<ottidmes>
laalf: looks good is not what we are after, if you start it manually does it run with that config? is it the same config on your other distros?
adamantium has joined #nixos
<laalf>
ottidmes:same config on other distros, doesnt run on nixos
<adamantium>
I didn't think it was possible to break nixos :(
<growpotkin>
lol
<growpotkin>
adamantium: happens all the time m8
<adamantium>
WAT
<growpotkin>
adamantium: so it looks like a python package is failing to build for you. Does that sound about right?
<adamantium>
NO CLUE man
<adamantium>
i only explicitly ask for one python package in my configuration.
<adamantium>
i could comment it
<growpotkin>
adamantium: "azure-mgmt-compute"
<growpotkin>
is the failing python pack
<adamantium>
i dont know what would even want that.
<adamantium>
I only use certifi really
<pbb>
Hi, since I reinstalled my laptop, my trackpoint and my physical buttons are not working anymore. The mouse pad works, but sometimes it responds very badly, as if there was no graphics acceleration (but there is, that's not the problem).
<growpotkin>
it looks like it is associated with a Virtual Machine software
<ottidmes>
laalf: I can't help you more, it should just be matter of standard debug steps. Like does is it only sxkhd that does not run but bspwm does, what do the logs say, what if you run it your self explicitly, what if you check the contents of the session commands file and try to run the command that they run yourself manually as root, etc.
<growpotkin>
adamantium: are you trying to install any kind of VM or remote desktop stuff?
<pbb>
I'm using the same configuration as before, so did a recent update in unstable break some input devices for someone else?
<adamantium>
growpotkin: literally makes zero sense. something is bad. this system doesn't use any virtualbox or libvirt. nothing
<adamantium>
gnome3 maybe does it pull in gnome boxes or something?
linarcx has joined #nixos
<laalf>
ottidmes: sxhkd says nothing.
<linarcx>
Hi guys. Why prezto does not exists in 18.09 channel?
<ottidmes>
linarcx: it just keeps running, no exit code?
<growpotkin>
adamantium: I couldn't tell you, I don't remember the exact command; but there is a nix-env command that lets you see dependencies. You can ask it what in your config depends on "azure-mgmt-compute" to help narrow down the culprit
<laalf>
ottidmes:yep.
<ottidmes>
linarcx: is that a zsh thing? if so, its under zsh-prezto
<linarcx>
No, it cannot find it.
<adamantium>
growpotkin: so this is the price i pay for using unstable channel
<adamantium>
growpotkin: hmm, not certifi gdm or gnome, i commented all related configurations and it still errs
Ariakenom has quit [Quit: Leaving]
<growpotkin>
adamantium: try `nix-store --query --references $(nix-instantiate '<nixpkgs>' -A pythonPackages.azure-mgmt-compute)`
<linarcx>
ottidmes: And it automatically find `. zpreztorc`? No need for extra configuration?
linarcx has quit [Remote host closed the connection]
<growpotkin>
the unfortunate thing here, is that it is a "setuptools" dependency, which is in almost every single python package
<growpotkin>
the good news is, it is specific to python2.7
<adamantium>
is unstable broken for everyone?
justanotheruser has quit [Quit: WeeChat 2.2]
<growpotkin>
yeah I mean unstable has packages that are "unstable"
justanotheruser has joined #nixos
<adamantium>
if it is, i dont mind waiting it out
<growpotkin>
but things only break if you try and install something that is messed up at that time
<adamantium>
but i wonder if *I* broke something that needs manually fixed
<growpotkin>
your issue is going to come down to one specific package that you are trying to isntall
<growpotkin>
install*
<growpotkin>
did you recently install something new?
<adamantium>
hard to say
<growpotkin>
or is this a result of moving your whole config over to unstable?
<adamantium>
i had auto upgrade enabled
<growpotkin>
but have you been on unstable for a while now?
<ottidmes>
laalf: if it runs, without an error, seems to imply that its not a matter of it not starting when starting the session, rather that it somehow does not load the config or some such, so if I were you, I would experiment with a simple config that just opens a terminal or whatever and supply that directly with sxhkd -c simple-config
<adamantium>
and maybe i ran a nixos-switch rebuild simultaneously during an auto ugprade, that is what I think happened
<adamantium>
i've been on unstable a 4-5 days upgrading daily
<growpotkin>
rebuilding and upgrading shouldn't be an issue, the store lock keeps it from conflicting with each other
<adamantium>
k
<growpotkin>
I would try updating your channels now, collecting garbage, and then trying a rebuild
<growpotkin>
just to get things "fresh"
drakonis has joined #nixos
<growpotkin>
and then I would start using "--show-trace" when you build to find the specific package's dependency chain that leads you to the crash.
<growpotkin>
when you find the culprit you can tell your config to install that package from the "stable" channel rather than unstable; while building everything else on unstable
<adamantium>
doing it now
<adamantium>
maybe i should try LUSTRATE ? i think it was called
<yl[m]>
Mic92: I've updated #56500 to address your comments. I have a question though: how does nox figure out which packages need to be rebuilt?
<justanotheruser>
(I fixed rootw, now it says root)
phreedom_ has quit [Remote host closed the connection]
phreedom has joined #nixos
<{^_^}>
[nixpkgs] @amazari opened pull request #56794 → chromium: Add optional vaapi hw video decoding support → https://git.io/fhxY8
silver has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @matthewbauer pushed to matthewbauer-patch-1 « ghc863: dont add extra libs »: https://git.io/fhxY2
<{^_^}>
[nixpkgs] @matthewbauer opened pull request #56795 → ghc863: dont add extra libs → https://git.io/fhxYa
ng0 has quit [Quit: Alexa, when is the end of world?]
dansho has quit [Ping timeout: 245 seconds]
xkapastel has quit [Quit: Connection closed for inactivity]
black-brick has joined #nixos
ddellacosta has quit [Ping timeout: 245 seconds]
ddellacosta has joined #nixos
<black-brick>
anyone of you is using vscode? I'm trying to edit but I need to have sudo rights so it pops a button saying "retry as sudo..." and when i click on it, it just reappears since it didnt work. My user has sudo rights
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fhxYP
<Myrl-saki>
losetup -P is broken.
lsyoyom has quit [Ping timeout: 255 seconds]
<Myrl-saki>
[myrl@myrl-lappy:/etc/nixos/nixpkgs]$ git status
<Myrl-saki>
HEAD detached at 07e2b59812d
qyliss^work has quit [Quit: bye]
qyliss has quit [Quit: bye]
Dedalo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Rusty1 has joined #nixos
<Myrl-saki>
Okay, maybe my desktop is broken. Wtf.
<Myrl-saki>
losetup works okay on my laptop.
<Myrl-saki>
Maybe it's the kernel version.
jluttine has joined #nixos
jluttine has quit [Ping timeout: 245 seconds]
<ottidmes>
black-brick: vscode user here, why would you need sudo rights, normal usage should not require that
<black-brick>
ottidmes: yes i know that! im starting vscode as my user but i want to edit a file and i need to be root to edit that file so when i try to save, it says "Failed to save : Insufficient permissions. Select 'Retry as Sudo' to retry as superuser."
<black-brick>
ottidmes: so i click on the retry as sudo button but it doesnt work
<ottidmes>
black-brick: ah, like so, let me try, I tend to just use nano for those files
<black-brick>
ottidmes: ohhh ok!
nikivi has quit [Quit: ZNC is awesome]
jluttine has joined #nixos
nikivi has joined #nixos
mbrgm_ has joined #nixos
<ottidmes>
black-brick: same issue for me, I had not even considered using vscode to edit files owned by root, most of the time its just small changes so so far, sudo nano has worked for me
lsyoyom has joined #nixos
qyliss has joined #nixos
<black-brick>
ottidmes: oh ok kinda glad it's not just me but i think it's annoying since it should work :(
qyliss^work has joined #nixos
mbrgm has quit [Ping timeout: 264 seconds]
mbrgm_ is now known as mbrgm
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
noonien has quit [Quit: Connection closed for inactivity]
phreedom has quit [Remote host closed the connection]
<CMCDragonkai>
During a nix-build, is there some way to acquire the current Git commit ID that the build is occuring from? Something like `builtins.readFile` but instead calls `git rev-parse HEAD`?
phreedom has joined #nixos
<CMCDragonkai>
I suppose I could read `.git/refs/heads/master` using `readFile`.
<slack1256>
CMCDragonkai: you mean the commitID of the nixpkgs you're bulding on top? or the commitId of the revision gotten from fetchFromGithub?
<CMCDragonkai>
No from my repo
<CMCDragonkai>
Like I have a project, inside it has `default.nix` and `release.nix`.
<CMCDragonkai>
I like to build it, and create something (and embed within itself the commit id of the checked out branch)
<slack1256>
I don't know if nix-build can do that, but you can pass the current revision as an argument to nix-build and use that.
<CMCDragonkai>
slack1256: It can certainly read `.git/refs/heads/master`.
<CMCDragonkai>
But this doesn't resolve the issue when the Git repo is unclean. Necessitating the calling of some shell function to check if the repo is unclean or not.
<__monty__>
And to actually remove the data on disk you need to run nix-collect-garbage afterwards.
<linarcx>
Oh. cool. last night i used `nix-collect-garbage`, but still i can see list of all generations in grub menu. why? i want to delete old ones and keep only last 3 generations.
<__monty__>
Did you --delete-generations before nix-collect-garbage?
<linarcx>
no
johanot has joined #nixos
<__monty__>
Then they can't be garbage collected yet, cause you're basically stating you don't want them to.
<infinee>
tells you exactly how to delete generations
fusion809 has quit [Remote host closed the connection]
<__monty__>
What I miss is a way to --delete-generations --keep 3 or something.
<__monty__>
Guess it wouldn't be that hard to script though.
thc202 has joined #nixos
<immae>
there is one
<immae>
__monty__: look at the nix-env manual in the generations section, there are examples :p
linarcx has quit [Ping timeout: 245 seconds]
<immae>
(last n days, last n generations, ...)
<immae>
sorry, I jumped too quick in the conversation, the message was for linarcx
<__monty__>
immae: Can't find those examples anywhere. It's either listing them manually, older than x days or "old" which is everything but current.
<immae>
I have an example with `nix-env --delete-generations +5` for everything but the last 5
linarcx has joined #nixos
<immae>
I don’t know how recent the "feature" is though
<__monty__>
Ah, yep. That's in the manual. Couldn't find it by "last n days"
<immae>
ok :)
<__monty__>
It's not explained very well though. "or a value such as. +5 to only keep the specified items older than the current generation."
<__monty__>
Sounds like it only deletes generations you specify with + *if* they're older than the current generation.
<immae>
Well, to me it looks clear, but since I was looking specifically for that I’m a bit biased :)
<__monty__>
No, the reverse. It only *keeps* them if they're older. Which is a pretty weird feature admittedly but it doesn't sound like "keep the last 5 gens."
<immae>
no no I use it very often and it does it right :)
<immae>
If anything, only the formulation is misleading :)
Ariakenom has joined #nixos
<linarcx>
infinee: thank you so much.
<linarcx>
__monty__: thank you too.
civodul has joined #nixos
arjen-jonathan has joined #nixos
jomik has joined #nixos
__Sander__ has joined #nixos
phreaky has joined #nixos
<linarcx>
I want to use i3nagbar. but it seems that there isn't exists in channels.
<phreaky>
Hey guys , how does one check all of the supported versions of a given package ( if I want to get other versions if possible )
gilligan_ has joined #nixos
<__monty__>
linarcx: Then you'll have to package it yourself.
<linarcx>
__monty__: Any tutorial to start?
<__monty__>
phreaky: Afaik one channel only has a single version of any package. Sometimes there's package_x_y_z with a different version because some packages depend on a specific older version. But there's not really a good way to list those, except for grepping the name.
_kwstas has joined #nixos
<__monty__>
linarcx: The nix and nixpkgs manuals.
_kwstas has quit [Remote host closed the connection]
jomik has quit [Quit: WeeChat 2.4]
johanot has quit [Ping timeout: 245 seconds]
<phreaky>
__monty__, thanks a lot
johanot has joined #nixos
justanotheruser has quit [Ping timeout: 240 seconds]
<__monty__>
np
vk3wtf has joined #nixos
Yaniel has joined #nixos
linarcx has quit [Remote host closed the connection]
mariel__ has joined #nixos
<{^_^}>
[nixpkgs] @jpathy opened pull request #56809 → Zotero: Fix to use patchelf instead of FHS Env. → https://git.io/fhxZN
simukis has joined #nixos
<{^_^}>
[nixpkgs] @peti pushed 3 commits to haskell-updates: https://git.io/fhxZx
phreaky has quit [Quit: Leaving]
<{^_^}>
[nixpkgs] @peti merged pull request #56795 → ghc863: dont add extra libs → https://git.io/fhxYa
<{^_^}>
[nixpkgs] @peti pushed 2 commits to haskell-updates: https://git.io/fhxZp
<{^_^}>
[nixpkgs] @peti pushed 0 commits to matthewbauer-patch-1: https://git.io/fhxZh
<sphalerite>
ramses_: Modules are what do this for nixos
<sphalerite>
ramses_: it's a bit of a complicated quesiton because you have to define how to merge stuff
<sphalerite>
modules are the heavy-weight and versatile solution, where you define types and stuff
<sphalerite>
there isn't really a simple way AFIAK
<sphalerite>
AFAIK
<tilpner>
Could try zipAttrsWith
<ramses_>
sphalerite: well, I have a module actually (https://github.com/MSF-OCB/NixOS/blob/master/users.nix) now I wanted to be able to merge two users and have their extraGroups combined. But right now the rhs takes precedence
LysergicDreams has quit [Remote host closed the connection]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<tilpner>
Hey thoughtpolice, what are you using firecracker for, and did you try building it from source (and if so, what was the issue that caused you to stop)?
lopsided98 has quit [Quit: No Ping reply in 180 seconds.]
<tilpner>
Ahh, the earliest commit message answers it
<{^_^}>
[nixpkgs] @xeji pushed commit from @jb55 to master « clightning: 0.6.3 -> 0.7.0 (#56788) »: https://git.io/fhxWW
<therealwaphire[m>
guys how do you deal with programs that are expecting traditional FHS?
<tilpner>
steam-run and friends
<therealwaphire[m>
for eaxmple, running taffybar outputs 3 of "Directory does not exists:" warnings
<tilpner>
Oh, not for those
<therealwaphire[m>
yeah
<tilpner>
Which directories?
<etu>
azazel, rkav: I did some work on the nix-mode for emacs during January this year. We improved it quite a bit but there's still work to do with corner cases etc.
<therealwaphire[m>
`/usr/share`, `/usr/loca/share` and `/etc/profiles/per-user/user/share`
<simpson>
therealwaphire[m: If it's not unfree, patch it until it stops misbehaving. FHS user environments are a last resort.
<tilpner>
And what does it look for in there?
<etu>
azazel, rkav: But I've introduced regression testing for indentation so we have test cases for it. So if you feel up to it. Write a testcase and a fix and I'll look into it :-)
<therealwaphire[m>
<freenode_til "And what does it look for in the"> doesn't say anything in particular
<etu>
azazel, rkav: Also, bug reports are very welcome as well. But no promises that it will be fixed very fast.
<therealwaphire[m>
just `Directory does not exist: `
<tilpner>
You can probably find out with strace
<tilpner>
But yeah, patch it or ignore it, if it doesn't obviously break anything :/
<therealwaphire[m>
yeah i guess
<therealwaphire[m>
has anyone managed to run `throttled` btw?
<therealwaphire[m>
for undervolting on the lenovo x1c6?
<azazel>
etu: yes, I saw your work. great job! Certainly I will report bugs, fixes... we will see :-P
<azazel>
therealwaphire[m: `throttled`?
<tilpner>
thoughtpolice: So... I got it to build from source. But it aborts with "invalid system call"
<azazel>
therealwaphire[m: there's a specific module for x1 gen 6 in nixos-hardware
<azazel>
therealwaphire[m: the nixos-hardware module runs a much simpler, and maybe less efficient, short-term fix taken from the specific page on archlinux wiki
<rkav>
etu: Awesome! I see now that nix-indent-line is marked as buggy, but it's certainly seems better than ident-relative. I'll use it for a bit and keep my out for flaws
<joko>
therealwaphire[m: but it's WIP, it still needs you to add /etc/lenovo_fix.conf
<azazel>
etu: keep up the good work!
<therealwaphire[m>
oh thanks, man :)
<joko>
If you don't want to undervolt, then what azazel proposed is better, I guess
das_j has quit [Remote host closed the connection]
<pie_work>
of course i just had to ask, looking at settings in pavucontrol showed that it was switched to some dock audio output, this wasnt synced properly with whatever plasma was showing me
<pie_work>
so now i fixed it...
das_j has joined #nixos
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « pandoc: update overrides for the new version »: https://git.io/fhxla
<phreaky>
the output of `nix-env -qa | grep solc` gives me only 0.5.3 version which I think was in the unstable channel when I need 0.4.24 which is supposed to be in the stable one
<noonien>
does anyone else have problems when ssh-ing into a nixos machine?
<noonien>
ssh can't seem to create a terminal
linarcx has joined #nixos
<linarcx>
Hi. i installed mpv and used vaapi for decoding. but it give me an error:[vaapi] Failed to initialize VAAPI: unknown libva error
<linarcx>
Error opening/initializing the selected video_out (--vo) device.
<linarcx>
also i installed libva package.
<linarcx>
what should i do?
<waleee>
noonien: have you changed the default configuration.nix? Per default nixos doesn't have any ports open from the outside, so ssh will be blocked if not enabled explictly
<ramses_>
sphalerite: well, I have a module actually (https://github.com/MSF-OCB/NixOS/blob/master/users.nix) now I wanted to be able to merge two users and have their extraGroups combined. But right now the rhs takes precedence. Do I understand correctly that I should implement a proper merge strategy for the loaOf option?
<waleee>
noonien: and you're a 100% sure zsh is enabled?
<linarcx>
callahad: Rebuild the configuration. But nothing changed. Reboot the system?
<noonien>
waleee: executing zsh without a terminal works, zsh gets executed, but not in a terminal, more like a pipe
<aswanson>
good morning folks, I have a weird thing bugging me right now. In my configuration.nix I'm fetching a tarball from gitlab with `builtins.fetchTarball`. This works fine with `nixos-rebuild` but the nixos upgrade service is choking on it. I can see a line in the journal saying that lbzip2 is not available to tar when the service tries to upgrade, but I don't understand why nixos-rebuild works fine.
<callahad>
linarcx: Hm. I know very little about this (other than mpv w/ vaapi is working for me locally).
<callahad>
linarcx: I also have this in my configuration.nix: hardware.opengl.extraPackages = with pkgs; [ vaapiIntel vaapiVdpau libvdpau-va-gl ];
<{^_^}>
[nixpkgs] @xeji pushed commit from @herrwiese to master « gogs: 0.11.66 -> 0.11.86 (#56819) »: https://git.io/fhx4u
<andi->
aswanson: I could be wrong but builtins.fetchTarball could also impurely use the current runtime environment much like `builtins.fetchGit` fails when git is not "installed"/available in the current environment.
<{^_^}>
[nixpkgs] @xeji pushed commit from @herrwiese to release-19.03 « gogs: 0.11.66 -> 0.11.86 (#56819) »: https://git.io/fhx4a
jomik has joined #nixos
<{^_^}>
[nixpkgs] @Mic92 merged pull request #56500 → go_1_12: init at go 1.12 → https://git.io/fhNyn
<aswanson>
andi-: that tracks with the behavior I'm seeing but that raises the question of how to get lbzip2 into the environment that the upgrade service is using
<jomik>
Hey guys - is there a nice/simple way to check if there is an update available for my nix channels?
<linarcx>
callahad: Oh, new error arised: [vaapi] libva: /run/opengl-driver/lib/dri/i965_drv_video.so has no function __vaDriverInit_1_0
jabranham has joined #nixos
<tilpner>
noonien: Did you do programs.zsh.enable on the host?
<andi->
aswanson: systemd.services.nixos.upgrade.path = [ pkgs.… ]; should do the trick
<andi->
but e.g. gnutar is already in there.. so you might want to play with that
<aswanson>
andi-: thanks for the pointer, I'll give it a shot!
kasimon has joined #nixos
<noonien>
tilpner: i've only got `shell = pkgs.zsh;` in my users config
<tilpner>
noonien: Well, then try what I suggested
<noonien>
i don't see how that makes a difference, since ssh seems at fault here
<noonien>
i get the same behavior with bash as well
<noonien>
and `less`, and `cat`
<gchristensen>
noonien: you might need to do shell = "${pkgs.zsh}/bin/zsh"
drakonis_ has joined #nixos
<tilpner>
No, they're equivalent
<gchristensen>
"Don’t forget to enable your shell in programs if necessary, like programs.zsh.enable = true;. "
<tilpner>
Which they don't want to try
<tilpner>
For... reasons
<kasimon>
Hi, does anyone know how big the current stable binary cache is? If I would want to mirror one for internal use, how much space would I need?
<gchristensen>
kasimon: I have this vague idea that it is roughly 80gb but that is probably wildly out of date
linarcx has quit [Ping timeout: 240 seconds]
<tilpner>
kasimon: It depends on if you also want to mirror all the history, or just one evaluation
<kasimon>
tilpner: just one evaulation
<noonien>
gchristensen: i will try, but, `ssh <host> -t cat x` yields the same result
<gchristensen>
noonien: did you do programs.zsh.enable = true?
<adamt>
kasimon: We've been caching everything we've used for about a year, with updates from stable every two weeks, and it takes up like 300GB disk space
<sphalerite>
ramses_: whoops sorry, missed it earlier. I think what you probably want is to just use multiple modules to define your users, then they'll get merged appropriately by the module system without any further code on your part
waleee has quit [Quit: WeeChat 2.4]
<tilpner>
kasimon: Note that adamt might be using only small parts of the evaluation
<noonien>
gchristensen: i just did, same outcome, `ssh host -t cat x` hangs
<kasimon>
adamt: Thanks, that gives me an idea how much to expect.
<noonien>
`ssh host cat x` works just fine
<adamt>
Yeah I know it didn't really answer the specific question, but I have this feeling that kasimon doesn't really need to store all binaries for one eval either. :-)
<noonien>
and so does `ssh host zsh -i`
<noonien>
(i don't get a terminal in either cases)
<tilpner>
Can you get logs from the host?
<noonien>
hmm, didn't think of that, just a sec
<kasimon>
tilpner: Is there a way to request the size of a path without downloading it?
<tilpner>
kasimon: Yes, you can query the .narinfos of each from Hydra
<tilpner>
(From cache, no hydra needed)
<kasimon>
great, thanks.
<gchristensen>
hmm ssh without `-t` to spawn an interactive shell will get you a weird terminal
<noonien>
gchristensen: yeah, more akin of just a two-way pipe, not a terminal really
<ptotter[m]>
systemd.services.*.serviceConfig.ExecStart can contain variable interpolation ${...}, but can it do more? how much templating can be done? loops? conditionals?
<gchristensen>
ptotter[m]: you can do anything you want as long as Nix evaluates it to a string
<gchristensen>
${...} is done by Nix
<tilpner>
ptotter[m]: Anything Nix can evaluate, including conditionals and I guess "loops" too
<noonien>
gchristensen: do you get the same behavior? or is it just my machine?
<callahad>
linarcx: Any luck?
<tilpner>
noonien: Does it also happen when you remove the shell = pkgs.zsh?
<ptotter[m]>
is something like ${if foo} blah ${else} blop ${end} possible?
<tilpner>
noonien: I have ssh and zsh working on all my machines and servers
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<adamt>
ptotter[m]: if-statements are expressions, so yes, they can end up in your ExecStart
<tilpner>
ptotter[m]: It would be ${if foo then "blah" else "blop"}
<Taneb>
ptotter[m]: I think that's a syntax error but you can do what tilpner just suggested
<noonien>
tilpner: yes, i get the same behavior. odd that it works for you, latest 18.09?
<ptotter[m]>
thank you
<tilpner>
ptotter[m]: Note that "blah" and "blop" can also be ''strings like this'', which can again have ${} interpolation
<noonien>
the ssh client is running on a raspberry pi zero w, with a dated 19.04 sd image, this MIGHT be the issue, i don't have another machine to ssh from, i'll try from my phone
<tilpner>
... dated 19.04? That must be a typo
<noonien>
19.03, sorry
<ramses_>
sphalerite: you mean to create a module per user?
linarcx has quit [Ping timeout: 268 seconds]
<tilpner>
Still, 19.03 isn't even released yet, so hardly dated :)
<ramses_>
That seems a bit heavy syntax-wise though..
<sphalerite>
ramses_: yeah. It's quite simple, if you want it in one file you can do { imports = [{module1} {module2}] }
<ptotter[m]>
tilpner: excellent, better than bash `` substitution (and more like $())
<sphalerite>
ramses_: without declaring any options in the modules of course, just the values for them
<noonien>
i'm talking about dezgeg's sdimage, which is from 2018-09-13 05:14
<{^_^}>
[nixpkgs] @dtzWill opened pull request #56824 → skawarePackages: skalibs 2.8.0.0 and updates all around → https://git.io/fhx4x
<noonien>
yeah, i think it might be the nixos machine that i'm using as a ssh client, sshing from my phone works just fine
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<noonien>
hmm, i wonder, do you guys do most of your user config from configure.nix, and try to have a clean $HOME, do most of your user config in $HOME, or a balance of both?
kasimon has quit [Ping timeout: 256 seconds]
kreisys has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<ottidmes>
tilpner: then you should not have said "but ramses_ seems to want to avoid complexity (and particularly modules)"
<wedens>
jomik: I'm not sure when it'll return `null` though :)
<tilpner>
ottidmes: I wasn't here either for the original problem description, I just saw sphalerite suggesting modules
<tilpner>
ottidmes: And sphalerite suggested multiple modules, which would be too much for ramses_. You're correct that it wasn't very clear, I don't know ramses_ exact restrictions
<ottidmes>
tilpner: I just tuned in and saw this contradictionary conversation, "it should not use modules", "let's use modules, it's fine", but I agree, if modules are allowed, mkMerge is probably what you want, otherwise you would just have to copy over the implementation of recursiveUpdate and tweak it to behave the way you'd want
<therealwaphire[m>
anybody using filebot on NixOS?
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « ghc: Drop outdated compiler versions 8.6.1 and 8.6.2. We have 8.6.3. »: https://git.io/fhxBN
<tilpner>
ottidmes: There are three different levels of "using modules" in this conversation, and we haven't detailled which we mean. ramses_ is using one module in their configuration, sphalerite suggested writing multiple modules AFAICT, and clever suggested using the modules system with lib.evalModules
<tilpner>
These are obviously very different options when compared by complexity
<sphalerit>
what I meant was to use one module to implement the options and multiple modules to set them
<tilpner>
And what I suggested doesn't even work...
<sphalerite>
ramses_: what about a function like makeUser, which takes a list of roles or something, like yusuph = mkUser { roles = ["admin"]; extraGroups = ["docker"]; }
<sphalerite>
ramses_: which then adds the manually specified groups to the role groups
<{^_^}>
[nixpkgs] @lheckemann pushed commit from @jtojnar to release-19.03 « nginx: build with openssl 1.1 »: https://git.io/fhxRE
crmlt has joined #nixos
<jomik>
weden, sphalerite: So I can get the revision with nixos-version - but, I'd probably rather that I was able to specify the channel to get the revision from. I guess I could just take the prefix of the `.version` attribute on lib. But is there a nice way to just get the whole revision?
<sphalerite>
jomik: I don't think I understand what you want
<jomik>
I want the commit sha of my current <nixpkgs>
<jomik>
or my <unstable>
<jomik>
the revision.
<sphalerite>
jomik: does `cut -d . -f 3 <$(nix-instantiate --find-file unstable)/svn-revision` do what you want?
<jomik>
Hmm, that just returns a path sphalerite ?
<sphalerite>
jomik: did you run the entire command between the `s?
<jomik>
Ah, adapted it wrongly - I am using fish :P Well, it sort of does.
<jomik>
It's the same as `nix-instantiate --eval '<unstable/lib>' -A version | cut -d . -f 3`, I guess?
<jomik>
Only gives me a prefix of the hash though, I'd prefer the whole thing.
ThatDocsLady has quit [Ping timeout: 258 seconds]
<wedens>
jomik: I've sent `lib.trivial.revisionWithDefault null` earlier
<ramses_>
sphalerite: that might be a rather simple suggestion, I'll give that a go and see how I like the resulting code
<ramses_>
tilpner: will mkMerge merge the lists within an attrset? Is there any documentation about this function?
<sphalerite>
jomik: ah, try .git-revision in the dir then :)
<jomik>
So that would be `nix-instantiate --eval -E 'with import <unstable> {}; lib.trivial.revisionWithDefault null'` as the winner?
<jomik>
Though, version seems to always be present right? And that last part seems to be a prefix of the revision
<ottidmes>
jomik: I just checked with: nix eval --raw '(with import <nixpkgs>; lib.trivial.revisionWithDefault null)' it works, but it fails for my unstable checkout, since its an added git worktree
<jomik>
I.e. "19.09pre171582.26d8a8c0eb2" is what version always looks like?
<jomik>
ottidmes: Oooh.
<jomik>
So what is your `lib.version`?
<tilpner>
ramses_: No, I was wrong
<jomik>
WELL, to be honest. I guess I just need to make it work for my setup! Lol
<ottidmes>
jomik: "19.03pre-git"
juliendehos has quit [Quit: Leaving]
nschoe has joined #nixos
<jomik>
ottidmes:Thanks :)
<ottidmes>
jomik: yep, so whatever works for you, works :P for me I would have to use what I suggested, but thats because I use git worktrees for my channels
<jomik>
ottidmes: git -C ... does not seem to work for me.
<ottidmes>
jomik: but you are using nix-channel, right? then it won't work
drakonis has joined #nixos
<jomik>
Ah, yeah. That makes sense. Lol.. :D
<ottidmes>
jomik: so for your use case, what sphalerite suggested is the most direct, the other solution within lib.trivial.revisionWithDefault depends on .git as well, which AFAIK is not present with nix-channel
gagbo has quit [Ping timeout: 255 seconds]
<jomik>
the revisionWithDefault works for me and gives the correct rev
infinee_ has quit [Quit: leaving]
drakonis_ has quit [Ping timeout: 250 seconds]
<wedens>
ottidmes: `revisionWithDefault` works with channels
gagbo has joined #nixos
<ottidmes>
wedens: never said otherwise, I only said it does not work for git worktrees. I meant to say with my last setence that you could just as well include .git-revision directly since the other check done in revisionWithDefault is for .git, which is not present in nix-channel directories AFAIK
<wedens>
I'd rather avoid using `.git-revision` directly, as I consider it "implementation detail". just my personal preference ;)
<yl[m]>
domenkozar: does this work with cachix: `nix-instantiate default.nix | cachix push yl` ?
<ottidmes>
wedens: well, since it fails for me, its already a bad abstraction, but I get your point
<yl[m]>
`nix-build` timesout for me on CircleCI, but nix-instantiate still builds some stuff that I'd like to push
<wedens>
ottidmes: what's the output of `nixos-version --revision` with your setup?
<ottidmes>
wedens: that wont give you any interesting results, its my unstable channel git worktree that gives no useful revision info
Taneb has quit [Quit: I seem to have stopped.]
<ottidmes>
which is just an extra channel, not my main channel, in my setup
<wedens>
ahh, ok
atriq has joined #nixos
<jomik>
is there a font with a NixOS symbol glyph ? :D
atriq is now known as Taneb
<wedens>
yes
drakonis_ has joined #nixos
orivej has quit [Ping timeout: 250 seconds]
<jomik>
,locate xfd
<jomik>
wedens: Which? :O
<therealwaphire[m>
is anybody using filebot on NixOS?
<{^_^}>
Found in packages: texinfo, texinfo5, wallabag, zafiro-icons, xfce.xfdesktop, xlibs.xbitmaps, iconpack-obsidian, papirus-icon-theme, texinfoInteractive, xfce4-13.xfdesktop, numix-icon-theme-circle, numix-icon-theme-square, elementary-xfce-icon-theme, perl526Packages.TextUnidecode, perl528Packages.TextUnidecode, perldevelPackages.TextUnidecode
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
Guest68667 has joined #nixos
infinee_ has joined #nixos
<jomik>
Umm, did I just install all nerdfonts by just installing `nixpkgs.nerdfonts`? :O
<srhb>
jomik: Yes.
<srhb>
jomik: And the github release tarball hosting is.. Sloow :)
<jomik>
LOL
<jomik>
YES
fendor has joined #nixos
<jomik>
I was like, this is taking a long time.
<jomik>
Then I checked fc-list... XD
<srhb>
Yeah, expect 20 minutes or so.
<pie_work>
Big Fonts
<infinee_>
I just added an overlay file for a previously installed `environment.systemPackage`. nixos-rebuild switch didn't seem to recompile that pkg from what I can tell. Is there a way to forcefully rebuild that package?
<srhb>
infinee_: If it didn't rebuild, either your change does not cause the change you expect, or your overlay is not picked up.
<jomik>
How do you use `withFont` in the nerdfonts package? Do I provide a font name, or what? :O
<wedens>
jomik: which font do you use?
<srhb>
jomik: Looks like it just passes that string to ./install.sh -- so whatever that takes.
<jomik>
Iosevka in this case. so I can do `nerdfonts.override { withFont = "Iosevka"; }` ?
<{^_^}>
nix#1930 (by ElvishJerricco, 1 year ago, open): Nix 2.0: Print out path to stdout with nix build
<tokudan[m]>
srhb: ah well, going back to nix-build then...
<tokudan[m]>
slabity: unrecognised flag '--verbose=5'
<tokudan[m]>
slabity: i'll just use nix-build
<tokudan[m]>
thanks!
<srhb>
tokudan[m]: Note that there's some sensibility to this: Without keeping a link around, the path might simply disappear whenever (on a system where gc can happen)
<jomik>
wedens: ah, yeah, cuz I have to fetch it all :P
<slabity>
tokudan: Sorry, that option is for nix-build
<srhb>
tokudan[m]: So the logic is something like: Do the build, keep the link around, manipulate the path however you need, then remove the root.
<tokudan[m]>
srhb: i don't care about any link, i want to build output while it's compiling and not just thge last 10 lines. not a path.
johanot has quit [Quit: WeeChat 2.2]
<srhb>
Not understood, but as long as you've got a working solution I guess :)
<tokudan[m]>
srhb: nix build only shows me the last 10 lines of the output on a failed build. that is not enough to identify the real problem
_kwstas has joined #nixos
_kwstas has quit [Remote host closed the connection]
<tokudan[m]>
srhb: nix-build gives me the full output during the build process, which allows me to figure out what's going wrong
__Sander__ has quit [Quit: Konversation terminated!]
<srhb>
tokudan[m]: nix log works for the first case though
<srhb>
Oh, no it doesn't.
<srhb>
Nevermind :)
<tokudan[m]>
srhb: got it covered with nix-build, so I'm fine :)
<srhb>
Great :)
orivej has joined #nixos
Neo-- has quit [Ping timeout: 264 seconds]
Thra11 has quit [Ping timeout: 255 seconds]
rpg has joined #nixos
rpg has quit [Client Quit]
rpg has joined #nixos
erasmas has joined #nixos
pie_work has quit [Ping timeout: 255 seconds]
arjen-jonathan has quit [Ping timeout: 255 seconds]
ddellacosta has joined #nixos
Thra11 has joined #nixos
<ldlework>
Anyone got some sweet dropbox nix expressions maybe including mounting some authenticated folder or something?
<srhb>
lejonet: I took a stab at trying to nixify the node+angular build of the mgr frontend. It's... Not fun. Might just be because I don't know the ecosystem, but I think we need to start thinking about alternatives. :P
<lejonet>
srhb: its node+angular? Count me out... xD
<srhb>
lejonet: Indeed...
owickstrom has quit [Quit: Connection closed for inactivity]
<lejonet>
I didn't spend the past 4 years maintaining a node+angular project and finally managed to drop it last month to take up another one... xD
<srhb>
Uh, but then you clearly know how to do it ;-)
<Izorkin>
yl[m]: checked PR datadog?
<lejonet>
Nope, I still don't know how to maintain it beyond "Don't Touch It!"
<lejonet>
xD
<srhb>
Ah, okay :P
<lejonet>
There really aren't any good ways to maintain them, unless you're actively coding on them
<lejonet>
Because the npm ecosystem is a... crapfest
Guest68667 has quit [Quit: Guest68667]
<lejonet>
and wtf redhat, why the hell does virt-manager now depend on dconf to save configured connections?
<srhb>
lejonet: It's a bit frustrating. I feel like I have all the deps specified and such, and yet I have no idea when the errors will stop coming. And of course, the dev cycle is.. Long.
<lejonet>
what happened to plain old files...
<lejonet>
srhb: yeah, compiling ceph isn't short :/
endformationage has quit [Ping timeout: 240 seconds]
<srhb>
Well I'm not there yet. I've stripped out the frontend build for sanity
<srhb>
But yeah, what do we do? Check if someone else built it for us so we can just shim it in? Write a release note that we've dropped support for it?
<ldlework>
The dropbox stuff on the wiki doesn't seem to wokr
<lejonet>
Well, in another, but related, note, I've deployed 13.2.4 to 3 different libvirtd servers, and had it running on one of them for at least a week or so, with no issues whatsoever
<srhb>
lejonet: Nice. Yeah, I feel this is probably the only outstanding issue.
<lejonet>
srhb: atm I'm mainly leaning towards the latter, while we search for the former
<lejonet>
but if its a node+angular story, it might be impossible :/
<lejonet>
node really wasn't designed to be packaged :(
<lejonet>
Because honestly, unless there is some node/npm version of elixirs distillery/exrm, I doubt there is a way to package it as a package
shabius has quit [Quit: Leaving]
<srhb>
lejonet: What's that?
<srhb>
Also I don't believe it's impossible, I'm just finding it hard to imagine a solution where the next (even minor) bump won't be a pain in the ass...
<lejonet>
distillery/exrm is just a packaging tools that takes a elixir project (erlang wrapper language more or less) and makes a release tarball
<srhb>
I see.
<lejonet>
s/ a /
<lejonet>
/
<lejonet>
screw my fingers today...
<srhb>
I wonder if we can just dissect a deb or something
<lejonet>
so maybe just grabbing that package might work without too much headache
<srhb>
Indeed. I'm assuming the frontend is fairly relocatable
<srhb>
Though honestly I'm not sure whether the backend is working yet.
<srhb>
Did you have luck with py3?
<lejonet>
I haven't looked more at it after it refused to locate the six module
<srhb>
Gotcha.
<lejonet>
Been busy nixosifying my infra :D all my machines that are supposed to be converted to nixos have been converted :D
nschoe has quit [Quit: Program. Terminated.]
<srhb>
Good job!
<drakonis>
conversely, check out fedora's ceph rpm spec
<lejonet>
so it seems like debian uses py3 for its ceph package, and explicitly turns off the dashboard frontend when building ceph, so its definitively a thing to look closer at
<drakonis>
the longest thing there is the changelog
<srhb>
drakonis: Looks like they also disable the frontend
dweller has quit [Ping timeout: 246 seconds]
<srhb>
Maybe it'll turn out the thing just isn't buildable
<srhb>
Gentoo also turned it off recently.
<srhb>
:-)
<lejonet>
Well, the dashboard is definitively not sandbox buildable as-is :P
dweller has joined #nixos
orivej has quit [Ping timeout: 250 seconds]
<drakonis>
yes.
<drakonis>
debian's way to do packages is baffling tho
<lejonet>
I know too little of both rpm and deb build processes, so I can't deduce how they actually manage to create the ceph-mgr-dashboard rpm/debs, but both specs clearly have it as separate packages
rpg has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Ariakenom has quit [Quit: Leaving]
jomik has quit [Ping timeout: 245 seconds]
<lejonet>
But it does indeed seem like none of em actually build the frontend, just the backend
<lejonet>
Because I see no nodejs deps in the ceph-mgr-dashboard package of either debian or rpm
Ariakenom has joined #nixos
kreisys has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<thoughtpolice>
tilpner: Mmmm, do you have your changes anywhere I could test them? I have used firecracker myself but only minimally...
<lejonet>
haha yeah, brb while I go and punch my punchingbag... xD
<srhb>
lejonet: OK, so now it's "just" a question of how to graft it in.
<srhb>
Actually that may already have happened too...
<srhb>
Hm, nope. Oh well, close
<lejonet>
srhb: so just grabbing the release tarball instead of git then?
<srhb>
lejonet: We already are doing that :)
<srhb>
So it's there, we just need to point the backend at it somehow.
<JonReed>
What does `${PATH:+':'}` in bash does? The `wrapProgram` script uses it inside like `export PATH=$PATH${PATH:+':'}<suffix>`, why not `export PATH="$PATH:<suffix>"`?
<symphorien>
It does not add any colon if PATH is initially empty
<JonReed>
symphorien: Ahh, ok, thanks.
<lejonet>
srhb: which probably "just" means finding what directory it expects to find those files
<srhb>
lejonet: Yep, looks like we need to pull out the prebuilt stuff (trivial path drv) and point module.py at it somehow
<srhb>
lejonet: return os.path.join(current_dir, 'frontend/dist') -- shoot me now
<srhb>
Ah well, trivial to fix.
<tilpner>
thoughtpolice: I attached a strace, but the exact KVM-related syscall before SIGSYS changes, so it might not actualy be KVM-specific
dweller has quit [Read error: Connection timed out]
<ldlework>
wow great, dropbox doesn't support xfs
dweller has joined #nixos
<srhb>
ldlework: They explicitly disabled everything but ext4 recently, iirc
yl has joined #nixos
<lejonet>
srhb: oh c'mon, can they stop doing lovecraftianesque stuff?!
<lejonet>
srhb: and I think I found why ceph-mgr craps out with py3 too, there is apparently ANOTHER cmake var you need to set for ceph-mgr to realize that we're using py3 besides WITH_PYTHON3
<srhb>
lejonet: Remind me, how do I check the DWITH default values?
<lejonet>
srhb: you dig through the CMakeFiles.txt xD (I know there is a cmake command to do it, but my cmake is really, really rusty)
<srhb>
Right, same here. Digging into the files it is :-P
<lejonet>
"Cowabunga it is" ;)
<srhb>
Accurate.
infinee_ has quit [Client Quit]
shabius has joined #nixos
<srhb>
lejonet: Did you try just enabling the dashboard on your..um, test cluster? :-)
gagbo has quit [Ping timeout: 240 seconds]
infinee_ has joined #nixos
<lejonet>
srhb: no xD let me try it :P
ramses_ has quit [Ping timeout: 256 seconds]
<lejonet>
This is a bad idea, lets do it!
edouard has joined #nixos
<lejonet>
well, enabling it went just fine, lets access it and watch the fireworks
<srhb>
lejonet: Because this path is present and correct for me: /nix/store/kj6p315vpblxyca0a8213qflqm14fppc-ceph-13.2.4-lib/lib/ceph/mgr/dashboard/frontend/dist
<lejonet>
srhb: well, enabling it didn't explode anything, and doing a curl localhost:7000 actually returned what looks like the site :O
<srhb>
lejonet: sigh. :)
<lejonet>
have we been digging unecessarily? xD
<srhb>
I believe so.
<{^_^}>
[nixpkgs] @dasJ opened pull request #56835 → nixos/nginx: Support additional listen parameters → https://git.io/fhxzL
<srhb>
I still want to know what that random python error is about, but..
erictapen has quit [Quit: Lost terminal]
<iclanzan>
Does anyone know how to install a package from a remote tarball using nix-shell? Doing `nix-shell https://url-to-tarball` installs the dependencies of that package not the pacage itself. I guess I need something like the -p flag but that works with remote tarballs.
<lejonet>
srhb: its cherrypy whining that its version is unknown, for some unknown reason... I dug a tiny bit in it, but I couldn't find how cherrypy.__version__ isn't a correct version
<srhb>
lejonet: Oh yeah, I remember now. That was just a search and replace, I think..
<ldlework>
arghhhh
<lejonet>
srhb: dashboard works just fine :)
<ldlework>
I'm just going to use a USB key I guess
<srhb>
lejonet: Well, ok.. Figure out the error message, check if we have changes that aren't documented in release notes, then good to mrge.. :-P
<srhb>
I've lost track of the changes..
sigmundv has quit [Ping timeout: 250 seconds]
<Izorkin>
aanderse: update nginx. Checked in virtualpc - normal running
<lejonet>
srhb: which changes do you mean we should look for?
<srhb>
lejonet: Any that require NixOS users to change config
<lejonet>
srhb: for the cluster you mean?
<srhb>
lejonet: Yep. I think there's nothing, really.
<srhb>
But I'm not totally sure.
<lejonet>
There shouldn't be any changes in conf, there would actually be relaxations in the conf because the requirement of a mgr is gone
<srhb>
lejonet: What, didn't they just make it mandatory? :P
<srhb>
In 12 I mean.
<lejonet>
srhb: Yes, they made it mandatory for there to be a mgr SOMEWHERE, the module I made made it mandatory everywhere there is a ceph daemon enabled xD
rpg has joined #nixos
<srhb>
Ah, right.
<lejonet>
Because I only tested it on a all-in-one cluster :P
<srhb>
lejonet: Something is still fishy in the test: mgr[py] Module not found: 'dashboard'
<srhb>
lejonet: Did you change something else?
<lejonet>
Not for dashboard
johanot has joined #nixos
<lejonet>
I'm getting gcc segfaults when compiling, so I'm gonna set my RAM down to 2800 again, brb
<srhb>
Ouch :P
<lejonet>
Eh, known hardware fault with first gen ryzens and it hasn't hit me until now (I recently got a bios version on my mobo that makes it not crash with the memory set to 2933 so :P )
<lejonet>
I haven't really been bothered by it, so I never sent in my CPU to get it replaced
_kwstas has quit [Remote host closed the connection]
<lejonet>
srhb: I think the solution to the cherrypy problem, in its most trivial form is: Find the version that was pulled into the python env (should be fairly easy I think) and then do a search/replace in the __init__.py file of cherrypy so that the fallback gets the right version too
_kwstas has joined #nixos
linarcx has joined #nixos
<tilpner>
ldlework: Did you reboot?
<linarcx>
Hi. when i want to list all generations it actully return 5 entries(40, 39, 38, 37, 36). but after i restart the system in grub menu i see 2 entry(27, 28). why?
arjen-jonathan has quit [Ping timeout: 240 seconds]
<lejonet>
linarcx: seems like your /boot wasn't mounted when building the past 12 generations
<linarcx>
lejonet: why? i have only one partition. /boot and install nixos on it.
<linarcx>
how it can be possible?
<lejonet>
That is weird indeed
<linarcx>
bug report it?
<lejonet>
What bootloader you use?
<linarcx>
grub
<linarcx>
and arch and windows installed on same hard disk
<Denommus>
I tried to change the pythonPath in my config.nix, but somehow it's still trying to build azure-related packages
<Denommus>
what am I missing?
<lejonet>
But you said you only had 1 partition? You mean only one partition for nixos?
<linarcx>
yes /dev/sda11 is for nixos(46G)
<lejonet>
linarcx: you have uefi or bios?
<linarcx>
bios
<lejonet>
but as you have 2 other OSes on the same harddisk, there must be a /boot partition outside of that nixos partition, which is what actually gets run
<linarcx>
yes indeed.
<Denommus>
oh, found it
<ldlework>
tilpner:that was it!
<lejonet>
linarcx: that partition needs to be mounted somewhere for nixos-rebuild to be able to update the generations
<linarcx>
lejonet: How can i do that?
<lejonet>
linarcx: how does your hardware-configuration.nix look?
<linarcx>
wait a second..
<srhb>
lejonet: It's a bit more complicated than that. Under normal circumstances, cherrypy reports the version correctly, so _something_ about the import/whatnot in mgr causes setuptools to barf out.
<lejonet>
linarcx: basically you want to add an entry for /boot in hardware-configuration.nix that mounts that partition to /boot
<linarcx>
lejonet: it's part of it: ` fileSystems."/" = { device = "/dev/disk/by-uuid/a43e3ab2-1915-4e8a-b83b-d3f9d6874478"; fsType = "ext4"; };`
<lejonet>
srhb: yeah, hence the most trivial way to solve it is to search/replace it, not saying that is the best way, because it might break other things, but it will remove the error :P
<srhb>
lejonet: Ah, sure, yes.
<lejonet>
linarcx: which will put all the generations in /boot on /dev/sda11, but that is not the device, seemingly, that the grub that is run references for nixos
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<lejonet>
srhb: I didn't look at the code long enough, to actually figure out what the code that tries to fetch the version actually tries to do
<lejonet>
srhb: but its almost like it tries to access the version like a class method or similar
<srhb>
lejonet: There's a lot of try: except: that makes it a bit harder.
<srhb>
It might even be a barfed import somewhere..
<lejonet>
srhb: indeed, and I spent maybe 5 mins looking at it
<{^_^}>
[nixpkgs] @andir pushed to openssl-1.1 « openvswitch: use openssl 1.0.2 for the time being »: https://git.io/fhxg0
maximiliantagher has joined #nixos
<linarcx>
lejonet: What should i do? change that device uuid?
<DigitalKiwi>
what does this mean "/etc/kbd directory contains user files. Symlinking may fail. at /nix/store/###-setup-etc.pl line 114." repeated for other than kbd too
<thoughtpolice>
tilpner: Are you sure this is KVM related and not seccomp related? SIGSYS is also what is sent by the firecracker seccomp tooling to the process
<thoughtpolice>
If you built it with glibc, while they use musl, a seccomp-related failure would not be surprising (they may use different calls internally over time to implement some API... it's one of the nastiest issues with seccomp in general)
nckx has quit [Quit: Updating my GNU Guix server — https://guix.info]
<lejonet>
linarcx: I haven't dual-booted nixos, so I dunno exactly how to solve it. What partition does Arch linux use for /boot?
<linarcx>
lejonet: /dev/sda8
_kwstas has quit [Read error: Connection reset by peer]
<lejonet>
linarcx: when you boot, do you get the option of either booting arch or booting nixos?
_kwstas has joined #nixos
<linarcx>
lejonet: Yes. i have a list of options: `1. nixos-defalult 2.nixos-configurations-list 3.arch-linux 4.windows`
<linarcx>
another wired behaviour is that last time i boot into nixos and wana to list generations, it return nothing :|
<linarcx>
but now it return 5 last generations properly.
<tilpner>
thoughtpolice: No, I'm not sure at all
<lejonet>
linarcx: I dunno how you've got your grub setup to point to all of those at the same time, was too long since I fiddled with grub, but you need to somehow manage nixos-rebuild to be able to write to the location which hosts your grub configuration
jtojnar has quit [Read error: Connection reset by peer]
<lejonet>
Its sadly outside of my knowledge atm :(
jtojnar has joined #nixos
<linarcx>
lejonet: No problem. thank you so much for your help. :)
<lejonet>
srhb: I think what we should look at to debug the real problem is lib/ceph/mgr/mgr_module.py, when I switched to py3, even if I include six in the python env, that one can't import it
<lejonet>
linarcx: good luck :)
linarcx has quit [Remote host closed the connection]
<srhb>
lejonet: Huh, why do you suspect that?
<srhb>
I mean, why would it affect cherrypy.__version__
<lejonet>
srhb: because I think its mgr_module.py which actually loads all the modules, kind of how gentoo does with python-exec
<Denommus>
the original preConfigure of both packages is still trying to run
the-kenny has quit [Quit: WeeChat 2.3]
<lejonet>
srhb: so my guess is that the whole switcheroo around screws up pythonpath or something
<srhb>
lejonet: The weird thing is that I'd expect cherrypy to fail to import then. Which isn't happening.
<srhb>
Denommus: overridePythonAttrs probably?
<lejonet>
srhb: oh, now I see why it needs the cherrypy version, oh god, nuke it from orbit
* lejonet
dies
<srhb>
Denommus: But I think you're nuking more than you want to in that approach? I think there's python override examples in the nixpkgs manual that you can refer to
<{^_^}>
undefined variable 'adding' at (string):220:1
aw has joined #nixos
<catern>
LICENCEEEEEEEEEE
<Denommus>
isn't it great when whenever you reach a milestone in a nix overlay you get another error right after that?
<catern>
why would someone have multiple Es at the end of LICEN[CS]E?
<ddellacosta>
(addendum: there isn't actually any documentation at the link in the README, as far as I can tell: https://github.com/NixOS/nixpkgs)
spacefrogg has joined #nixos
<srhb>
catern: Or none!
<catern>
true, true
jomik has quit [Quit: WeeChat 2.4]
<catern>
LICENS
<catern>
lol
<srhb>
Arguably that's a more fair situation.
<samueldr>
catern: LICENSE.rtf maybe?
<catern>
licenz
<srhb>
Oh, it's a glob :P
<catern>
wait, [CS] is valid in globs?
<samueldr>
probably not, probably a glob-like syntax
<srhb>
catern: yes.
<srhb>
Well, it is in my shell at least..
<samueldr>
not in mine :)
<catern>
huh, yeah, it's valid in bash
<catern>
wow, I didn't know that
<srhb>
zsh too.
<samueldr>
oh, I was had by nullglob
<catern>
I usually would do {C,S} instead
<srhb>
Ditto.
<catern>
(which I think is equivalent? although interestingly the output filenames are listed in a different order with each technique...)
<catern>
(aha, I think it's because [CS] works by matching, while {C,S} generates two separate globs - it's not actual part of the glob syntax)
jomik has joined #nixos
<srhb>
Oh yeah, you're right.
<srhb>
Time to relearn I guess >_>
jomik has quit [Client Quit]
[Leary] has joined #nixos
mmlb has joined #nixos
jomik has joined #nixos
jonafato has quit [Ping timeout: 246 seconds]
<jomik>
Anyone here using polybar? I am having an issue where the polybar started by systemd service does not show all my modules, but when I start it by executing the script that ExecStart points at, it works... Any idea why?
jonafato has joined #nixos
Lears has quit [Ping timeout: 246 seconds]
<jomik>
No errors in journalctl from what I see..
gagbo has joined #nixos
jomik has quit [Client Quit]
orivej has joined #nixos
jomik has joined #nixos
<jomik>
Oh, there is a polybar IRC :P
<srhb>
jomik: Is it possible that it's an impurity in your environment?
<tilpner>
jomik: I eventually gave up and set ExecStartPre = sleep 1
<tilpner>
:(
oldandwise has joined #nixos
<oldandwise>
i need a nix expression to compile a C++ program that uses pthread
<catern>
oldandwise: does this C++ program already use some build system?
<srhb>
lejonet: Gutting out that check does indeed work. Bedtime.
<gchristensen>
srhb: sleep well!
<oldandwise>
yes,.. i just want to setup the sorrounding environment giving it cmake, c++ and python etc...
<catern>
oldandwise: pthread should already be present with stdenv.buildDerivation, try running a simple compile using pthread inside "nix-shell" and see if that works
<oldandwise>
i,m using, stdenvNoCC
<oldandwise>
to control c++ version
civodul has joined #nixos
gagbo has quit [Ping timeout: 244 seconds]
<jomik>
srhb: The script looks like this : https://git.io/fhx2H My ExecStart points to that, which causes missing modules, but it works if I execute the script directly in my shell. So maybe.
<catern>
okay, not sure if that one will have pthread, but it might, so still try nix-shell and a test compile
<{^_^}>
[nixpkgs] @Infinisil merged pull request #56550 → nixos/xserver: Point to man page for options available in monitorConfig → https://git.io/fhAL5
maximiliantagher has quit [Remote host closed the connection]
maximiliantagher has joined #nixos
jomik has quit [Quit: WeeChat 2.4]
barryfm has joined #nixos
jomik has joined #nixos
<{^_^}>
[nixpkgs] @mquinson opened pull request #56849 → Fix the upstream project URL → https://git.io/fhxVz
alex``` has quit [Read error: Connection reset by peer]
<jomik>
So I am unable to see output from my user systemd services with `journalctl --user --unit` but I can find it with `sudo journalctl _SYSTEMD_USER_UNIT=chatty.service`? Why?
<jomik>
disregard that, what the sudo one shows me is old I thing :D
<jomik>
s/thing/think
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
zupo has joined #nixos
alex`` has joined #nixos
alex`` has quit [Read error: Connection reset by peer]
ATuin has quit [Ping timeout: 255 seconds]
shabius has joined #nixos
<barryfm>
Is the 19.03 channel active? Its not being updated by the 19.03pre tested updates
<ottidmes>
barryfm: the 19.03 already exists, but its still in beta, not yet the official new stable
jomik has quit [Quit: WeeChat 2.4]
<barryfm>
The successfull updates for 19.03 hasn't been made active
alex`` has joined #nixos
ddellacosta has joined #nixos
jomik has joined #nixos
<{^_^}>
[nixpkgs] @veprbl pushed to master « simgrid: Fix the upstream project URL »: https://git.io/fhxVP
rpg has quit [Ping timeout: 240 seconds]
<suzu>
hmm
<suzu>
maybe that's why my haskell packages are not coming from nixos cache..
<disasm>
I'm not sure if it would be considered an upgrade or downgrade if I went to 19.03. I'm on unstable but last updated a couple weeks ago I think.
<jomik>
19.03 is not released yet, it's in pre-release :P
<jomik>
tilpner: Lol, I seem to have deduced that my issue arises when the service is forking... So I could just generate a service per bar that is simple instead ....
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Moredread[m]>
Is there a way to check if a file somewhere on the fs exists in a nix derivation? I want to set releasePath for renoise, which requires downloading it manually and setting the path to the downloaded file.
<suzu>
you mean you want to search the nix store for a file?
<Moredread[m]>
But I share the config on multiple PCs and haven't the file always uptodate on all.
gagbo has joined #nixos
maximiliantagher has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @kalbasit pushed to nixpkgs_fix-go-meta-branch « go_1_12: set meta.branch to 1.12 »: https://git.io/fhxVQ
<Moredread[m]>
suzu: No, the file is outside of the store.
maximiliantagher has joined #nixos
<Moredread[m]>
I currently have an overlay with "renoise = pkgs.renoise.override { releasePath = ~/Downloads/rns_3_1_1_linux_x86_64.tar.gz; };", but I want to only do the override if the file actually exists.
jomik has quit [Quit: WeeChat 2.4]
<Moredread[m]>
Renoise is a particular case because the file is different for each user, so there is no canonical sha hash for that file.
Denommus has quit [Remote host closed the connection]
<suzu>
what in the fuck why is nix building haskell packages manuallyu
<Moredread[m]>
That is why the releasePath is somewhere outside of the store.
jomik has joined #nixos
bvdw8 has joined #nixos
maximiliantagher has quit [Ping timeout: 246 seconds]
eyJhb has quit [Quit: Bye]
Aleksejs has quit [Ping timeout: 272 seconds]
<simpson>
suzu: That's the default mode of operation for Nix; the binary cache is a lucky optimization, not a guarantee.
phire has quit [Read error: Connection reset by peer]
rev_strangehope has quit [Ping timeout: 272 seconds]
bvdw has quit [Ping timeout: 272 seconds]
Thra11 has quit [Ping timeout: 268 seconds]
johanot has quit [Quit: WeeChat 2.2]
rev_strangehope has joined #nixos
Aleksejs has joined #nixos
maximiliantagher has joined #nixos
<{^_^}>
[nixpkgs] @kalbasit pushed 0 commits to nixpkgs_fix-go-meta-branch: https://git.io/fhxVh
<{^_^}>
[nixpkgs] @kalbasit opened pull request #56852 → go_1_12: set meta.branch to 1.12 → https://git.io/fhxwv
<simpson>
Yeah, that manual should still be relevant even on non-NixOS.
fendor has quit [Ping timeout: 250 seconds]
<Moredread[m]>
suzu: have you looked into cachix if you are interested in sharing the build results on multiple machines?
<simpson>
You basically *are* making your own cache when you compile. If you're compiling a lot of stuff, then it likely means that whatever you're building isn't something that's built by nixpkgs Hydra.
<suzu>
are there any cachix caches that contain haskell packages that i can subscribe to?
<yl>
suzu: you can simply create your own and push to it
ddellacosta has joined #nixos
<suzu>
i suppose. ok
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
slack1256 has joined #nixos
<rnhmjoj>
do you know of any changes related to openGL recently? i have a package that stopped building due to undefined references to openGL functions like glTexEnvi and glClearDepth. i can't pinpoint the commit because it also broke due to an unrelated bug in SDL
<rnhmjoj>
(now fixed)
<{^_^}>
[nixpkgs] @Profpatsch merged pull request #56824 → skawarePackages: skalibs 2.8.0.0 and updates all around → https://git.io/fhx4x
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<yl>
WilliamHamilton[: you should override the patches as well
<pie__>
question: i think i saw something at some point where you could have a template file and substitute named things in it with nix somehow, do i remember wrong, or what was it?
<WilliamHamilton[>
yl: so I have to figure first how that patch has to be modified, right?
<pie__>
symphorien, doh, i did not think this is osmething the basic substitute thing does
kvda has joined #nixos
<pie__>
ugh this feels kind of ugly for some reason though
<pie__>
like, the semantics of env vars and stuff gets all mixed up in there
nschoe has quit [Ping timeout: 245 seconds]
<pie__>
theres no nix level substitute?
<pie__>
use case: I want to generate a configuration file from a template
<pie__>
well...i guess if nix has basic string substitution it cant be that hard to implement
<pie__>
replaceStrings it is
<clever>
pie__: would be easyer to do it at the bash level, with sed for example
<pie__>
but then id have to like...generate sed
<pie__>
(which may not be all that difficult in this case but also feels ugly)
orivej has quit [Ping timeout: 245 seconds]
<WilliamHamilton[>
yl: eventually I reviewed that patch and since I'm not interested in python bindings, I just set `patches = [];`. But I'm still curious on how one would update that patch if he was interested in it
<pie__>
clever, i mean on one hand, thats kind of cool, on the other hand, ehhh
<pie__>
nixpkgs is pretty amazing
<pie__>
clever, also im increasingly feeling like most of nix is built on the same few files
<pie__>
10% of the code builds 90% of the pckages? :P (those numbers probably arent extreme enough)
alex`` has quit [Ping timeout: 240 seconds]
<JonReed>
Do urxvt plugin packages like `urxvt_font_size` have any outside of `rxvt-unicode-with-plugins`? That it can they be used by themselves or they will ultimately be always only used by `rxvt_unicode-with-plugins`?
<JonReed>
s/have any/have any use
thc202 has quit [Ping timeout: 246 seconds]
<JonReed>
(I'm trying to add a new urxvt plugin and I want to patch it inside the `wrapper.nix` of `rxvt_unicode-with-plugins`, which make the plugin itself useless if it's used by anything but this wrapper. But all urxvt plugins are still available in `all-packages` individually, so I wonder whether they are somehow used outside that wrapper
<suzu>
this has been building for hours wtf!
<suzu>
theres gotta be a bin cache of haskell pkgs someplace
<simpson>
suzu: What are you building?
<suzu>
a haskell pakcage that im working on
<suzu>
and it is still building every fucking dep ever
<suzu>
manually
<pie__>
are you building with a custom GHC or something for some reason
<suzu>
no
<suzu>
i use stack for the project
<Moredread[m]>
are you on the stable channel?
<suzu>
so it's all packages from the lts
<suzu>
no. im on the unstable channel
<pie__>
theres a binary cache but it sounds like you somehow ended up building the world.
<pie__>
if the unstable had an update recently or something to haskell stuff then the binary packages might not have caught up
<suzu>
how can i switch to the stable and try?
<pie__>
i dont have any snippets on hand (maybe someone else can help with that) but you probably want to look into pinning
<pie__>
which is to say pinning your build to a specific nixpkgs version
<pie__>
(this is good to do anyway IMO)
<Moredread[m]>
suzu: you can try it with the -I flag for example, e.g. "nix-build -I nixpkgs=channel:nixos-18.09"
<pie__>
doh, thats simpler ^
<suzu>
also to be precise: it's a project that i made with stack using a stack lts. no fancy stuff here, i have a few deps that are cloned github repos, a few from hackage, and then 999 from stackage
<pie__>
(pinning isnt that hard either)
<suzu>
oh yay it just died becuase i ran out of diskspace
<suzu>
sigh
<suzu>
ok i will try that build flag and see if something different happens
<suzu>
oh yes and i also made my .nix by running stack2nix
<suzu>
and i'm just building a (callPackage ./stack2nixout.nix).mypkg
<Moredread[m]>
suzu: did you use dependencies from nixpkgs? I'm not familiar with the haskell workflow, but it might be that your build tool pulls the deps instead. Can you post your derivation e.g. on https://gist.github.com/?
gagbo has joined #nixos
<jabranham>
I need a browser with flash. Anyone know what still has flash these days?
<suzu>
sure, one sec while i paste
<pie__>
oh you said youre using stack
<pie__>
what jabranham also sounds possible
<pie__>
i cant remember if stack has nix integration enabled by default but tbh i dont even know what it does
<clever>
suzu: if you manually use callCabal2nix on just one or 2 things, you wont have to rebuild the world, but you will also not be using the same versions stack.yaml defined
<suzu>
hmm
<suzu>
but the point of stack is to pin the versions of dependencies that all build together
<suzu>
if i just build everything with the latest avail vers of all the deps from nixpkgs, i will likely break all my shit
alex``` has joined #nixos
<clever>
suzu: so you have the choice between:
<clever>
a: build a few things, but risk breaking it
<clever>
b: build all the things (with nix)
<clever>
c: build all the things (but with stack)
<clever>
i generally just ignore option c
<suzu>
hmm alright
<suzu>
how do i go full-on nix with option b?
<suzu>
does that play nicely for local development?
<clever>
thats what stack2nix does
<suzu>
but stack2nix builds /everything/ from scratch
<suzu>
:(
<clever>
what is an example of something its building?
<suzu>
aeson, sha
<suzu>
containers
<suzu>
basically everything in that file i pasted i suppose
<suzu>
gets built
<clever>
stack would have built those from source anyways, so its not really any worse
<suzu>
and with no help from binary caches
<clever>
the simplest thing you can do is to setup a hydra box, to pre-build things, and cache them for many users
<suzu>
would it have? it seems like this nix build has been actually chugging all day
<clever>
cachix lets you skip having to run your own machine for that
<WilliamHamilton[>
clever: thanks, but even with that I get the infinite recursion; I'm issuing "nix-shell" "-p" "with import <nixpkgs> {}; pkgs.haskellPackages.ghcWithPackages (p: [(pkgs.haskell.lib.dontCheck (pkgs.haskell.lib.dontHaddock p.pretty-show)) (pkgs.haskell.lib.dontCheck (pkgs.haskell.lib.dontHaddock p.z3))])" "--run" "ghci"
<clever>
the cabal file defines what the deps are, but versions are still based on nixpkgs
<clever>
WilliamHamilton[: i suspect you may also have super and self backwards in one or more files, try swapping them around
<suzu>
so if i upgrade nixpkgs i may need to upgrade my project to accomodate the new vers of the deps
<clever>
WilliamHamilton[: i can never remember which way around it goes
<clever>
suzu: i try to also pin nixpkgs, so when i update my channel, the project doesnt update
<suzu>
ok. and how do you do local development?
<clever>
nix-shell and nix-build
<clever>
for haskell packages, you need to use the .env attribute to get a usable shell
slack1256 has quit [Remote host closed the connection]
<suzu>
what about things like ide integration?
<clever>
suzu: i just use plain vim, and a ghcid running in nix-shell
slack1256 has joined #nixos
<clever>
ghcid -c "runhaskell Setup.hs repl"
<clever>
suzu: this tells ghcid to run "cabal repl" to get a ghci
<suzu>
hmm. i use spacemacs with intero
<clever>
and then cabal deals with all the special args ghci needs to start
<WilliamHamilton[>
clever: that was it, thanks!
<suzu>
hmm
<suzu>
okay
<suzu>
i think i should just ditch stack then
<suzu>
and nix all the things up
<suzu>
and then i can profit from the cache
<WilliamHamilton[>
clever: btw, I have a lot of problem because cabal generates `.ghc.environment.x86_64-linux-8.6.3` files, and I do not always remember to clear them. is there a way to disable them?
<clever>
WilliamHamilton[: none that i know of
<WilliamHamilton[>
clever: got it, thanks!
<suzu>
does cabal new-build use the nix cache?
<clever>
suzu: i dont think it does
<suzu>
okay
<suzu>
yolo
<clever>
suzu: and if there is even a slight mismatch between what nix-shell provides, and what cabal wants, cabal will just build its own copy
<suzu>
how can i get started nixing up my project?
<clever>
suzu: start with a nix file similar to the extra-statsd one