joined #nixos
<seven-eleven>
hi
<seven-eleven>
how can I install packages only if they haven't been installed yet?
plutes has joined #nixos
<seven-eleven>
maybe? nix-env -i --preserve-installed fzf.... seems that still creates new symlinks but doesn't install new files?
<patagonicus>
seven-eleven: I'm not quite sure what you mean (but I'm also somewhat new to Nix). Running nix-env -i for a package that is already installed shouldn't do much. It should take a fraction of a second to figure out what needs to be installed, see that it's already there and then quit.
<hoverbear>
Oh, I can't build libgpg-error on musl? :(
<patagonicus>
Is there an easy way to see which package nix is currentyl building? In this case specifically for a single user install, but I'd also be interested for how to do it on NixOS if it works differently there.
<hoverbear>
patagonicus I think `--show-trace` and `--verbose` might get you there?
<seven-eleven>
symphorien, when i run -iA it installs the file again even though it is installed already? https://dpaste.org/kWYG
<hoverbear>
I can see mine during a build...
<cole-h>
You'd want --verbose; --show-trace is for errors that pop up during build :P
<patagonicus>
Oh, huh. That might be it. I get the log messages, but it's not telling me which package they belong to and sometimes I can figure it out scrolling back, but it can be cumbersome.
<cole-h>
(There are varying levels of verbosity, so keep adding until you see what you like)
<seven-eleven>
symphorien, when I run `apt install fzf` apt says it's installed already, nix doesn't have that feature?
<srk>
patagonicus: nix-top :)
<symphorien>
seven-eleven: that was just a general remark, only remotely related to your program
<seven-eleven>
patagonicus, yeah but why reinstall it if its already installed
<cole-h>
For single-user, you can ls ${TMPDIR-/tmp} and look for nix-* directories
<seven-eleven>
symphorien, -A helped a lot, much faster, just have to specify nixpkgs.fzf
<patagonicus>
srk: That sounds like what I want, but the version from unstable seems to break because I don't have build_users (as this is a single user install).
<cole-h>
hoverbear: How did you get it to have that extra stuff? :o I've never seen `pkg> ....` before
<symphorien>
patagonicus: ls /tmp
<seven-eleven>
i think nix is not reinstalling it when its already installed, I think it informs the user wrongly by saying "installing"
<symphorien>
The temp dir for each build contains the name of the package
<cole-h>
Ahh, it's the nix(1) build, not nix-build
<hoverbear>
Yeah `nix` feels much more reasonable to use
<seven-eleven>
how do I update all installed packages? is this only possible by adding the package to the channel? if yes how can I add a package to the channel by default when installing with -iA?
<cole-h>
(and also much more experimental -- it could break at any time ;^) )
<hoverbear>
I've generally had a terrible time using `nix-build` and stuff
<cole-h>
(and will be disabled by default in 2.4)
<symphorien>
seven-eleven: nix-env -u
<hoverbear>
cole-h What. Why?
<LnL>
read the last like of nix --help
<LnL>
line*
<seven-eleven>
symphorien, interesting, then I dont need to use channels at all, thanks :-)
<patagonicus>
symphorien: That doesn't show any nix builds for me. Not sure where nix is putting it's tempfiles for me. (I'm using nix-build currently as I'm trying to build an installation image. Also cross compiling if that matters. And, as I've said, single user install).
<cole-h>
hoverbear: Because it's unstable and people shouldn't rely on it. You can still get it by adding `nix-command` to `experimental-features` (or whatever it's called) as an explicit opt-in
<hoverbear>
LnL: I get that, I mean more "Why is it being disabled?"
<srk>
patagonicus: yes, nix-top needs daemon
<symphorien>
seven-eleven: also, may I recommend replacing all usages of nix-env by home-manager ?
<cole-h>
hoverbear: So it's more of a "if it's opt-in, you accept any and all responsibility for this breaking your workflow" etc
<symphorien>
patagonicus: yes I only tried that on nixos
<hoverbear>
cole-h So my user swon't be able to use `nix` by using the curl install without some manual system intervention?
<LnL>
hoverbear: because it's experimental and it's behaviour will be changing in the next releases
<hoverbear>
LnL: Yes I understand what experimental means thank you. :)
<patagonicus>
symphorien: Still helpful. Most of my systems are NixOS and this one will be migrated eventually. :)
<hoverbear>
I don't understand why "Experimental" means "We're disabling it despite it being default enabled for some time"
<LnL>
because people are using it in automation
<LnL>
like say a makefile ;)
<hoverbear>
Which is exactly what I will continue to do because `nix build` is a clear upgrade
<seven-eleven>
symphorien, just read home-manager's github README.md page and im not sure what advantages it gives over nix-env
<hoverbear>
:-P
<symphorien>
seven-eleven: nix-env is full of traps and oddities, home-manager avoids them
Neo-- has quit [Ping timeout: 260 seconds]
<patagonicus>
symphorien: Oh, actually, maybe I just run ls exactly between builds, now I'm seeing /tmp/nix-build-* dirs.
kleisli has joined #nixos
<symphorien>
For example if you install jd8, nix-env -u will update it to jdk10 or whatever
<symphorien>
There are workarounds, but they more or less reimplement what home-manager does
<seven-eleven>
ohh i see
<seven-eleven>
it should just update e.g. jdk8-0.0.1 to jdk8-0.0.3 but not to jdk10 thats not good yeah
orivej has quit [Quit: No Ping reply in 180 seconds.]
<seven-eleven>
but why would nix-env do that? consider jdk10 is a totally different package?
<seven-eleven>
considering*
<symphorien>
It does not know where the package comes from any more
<seven-eleven>
oh i see
<seven-eleven>
so I will use home-manager then
<symphorien>
The only metadata available at upgrade time is name-version
<ornxka_>
oh nix-shell -p ruby lets you just run gem and install to ~
simba1 has joined #nixos
omneh`s_ghost has joined #nixos
bastion-tester has quit [Quit: ZNC 1.7.4 - https://znc.in]
bastion-tester has joined #nixos
jmeredith has joined #nixos
<omneh`s_ghost>
Hey there, I'm having some recent trouble with rust. I'm using Mozilla's nightly rust channels, and today's upgrade gives me a libz.so.1 not found error when I try to run rustc, has anyone else run into this yet?
<drozdziak1>
I try to define some overlays which add some Python dependencies in sequence. https://paste.q3k.org/paste/RnHqM7EP#keLeY7b56ZIWTTvvCNR3q2SA+5i0h7UgNT6XeocJ5i6 When I try to use them with "pkgs.python3.withPackages (ps: [ps.pdfplumber])" I get "infinite recursion encountered". What am I doing wrong?
kreyren has joined #nixos
<energizer>
show your overlay
<drozdziak1>
I use the file under the link and instantiate pkgs with "import <nixpkgs> {overlays = import ./the_file_from_link.nix}"
<rycee>
energizer: Phew, must be some local routing problem. I get a server not found.
<energizer>
drozdziak1: super.python3.override not self
<drozdziak1>
energizer: thanks, I think it moved forward, now it can't see `super.my-pdfminer` on line 30
<energizer>
drozdziak1: self.my-pdfminer
cosimone has joined #nixos
<energizer>
also, you dont need two different overlays there, you an combine them
<drozdziak1>
energizer same. There's nested closures, are you referring to inner or outer self?
<energizer>
propagatedBuildInputs = [super.my-pdfminer]; should be self.
orivej has quit [Quit: No Ping reply in 180 seconds.]
cfricke has quit [Quit: WeeChat 2.8]
orivej has joined #nixos
jkachmar has joined #nixos
<energizer>
how do i make stage 1 prompt me to decrypt a certain disk before proceeding? is that neededForBoot?
alp_ has quit [Ping timeout: 260 seconds]
cosimone has quit [Quit: Quit.]
Thra11 has quit [Ping timeout: 260 seconds]
<patagonicus>
energizer: Should be boot.initrd.luks.devices . I've only used it for a device containing an LVM volume before, but it supports multiple devices and shouldn't need / to be on whatever you're unlocking.
<energizer>
patagonicus: thanks
o1lo01ol_ has joined #nixos
<pikajude>
is there a sensible way to get an openvpn service to accept TTY input for the password?
<pikajude>
the one i'm connecting to requires an OTP
<pikajude>
a systemd service*
kleisli has quit [Remote host closed the connection]
kleisli has joined #nixos
o1lo01ol1o has quit [Ping timeout: 240 seconds]
FRidh has quit [Quit: Konversation terminated!]
<jkachmar>
Is there a particularly good place to look for documentation or examples on packaging + making a NixOS service module for applications that are entwined with the FHS env?
hazel has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @dminuoso opened pull request #91316 → Support virtio_scsi devices on nixos-generate-config → https://git.io/JfxeR
orivej has quit [Ping timeout: 256 seconds]
orivej has joined #nixos
<jkachmar>
Specifically, I'd like to pick up where someone had left off on packaging up Pi-hole for nixpkgs and NixOS.
<jkachmar>
The "core" administrative CLI tool is basically just a big ol' bash script that points to particular locations on the filesystem.
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
<clever>
jkachmar: id patch the script to take paths via env vars and then have systemd set those env vars
<seven-eleven>
what command do i have to run after I created my /etc/nixos/configuration to install the packages for all users?
<clever>
seven-eleven: nixos-rebuild switch
<abathur>
<3 asymmetric
<{^_^}>
asymmetric's karma got increased to 6, it's a crit!
<seven-eleven>
clever, thanks
<asymmetric>
abathur: :)
<seven-eleven>
clever, wait, im not using nixos, im using only nix on ubuntu
<asymmetric>
i moved to gh actions but it will still be nice to get travis working again
<clever>
seven-eleven: /etc/nixos/configuration.nix is only for nixos
<seven-eleven>
aaaah
<clever>
seven-eleven: if you want to install something system-wide on ubuntu, you just run nix-env -i as root
<seven-eleven>
oh thats clever, i read something like that in the cheatsheet, thanks!
Thra11 has joined #nixos
zebrag has joined #nixos
Neo-- has joined #nixos
<jkachmar>
clever: I was thinking about something like that, but part of the utility of packaging up the administrative scripts is in the fact that a user can just get some info or temporarily modify/disable things from the CLI
<clever>
jkachmar: then youll want things somewhere like /etc/pihole/ and use environment.etc
alp_ has joined #nixos
<jkachmar>
Aaah, TIL `environment.etc` yeah okay
<dminuoso>
so... nixops on hetzner.
knupfer has quit [Quit: knupfer]
knupfer has joined #nixos
<abathur>
asymmetric: nod, I saw that; don't blame you; luckily I haven't had bandwidth to make enough progress on anything I've got set up with nix+travis to really hurt for the lack of it yet
<dminuoso>
There appears to be absolutely no frigging documentation how nixops plugins should be used...
<dminuoso>
Am I missing something?
<dminuoso>
There appears to have went so much work in it, but no documentation?
<hoverbear>
How can I override the ld/cc for the `buildRustPackage` command? Seems I need like `(overrideCC stdenv (wrapCCWith { .... }}).mkDerivation`?
<clever>
dminuoso: though that is complicating it some to force specific versions of plugins, i think you just want `p = ps: [ ps.nixops-packet ];` for ex
drakonis1 has joined #nixos
drakonis_ has quit [Read error: No route to host]
ATuin has quit [Client Quit]
drakonis_ has joined #nixos
ATuin has joined #nixos
drakonis has quit [Ping timeout: 272 seconds]
pjt_mpt has joined #nixos
drakonis_ has quit [Read error: No route to host]
<Thra11>
One of my systems has selected a DNS server which won't resolve github.com -_-
drakonis1 has quit [Ping timeout: 240 seconds]
<gustavderdrache>
hrm... can anyone else build kubernetes-helm as of nixpkgs-20.09pre230446.a84cbb60f02? i'm getting a weird error and i want to check if it's just me before i file an issue
<hoverbear>
Hmmm I am having a lot of trouble using musl to build a rust package. :(
<seven-eleven>
should i install `curl -L https://nixos.org/nix/install | sh` only to one user and run it from root with /home/some-user/.nix-profile/bin/nix or should i install it to root too?
drakonis has joined #nixos
<clever>
seven-eleven: if you do a multi-user install, it should setup root automatically for you
<smiles>
Hey I am trying to use a remote ubuntu machine with nix install to build a arm7 nixos when I try to do it it says it is a x86_64-linux and cannot do it what do I need to add to the ubuntu machine so it can build arm7 binary?
<clever>
smiles: you either need an arm machine with nix already installed, or switch to cross-compiling
<seven-eleven>
clever, i looked it up in the docs "sh <(curl https://nixos.org/nix/install) --daemon", but that didn't install anything. i think i did it incorrectly
<smiles>
that is my other issue where do I find the config file on the ubuntu machine?
<hoverbear>
clever Is that preferred over `pkgs.cross`?
<seven-eleven>
clever, (for multi user installation that is)
<sphalerite>
aah ok, then the firmware isn't configured right to boot it. Check if UEFI is enabled and Secure Boot is disabled
<energizer>
ok will check that, brb
emmanuel` has joined #nixos
orivej has quit [Quit: No Ping reply in 180 seconds.]
orivej has joined #nixos
asheshambasta has quit [Ping timeout: 246 seconds]
<emmanuel`>
So I was just trying this command in the repl at the suggestion of a README "nix-build -f. binaries". I looked in the manpage and didn't find anything that could help me solve my problem. Does anyone here know what was trying to be achieved here?
<cole-h>
nix-build doesn't have an -f flag.
<emmanuel`>
What if you do "nix build"
<emmanuel`>
I realized that "nix build" was what I should have entered.
<cole-h>
That would work. But nix(1) is experimental, and will even be disabled by default in 2.4.
<emmanuel`>
How is that any different?
ATuin has quit [Ping timeout: 260 seconds]
<cole-h>
Maybe you could explain what problem you're actually experiencing?
<cole-h>
Well, it's different because they are two different commands.
<emmanuel`>
Right, so I'm trying to build some binaries.
<emmanuel`>
This was the command I was given (I'm still relatively new to actually using nix for devops)
boxscape has quit [Quit: Connection closed]
<seven-eleven>
i installed nix with --daemon for all users and the installer says I should run now `nix-shell -p nix-info --run "nix-info -m"` but there's no command nix-shell in my system
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/JfxUv
<sphalerite>
seven-eleven: you may need to start a new shell
boxscape has joined #nixos
<seven-eleven>
sphalerite, yeah did that, but in a new shell there's also not nix-shell
<sphalerite>
seven-eleven: ah ok. Does ~/.nix-profile/etc/profile.d/nix-daemon.sh exist? If so, try sourcing that and then try again
<seven-eleven>
sphalerite, nope
<seven-eleven>
sphalerite, there's a systemd unit and /etc/profile.d/nix.sh but in $HOME theres nothing
alp_ has quit [Ping timeout: 272 seconds]
<sphalerite>
seven-eleven: what about /nix/var/nix/profiles ?
<seven-eleven>
sphalerite, yeah that exists
<cole-h>
Maybe single user install?
<seven-eleven>
single user install works for me, this time i tried --daemon for multi user
<hoverbear>
cole-h Are you familiar with how to get the rust overlay to install other targets? It seems to be doing it but I don't have a core for the other target?
<cole-h>
hoverbear: Unfortunately not 😬 When I was building a project for musl, I just used rustup to impurely install that target, and then did `cargo build --target <musl triple>` or something like taht
<cole-h>
s/taht/that
<seven-eleven>
i have this in ~/.bashrc [[ $- != *i* ]] && return, could that prevent /etc/profiled.d being loaded?
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<hoverbear>
cole-h Yeah I see a lot of people saying that but I need a real solution ot a nix-shell hack :S
<hoverbear>
cole-h I would personally be fine with that but I tihkn that's ultra-cache-unfriendly
<cole-h>
Probably true.
<cole-h>
s/Probably/Most likely/
knupfer has joined #nixos
<sphalerite>
seven-eleven: hm, not sure what's going wrong there, but what you'll want is to have /nix/var/nix/profiles/default/etc/profile.d/nix-daemon.sh sourced in shells where nix will be used
<hoverbear>
cole-h: I did notice that buildrusttarget seems to be calling gnu related stuff so I'm very suspicious
<{^_^}>
nmattia/naersk#110 (by Hoverbear, 2 days ago, open): Git submodules not being initialized?
orivej has joined #nixos
<energizer>
actually no, the displays are asleep, i guess it's not displaying anything
<cole-h>
hoverbear: Right, I remember this. Well, I don't think that issue will be solved soon (nmattia seems busy), so would it really hurt if you used mkDerivation and cherry-picked stuff from buildRustPackage until it works? :P
<hoverbear>
cole-h Nope I actually think it's probably the right way to go about it, I just didn't want to have to solve caching crates in the nix store xD
<puck>
serial number was in the 11'xxx'xxx range, the yubikey 5 does 10'xxx'xxx
alp_ has joined #nixos
<seven-eleven>
how can I run the nix package manager installer script unattended `sh install --daemon`?
fresheyeball has quit [Ping timeout: 265 seconds]
<cole-h>
yes | sh ...
asymptotically has quit [Quit: Leaving]
fresheyeball has joined #nixos
<seven-eleven>
hmm let me try that, because I thought it asked me for my password
justanotheruser has quit [Ping timeout: 264 seconds]
andreas303 has quit [Remote host closed the connection]
<cole-h>
It will ask for your sudo password at some point. You can just do `sudo echo` and then it should be fine for the rest of the installation.
<cole-h>
*just do `sudo echo` prior to installing
Guest78 has joined #nixos
<seven-eleven>
oh that's a great trick
Guest78 has quit [Client Quit]
andreas303 has joined #nixos
<seven-eleven>
i will try it tomorrow and will see if I can get it glued together with ansible, not sure if ansible issues tasks in the same session, will see :-)
<hoverbear>
cole-h: Yeah still has that pre-build step :(
ddellacosta has quit [Client Quit]
<cole-h>
:(
ATuin has quit [Ping timeout: 264 seconds]
justanotheruser has joined #nixos
erasmas has quit [Quit: leaving]
Rusty1 has joined #nixos
pingiun has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
kenji has quit [Ping timeout: 260 seconds]
<energizer>
update: turns out my mobo firmware is too old to boot from an nvme drive and requires flashing a custom modded bios, which might be a step too far for me
_deepfire has quit [Remote host closed the connection]
absoluutely has quit [Remote host closed the connection]
absoluutely has joined #nixos
justache has joined #nixos
_deepfire has joined #nixos
annadane has joined #nixos
alp_ has quit [Ping timeout: 272 seconds]
<energizer>
i was messing around in my bios/uefi system and the "Linux Boot Manager" boot option that used to be in the list is no longer there so i can't boot into nixos. how do i get it back?
<energizer>
i tried resetting my mobo settings but that didnt help
philr_ has joined #nixos
o1lo01ol1o has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]