<clever>
which is why i mentioned needing 2 service entries
<codygman>
got error "called with unexpected argument 'src'
<clever>
codygman: you need to use .overrideAttrs
<iqubic`>
clever: I've given up trying to make this service run periodically.
concatime has joined #nixos
<concatime>
Hi. How do we aloud a build to access to internet?
<iqubic`>
I just need to have it run once per boot, before display-manager.
<clever>
codygman: make it fixed-output, by telling nix what the hash of the result is
iqubic` is now known as iqubic
codygman has quit [Ping timeout: 260 seconds]
<sir_guy_carleton>
if i rollback a system profile and use nixos-rebuild to build a new system profile, would it override the rollbacked system profile?
<clever>
sir_guy_carleton: it would build a new one based on the current configuration.nix
<clever>
which will upgrade things again
<clever>
and also the current nixpkgs
<sir_guy_carleton>
so i've rollback from 45 to 44 and if i nixos-rebuild it would make 46, right?
<clever>
yeah
<clever>
so the rebuild would switch you from 44 to 46
<clever>
and confusingly, a rollback would put you back on 45
<jtojnar>
I am getting "syntax error, unexpected $undefined, at /nix/store/xh6ghidxl984qffakv5rkny59m5m788j-nixos-unstable.tar.gz:1:1"
<iqubic>
How do I check if a certain systemd service is running. I just want to make sure that my custom service started alright.
<iqubic>
s/running./running?/
<Myrl-saki>
iqubic: systemctl status?
<iqubic>
Yeah, that's showing me that it is inactive.
<iqubic>
Or, wait, no I read that wrong. We're cool.
xcmw has joined #nixos
<{^_^}>
[nixpkgs] @aszlig opened pull request #40053 → nixos/bind: Allow to set extra options → https://git.io/vprxt
andreabedini has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
andreabedini has joined #nixos
<sauyon>
what's a quick solution to the locale issues that were introduced
codygman has quit [Ping timeout: 260 seconds]
sir_guy_carleton has quit [Remote host closed the connection]
wchresta_ has joined #nixos
nuncanada has quit [Quit: Leaving]
leary` has quit [Remote host closed the connection]
wchresta_ has quit [Ping timeout: 240 seconds]
andreabedini has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<{^_^}>
[nixpkgs] @aszlig opened pull request #40054 → nixos/nsd: Allow to configure root zone → https://git.io/vprxx
leary` has joined #nixos
freeman42x]NixOS has quit [Ping timeout: 268 seconds]
astrofog has joined #nixos
jethroksy has joined #nixos
<iqubic>
So my Fn lock key works fine, but not the num lock key.
<iqubic>
the num lock key is Fn+F5, and it's supposed to allow me to use the 789uiojkl cluster of keys without having to use Fn + i or Fn + o. But it's not working for me.
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Arcaelyx has quit [Read error: Connection reset by peer]
u0_a152 has joined #nixos
u0_a152 has left #nixos [#nixos]
<fearlessKim[m]>
I would like to conditionnaly import a file depending on a module existence (to be able to rebuild even if my module doesn't exist on nixos-unstable): `lib.optional (config.programs ? myCustomProgram)` generates an infinite loop
<fearlessKim[m]>
infinisil: thanks but I don't know which condition to use. basically I have this module that exists in one branch and doesn't in nixos-unstable. I want it to be imported when available . Maybe my only option is to use builtins.fileExists then
<infinisil>
Yeah, I don't know of anything else that could work
<infinisil>
Although, maybe you could use the options argument
<jD91mZM2>
I'm having issues with xmonad again :(. When using a single monitor, it's fine. But when I add my second monitor the rendering goes horrible
<jD91mZM2>
xmonad places the windows in the right location, but off-screen. It's like xmonad doesn't agree with my monitor settings
<jD91mZM2>
I can apparently take screenshots still
<stumble>
We can evaluate things here, can we?!
<clever>
jD91mZM2: i'm guessing that xmonad isnt aware of the screen changing size?
<stumble>
> 1 + 1
<{^_^}>
2
<stumble>
Awesome.
<jD91mZM2>
anyway, the mouse is weird as well. It selects things like 5cm above it
<jD91mZM2>
clever: Sounds right to me.
aarvar has joined #nixos
<jD91mZM2>
Their wiki wants me to check if xmonad is correctly compiled with xinerama, but because of nix's structure the ghc xinerama library isn't even installed
<leary`>
jD91mZM2: you may be suffering from the state bug if you're on non
<leary`>
non-git 0.13*
<infinisil>
stumble: Yup, I just added that a few days ago :)
<jD91mZM2>
leary`: state doesn't exist
<jD91mZM2>
ok now it does. but I've tried deleting it
<infinisil>
stumble: If you wanna spam the bot you can do so in #bottest
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/vpovO
<{^_^}>
→ d25607c7 by @megheaiulian: Fix pci_get_bus_and_slot removed in kernel 4.17
<{^_^}>
→ 52dac088 by @matthewbauer: Merge pull request #39989 from plumelo/feature/virtualbox-kernel-4.17-patch
<leary`>
I unless the version is way older I doubt it will make a difference. You could upgrade xmonad and contrib to git.
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/vpovs
<{^_^}>
→ b79d1e80 by @mogorman: platformio: add git to chroot so it is usable by platformio command
<leary`>
But having a working TTY or another WM to delete the statefile from is probably all you need.
<{^_^}>
→ 96ef188a by @matthewbauer: Merge pull request #39998 from mogorman/platformio
<{^_^}>
[nixpkgs] @matthewbauer merged pull request #39998 → platformio: add git to chroot so it is usable by platformio command → https://git.io/vpw2A
<leary`>
jD91mZM2: well I don't know what's causing the issue if the statefile isn't present, it just sounds so much like that bug, and it's easily one of the most common causes of trouble with 0.13 ... but ask on #xmonad, maybe someone will be able to help.
<leary`>
The channel is slow but people do check it.
<jD91mZM2>
I'll check again, maybe the statefile is there now... or something idk
<clever>
achambe: of note, nixos already forces nixpkgs to use nixpkgs.config and nixpkgs.overlays for the config and overlays, so the users config.nix and overlays wont apply
jethroksy has quit [Ping timeout: 264 seconds]
<achambe>
ah ok
<clever>
those values must instead be defined in the configuration.nix
<clever>
so only configuration.nix and nixpkgs themselves are loaded from NIX_PATH
<clever>
as a way of doing things in a more pure manner
<jD91mZM2>
now not even single-monitor is perfect for some reason
jperras has quit [Ping timeout: 240 seconds]
<jD91mZM2>
this is confusing
<jD91mZM2>
I'll try disabling my monitor setup and see if using the default xrandr settings makes it approachable
<achambe>
clever: is there a way to pass the hash in from -I ?
rosa has joined #nixos
<achambe>
i mean, not the git-revision but the output hash
<clever>
achambe: when using a URL like that, it is technically impure, nix will check for updates if it has been over an hour since you last used it
<achambe>
that is pretty close to what I want though tbh.
<achambe>
oh I see
<clever>
achambe: but as long as github isnt cheating, it will return the exact same thing every time
<achambe>
thanks, that is something I never understood
jD91mZM2 has quit [Client Quit]
<clever>
[clever@amd-nixos:~]$ ls -ltrh ~/.cache/nix/tarballs/
<clever>
achambe: nix uses this directory to cache the results of fetching such things
<clever>
0lmg3b0f50wmfbh3fgbfz80g2lhzjy04cps1mm2ziyxmafkiqlmp.info and 0lmg3b0f50wmfbh3fgbfz80g2lhzjy04cps1mm2ziyxmafkiqlmp-file are based on a hash of the URL, and contain metadata on it
<clever>
and based on the metadata in the .info, it will know how old the -file is, and if it should check again, and what cache-control magic is at play
<clever>
and if things are unpacked, there is also a hvm7h0ydc68hqa9x5657f5h8lh8437cr-4975378b2139f24b1368f44e9e3d44e34c3e48ad.tar.gz-unpacked symlink
<clever>
which is based on a hash of the .tar.gz the url gave
<achambe>
right, it would be nice if i could put the output hash so it never needs to recheck, but i do trust github mostly.
<achambe>
to return the same stuff
<clever>
so if 2 URL's give binary-identical .tar.gz files, it can reuse the -unpacked variant
jD91mZM2 has joined #nixos
coot_ has joined #nixos
<clever>
you could also set the sha256 directly, if you add another larger nix file into the mix, but then <nixpkgs> entries in nixpkgs (which is a bug) would fail
coot_ has quit [Remote host closed the connection]
<leary`>
BTW jD91mZM2 I could be wrong about this, but I have a theory. If your monitor setup is being configured by a script after xmonad loads, then restarting xmonad statefully might not suffice to make it see than config, and deleting state from outside an xmonad session wouldn't help because the conditions are recreated every time you log in.
<leary`>
jD91mZM2: If that's what's happening, you might be able to fix it by configuring your screens with xorg config before xmonad sees them.
<leary`>
Though it would be better to just update xmonad instead.
<jD91mZM2>
Seems like I'm using the vesa drivers. Shouldn't nouveau be the default for nvidia systems?
rosa has quit [Quit: rosa]
coot has joined #nixos
<clever>
jD91mZM2: the vesa drivers will be available even if your not using them
<etu>
jD91mZM2: On live-usb it's not nouveau and I'm very happy for that :D
<clever>
jD91mZM2: you need to check the logs for Xorg to see which drivers it used: sudo journalctl -u display-manager -b 0
<etu>
Cuz my nvidia-card doesn't work with nouveau at *all*. I don't even get a tty. It's just blackscreen. So NixOS live-usb is more or less the only one that works out of the box without adding kernel params to disable nouveau :)
<jD91mZM2>
clever: That's just showing me xorg crashes lol
<fearlessKim[m]>
I am writing a module for telnet with its own user telnetd. On start I have `bind: Permission denied`, how can I grant the telnetd user the bind persmissions ? add him to "network" group ?
<jD91mZM2>
Oh noes, I tried to set hardware.nvidiaOptimus.disable, but it failed
<jD91mZM2>
-bbswitch-0.8-4.14.39/bin/discrete_vga_poweroff: line 3: /proc/acpi/bbswitch: No such file or directory
<tilpner>
fearlessKim[m] - Can try serviceConfig.CapabilityBoundingSet = "CAP_NET_BIND_SERVICE";
<jD91mZM2>
I'm also getting stuff like "setup: line 313: pkgBuildAccumVars: readonly variable"
lonokhov has joined #nixos
<GiGa|Laptop>
fearlessKim[m], what's the use case for an insecure protocol such as telnet?
<GiGa|Laptop>
jD91mZM2, would adjusting a setting on bumblebee help?
<fearlessKim[m]>
GiGa|Laptop: I am not sure but it seems mininet (a network simulator) uses it to run commands on netns nodes
<jD91mZM2>
GiGa|Laptop: I'm not sure, haven't really done anything else than set that setting
rosa has quit [Quit: rosa]
<jD91mZM2>
running `sudo systemctl start bbswitch` manually works
<jD91mZM2>
I'll try rebooting
jD91mZM2 has quit [Quit: WeeChat 2.0]
rosa has joined #nixos
<sphalerite>
fearlessKim[m]: the process needs to have CAP_NET_BIND_SERVICE to bind port 23. I think a nicer solution though is using systemd socket activation, if possible
<sphalerite>
oh, tilpner already said the cap thing
<{^_^}>
→ 69bf313e by @Mic92: nodePackages{v6,v8}: update
<{^_^}>
→ 0e5d6a66 by @Mic92: nodePackages.imapnotify: init at 0.4.1
<{^_^}>
→ f343efcf by @Mic92: nodePackages.elm-oracle: init at 1.1.1
<sphalerite>
You shouldn't need to change videoDrivers at all really
<jD91mZM2>
Okay so what exactly do I want to achieve here? I want my display to stop being weird when starting. They kept being weird even when using only the intel driver. So apparently that's not the solution.
<sphalerite>
you're on unstable, right?
<jD91mZM2>
Nope
<jD91mZM2>
I wanted to, but since X crashed I just realized that unstable maybe wasn't the best right now
<{^_^}>
→ 1c67dc2f by R. RyanTM: libfm-extra: 1.3.0 -> 1.3.0.2
<{^_^}>
→ 4a0832d6 by @Mic92: Merge pull request #40090 from r-ryantm/auto-update/libfm-extra
<sphalerite>
an issue with xmonad is a lot easier to work out than an issue that may be in X, X's video driver, the kernel graphics drivers, mesa, …
rosa has quit [Quit: rosa]
<clever>
ive had fun trying to debug why page flipping didnt work in the android drivers for framebuffers on the rpi before
<sphalerite>
so when did this issue start appearing? AFAIU you weren't having it yesterday?
pie___ has joined #nixos
<sphalerite>
clever: what happened to sleeping? :p
<clever>
android implements page flipping by configuring the framebuffer to have a virtual height that is twice the screen height
<clever>
and then it scrolls up/down 1 whole screen, to flip
rosa has joined #nixos
<clever>
years after giving up on fixing that i found the problem, when you use the GPU function to reconfigure the framebuffer (to adjust scroll), it re-allocates the framebuffer ram
<clever>
clearing the whole screen
<jD91mZM2>
sphalerite: Worst part was that overriding with -Wno-error didn't work either because the wrapper adds their flags last
<clever>
so the act of making a given frame you just render visible, also clears the frame, right as you make it visible, lol
<sphalerite>
jD91mZM2: when you're back — what hardware is this on?
<{^_^}>
[nixpkgs] @fpletz pushed commit from @Ma27 to release-18.03 « shutter: wrap `XDG_DATA_DIRS` with `hicolor-icon-theme` »: https://git.io/vpocv
<{^_^}>
[nixpkgs] @fpletz pushed commit from @bjornfor to release-18.03 « shutter: add missing dependencies for Screenshot->Export (Ctrl+U) support »: https://git.io/vpocU
<jD91mZM2>
I think that's what the X550LD is. An ASUS.
<sphalerite>
ah ok
wchresta_ has quit [Ping timeout: 268 seconds]
<sphalerite>
oh right, you were using a different linux distro on it previously, right? And I'm guessing you didn't have this issue
<jD91mZM2>
Yeah. In fact, I didn't have this issue yesterday o_o
<sphalerite>
:/
<jD91mZM2>
And I was using NixOS then. Same channel. Same everything.
<sphalerite>
have you tried booting an older generation?
<jD91mZM2>
Not really :P
<sphalerite>
That might be worth doing :p
<sphalerite>
(I hope you haven't gc'd them)
<jD91mZM2>
What would I do even if it worked?
<sphalerite>
celebrate that some progress in finding the cause has been made xD
Piece_Maker has joined #nixos
Acou_Bass has quit [Ping timeout: 255 seconds]
Piece_Maker is now known as Acou_Bass
Synthetica has joined #nixos
<jD91mZM2>
sphalerite: How do I know which generation to boot into? I don't remember all the numbers!11
<sphalerite>
at that point you could look at the git log for nixpkgs between the two versions (because there won't havve been that many changes presumably) to see if there's anything obvious there, and diff the system generations themselves
<sphalerite>
jD91mZM2: the last one from yesterday maybe?
<jD91mZM2>
Well I mean this started happening from that one
<jD91mZM2>
first boot this morning
<sphalerite>
so the one before that!
Neo-- has joined #nixos
Ariakenom has quit [Ping timeout: 240 seconds]
<jD91mZM2>
And then bisect backwards one by one?
rosa has quit [Quit: rosa]
<sphalerite>
Well yeah actually bisecting would probably be more efficient :p
<jD91mZM2>
I'll just look at the git history for any big changes. But I doubt that any of that matters.
<sphalerite>
Actually maybe boot the oldest generation you have and see if the problem exists there
<sphalerite>
if it does, we know it's not a system-level change that broke it
andrewmiller1 has joined #nixos
<andrewmiller1>
halp
<andrewmiller1>
baloo is taking all resources
<andrewmiller1>
my fans are melting
<jD91mZM2>
andrewmiller1: Kill it, disable it, burn it alive
<andrewmiller1>
i've tried
<jD91mZM2>
pkill baloo?
<jD91mZM2>
sphalerite: Is that really necessary though? pls say no
<andrewmiller1>
you're my savior
<sphalerite>
ah, good ol' desktop search indexing tools. Baloo, Tracker, they're all horrendous
<andrewmiller1>
i love you
<sphalerite>
jD91mZM2: well if you want to find out the cause of the problem and get a chance at fixing it it'll be helpful :p
<jD91mZM2>
sphalerite: My first generation didn't even have a working VPN connection... I know it's weird, but I hate rebooting, and I think I'd rather stay with one monitor
<sphalerite>
wow
<andrewmiller1>
sphalerite: i'm willing to try anything you guys tell me. i have no idea what was different. my .nix-* files in my home folders, even in root, were deleted and i want to suspect baloo
<sphalerite>
o.O
<andrewmiller1>
but my computer is finally resting. it's been at max cores for over 24 hours
<andrewmiller1>
i thought i'd give baloo some time and space and went to bed with it running
<andrewmiller1>
my trust has been stepped on and hurt
<andrewmiller1>
baloo will never live on my computer again
<{^_^}>
→ 834b6049 by R. RyanTM: snd: 18.2 -> 18.3
<{^_^}>
→ 78b1ce84 by @Fuuzetsu: Merge pull request #40061 from r-ryantm/auto-update/snd
<jD91mZM2>
Is there anything wrong with using crontab on nix? Will any configs be uninstalled correctly if I ever uninstall it so my system doesn't get filled with crap?
<andrewmiller1>
services.cron.cronFiles is in man 5 config so
<andrewmiller1>
jD91mZM2: is that different from what you're wanting?
<jD91mZM2>
andrewmiller1: I was thinking if there's something wrong with simply using `crontab -e`
<jD91mZM2>
I don't want to commit my changes to my dotfiles repo, nor have them nixified
<kuri0>
what hashing function is used for the hashes in /nix/store ?
griff_ has joined #nixos
<andrewmiller1>
jD91mZM2: ah interesting. i don't know cron jobs. maybe someone who does can chime in
<Mic92_>
only works in nixos-ofborg for now as it seems
<Mic92_>
#nixos-borg
<infinisil>
It does occasionally exit for some reason, need to fix that
<andrewmiller1>
embarrassing, but i don't want to mislead any noobies searching one day. "warning: Nix search path entry '$HOME/.nix-defexpr/channels' does not exist, ignoring" only happens if you do use `sudo nixos-rebuild switch`. being root first, then running `nixos-rebuild switch` doesn't run into this message
<sphalerite>
I'm guessing it would be something like ssh-ng://user@host?root=/home/user/foo but I'm not sure what "root" should be
<clever>
sphalerite: i think Myrl-saki recently set that up
<srhb>
sphalerite: What do you mean? That looks right.
<srhb>
sphalerite: What do you mean "root" specifically
<sphalerite>
srhb: it says "unknown setting 'root'"
<srhb>
Oh. :|
<srhb>
Well, that's inconsistent...
<sphalerite>
I tried ?store=/home/user/foo as well since that's what you'd do locally, but that complains that the path isn't in the nix store
<sphalerite>
so I'm guessing that changes the store path rather than doing a chroot store
<zybell>
normally ssh doesnt support chroot over that what is configurable in /etc/passwd. With ng that may be diff.
raynold has quit [Quit: Connection closed for inactivity]
<sphalerite>
zybell: this is nix-specific functionality, nothing to do with ssh per se
<clever>
sphalerite: i worked with Myrl-saki a few days ago to create a fake nix-store bash script that would use nix-user-chroot to enter a chroot, and run the real nix-store
hiroshi has joined #nixos
<sphalerite>
clever: I don't need nix-user-chroot, just nix's built-in chroot stores
s33se has quit [Quit: s33se]
orivej has joined #nixos
<zybell>
then set them up in /etc/passwd or sshd doesnt see them
<sphalerite>
zybell: again it's got nothing to do with ssh, it's a nix featureb
<sphalerite>
nix is supposed to do the chrooting
shachaf has quit [Ping timeout: 255 seconds]
s33se has joined #nixos
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<zybell>
sphalerite:if you access sth over ssh you get sshd on the remote side. Whatever you want,sshd must allow it. If you login as user at the sshd,game over. If you login as root however,you can go fancy.
<{^_^}>
→ 9ee600a0 by R. RyanTM: insomnia: 5.16.1 -> 5.16.2
<{^_^}>
→ 3273f0d2 by @Mic92: Merge pull request #40102 from r-ryantm/auto-update/insomnia
<sphalerite>
freusque: the idea is that the .drv depends on the .drvs for all the dependencies, but the outputs of *those* won't be kept unless you have keep-outputs set.
<{^_^}>
→ 184cb1e8 by @markuskowa: soapysdr+plugins: init at 0.6.1
<{^_^}>
→ e0d33717 by @markuskowa: limesuite: init at 18.04.1
<{^_^}>
→ 7ba27541 by @markuskowa: welle-io: add soapysdr support
<freusque>
I'll try this now, thanks.
<sphalerite>
robstr: builtins.trace is the underlying thing, which prints its first arugment and returns its second one. traceVal from nixpkgs's lib just calls builtins.trace with the same argument twice, so it prints the same value it returns
<sphalerite>
freusque: also, you can limit the amount of "damage" the GC does while you're testing by passing --max-freed 1M or something so it won't delete all the stuff
<freusque>
ah superb thanks.
<viric>
Dezgeg: what is this thing in release-lib.nix? :) crossMaintainers = [ maintainers.viric ];
<{^_^}>
→ fcb6a0a5 by @Synthetica9: spotify: remove/update gnome2 references
NitroZ has joined #nixos
<{^_^}>
→ 2dbb29c3 by @Synthetica9: spotify: remove platform assertion
<{^_^}>
→ ddf90250 by @Mic92: Merge pull request #40116 from Synthetica9/spotify-nognome2
<robstr>
sphalerite: yea I need to fiend a better workflow / understanding of this. Watched Gabriels talk about haskell and nix. It's not magic often I struggle with basic stuff like `.env` I dont know sometimes what's inside all things :)
NitroZ has quit [Client Quit]
<Dezgeg>
I don't know if it affects anything
<sphalerite>
freusque: oh yeah, or you can use nix-store --gc --print-dead to have it not delete anything at all, just tell you what it would delete
<{^_^}>
[nixpkgs] @zimbatm closed pull request #9722 → stage-1-init: add support for the ability of specify a name of squashfs image → https://git.io/vZmJ9
<{^_^}>
→ 3b4ee14a by R. RyanTM: pspg: 1.1.0 -> 1.1.1
<{^_^}>
→ d43f1dda by @Mic92: Merge pull request #40070 from r-ryantm/auto-update/pspg
tmaekawa has quit [Client Quit]
<clever>
freusque: it has to be in nix.conf when you run the GC
<freusque>
aha
<freusque>
thanks clever, I will try that
endformationage has joined #nixos
jperras has quit [Quit: WeeChat 2.1]
jperras has joined #nixos
<freusque>
so that would be gc-keep-outputs we are talking about
genesis has joined #nixos
Neo-- has joined #nixos
<freusque>
aaand that works.
<freusque>
thanks a lot.
andrewmiller1 has joined #nixos
<jD91mZM2>
zybell: Sorry, I had to go. Isn't school great? /s Anyway, the question was: Does NixOS generally dislike contabs the same way it'd dislike editing /etc/lightdm/lightdm.conf manually?
<{^_^}>
→ d813ba2d by R. RyanTM: libx86emu: 1.12 -> 1.14
<{^_^}>
→ ff13b781 by @Mic92: Merge pull request #40097 from r-ryantm/auto-update/libx86emu
selaux has quit [Remote host closed the connection]
infinisil_ has quit [Quit: Configuring ZNC, sorry for the join/quits!]
<jD91mZM2>
spacefrogg: I just wanted something quick and dirty for a stupid script I'm forced to run but I really don't care about
winem_ has joined #nixos
rain1 has joined #nixos
infinisil has joined #nixos
<jD91mZM2>
spacefrogg: That's why I liked cron. I set it up using one line of code and then I forget that stupid thing ever existed
<etu>
yeah, that's what's bad about cron :D
<etu>
You don't know if things runs as you want unless you have something around it reporting it to you
<etu>
systemd timers are nice
<spacefrogg>
jD91mZM2: systemd infrastructure is supported in nixos which is why it is a two-liner in /etc/nixos/configuration.nix, as well.
xcmw has joined #nixos
<jD91mZM2>
spacefrogg: Yeah but then I need to be reminded that that stupid thing exists every time I open my configuration.nix
<andrewmiller1>
am i crazy or is node.js 9.x taboo or something? i can't find any mention about it in the docs and it's almost like it was avoided on purpose without saying why
<jD91mZM2>
spacefrogg: also it'd be commited to git since I keep my dotfiles there, and it's a private thing
<andrewmiller1>
trying to add a package from npm
<spacefrogg>
jD91mZM2: The concept of Nix/NixOS is to not have implicit state. If you want that, NixOS is probably not what you seek.
<jD91mZM2>
spacefrogg: I mean I want the declarative state for everything else. I just don't want it for this thing because I hate it and don't want to be reminded that it exists.
<jD91mZM2>
But I got permission to not run the script every day for now, so that means I could uninstall cron and get rid of the crap. So all is well. yay
<spacefrogg>
jD91mZM2: Social comment. :)
johann__ has quit [Quit: Leaving.]
<Myrl-saki>
clever: sphalerite: Pong. From what I understand, what clever is saying is to use my wrapper script to SSH a chroot session. It requires hacking both the SSH config and a custom "profile" file.
<jD91mZM2>
spacefrogg: What do you mean? Sorry if I'm sounding agressive, I'm just a little annoyed by my school. Basically, they (used to) force me to send an email every single day. So I automated this task. But I still hate it with passion :P
<infinisil>
Huh, what's that for jD91mZM2 ?
ryantrinkle has quit [Ping timeout: 248 seconds]
<spacefrogg>
jD91mZM2: Nothing, I figured something like that. :)
<jD91mZM2>
infinisil: I have no idea. I only have to go to school a few times a week, but the rest of the week they want me to be reminded that it still exists or something. It's completely stupid :P
<aminechikhaoui>
openssh -> openssl
<infinisil>
That sounds odd
iqubic`` has left #nixos ["ERC (IRC client for Emacs 25.3.1)"]
<jD91mZM2>
infinisil: That's because it is. Welcome to the world of stupid concepts and idiot practices where I live :^)
<infinisil>
Weee \o/
iqubic has joined #nixos
<iqubic>
is it possible to have a systemd service that runs after internet connections have started, but before display-manger.service?
<infinisil>
iqubic: I think just `before = [ "display-manager.service" ]` and `after = [ "network.target" ]` should work
wchresta_ has joined #nixos
<iqubic>
But is that a time that exists?
xAFFE has left #nixos ["Error from remote client"]
<{^_^}>
→ 5fada46e by @jD91mZM2: powerline-rs: add bash & fish completion
<{^_^}>
→ 4893356b by @joachifm: Merge pull request #40033 from jD91mZM2/powerline-rs
xAFFE has joined #nixos
<iqubic>
Does "network.target" run after "display-manager.service"?
phreedom has joined #nixos
<jD91mZM2>
Is there a good way to list files in root that nix doesn't own? Something like "oh no you created /var/cron, you shouldn't have done that"?
silver has joined #nixos
<jD91mZM2>
iqubic: I don't think so, since my VPN starts up long before X. And I guess internet has to be set up waay before that as well :P
xAFFE has left #nixos [#nixos]
xAFFE has joined #nixos
<iqubic>
jD91mZM2: Cool. Thanks for the info.
coot has quit [Read error: No route to host]
coot_ has joined #nixos
<infinisil>
iqubic: What do you mean by "a time that exists"?
logzet has joined #nixos
<{^_^}>
[nixops] @AmineChikhaoui pushed to master « seems there was a typo in 23a039a0, we should make openssh available »: https://git.io/vpoXM
<bebarker>
Maybe this isn't possible due to it being impure, but is there a way to reference an environment variable from a nix expression like: with import "${HOME}/workspace/nixpkgs" { }; ? I get error: undefined variable 'HOME' at /nixenv/brandon/scala-default.nix:2:16
<bebarker>
I should probably just pass it in as an argument
<infinisil>
builtins.getEnv "HOME"
<iqubic>
infinisil: Well, if display-manager.service runs first, then it would be impossible to have a service that runs after network.target and before display-manager.service
<infinisil>
iqubic: Yeah, the after and before only applies if they happen to start at the same point
<iqubic>
Oh. So what I want to do is say that my service requires network.target
<spacefrogg>
jD91mZM2: There's a file /etc/.clean that holds files copied to/created in /etc. And there are soft-links pointing to /etc/static. The difference between those and all files in /etc should be the list of files not under control of nix. No tooling available to my knowledge.
<iqubic>
Is that possible to do?
<infinisil>
With services I think you'd use `requires = [ "foo.service" ]`, but network.target is a bit special probably
<spacefrogg>
iqubic: better use wants
<spacefrogg>
iqubic: and after
<jD91mZM2>
spacefrogg: Interesting, thanks
<iqubic>
what is the difference between requires and wants?
<infinisil>
iqubic: man systemd.unit
<spacefrogg>
iqubic: No. requires makes your service fail if your requirements are not met. wants is weaker. Neither of them order the services. Use before/after to establish an order
<iqubic>
Alright then. How do I set "wants" for my service?
wchresta_ has quit [Ping timeout: 264 seconds]
<domenkozar>
aminechikhaoui: uhh how the hell that happened :)
<jD91mZM2>
spacefrogg: `man configuration.nix` says "Start the specified units when this unit is started.". Seems like `wants` does actually order services, no?
<domenkozar>
aminechikhaoui: well I know how :(
<jD91mZM2>
spacefrogg: Wait nvm I misread your message
<{^_^}>
→ faa0d2fa by @Ekleog: todiff: 0.4.0 -> 0.5.0
<jD91mZM2>
--arg is for non-string arguments, like integer
<{^_^}>
→ e7559860 by @xeji: Merge pull request #40045 from Ekleog/todiff-0.5.0
<steveeJ>
WilliamHamilton, jD91mZM2: it's for providing nix expressions, they can still evaluate to a string
nick_l has quit [Ping timeout: 260 seconds]
<steveeJ>
the point is, they are evaluated
fare__ has quit [Ping timeout: 276 seconds]
<WilliamHamilton>
steveeJ: but what if I said `nix-shell --arg compiler "ghc802"`. Isn't "ghc802" a nix expression, which is evaluated to itself?
<steveeJ>
WilliamHamilton: no, in that case the enclosing " would not make it to the evaluator, and it would try to find a variable with the name ghc802
dbe has joined #nixos
__Sander__ has quit [Quit: Konversation terminated!]
<WilliamHamilton>
steveeJ: I see, it's clear now, thanks!
<WilliamHamilton>
and then try to use those packages from a nix-shell, I get an error along the lines of `error: attribute 'packageOverrides' missing`
<WilliamHamilton>
I don't quite understand where packageOverrides comes from
<jD91mZM2>
nix-shell doesn't really run packages very well. you might wanna create a shell.nix file that runs `(import <nixpkgs> {}).callPackage ./path/to/package.nix`
<srk>
is there a way to install riscv gcc (riscv32-unknown-elf-gcc)?
primeos has quit [Ping timeout: 260 seconds]
hotfuzz_ is now known as hotfuzz
<jtojnar>
oh, cool, the first character in the base32-encoded sha256 hashes encodes exactly one bit
<WilliamHamilton>
jtojnar: would that explain why I can do that with haskellPackages instead of haskell.packages.ghc802?
wchresta_ has quit [Ping timeout: 264 seconds]
<ryantm>
jtojnar: This past weekened I made progress on moving the error control flow of nixpkgs-update away from shelly's exception throwing into Either
<jtojnar>
yes, haskellPackages is a single attribute
<WilliamHamilton>
jtojnar: I *think* I see the issue; what would be a workaraound for this?
<WilliamHamilton>
jtojnar: no, thinking again, I don't see the issue and I'd appreciate if you could explain it to me
phreedom has quit [Ping timeout: 255 seconds]
<jtojnar>
WilliamHamilton: haskell.packages is an attribute set, it contains packageOverrides attribute among others
jD91mZM2 has quit [Ping timeout: 240 seconds]
<jtojnar>
you are replacing the attribute set with an attribute set that only contains ghc802 attribute
<{^_^}>
→ ce9c8bee by @xeji: Merge pull request #40032 from yrashk/sit-0.3.2
troydm has quit [Quit: What is Hope? That all of your wishes and all of your dreams come true? To turn back time because things were not supposed to happen like that (C) Rau Le Creuset]
<jtojnar>
jD91mZM2: it is pretty reasonable, realizing that `{ foo.bar.baz = ...; }` is a syntax sugar for `{ foo = { bar = { baz = ...; }; }; }`
hakujin has joined #nixos
<jtojnar>
but it can be counterintuitive, which is why for example Elm language does not have this, even when it is frequently requested
vaninwagen has quit [Ping timeout: 268 seconds]
<jtojnar>
ryantm: cool
<jtojnar>
ryantm: I probably will not get to the changelogs until summer :-/
<ryantm>
jtojnar: The reason I'm working on this flow control stuff is because without it, it's hard to extract parts of updatePackage into their own functions. updatePackage got a lot smaller :)
<jtojnar>
ryantm: I understand, in the end I would like to get rid of Shelly completely
<{^_^}>
→ 046e374c by @flokli: virtualisation/qemu-vm.nix: use lib.optionalString
<{^_^}>
→ 36f4b7f6 by @flokli: virtualisation/qemu-vm.nix: remove ttys
<{^_^}>
→ 70c57fe3 by @flokli: qemu: fix spaces in postInstall
Ariakenom has quit [Ping timeout: 260 seconds]
coot has quit [Quit: coot]
<sphalerite>
So I'm having some odd issues where any realisations inconsistently fail with error: writing to file: Broken pipe
<sphalerite>
I think it's caused by an ssh-ng: substituter I have configured, but I don't really know how to debug it. I've been running strace on it but haven't been able to work out where the fd that the write fails on comes from
<sphalerite>
does nix have something that will show a C++ backtrace on errors?
<ryantm>
sphalerite: I don't actually know the answer to that, but have you tried GDB?
<Dezgeg>
stick abort() there and look in coredumpctl?
<jD91mZM2>
sphalerite: For a backtrace you can run it in GDB and type `bt`, no?
<Dezgeg>
maybe with a clever enough conditional breakpoint you can avoid changing the source
jrolfs has joined #nixos
nuncanada has joined #nixos
<sphalerite>
ryantm: jD91mZM2: Dezgeg: I'd rather not go to the fuss of rebuilding nix, this isn't the most powerful of machines :/
<sphalerite>
maybe I could use a catchpoint to find the failing write syscall
<jtojnar>
ufff, #botbot is full of  characters, which telepathy logger stores but when it tries to load it, it crashes
<sphalerite>
oh yeah, and debugging nix with gdb is a pain because of all the forking
<jD91mZM2>
sphalerite: Why would you need to rebuild nix? You mean in order to install gdb?
<jtojnar>
s/botbot/bottest
<sphalerite>
jD91mZM2: in order to get it to actually stop on the bit where the error is occurring
ihar has joined #nixos
<jD91mZM2>
sphalerite: I'm confused but oook
jrolfs has quit [Ping timeout: 240 seconds]
<tmplt>
Is it possible to check how much space a derivation and all its dependencies take up?
xcmw has joined #nixos
fare__ has joined #nixos
<sphalerite>
tmplt: du -chd0 $(nix-store -qR /nix/store/hash-foo)
mtjmullen has joined #nixos
<sphalerite>
jD91mZM2: it's not a segfault or anything so I need to actually set a breakpoint
<sphalerite>
and I don't know where
lonokhov has quit [Quit: WeeChat 2.0]
<tmplt>
sphalerite: thanks
<sphalerite>
tmplt: oh actually with 2.0 there's nix path-info -S /nix/store/hash-foo as well
<sphalerite>
Which should be faster, and it's a bit less of an eyesore :p
waleee has joined #nixos
Neo-- has quit [Ping timeout: 240 seconds]
jrolfs has joined #nixos
Sonarpulse has joined #nixos
asuryawanshi has joined #nixos
<waleee>
I have stumled on an extremly weird behavior when installing the ion-shell with nix-env. When upgrading all nix-env installed packages, ion is "upgraded" to the ancient ion3 window-manager
johann__ has joined #nixos
<sphalerite>
waleee: basically, nix-env -u is stupid
<infinisil>
jtojnar: I'll look into stripping the escape codes
<leary`>
Ooh. The SHA for a fetch in one of my overlays starts with 1234. I feel special.
<{^_^}>
[nixpkgs] @xeji opened pull request #40131 → nixos/release-combined.nix: make test set consistent with release-small.nix → https://git.io/vpoNj
<sphalerite>
waleee: nix-env -u works like nix-env -i without -A, so you're probably best off using an overlay like the one linked in the FAQ above
<sphalerite>
leary`: pff. Tell me when it starts with 31333337 :p
<jtojnar>
infinisil: that was the worst thing about the spam attacks, they used escape codes a lot
asuryawanshi has quit [Remote host closed the connection]
<infinisil>
> 16*16*16*16 # leary`: chance of this happening:
<{^_^}>
[31;1merror:[0m syntax error, unexpected $end, expecting ')', at [1m(string)[0m:10:21
<infinisil>
Oh no, sry jtojnar
tylerjl has quit [Remote host closed the connection]
asuryawanshi has joined #nixos
leothrix has joined #nixos
<waleee>
sphalerite: Ok. The other question would be why ion3 still is in nixpkgs, the developer abandonded it in 2009
<infinisil>
jtojnar: string handling still an unsolved problem in 2018 xD
<sphalerite>
waleee: nixpkgs doesn't tend to trim much fat :p if you want to make a PR removing it it would probably be accepted though
civodul has quit [Quit: ERC (IRC client for Emacs 25.3.1)]
obadz- has joined #nixos
<jtojnar>
infinisil: this would not happen with proper error handling (sum types)
<jtojnar>
or smarter compilers
<sphalerite>
naturally, when I'm trying to reproduce the error in gdb it works fine
<infinisil>
Everybody should just use haskell or other languages with good type systems
<jD91mZM2>
infinisil: Rust! Rust! Rust!
<infinisil>
Yes!
obadz has quit [Ping timeout: 256 seconds]
obadz- is now known as obadz
<infinisil>
But Haskell is on whole nother level regarding the type system than Rust
<jD91mZM2>
HOW DARE YOU
<jD91mZM2>
Rust's type system is 110% perfect and can't get any better
<sphalerite>
infinisil: but Haskell doesn't have lifetime tracking!
<Synthetica>
capisce: But it should have no monads and just be a C clone
<jD91mZM2>
infinisil: You can still do pretty much everything you need with Rust's existing type system
<jD91mZM2>
Just look at iterators - lazy loaded
<jD91mZM2>
no monads or anything there
leothrix has quit [Ping timeout: 260 seconds]
<infinisil>
jD91mZM2: well yeah, some instances of monads are there in many languages: lists, Optional values, Result style error handling, etc.
<infinisil>
But the monad concept can't be generalized
<jD91mZM2>
sphalerite: OMG multiple monitors work if I use the xorg config!
dbmikus has quit [Quit: WeeChat 2.1]
<taktoa>
higher-kinded polymorphism is really useful and obvious, idk why people don't put it in their languages
dbmikus has joined #nixos
<sphalerite>
taktoa: because it's scary and hard to implement, I'm guessing?
<jD91mZM2>
sphalerite: which is really weird because I can't see any new files inside /etc/X11/xorg.conf.d
<taktoa>
if you already have a hindley-milner style type system, it seems pretty easy to implement to me
<sphalerite>
jD91mZM2: the X server is called directly with the generated config as an argument
leat has joined #nixos
<sphalerite>
jD91mZM2: pgrep -a X
<sphalerite>
you'll see the path to the config on its cmdline
<infinisil>
taktoa: why does then only haskell have it? There's lots of languages using hindley-milner
<ghostyy>
infinisil, what kidn of trouble did you have with implementing monads?
<ghostyy>
dont you just need functions?
smichel has joined #nixos
<taktoa>
like idk maybe this is my haskell experience showing but imo it's not some ivory tower shit, it's something you can easily note if you are writing java or whatever
<infinisil>
ghostyy: I can't remember exactly, but try to convert Monad to it and I'm sure you'll have something that doesn't work
<ghostyy>
ah alright then
<Cale>
taktoa: It's not quite as simple as you might imagine to really get it right and not just require type annotations everywhere.
<taktoa>
the ability to be polymorphic over type constructors in addition to fully saturated types is just a basic aspect of system f IIRC
<Cale>
Oh, wait, higher *kinded* polymorphism
<taktoa>
yeah
<Cale>
I thought you said higher rank polymorphism, my bad
<taktoa>
higher rank I can understand not wanting your language
<Cale>
Yeah, there's no excuse.
<jD91mZM2>
sphalerite: Ok so xrandr still doesn't work but changing the Xorg config does lol what is this
fogbugz123[m] has joined #nixos
<taktoa>
like I was kind of shocked when I heard that ocaml doesn't have it
<infinisil>
Whoa really
jrolfs has quit [Ping timeout: 268 seconds]
<taktoa>
they have module functors which you can use to kind of hack it in but otherwise yeah, no HKT
<infinisil>
You mean similarly how Go has hacked in generics for basic types but otherwise doesn't? Hah
tylerjl is now known as leothrix
<maurer>
If you want to see the OCaml way of doing HKT, go look at Jane Street's "Core" library
<maurer>
The module signatures they have there are how they do it
<maurer>
Also, it was explicitly admitted at one point htat the reason rust didn't have HKT in its initial design was to avoid monads
<maurer>
because they didn't want the haskell community to filter in and start yelling about category theory to all the systems programmers
<taktoa>
yeah I was going to point that out
<taktoa>
seems like a social thing mostly
<fogbugz123[m]>
Hi, i'm migrating to nixos and i've hit a little roadblock. StumpWM is marked as broken in the repos
<maurer>
and by now they've dug themselves into a bit of a hole where they can't seem to figure out how to put HKT in _now_ given their other decision
<maurer>
It'll show up eventually, but it'll take some time
<fogbugz123[m]>
but there's no open issue or anything
periklis has quit [Ping timeout: 256 seconds]
<capisce>
the JavaScript folks also ignored the category theory advice about the Futures API
<capisce>
Go is an abomination imo
<goibhniu>
hi fogbugz123[m], you can use nixpkgs.config.allowBroken = true; to see what breaks
<infinisil>
maurer: interesting, but that seems like an odd reason to not add a feature which clearly would've been appreciated eventually
smichel has quit [Quit: smichel]
owi has joined #nixos
<infinisil>
A systems language competitor to C with higher-kinded types, imagine that
<capisce>
maurer: that's why I kind of doubt the Go strategy to "start small" and add on features later
<capisce>
some features are not easily bolted on after the fact
katona has joined #nixos
<capisce>
but should be considered as part of the entire design
<taktoa>
I think the right way to do it would have been to write the type system in a way that makes adding HKT easy, but at the same time obfuscate it such that only the people who wrote rustc could figure out exactly _what_ to change
<taktoa>
:^)
<sphalerite>
ugh, the USB port on my ereader is broken. So now I can't charge it.
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<infinisil>
(Maybe this typing conversation would be better fit for #nixos-chat)
<taktoa>
agreed
<maurer>
taktoa: So, the original rust that HKT would have been "easy" on also had less flexible lifetimes and optional garbage collected pointers
<maurer>
taktoa: the thing that makes HKT nontrivial is how it interacts with the lifetime calculus
<taktoa>
maurer, let's take this to #nixos-chat
<sphalerite>
fogbugz123[m]: this is what happens when nobody's maintaining the package, its build breaks, and a release is coming so we need "zero hydra failures" (i.e. all the packages not marked as broken building successfully)
<sphalerite>
fogbugz123[m]: You can try setting NIXPKGS_ALLOW_UNFREE=1 and building it to see what's actually wrong with it
<owi>
Not sure if this is the best channel to ask, but I'll try. I'm running NixOS 18.03 and using it primarily for Haskell development. Generally I'd like to use GHC 8.4.2. I find GHC 8.4.1 in my nix-env, but not the latest version. Also, when I use GHC 8.4.1 it seems to build from source. Are there options for using recent versions of GHC that are available in caches, without upgrading the entire NixOS version?
<owi>
Thanks for any tips!
<Cale>
Sonarpulse: ^^
<sphalerite>
owi: you can use the unstable channel for your user env
<taktoa>
yeah, it took me a while to realize you could just have multiple channels at the same time
<owi>
sphalerite: OK. I found docs for using channels for nix on other operating systems, but haven't seen anything for NixOS. Do you have any links?
<sphalerite>
owi: basically the same applies as on non-nixos
<owi>
I'll try that.
<jD91mZM2>
sphalerite: Ok so fun fact: Selecting smaller screen sizes work even using xrandr. It's like xorg only messes up if my total screen canvas thing expands
<Robbiedobbie>
Is it possible to host your own binary cache without using nix-serve?
<{^_^}>
[nixpkgs] @rycee pushed to master « lib.types: fix loaOf behavior for long lists »: https://git.io/vpKUF
<{^_^}>
[nixpkgs] @rycee closed pull request #37734 → lib.types: fix loaOf behavior of long lists and many merges → https://git.io/vxBxW
<Robbiedobbie>
I want to host a binary cache for a custom channel that I will use on multiple computers, however, the only suitable place for me to host it on will only be able to serve a directory with files through http
<Robbiedobbie>
So I was wondering if, and if so, how, I can host a custom binary cache there
<sphalerite>
Robbiedobbie: yes
<sphalerite>
Robbiedobbie: you can create a flat-file binary cache using nix copy --to file:///some/path, and just serve that with a regular HTTP server
<Robbiedobbie>
Ah ok, from the documentation I wasn't sure if the copy would work just fine on a normal directory
<sphalerite>
the file:// is critical for making it create a binary cache rather than a chroot store
<jD91mZM2>
Robbiedobbie: Seems like all nix-serve does behind the schenes is give you files from /nix/store. So you can probably just serve that directory somehow
<sphalerite>
jD91mZM2: not really, since it needs to serialise them to nar files and narinfos
<jD91mZM2>
sphalerite: Ah
<Robbiedobbie>
Thanks for the help. I will try it out
<jD91mZM2>
sphalerite: ooo `nix copy` supports uploading binary cache to s3 built in? Wow that's cool
<sphalerite>
jD91mZM2: yep, it's what's used for hydra AFAIK
MichaelRaskin has joined #nixos
<elvishjerricco>
jD91mZM2: Yea it's sweet because it provides a way to have a secret cache without crappy HTTP auth. Upload to S3 and your team can just specify it as a substituter if they have AWS credentials configured on their machine.
<jD91mZM2>
Wait why aren't we using `nix build` instead of nix-build more often? A-a-a-and `nix run`. omg the `nix` command is amazing
<elvishjerricco>
jD91mZM2: I think we just need `nix shell` and `nix install` and ill completely ditch the old commands
<{^_^}>
→ c7577fa1 by @Mic92: mbedtls: remove darwin patch
<{^_^}>
→ 08db5514 by @Mic92: Merge pull request #40137 from Mic92/mbedtls
<elvishjerricco>
The only gripe I have is that I constantly have to `-f .`
<jD91mZM2>
Are channels completely deleted with `nix`? Is it using NIX_PATH now?
<sphalerite>
elvishjerricco: I just had an idea for a UI that could work for that — what would you think of doing `nix build .foo` to build the foo attr of default.nix?
<sphalerite>
jD91mZM2: it is using NIX_PATH, but that doesn't mean channels are gone
jtojnar has quit [Quit: jtojnar]
<jD91mZM2>
I mean I can't use `nixos.<whatever>`, but I can use `nixpkgs.<whatever>`
<elvishjerricco>
sphalerite: Eh. That seems to stomp on the whole "consistency" goal
<sphalerite>
jD91mZM2: ~/.nix-defexpr isn't really superseded by it yet because there's no nix-env anyway, but that's the idea
<sphalerite>
elvishjerricco: I don't see why — it would work for all the other tools as well
<sphalerite>
elvishjerricco: and since it's not a valid attrpath standalone, it doesn't conflict with any of the other ways of specifying an installable I think?
<elvishjerricco>
sphalerite: But it's not an attr path. It's a weird kind of path that's inconsistent with the Nix language
<sphalerite>
elvishjerricco: nor is /nix/store/hash-foo.drv :p
<elvishjerricco>
sphalerite: I'd be ok with a sufficiently convenient path that joins the default top level paths representing pwd
<elvishjerricco>
And I know I can do that manually by adding . to `NIX_PATH`, but I think this ought to be a built in behavior so that scripts can rely on it
<sphalerite>
it still doesn't make building ./release.nix convenient though :/
<elvishjerricco>
sphalerite: Oh well that's always going to require specification. `default.nix` is the concern for m
<elvishjerricco>
Ah! I got it!
<elvishjerricco>
`nix build default.foo`
<elvishjerricco>
Make a top level attr `default` that's just `./default.nix`
wchresta_ has joined #nixos
jtojnar has joined #nixos
<sphalerite>
well nix build doesn't (yet?) auto-import and auto-call
<cheshircat_>
hey, I'm having trouble packaging something that calls git in its build script
<cheshircat_>
is there a way to work around this?
<elvishjerricco>
sphalerite: Right, I meant `default.foo` would be the same as `-f default.nix foo`
<sphalerite>
cheshircat_: because it's trying to access the network?
<jD91mZM2>
cheshircat_: What exact git command are you running?
<sphalerite>
cheshircat_: put whatever in place using a fetchgit (or fetchFromGitHub or whatever) derivation before it runs, and patch out the command from the script
<jD91mZM2>
sphalerite: fetchFromGitHub doesn't preserve .git. Not sure if fetchgit does either to be honest
<sphalerite>
jD91mZM2: it doesn't by default, but you can tell it to.
<jD91mZM2>
oh
wchresta_ has quit [Ping timeout: 240 seconds]
<cheshircat_>
yea, it's trying to access the network
<sphalerite>
jD91mZM2: but if it's a clone, it's not entirely unlikely that the script doesn't *actually* need .git
<jD91mZM2>
Yeah and it can save time using GitHub's .tar.gz files
<cheshircat_>
How do you fetchgit into the source of another fetchgit?
<jD91mZM2>
I was using a custom modeline, but now it seems to work fine. Lemme try
<sphalerite>
oh yeah, it only puts RightOf in if previous is unset for the monitor
vaninwagen has joined #nixos
<jD91mZM2>
Seems like that wasn't the cause of my issue though. Phew
jrolfs has joined #nixos
<jD91mZM2>
... aaannndd FINALLY
<jD91mZM2>
MY MONITORS WORK AGAIN
<jD91mZM2>
Only took the whole day :^)
<sphalerite>
>_<
<{^_^}>
[31;1merror:[0m syntax error, unexpected ')', at [1m(string)[0m:10:11
<sphalerite>
lol
<sphalerite>
I'm glad nix is convincing enough for you to deal with all this bullshit :p
<jD91mZM2>
I mean I would go back to arch but I don't know if I would manage to install it a second time /s
<sphalerite>
hahaha
<jD91mZM2>
If I want to move my services.xserver.xrandrHeads to a different file, how would you recommend I do it? Using the `imports` array (which would cause a "duplicate key" thing on services.xserver which I would somehow need to fix), or simply set `xrandrHeads = (import ...)`?
<sphalerite>
imports would work, there's magic that merges all the modules together
<jD91mZM2>
IIRC I tried that before. The magic only works on completely separate modules.... I think. Not sure.
<sphalerite>
in fact, your configuration.nix and, for example, nixos/modules/services/x11/xserver.nix, are conceptually the same thing
<jD91mZM2>
Ok lemme try
<jD91mZM2>
You're right. I remember incorrectly
<jD91mZM2>
Thanks for everything sphalerite! I couldn't have done this without you - you're a great person. Good night :)
<Ralith>
dtz: you mean pkgs/test? also there totally was a clang test, I just missed it
<Ralith>
still don't see any libc++ stuff anywhere
tester123 has joined #nixos
<LnL>
Ralith: the cc-wrapper tests run with all stdenv variants
<tester123>
I currently have `texlive.combined.scheme-full` in my nix config. But I'm trying to add the tex package "xcolor" to my system. How do I do this?
<sphalerite>
[3/365/42431 copied (1159.4/4429.4 MiB)] copying path '/nix/store/2vl7hxwf9q0axx61n9pgd33g6hvzxrz9-iana-etc-20180108' from 'ssh-ng://192.168.99.14' to 'file://cache/' … this might take a while
<dtz>
(sorry that link is against master so won't last forever but hopefully works for now)
<dtz>
(grumble grumble github needs a $#@@# permalink option)
<oh_lawd>
that I guess should enable thinkpad battery charge thresholds
<oh_lawd>
the default tlp doesn't support them without a module
<sphalerite>
srk: I'm making a binary cache and being a bit indiscriminate about what I'm including :p nix copy --from ssh-ng://192.168.99.14 --all --to file://cache/
<sphalerite>
error: cannot open connection to remote store 'ssh-ng://nobody@192.168.99.13': serialised integer 7161674624452356180 is too large for type 'j'
<sphalerite>
note to self: don't try to use the nobody user
tester123 has quit [Quit: Page closed]
<hodapp>
o_O
<hodapp>
that... is a strange error
<sphalerite>
yep. Only occurs if I try to use nobody though, I'm guessing it's nobody's uid
<sphalerite>
hm no, it's 65534
flop has joined #nixos
jtojnar has quit [Quit: jtojnar]
<Dezgeg>
I think it's some part of an error message in ascii
muzzy has joined #nixos
muzzy_ has joined #nixos
muzzy__ has joined #nixos
<{^_^}>
[nixpkgs] @7c6f434c merged pull request #40141 → iml: use openblas instead of atlas → https://git.io/vpKsc
<{^_^}>
→ b0d70d33 by Timo Kaufmann: iml: use openblas instead of atlas
<{^_^}>
→ b1f064c9 by @7c6f434c: Merge pull request #40141 from timokau/iml-openblas
jtojnar has joined #nixos
orivej has quit [Remote host closed the connection]
<flop>
Hey, guys! If I want to update existing nixpkg to the newer version, should I mark myself as a maintainer?
<sphalerite>
flop: you don't have to, but it's appreciated
<flop>
thanks!
<sphalerite>
flop: and what it means (not sure if it's actually formally defined anywhere, but practically) it means you'll get emails from hydra if it breaks, and might get pinged on github if someone has an issue with the package
<simpson>
So many emails.
<sphalerite>
simpson: I haven't got one in ages, you clearly maintain too many packages :p
<simpson>
It's actually not that bad, but sometimes if you maintain a package that has an oft-broken dependency, then that can be irritating.
<Dezgeg>
I think the emails were disabled at some point
<simpson>
sphalerite: Some leaves of the Python tree, for example, do this.
<Synthetica>
Maybe I should've asked here instead of in #nixos-devel
<Synthetica>
*dev
<dtz>
okay! Just wanted to stay on top of this, I have some failures/etc to report but don't really have a good place to report them--and definitely don't want to double-report things that are known
xy2_ has quit [Read error: Connection reset by peer]
xy2_ has joined #nixos
<sphalerit>
gchristensen: could you reset the topic? The matrix bridge seems to have been down when you last did it, so the old logs link is still shown on matrix
hamishmack has joined #nixos
<MichaelRaskin>
\\\
<ij>
Will services.openvpn restart on a network change?
dbmikus__ has quit [Ping timeout: 276 seconds]
dbmikus__ has joined #nixos
rosa has joined #nixos
rycee has quit [*.net *.split]
WilliamHamilton has quit [*.net *.split]
suvash_away has quit [*.net *.split]
knupfer has quit [Remote host closed the connection]
rosa has quit [Client Quit]
<elvishjerricco>
Has anyone tried to port the NixOS systemd configuration to other systemd linuxes? Would be sweet to configure systemd stuff with Nix on Ubuntu. Plus you could probably import a lot of the NixOS modules as is
<Ralith>
does `nix build` have a way to process an expression?
<Ralith>
on the command line, that is
kier has quit [Ping timeout: 248 seconds]
barrucad1 is now known as barrucadu
<Synthetica>
Oh, dtz: I'd just file an issue I guess; I'd say it's better to have it over-reported then under-reported
olto has quit [Ping timeout: 240 seconds]
rycee has joined #nixos
jrolfs_ has quit [Ping timeout: 260 seconds]
WilliamHamilton has joined #nixos
dbmikus__ has quit [Ping timeout: 248 seconds]
asuryawanshi has quit [Ping timeout: 268 seconds]
szicari has quit [Quit: szicari]
kier has joined #nixos
<elvishjerricco>
Ralith: The `INSTALLABLE` syntax will recognize parenthesized Nix code
<elvishjerricco>
E.g. `nix build '(with import <nixpkgs> {}; hello)'
blankhart has quit [Ping timeout: 256 seconds]
kisik21 has joined #nixos
<Ralith>
ah, right, I remember now
<Ralith>
thanks
<Ralith>
hmm
<kisik21>
Hi. What's wrong with this line in environment packages list?
<kisik21>
(python3.withPackages (python-packages: with python-packages: [ pillow ];)
<Ralith>
elvishjerricco: I have a script that uses `nix-build -E expr -A foo -A bar`; any idea how to do that in nix 2 without repeating `expr`?
<Ralith>
iow, I want to get multiple outputs from a multi-output derivation
<Ralith>
`nix-build -f . baz.foo baz.bar` is *almost* what I want, but I want to apply an override to baz first
lord| has joined #nixos
<{^_^}>
[nixpkgs] @7c6f434c pushed to master « love_11: init at 11.1 »: https://git.io/vpKEc
Piece_Maker has joined #nixos
<MichaelRaskin>
elvishjerricco: Note that you can even export a NixOS service as a runner script
<Ralith>
er, sorry, `nix build -f . baz.foo baz.bar`
Acou_Bass has quit [Ping timeout: 268 seconds]
Piece_Maker is now known as Acou_Bass
<MichaelRaskin>
And surely you can export the unit file
hotfuzz has quit [Ping timeout: 240 seconds]
<kisik21>
Seems like Python 3 environment derivation is independent from Python itself, so one could include python3 and python3.withPackages in the same configuration. Nice.
hiratara has joined #nixos
hotfuzz has joined #nixos
<kisik21>
Or not so independent. Who knows.
<kisik21>
Nope. Collision.
<srhb>
kisik21: Hopefully you wouldn't need python3 itself when you're using withPackages?
aarvar has joined #nixos
<kisik21>
I just discovered that these are conflicting, and interestingly, the less featured one (bare python3) ended up in /run/current-system/sw/.
<kisik21>
srhb: how to suppress installing python3 when withPackages option is used?
<srhb>
kisik21: For that, you should probably be getting packages from python3Packages instead.
<srhb>
But I'm not sure what you're doing.
<srhb>
The purpose of withPackages is to bundle up packages with the interpreter after all.
<kisik21>
I want to have Python 3 with Pillow installed in site-wide config, and for common use-cases have a bare interpreter.
<srhb>
I don't see how you'd avoid having python3 conflict with python3 in either case.
<kisik21>
Could I install python3packages.pillow in environment.systemPackages and be sure it will be used by python3?
<srhb>
I recommend using a nix-shell instead of installing those. You can always add a gcroot.
dbmikus__ has joined #nixos
<kisik21>
wait... hm, I now think of nix-shell like a virtualenv for non-python packages, including python interpreter itself. Am I right?
<simpson>
Sure.
<srhb>
kisik21: or python.withPackages
<srhb>
kisik21: It'll do basically anything.
<kisik21>
ok, as a workaround, I disabled bare python3 in common.nix file included from site-wide configuration.nix, now python3 needs to be installed in configuration.nix...
doyougnu has quit [Ping timeout: 256 seconds]
wchresta_ has quit [Ping timeout: 256 seconds]
<kisik21>
Would it be correct to install (python3.withPackages (python-packages: with python-packages; [])) for convenient editing?
suvash_away has joined #nixos
johnhamelink has joined #nixos
<simpson>
Aw, the `trickle` utility doesn't work on Nix-patched executables. That's too bad. Are there any good alternatives for per-process traffic shaping?
dbmikus__ has quit [Ping timeout: 256 seconds]
<srhb>
kisik21: I wouldn't. I don't know whether it's correct or not (I don't use python much)
<srhb>
kisik21: But if you're doing that, I think that's the same as just having a bare python3.
<simpson>
kisik21: Well, don't worry about 'correct'; if it builds and runs, then it works for you. But I would *not* think of 'installing' any longer. Think instead of temporary environments.
<srhb>
^
<simpson>
I have a `pysh` helper in my .bashrc which is an alias for nix-shell'd Python with readline.
<kisik21>
Nix is both a blessing and a curse for developers, as it seems...
<srhb>
It's a really "buy-in" thing. The more you embrace the philosophy (in this case, ephemeral environments) the better it works.
<srhb>
Usually. :)
stumble has quit [Ping timeout: 240 seconds]
<kisik21>
though it's more of a blessing for autoconf-based C programs and their creators, and a living hell of environments for developers. Though Python was always a bunch of custom environments.
<srhb>
I'd take a properly written shell.nix over whatever magic virtualenv does any day though. :D
<kisik21>
But how do you install pyfiglet from git, for example? Writing a derivation that will build it from master every time?
<srhb>
I just override the src attribute (presumably, again, not python-fu strong)
<kisik21>
pip clearly trumps nix-shell in this case... Only one single line in requirements.txt
<srhb>
We absolutely _can_ do that (see builtins.fetchGit) but I prefer to always be version locked for reproducibility.
<srhb>
And make a convenient way of upgrading the bound to whatever HEAD is instead.
<Synthetica>
TIL: TOFU is not just a weird thing I do
<Synthetica>
!tofu
<{^_^}>
To get a sha256 hash of a new source, you can use the Trust On First Use model: use probably-wrong hash (for example: 0000000000000000000000000000000000000000000000000000) then replace it with the correct hash Nix expected.
<srhb>
Synthetica: I think that's cargo culted to immortality. :-)
<{^_^}>
[nixpkgs] @tazjin opened pull request #40147 → bluemix-cli: init at 0.6.6 → https://git.io/vpKzt
<srhb>
Synthetica: 52i0 ftw
<simpson>
TOFU is an interesting pattern. It shows up often.
<srhb>
It even got better in Nix 2.0 now that you don't have to download twice all the time...
<kisik21>
Ok, goijng the nix-shell way, could anyone tell me how to make a shell.nix that will install bash and python3.withPackages and will Do The Right Thing (TM) of launching a bash script which itself just bootstraps a python module?
<kisik21>
can't find boilerplate for this on wiki
<srhb>
I think it's even in the manual. Sec, just working on getting a working browser.
Arcaelyx has joined #nixos
rosa has joined #nixos
fare__ has quit [Ping timeout: 276 seconds]
<kisik21>
srhb: and what's up with your browser, why it doesn't work?
<srhb>
kisik21: I did bad things to my nixpkgs checkout. :-)
<kisik21>
Did you modify nixpkgs directly? Shouldn't these things be exported to some sort of external file, like with my lurch (pidgin plugin) derivation?
<srhb>
kisik21: Sure, but my browser is impute and pollutes my homedir with pesky state, and something Bad Things ruin that state.
<kisik21>
firejail?
<kisik21>
could sandbox anything into private homedir
<srhb>
kisik21: Oh cool, that sounds like the sort of thing I'd want, thank you.
<kisik21>
needs a little bit of configuration though
<srhb>
Yep, checks popped up on all repos for me today..
<WilliamHamilton>
hi all, has anyone tried to use ghc plugin in nixos? When trying out a ghc plugin, it seemed not to be working despite the author's suggestion; I wonder if ghc has been patched in a particular way for nixos
<simpson>
Nope, it's a POS. The `bx` plugins don't work at all on NixOS.
<Mic92_>
Ok. I have not tested those.
<simpson>
Oh, this isn't the bx part. I don't know these tools at all, sorry.
<Mic92_>
in the worst case, you can docker your problems away.
<{^_^}>
[nixpkgs] @rkoe opened pull request #40148 → ext4magic: init at 0.3.2 → https://git.io/vpKaT
iqubic has joined #nixos
joehh has joined #nixos
<ghostyy>
how do i append to a list in nix expression language? i want to append "skein" to boot.initrd.luks.cryptoModules
logzet has quit [Remote host closed the connection]
<Mic92_>
ghostyy: ++
<{^_^}>
ghostyy:'s karma got increased to 1
<iqubic>
what's the command to install packages for a single user, without going through the hassle of adding it to the right config file.
<Mic92_>
now actually the operator ++
<ghostyy>
oh thanks Mic92_
<iqubic>
ghostyy: --
<iqubic>
Looks like karma only goes up.
<Mic92_>
karma
<Dezgeg>
in case of a nixos option you'd only say boot.initrd.luks.cryptoModules = ["skein"] though
<ghostyy>
what if i need the rest of them?
<rain1>
how do i increment a variable in C?
<Mic92_>
rain1: nice try
<ghostyy>
rain1, ++
<{^_^}>
rain1,'s karma got increased to 1
<ghostyy>
obviously
<iqubic>
So I want to install xmodmap just for my user, and not for everyone else. I think it's like "nix-env -iA..."
<iqubic>
iqubic: ++
<{^_^}>
iqubic:'s karma got increased to 1
<ghostyy>
haha
<iqubic>
Well, I found a bug in the system.
<ghostyy>
iqubic, good job for finding that, why dont you give yourself a pat on the back
<srhb>
It's not a bug, it's a feature. The bot was not exploited!
<ghostyy>
or an additional karma
<iqubic>
ghostyy: ++
<{^_^}>
ghostyy:'s karma got increased to 2
<ghostyy>
:P
<{^_^}>
[nixpkgs] @knedlsepp opened pull request #40149 → qt5.qtwebengine: Fix darwin build → https://git.io/vpKaE
<Mic92_>
iqubic: its xlibs.xmodmap
<iqubic>
Mic92_: What's the full command? I don't actually know that.
<Mic92_>
iqubic: nix-env -iA xlibs.xmodmap or nix-env -f '<nixpkgs>' -iA xlibs.xmodmap if you do not use channels
<jtojnar>
😮 Instead of binary pass/fail commit statuses, integrators can now report more fine-grained outcomes, such as a neutral conclusion
<kisik21>
yeah, I now know why python3 doesn't work...
<kisik21>
xonsh adds its own python3 in the path
<kisik21>
and it collides with the python3 from /run/current-system/sw/bin/
<kisik21>
I feel some serious hackery down there in xonsh derivation...
<Mic92_>
kisik21: how are you using xonsh?
<iqubic>
WHat is xonsh?
<Mic92_>
a shell
<Mic92_>
kisik21: I guess the python wrapper of our python infrastructure is adding that.
<kisik21>
Mic92_: programs.xonsh.enable = true; and a patched script in /usr/local/bin so xonsh would find Python user site.
<kisik21>
yes, your wrapper script adds python in the path
<kisik21>
wait a sec... I think I could change it
<fogbugz234>
services.xserver.windowManager.stumpwm.enable is using nixpkgs.stumpwm, which is marked as broken and which I've not been able to fix. However, nixpkgs.lispPackages.stumpwm seems to work fine. Is there any reason to prefer the former, rather than the latter? Would you accept a pull request modifying this?
<Mic92_>
kisik21: you could use xonsh.override { python = ...; } maybe
<iqubic>
The service I wrote earlier is working just as expected.
<kisik21>
Mic92_: in override as in programs.xonsh.override?
<{^_^}>
[nixpkgs] @srhb opened pull request #40150 → zoom-us: liblgvnd fix for libGL.so in libpath → https://git.io/vpKVY
<kisik21>
please remind me, overriding packages does allow to add to propagatedBuildInputs?
<mightybyte>
What's the syntax for the URL argument to builtins.fetchGit?
<mightybyte>
...specifically for private github repos.
<srhb>
mightybyte: It supports multiple, iirc, but if you're asking I guess you're looking for ssh://user@host/file.git -- I *think*
hlolli has quit [Ping timeout: 240 seconds]
<mightybyte>
srhb: I've got that, but it's not working.
<mightybyte>
I've tried every variation I can think of.
<stumble>
fogbugz234: I'm a newbie, but since nobody else is replying: using lispPackages.stumpwm sounds sensible to me. That's the one people see if they search Nix packages.
<srhb>
mightybyte: I just tested with builtins.fetchGit { url = "ssh://git@github.com/srhb/myprivaterepo.git"; }
<srhb>
mightybyte: Appears to work?
<iqubic>
What happens if a "oneshot" systemd service runs a command that ends up failing?
<kisik21>
iqubic: I'd assume it is marked as failed.
<kisik21>
and it will not be restarted, since it's oneshot.
joehh has quit [Ping timeout: 264 seconds]
<srhb>
mightybyte: can you elaborate on not working? Is it just very, very, very, very slow?
<iqubic>
That's good enough for me.
<iqubic>
I mean I assume that things which rely on the internet will have some sort of fallback mechanic for if the internet is out.
<mightybyte>
I'm getting "fatal: not a tree object"
<srhb>
mightybyte: Do you have no master branch?
<mightybyte>
I was specifying a rev.
<mightybyte>
I wonder if that was the problem.
<srhb>
You need a ref, too.
<srhb>
In case the answer was "no" at least :-)
<srhb>
At least, I think that's what I concluded.
<mightybyte>
Ok, when I remove rev it works
<mightybyte>
Progress!
<srhb>
Woo
<mightybyte>
Now what's the difference between rev and ref?
fare__ has joined #nixos
stumble has quit [Read error: Connection reset by peer]
<mightybyte>
I want to point it at a particular commit hash just like with fetchGitHub
stumble has joined #nixos
<srhb>
ref is the... um, ref from which to find the rev.
<LnL>
revision (commit) vs a git ref (pointer to a commit)
<srhb>
(Admittedly a poor and also recursive definition) :-)
<mightybyte>
What does that mean?
<LnL>
git ls-remote will show you the refs of a repository
<srhb>
In effect, that fetchGit requires a ref (usually a branch) in which to go look.
<dtz>
many instances of fetchFromGitHub pretend a ref is a rev, and IIRC fetchFromGitHub doesn't support using 'ref' anyway?
<LnL>
things like branches, tags, ... are refs
<mightybyte>
LnL: git ls-remote doesn't help me either. It gives me both commit hashes and branch names.
<LnL>
isn't this builtins.fetchGit?
<dtz>
so just FYI as far as fetchFromGitHub goes its use is not a good indicator of what rev/ref are meant to be lol
<dtz>
if you have a commit hash, you want 'rev', should work with fetchGit ?
jrolfs_ has joined #nixos
<mightybyte>
dtz: Yeah, that's what I was doing.
<mightybyte>
But it wasn't working.
<LnL>
only if it's part of HEAD
<mightybyte>
Ahh
<dtz>
really need to fix this--is there a tracking issue on NixOS/nix about this?
<mightybyte>
So if it's on a different branch I have to specify the branch in ref?
<LnL>
ie whatever the default branch is on the remote
<dtz>
mightybyte: thinkso, yes, but I'm not sure if that's intentional/desired behavior or not
<mightybyte>
dtz: Yeah, it's pretty sad that I couldn't even figure this out while reading the documentation.
<dtz>
suspect it's more working around what the implementation does than anything else
<mightybyte>
It really should work like fetchFromGitHub
<{^_^}>
[nixpkgs] @yrashk opened pull request #40151 → sit: better fix for aarch64 bug → https://git.io/vpKwz
xy2_ has quit [Ping timeout: 268 seconds]
stumble has quit [Read error: Connection reset by peer]
stumble has joined #nixos
<{^_^}>
[nixpkgs] @rkoe opened pull request #40152 → cups-utax: init at 20140115 → https://git.io/vpKwK
jrolfs_ has quit [Ping timeout: 260 seconds]
stumble- has joined #nixos
stumble has quit [Read error: Connection reset by peer]
stumble- has quit [Ping timeout: 268 seconds]
<fogbugz234>
stumble- thanks yes I think using nixpkgs.lispPackages.stumpwm instead of nixpkgs.stumpwm makes sense
jrolfs_ has joined #nixos
<{^_^}>
[nixpkgs] @davidak opened pull request #40153 → plank: init at 0.11.4 → https://git.io/vpKok
day|flip has joined #nixos
tertle||eltret has joined #nixos
jrolfs_ has quit [Ping timeout: 240 seconds]
ericsagnes has quit [Ping timeout: 256 seconds]
<day|flip>
I been trying to figure out how to get around with app theme. that normally read from /usr/share/appnamehere/themes/. question is how will i be able to add and edit a new theme. app be able to read from that folder. i tried using "environment.pathsToLink = [ path/to/app/themes ]" but problem is i can't edit or add due to read only folder and files