gchristensen changed the topic of #nixos to: NixCon 2018 CfP is now open! https://nixcon2018.org/ || Share the output of nix-shell -p nix-info --run nix-info to help us help you. || https://nixos.org || Latest NixOS: https://nixos.org/nixos/download.html || Latest Nix: https://nixos.org/nix/download.html || Logs: https://logs.nix.samueldr.com/nixos/ || #nixos-dev, #nix-darwin, #nixos-aarch64, #nixos-chat, #nixcon
philippD has joined #nixos
<{^_^}> [nixpkgs] @orivej opened pull request #45284 → msgpack: rename from libmsgpack → https://git.io/fAesX
klntsky has quit [Ping timeout: 250 seconds]
sm0rux has quit [Ping timeout: 256 seconds]
iyzsong has joined #nixos
klntsky has joined #nixos
<{^_^}> [nixpkgs] @kini opened pull request #45285 → makeself: backport megastep/makeself#142 → https://git.io/fAeGT
<sorbet> (In case anyone was curious, I learned you can do nix-store --query --referrers /nix/store/... to see what's using that entry.
<sorbet> I realize now I should be overriding the python2 package to include my module.)
Ariakenom has quit [Quit: Leaving]
<{^_^}> [nixpkgs] @Chiiruno opened pull request #45288 → qbittorrent: Fix missing desktop file → https://git.io/fAeGM
<ldlework> files = attrNames (builtins.readDir ./.);
<ldlework> file-filter = file: (file != "default.nix" && hasSuffix ".zsh");
<ldlework> filtered-files = filter file-filter files;
<ldlework> I get "value is a function where a boolean was expected"
<ldlework> But I can't percieve where hte mistake is...
<ldlework> I tried in repl, files is a list of strings, file-filter is a function...
<ldlework> sorbet: cool
Lisanna has quit [Quit: Lisanna]
fresheyeball has quit [Quit: WeeChat 2.0]
sm0rux has joined #nixos
<ldlework> oh I see it
<{^_^}> [nixpkgs] @Chiiruno opened pull request #45289 → vscode: Wrap with GAppsHook → https://git.io/fAeZf
fenedor has joined #nixos
alexteves has quit [Remote host closed the connection]
fenedor has quit [Ping timeout: 240 seconds]
acarrico has joined #nixos
<Enzime> gchristensen: https://github.com/NixOS/nixpkgs/pull/43754#issuecomment-413902598 did I do something wrong?
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<gchristensen> you have to specify what to build https://github.com/nixos/ofborg#build
<ldlework> Anyone here use ranger + kitty by change and have kitty's image preview working in ranger?
countingsort has quit [Ping timeout: 272 seconds]
<ldlework> Looks like the nixos package hacks up ranger to force it to use w3m
Lisanna has joined #nixos
<ldlework> i think technically all i need to get it to work is make sure that kitty is in the ranger derivation's inputs?
<ldlework> is there an easy helper command for that?
<sorbet> If I want to have a list of python modules available system-wide for python2, what do y'all think is the best approach to that? The problem I'm running into is that even if I throw (python2.withPackages...) into environment.Systempackages,
<sorbet> there's a python2 entry earlier in the path that doesn't have the modules (that everything else is using).
<sorbet> So I think I need to override it, but I'm having a hard time trying to figure out how to do that with my overlay (which I've mostly just pieced together looking at examples).
hamishmack has joined #nixos
<{^_^}> [nixpkgs] @jerith666 opened pull request #45290 → plasma5: improve kbuildsycoca activation script → https://git.io/fAenm
goibhniu has quit [Ping timeout: 256 seconds]
tertle||eltret has joined #nixos
Anton-Latukha has quit [Quit: Leaving.]
acarrico has quit [Ping timeout: 240 seconds]
erickomoto has joined #nixos
Mateon3 has joined #nixos
erickomoto has quit [Client Quit]
Mateon1 has quit [Ping timeout: 256 seconds]
Mateon3 is now known as Mateon1
lassulus_ has joined #nixos
lassulus_ is now known as Random_int
lassulus has quit [Ping timeout: 244 seconds]
Random_int is now known as lassulus
etu has quit [Read error: Connection reset by peer]
<samueldr> hmm, trying to *understand* why trunk jobset seemingly has freeciv for darwin *removed or with an evaluation error*, but ofborg built it
etu has joined #nixos
<samueldr> I see the last time it was a member of an evaluation seems to be 26th of june
fenedor has joined #nixos
<{^_^}> Channel nixpkgs-18.03-darwin advanced to https://github.com/NixOS/nixpkgs/commit/a4e068ff9c8 (from 11 hours ago, history: https://channels.nix.gsc.io/nixpkgs-18.03-darwin)
johnw has quit [Ping timeout: 272 seconds]
<Enzime> samueldr: check the failed build steps
Lisanna has quit [Quit: Lisanna]
<Enzime> looks like an issue with the Mac builders
Supersonic has quit [Ping timeout: 265 seconds]
<samueldr> Enzime: earlier (2 or three builds earlier) fail without builder-specific issues
<samueldr> and this doesn't explain how it stopped being a member of the jobset
<gchristensen> samueldr: happened between 5ac6ab0...9c1edbd
<samueldr> thank you gchristensen
<gchristensen> :)
<samueldr> TIL about hydraPlatforms
<samueldr> (I know it's not simple, but that would be helpful if hydra had some way to show the reasoning behind changes like this)
<gchristensen> yes
<gchristensen> what has you interested?
<samueldr> ? just closing PRs
<gchristensen> cool
Supersonic has joined #nixos
<samueldr> verifying whether it was building fine on darwin or it was a regression
MrCoffee is now known as NightTrain
fenedor has quit [Ping timeout: 244 seconds]
<kandinski> clever: I did as you suggested and made a PR of the fix I needed. Would you mind explaining to me what the process is now? For instance, someone changed the base from master to staging and back. A bot added a lot of tags. And there is now a conflict which suggests someone else already made the same fix before me, but also added a comment.
<{^_^}> #45213 (by candeira, 17 hours ago, open): Libtiff: fix failing fetch of Debian patch
<kandinski> I think the process now is I close the PR, because it became redundant since I made it.
<{^_^}> [nixpkgs] @candeira closed pull request #45213 → Libtiff: fix failing fetch of Debian patch → https://git.io/fNhhz
<{^_^}> [nixpkgs] @samueldr merged pull request #44779 → freeciv: 2.5.11 -> 2.6.0 → https://git.io/fN9uK
<{^_^}> [nixpkgs] @samueldr pushed 2 commits to master: https://git.io/fAect
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
acarrico has joined #nixos
hamishmack has joined #nixos
<gchristensen> is it possible to add options to services.nginx.virtualHosts.<name> ?
mayhewluke has quit [Ping timeout: 240 seconds]
<{^_^}> [nixpkgs] @samueldr merged pull request #44784 → gramps: 4.2.8 -> 5.0.0 → https://git.io/fN92I
<{^_^}> [nixpkgs] @samueldr pushed 2 commits to master: https://git.io/fAeCJ
countingsort has joined #nixos
acarrico has quit [Ping timeout: 272 seconds]
<samueldr> hmmm, it's a submodule... now I'm curious
<countingsort> how would i make nixos appreciate that i took the time to write a $HOME/.config/nixpkgs/config.nix, instead of ignoring it? or am i doing it wrong (minimal example): https://ptpb.pw/CyBe
iyzsong has quit [Ping timeout: 244 seconds]
<gchristensen> countingsort: those only work with NixOS not Nix and not ~/.config/nixpkgs
<gchristensen> so only etc/nixos/configuration.nix
<init_6> it`s possible to aplly custom patches on default nixos kernel?
<countingsort> nixos.org/nix/manual mentions ~/.config/nixpkgs/config.nix though
<countingsort> o
<countingsort> its without .xonfig, im dumb
<countingsort> nvm me then
iyzsong has joined #nixos
<gchristensen> samueldr: I don't think I can :/
<countingsort> still doesnt work though
<gchristensen> what doesn't?
<samueldr> I don't have a setup to test, and am about to do AFK stuff, but is it possible to re-define *one* option in a nixos module?
<countingsort> gchristensen: i expect that to change my PS1 to '> ', it doesnt though
<samueldr> (though I'm guessing it would be too easy if it was possible)
<gchristensen> you can only use those options in /etc/nixos/configuration.nix
<countingsort> whys that?
<gchristensen> because those are NixOS options, not Nix options, and NixOS options go in to /etc/nixos/configuration.nix
<countingsort> so i cant change the prompt on a user basis?
<countingsort> oof
<gchristensen> right
acarrico has joined #nixos
<countingsort> why would i do that instead of a bashrc then?
<gchristensen> if you wanted to set it from one place, with the rest of your settings, and wanted it to be the default for all users.
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<countingsort> so as a user i have no other choice but to just use a bashrc? whats with the .bash_profile than, i have to hack that on a system basis to use a user one?
<gchristensen> you would edit bashrc or bash_profile based on how Bash treats those specific files
<countingsort> https://github.com/NixOS/nixos/blob/master/modules/programs/bash/bash.nix seems to have a .bash_profile hard coded and it defenitely doesnt execute the one in my home dir
<gchristensen> it writes to /etc/profile, not .bash_profile
<countingsort> how do i get it to execute my .bash_profile
<gchristensen> you'll have to read the Bash man page: `man bash`
philippD has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]
init_6 has quit [Ping timeout: 256 seconds]
barrucadu has joined #nixos
<countingsort> gchristensen: according to the bash manpage (and bash on literally any system besides nixos) it should execute ~/.bash_profile on startup unless --noprofile is passed
<gchristensen> only if it is a login shell
<countingsort> it is though
<gchristensen> how are you starting bash?
<countingsort> logging into a tty
<gchristensen> so X11 is not turned on?
<countingsort> doesnt matter
<countingsort> tried both
<gchristensen> if you run `bash --login` does it run your ~/.bash_profile?
mayhewluke has joined #nixos
oldandwise has joined #nixos
<countingsort> yes
<gchristensen> ok, then whatever you were starting wasn't a login shell :P
<oldandwise> having problem on my 32-bit netbook with `nix-env -iA nixos.octave`. Error says, '/bin/ld: src/.libs/octave-cli: hidden symbol `__divmoddi4' in /nix/store/xxxxxx-gcc-7.3.0/lib/gcc/i686-unknown-linux-gnu'. Can anyone help?
<gchristensen> anyway, past my bedtime. good luck
<gchristensen> and good night
<countingsort> gchristensen: logging in isnt a login shell? huh
<countingsort> good night
<{^_^}> countingsort: Night!
vodurden has joined #nixos
<{^_^}> [nixpkgs] @orivej-nixos merged pull request #45288 → qbittorrent: Fix missing desktop file → https://git.io/fAeGM
<{^_^}> [nixpkgs] @orivej-nixos pushed commit from @Chiiruno to master « qbittorrent: Fix missing desktop file (#45288) »: https://git.io/fAeWv
aw has quit [Remote host closed the connection]
spacefrogg has quit [Remote host closed the connection]
aw has joined #nixos
spacefrogg has joined #nixos
c00w has joined #nixos
init_6 has joined #nixos
<c00w> Anyone able to help with translating a nix-build command to nix build? I think I'm missing something but I can't figure out what.
<init_6> it`s possible to aplly custom patches on default nixos kernel?
garbas has quit [Quit: WeeChat 2.1]
hamishmack has joined #nixos
sorbet has quit [Quit: WeeChat 2.0]
thc202 has quit [Ping timeout: 240 seconds]
<c00w> init_6 - I've definitely seen some options to let you modify the config file, so I'm sure it's possible. Not sure how.
<init_6> c00w: not config. i need to apply new custom patches with new configs over some defult kernel
<c00w> There definitely seems to be a kernelPatches option.
<ldlework> Anyone using Kitty? Are you able to start Ranger?
<init_6> c00w: line 436 kernelPatches = ? THX!!!
<c00w> yw
sir_guy_carleton has joined #nixos
<c00w> I think my problem is actually simpler - I just don't know how to translate <nixpkgs/nixos> to using a channel referenced by channel:nixos-unstable
<{^_^}> [nixpkgs] @kalbasit opened pull request #45291 → linux: 4.4.148 -> 4.4.149; 4.9.120 -> 4.9.121; 4.14.63 -> 4.14.64; 4.17.15 -> 4.17.16; 4.18.1 -> 4.18.2 → https://git.io/fAeWh
dmj` has quit [Quit: Connection closed for inactivity]
<{^_^}> [nixpkgs] @bhipple opened pull request #45292 → ecos: init at 2.0.6 → https://git.io/fAelR
MarkOtaris has joined #nixos
<{^_^}> [nixpkgs] @r-ryantm closed pull request #45202 → libuv: 1.21.0 -> 1.22.0 → https://git.io/fNhbx
<{^_^}> [nixpkgs] @ryantm merged pull request #45241 → containerd: 1.1.1 -> 1.1.2 → https://git.io/fNjOi
<{^_^}> [nixpkgs] @ryantm pushed commit from R. RyanTM to master « containerd: 1.1.1 -> 1.1.2 »: https://git.io/fAelw
<eqyiel> any ideas why the second test here (extern.nix) fails with `server# mount: mounting /dev/vda on /mnt-root/ failed: No such file or directory` https://gitlab.com/eqyiel/nixos-mailserver/-/jobs/90278230 ? relevant gitlab-ci.yml: https://gitlab.com/eqyiel/nixos-mailserver/blob/bddf0d10461a5e96eb99f2a5f59ed3ebf71aaa9c/.gitlab-ci.yml
<eqyiel> I'm thinking that it might be alright if I make each `nix-build` a seperate job, but I'd rather not if possible
<{^_^}> [nixpkgs] @thefloweringash opened pull request #45293 → libqmatrixclient: enable Darwin platform → https://git.io/fAelp
<{^_^}> [nixpkgs] @bhipple opened pull request #45294 → scs: init at 2.0.2 → https://git.io/fAe8J
troydm has quit [Ping timeout: 256 seconds]
<Lears> ranger seems to work fine on kitty for me, ldlework. Though I'm on 18.03 and I haven't updated for a while.
<techieAgnostic> can someone tell me what im messing up trying to get stack to work?
<DigitalKiwi> probably that you're trying to use stack
<emily> has anyone used node2nix? I'm not sure how to get a "nix-shell"/"nix run" working properly
<eqyiel> emily: could you show what you've tried / what's not working?
<eqyiel> basically you build the `shell` attribute of the default.nix that node2nix generates
<ldlework> Lears: i think it had to do with optirun..
<ldlework> I still get the locale error though
<emily> right. I guess I'm just not sure on the correct command-line syntax to do that or what to do with the result of that build ^^; sorry, I've only recently switched to Nix/NixOS and am trying to get a Reason npm project I had up and running
vodurden has quit [Read error: No route to host]
<adamantium> ldlework: did you get ranger images working with your terminal?
<eqyiel> emily: here's a more minimal example: https://gist.github.com/eqyiel/dfbf353ba25f965f9bc0652fe0440085
<emily> eqyiel: ah, thank you for the pointer! I didn't realise a shell.nix separate to default.nix was needed
<emily> I missed the filename
<eqyiel> 😅
<ldlework> adamantium: no
<ldlework> adamantium: well yes
<ldlework> but not with the kitty icat support
<ldlework> just with w3m which works out of the box
<adamantium> idk what kitty icat is
troydm has joined #nixos
iyzsong has quit [Ping timeout: 256 seconds]
<ldlework> kitty is a gpu accelerated terminal
<ldlework> icat is it's image protocol
<adamantium> alacritty
<ldlework> lol it doesn't even have scrollback
<{^_^}> Channel nixpkgs-unstable advanced to https://github.com/NixOS/nixpkgs/commit/c40af1fde3d (from 16 hours ago, history: https://channels.nix.gsc.io/nixpkgs-unstable)
iyzsong has joined #nixos
<ldlework> adamantium: do you get the local complaint when using ranger?
<ldlework> locale*
<ldlework> Ah interesting, I only get that when I use the unstable version
<ldlework> let me know if you want to see my module for ranger
sir_guy_carleton has quit [Quit: WeeChat 2.0]
<oldandwise> what's with this missing __divmoddi4 error in nixos and how to work around this?
<ashkitten> uh, i'm having trouble getting steam to use my gtk theme?
<ashkitten> it's also got an error for a taskbar icon
<ashkitten> and it says it can't find the adwaita engine while starting up
iyzsong has quit [Ping timeout: 244 seconds]
Guanin has quit [Ping timeout: 244 seconds]
Guanin has joined #nixos
iyzsong has joined #nixos
vodurden has joined #nixos
ghasshee has quit [Ping timeout: 272 seconds]
drakonis_ has quit [Remote host closed the connection]
<ldlework> how do I get from a list to an attrset?
<ldlework> nm
<oldandwise> what's wrong here, i can't get it to build: `nix build --builders ssh://192.168.2.120 nixpkgs.cscope --max-jobs 0`
<oldandwise> error is: `error: unable to start any build; either increase '--max-jobs' or enable remote builds`
<oldandwise> already have in /etc/nixos/configuration.nix `nix.distributedBuilds = true;
<ldlework> is there any way to add functions to lib?
Fare has quit [Ping timeout: 240 seconds]
orivej has quit [Ping timeout: 244 seconds]
mayhewluke has quit [Ping timeout: 244 seconds]
mayhewluke has joined #nixos
hyper_ch has quit [Quit: ZNC 1.7.1 - https://znc.in]
hyper_ch has joined #nixos
hyper_ch has quit [Client Quit]
hyper_ch has joined #nixos
justbeingglad has joined #nixos
<{^_^}> Channel nixpkgs-unstable advanced to https://github.com/NixOS/nixpkgs/commit/5cb9703a8fa (from 15 hours ago, history: https://channels.nix.gsc.io/nixpkgs-unstable)
hyper_ch has quit [Quit: ZNC 1.7.1 - https://znc.in]
<oldandwise> why can't i do remote build?
hyper_ch has joined #nixos
hyper_ch has quit [Remote host closed the connection]
hyper_ch has joined #nixos
justbeingglad has left #nixos [#nixos]
graphene has quit [Remote host closed the connection]
graphene has joined #nixos
sm0rux has quit [Ping timeout: 256 seconds]
jD91mZM2 has joined #nixos
Aexoden has quit [Remote host closed the connection]
Aexoden has joined #nixos
c00w has quit [Ping timeout: 256 seconds]
endformationage has quit [Quit: WeeChat 1.9.1]
NightTrain has quit [Ping timeout: 272 seconds]
aarvar has quit [Ping timeout: 268 seconds]
vidbina has joined #nixos
<{^_^}> [nixpkgs] @Mic92 merged pull request #44323 → qsstv: init 9.2.6 → https://git.io/fNoye
<{^_^}> [nixpkgs] @Mic92 pushed 3 commits to master: https://git.io/fAeRm
simukis has joined #nixos
philippD has joined #nixos
vodurden has quit [Remote host closed the connection]
oida has quit [Ping timeout: 250 seconds]
iyzsong has quit [Read error: Connection reset by peer]
b has joined #nixos
_rvl has quit [Quit: ZNC 1.6.5 - http://znc.in]
iyzsong has joined #nixos
philippD has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]
oida has joined #nixos
<Myrl-saki> How do I force a rebuild for a systemd service?
<Myrl-saki> oldandwise: __divmoddi4 is a gcc only thing.
_rvl has joined #nixos
<Myrl-saki> oldandwise: And that means that ssh://192.168.2.120 can't be reached
lawlesseel has quit [Quit: Leaving]
lawlesseel has joined #nixos
goibhniu has joined #nixos
petar has quit [Quit: Lost terminal]
petar has joined #nixos
Lisanna has joined #nixos
orbekk has quit [Quit: WeeChat 2.0]
<Myrl-saki> Seems like ReloadIfChanged is half what I want.
<Myrl-saki> OTOH, there's StopOnReconfiguration
<ldlework> Anyone wanna try my improvements to nix-mode?
<Myrl-saki> ldlework: Emacs nix-mode?
<ldlework> Myrl-saki: yeah
<ldlework> It has some indentation problems i think I have improved
<Myrl-saki> ldlework: I don't use Nix to manage Emacs, how should I try it out?
<ldlework> Well I don't think you can :)
<Myrl-saki> :(
<Myrl-saki> FeelsBad
<Myrl-saki> Emacs indirectly depends on systemd so building it is, um, fun.
<Myrl-saki> (Amongst other things)
<ldlework> Myrl-saki: ohhh
<ldlework> I read your message wrong
<ldlework> you use emacs?
<Myrl-saki> Yep.
<ldlework> ah ok then trying it out is easy
<ldlework> are you already using nix-mode?
<Myrl-saki> Yep!
<Myrl-saki> I'm also using `straight` as my package manager, but it should be easy to override.
<ldlework> nice
<ldlework> you can just eval some code :)
<Myrl-saki> Oh okay. :P
<ldlework> Myrl-saki: you might want to comment out the use-package call
<ldlework> and just manually eval the :config bits
<ldlework> Myrl-saki: try to find some code that you have problems with
<ldlework> before you eval it
<ldlework> so you can see the difference :)
<Myrl-saki> ldlework: I'm guessing I'm too used to manually adjusting tabs that not having to adjust tehm would feel wrong. :P
<ldlework> lol
<ldlework> Myrl-saki: you can use nix-mode-format to format the whole buffer
<ldlework> theoretically
<ldlework> it is working pretty good for me!
CrystalGamma has joined #nixos
<Myrl-saki> ldlework: Thanks. I'll take a look at this later, since I'm um, a bit lazy.
<ldlework> i commented it real good too, so hopefully other people don't have to spend much effort to understand how it works in order to improve it further
<ldlework> :(
<Myrl-saki> Mmm
<Myrl-saki> I already took a look at it. :P I'll try it out later*? :D
<ldlework> lol you just open a file, screw up the indentation and then M-x nix-mode-format
<ldlework> but hehe, I'm satisfied enough that it works for me
<CrystalGamma> how do I get X11 forwarding over SSH to work between NixOS machines? I have tried setting (and forcing!) services.openssh.forwardX11 and programs.ssh.setXAuthLocation on both machines, but there is still no X socket in /tmp/.X11-unix/
<Myrl-saki> ldlework: Do you use straright too? Since I see a straight line on use-package.
<ldlework> Myrl-saki: yeah I do
<ldlework> Myrl-saki: i evangelize where possible too
<Myrl-saki> :P
<Myrl-saki> ldlework: I copy and paste the non-use-package bits to a new file, eval it, replace straight-use-package with the use-package sexp?
<ldlework> Myrl-saki: straight has use-package integration
<ldlework> you might need (setq straight-use-package-by-default t)
<Myrl-saki> Right
<Myrl-saki> (straight-use-package 'feebleline)
<Myrl-saki> That's what I have, and mostly nothing else.
<ldlework> lol
<Myrl-saki> (In terms of configuring straight, I mean. :P)
<ldlework> just turn on use-package support
<ldlework> then you can run that whole file
<ldlework> if it complains that the recipe has changed, just C-x e at the end of each form after :config to execute those
<ldlework> then it should be working
<BlessJah_> can root have channel that won't be visible to other users?
<Myrl-saki> ldlework: (load "~/.emacs.d/nix-mode-fixes.el")
<ldlework> sure
<Myrl-saki> Oh, I see what you mean now.
<Myrl-saki> So replace all straight-use-package with use-package?
mayhewluke has quit [Ping timeout: 260 seconds]
<Myrl-saki> If I use the -by-default flag
<ldlework> if you use the by default flag
<ldlework> you don't have to do anything
Ariakenom has joined #nixos
<Myrl-saki> Oh
<ldlework> (setq straight-use-package-by-default t) means "allow natural (use-package) use"
<srhb> BlessJah_: The channel "visibility" is determined by $NIX_PATH. If you remove /nix/var/nix/profiles/per-user/root/channels from it, it won't affect other users.
<srhb> (Well, it won't affect the users that have that change in NIX_PATH...)
mayhewluke has joined #nixos
<viric> BlessJah_: good :) I'm a really bad maintainer
<ldlework> hi srhb
<srhb> ldlework: Hi!
<Myrl-saki> ldlework: I think I got it. Thanks.
<ldlework> awesome
<BlessJah_> viric: I've noticed that you've tried to remove yourself from maintainers in master, but that did not yet propagate to 18.03 yet
<ldlework> no more manually tabbing crap!
<srhb> BlessJah_: In general, master does not propagate to stable channels
<BlessJah_> srhb: I thought about backporting rather than actually propagating
<srhb> BlessJah_: Backporting is sporadic outside of security fixes. :)
<BlessJah_> 0.18.0 is required for yubikey4, openssh talks directly to pkcs11 so library and it seems libc between 18.03 and unstable is incompatible
<viric> BlessJah_: I should have done that years ago
<viric> BlessJah_: but that's part of being a bad maintainer
<BlessJah_> I had to 'stable.callPackage <unstable/pkgs/opensc> {}' in overrides
<BlessJah_> since I wanted this system-wide, I had to add nixos-unstable as root, add overlays as root and now I wonder if it's possible to reduce visibility of said channel
vidbina has quit [Ping timeout: 240 seconds]
<BlessJah_> viric: next release won't build, it tries to mkdir /etc/completion.d or sth else directly in /etc
<viric> ok
LnL has quit [Read error: Connection reset by peer]
LnL has joined #nixos
CrystalGamma has quit [Quit: Leaving]
vcunat has joined #nixos
<symphorien> BlessJah_: when something installs config to /etc you can use this installFlag: https://github.com/NixOS/nixpkgs/blob/b9fafcfc3020580be1cefac93d8fe4cefe9135f5/pkgs/applications/audio/abcde/default.nix#L40
justbeingglad has joined #nixos
justbeingglad has left #nixos [#nixos]
adisbladis has joined #nixos
<{^_^}> [nixpkgs] @vcunat pushed 7 commits to staging: https://git.io/fAeuY
aexoxea has joined #nixos
<{^_^}> [nixpkgs] @etu opened pull request #45296 → php71: 7.1.20 -> 7.1.21 → https://git.io/fAeuc
<{^_^}> [nixpkgs] @etu opened pull request #45297 → php71: 7.1.20 -> 7.1.21 → https://git.io/fAeu8
shabius has quit [Quit: Leaving]
aexoxea has quit [Quit: Goodbye!]
<{^_^}> [nixpkgs] @vcunat pushed 2 commits to staging: https://git.io/fAeuP
<{^_^}> [nixpkgs] @vcunat merged pull request #45237 → expat: 2.2.5 -> 2.2.6 → https://git.io/fNjYY
aexoxea has joined #nixos
<{^_^}> [nixpkgs] @vcunat pushed 7 commits to master: https://git.io/fAeuy
shabius has joined #nixos
<{^_^}> [nixpkgs] @vcunat pushed 2 commits to staging: https://git.io/fAeu9
<{^_^}> [nixpkgs] @vcunat merged pull request #45174 → p11-kit: 0.23.12 -> 0.23.13 → https://git.io/fNhS6
<{^_^}> [nixpkgs] @vcunat merged pull request #45291 → linux: 4.4.148 -> 4.4.149; 4.9.120 -> 4.9.121; 4.14.63 -> 4.14.64; 4.17.15 -> 4.17.16; 4.18.1 -> 4.18.2 → https://git.io/fAeWh
<{^_^}> [nixpkgs] @vcunat closed pull request #45219 → harfbuzz-icu: 1.8.2 -> 1.8.8 → https://git.io/fNjJ3
<Taneb> Is nixos-generate-config going to know better than I do whether I boot with UEFI or BIOS
sir_guy_carleton has joined #nixos
<vcunat> Taneb: no, I wouldn't expect that.
<Taneb> (Knowing better than me here is a very low bar)
thc202 has joined #nixos
__monty__ has joined #nixos
<techieAgnostic> oh my
<techieAgnostic> nix is so nice to use with haskell
<thblt> Taneb: Use UEFI.
<vcunat> If either way doesn't boot, you may just repeat with the other one. You won't need to redo almost anything. (The installer should reuse the files, you'll just edit the config, etc.)
ambro718 has joined #nixos
<ambro718> Has anyone tried to get Xilinx Vivado running on NixOS? Not sure if I should even bother and just run it in a windows VM.
<{^_^}> [nixpkgs] @vcunat pushed 4 commits to staging: https://git.io/fAezD
<{^_^}> [nixpkgs] @vcunat merged pull request #43190 → libyaml: 0.1.7 -> 0.2.1 → https://git.io/fNvMo
<{^_^}> [nixpkgs] @vcunat merged pull request #45224 → harfbuzz: 1.8.2 -> 1.8.8 → https://git.io/fNjkJ
thetet has joined #nixos
thetet has left #nixos [#nixos]
<thblt> I need to compile a program in NixOS, but in a nix shell, but this program depends on dylibs which won't be available after nix-collect-garbage. Is there a common way to ask ./configure to statically link a given library?
<thblt> Context: I'm managing my Emacs extensions from Emacs, including mu4e, which uses a native component (mu) which itself depends on a dynamically linked libgmime
<LnL> going from a nix-shell to a build isn't that big of a step
<thblt> LnL: the problem is the package manager, I'd rather keep doing this from Emacs.
<LnL> and that will keep the paths alive
<LnL> but you're using nix-shell already
<thblt> LnL: Doesn't seem too hard indeed. I just need to create a default.nix, right?
philippD has joined #nixos
<thblt> For now I've configured my package manager to wrap build steps (./configure, make, etc) in nix-shell invocations
iyzsong has quit [Ping timeout: 272 seconds]
<LnL> that's basically the same then, unless you need something incremental
<thblt> LnL: thanks!
<LnL> use src = builtins.fetchGit ./.; or src = ./.; if it's a local project and then make the installPhase (or make install) install whatever you want to keep
<{^_^}> [nixpkgs] @vcunat opened pull request #45298 → libgit2: 0.26.0 -> 0.26.6 (security, fixes) → https://git.io/fAegJ
__monty__ has quit [Quit: leaving]
wucke13 has joined #nixos
<wucke13> Hello. What is the best solution to deploy a package, if the bin file does only work when ran from a folder inside of share of the given package?
revtintin has joined #nixos
timokau has joined #nixos
ambro718 has quit [Quit: Konversation terminated!]
Anton-Latukha has joined #nixos
__monty__ has joined #nixos
<__monty__> Is there a command to have something built on an ssh-substitutor? So it can then be ssh-substituted?
stanibanani has joined #nixos
stanibanani has left #nixos [#nixos]
Anton-Latukha has quit [Quit: Leaving.]
<{^_^}> [nixpkgs] @timokau opened pull request #45299 → gf2x: disable hardware-specific code → https://git.io/fAeg5
alexteves has joined #nixos
inquisitiv3 has joined #nixos
<{^_^}> [nixpkgs] @thefloweringash opened pull request #45300 → quaternion: Darwin support, fix dependencies and installed outputs → https://git.io/fAe2T
<inquisitiv3> What Wireguard derivation should one install? I get serveral matchees when I search with `nix search wireguard`. The two last hits seems promising (`nixpkgs.wireguard` and `nixpkgs.linuxPackages.wireguard`), but I'm not sure which of them I should install.
Fare has joined #nixos
<symphorien> there is a module iirc
<symphorien> it should install exactly what is needed
vidbina has joined #nixos
sm0rux has joined #nixos
<{^_^}> [nixpkgs] @timokau opened pull request #45301 → ntl: 11.2.1 -> 11.3.0 → https://git.io/fAe2c
<inquisitiv3> symphorien: Which of them should I install? I can't see any differencse. The only thing that differs is the attribute name.
<symphorien> I mean networking.wireguard.interfaces = ... will do the job itself, don't install any ot them: https://github.com/NixOS/nixpkgs/blob/release-18.03/nixos/modules/services/networking/wireguard.nix#L278-L285
pie__ has joined #nixos
<inquisitiv3> symphorien: Aah, thanks! I'll take a look.
pie_ has quit [Ping timeout: 240 seconds]
<{^_^}> [nixpkgs] @timokau merged pull request #45299 → gf2x: disable hardware-specific code → https://git.io/fAeg5
<{^_^}> [nixpkgs] @timokau pushed 2 commits to master: https://git.io/fAe22
silver has joined #nixos
<{^_^}> [nixpkgs] @eqyiel closed pull request #45191 → modules/alertmanager: add clusterAdvertiseAddress option → https://git.io/fNh5M
<wucke13> Is it considered good practice to always do mkdir build; cd build?
lorilan has joined #nixos
<lorilan> Hi I'm learning haskell and how to use it with a "nix + cabal setup". I'm trying to use ghc-mod but cannot enter the nix-shell with this context https://gist.github.com/lgirault/3ce09ffe06dd4e9496e138ad2666a31a
<lorilan> it says `Setup: Encountered missing dependencies: base >=4.5 && <4.12, template-haskell >=2.2 && <2.14` but adding template-haskell in the cabal file does not change anything and pining a specific version of base give me an infinite recusion during evaluation of the nix expr
alexteves has quit [Remote host closed the connection]
<lorilan> I asked the question on #haskell, without answers so far. But since it's a bit of "cross-concern" I thought I would try here
<{^_^}> [nixpkgs] @primeos pushed to master « quiterss: Set QT_PLUGIN_PATH (fixes a few problems) »: https://git.io/fAe2F
<__monty__> Does ghc-mod support ghc 8.4.3 even?
inquisitiv3 has quit [Ping timeout: 256 seconds]
michalrus has joined #nixos
<lorilan> I don't know, it really is my first steps with the language
<lorilan> I'm looking into it
<{^_^}> [nixpkgs] @xeji merged pull request #45296 → php71: 7.1.20 -> 7.1.21 → https://git.io/fAeuc
<{^_^}> [nixpkgs] @xeji pushed commit from @etu to master « php71: 7.1.20 -> 7.1.21 (#45296) »: https://git.io/fAeaf
<michalrus> Hello, my QEMU machine (more/less `nixos-rebuild build-vm`), after updating to 18.03, is freezing in <<< NixOS Stage 1 >>> on mke2fs.
<__monty__> lorilan: My advice? Don't even mess with ghc-mod yet. Just stick with ghcid if anything.
<michalrus> What could be the reason?
<{^_^}> [nixpkgs] @xeji merged pull request #45297 → php71: 7.1.20 -> 7.1.21 [stable backport] → https://git.io/fAeu8
<{^_^}> [nixpkgs] @xeji pushed commit from @etu to release-18.03 « php71: 7.1.20 -> 7.1.21 (#45297) »: https://git.io/fAeak
<__monty__> lorilan: I gave up on ghc-mod because it always lags on ghc and even then ime it's a brittle setup.
<{^_^}> [nixpkgs] @primeos pushed to master « androidStudioPackages.{dev,canary}: 3.3.0.4 -> 3.3.0.5 »: https://git.io/fAeaY
<oldandwise> can I 32-bit use a 64-bit for remote builds?
<{^_^}> [nixpkgs] @eqyiel opened pull request #45302 → modules/alertmanager: add clusterAdvertiseAddress option → https://git.io/fAea3
<Dezgeg> michalrus: does it have the commit dda74d9e50dbd8a412de743a53e9cfd585407342 ?
<michalrus> It’s the freshest 18.03 (today’s nixpkgs-channels), so if that one is in 18.03, then yes, but let me check!
<michalrus> So perhaps no.
<Dezgeg> maybe cherry-picking fixes it then
<lorilan> __monty__, ok … I'll try ghcid but it's really frustrating to give up like this. Until now I feel I've more tweaked nix conf than actually code
<michalrus> Lemme try.
<__monty__> lorilan: The problem is haskell's tooling. It's not on par with what you're used to probably. Oh, I just noticed you're specifying ghc861, that's WAY too new for ghc-mod afaik.
<__monty__> lorilan: I actually don't really miss any IDE features though. I get completion and near-instantaneous type-checking. That's nicer than anything eclipse gave me for java.
<lorilan> __monty__, ok, completion is kind of 50% of the feature I'm looking for (the other 50% being a good rename refactoring)
<{^_^}> [nixpkgs] @markuskowa opened pull request #45303 → molden: 5.7 -> 5.8.2 → https://git.io/fAeaw
<lorilan> by the way what's up with these number ? ghc 8.4.3 has been released last may, what is this 861 ?
<ongy> hm, is there something I need to set up, before I can use nixos containers?
<lorilan> ho ! does the 6 stands for june ?
<__monty__> lorilan: It's GHC 8.6.1 and that's *very* recent.
<__monty__> lorilan: No, previous version was 8.2.2, I'm not sure how GHC versioning works, it does seem to skip odd second numbers.
<ongy> Trying to start a container I defined (and the nixos-rebuild switch) I get in the log: /nix/store/iw231wi0fpckbghbv8wy018mmd3na1yk-container-init: line 28: /nix/var/nix/profiles/system/init: No such file or directory
<__monty__> It's not dates though.
<lorilan> __monty__, it's what I thought but then when I saw on https://www.haskell.org/ghc/ that the last released was 8.4.3 I thought it's a bit weird
<__monty__> lorilan: For the rename, I would just combine global with substitute in vim. Haven't done that yet though. Almost all the renames I want to do don't cross module boundaries and substitute is ok for that purpose.
<lorilan> __monty__, is it like a kind of an "early access" version ? ^_^
<lorilan> I talk about the rename from my scala experience but I don't know if I'll need it for haskell
inquisitiv3 has joined #nixos
<__monty__> Do you use vim? I can list all the extensions I use but it won't be very useful if you don't use vim...
<lorilan> yes I use it, but until now I only used it for small tasks. I use it to edit haskell files but still learning it m.p
<__monty__> lorilan: Stable branches are numbered x.y, where y is even. Apparently the second component is even *is* a thing.
<lorilan> :p
<lorilan> ok, noted !
<{^_^}> [nixpkgs] @eadwu opened pull request #45304 → Linux update on 8.18.18 → https://git.io/fAeaA
<{^_^}> [nixpkgs] @primeos pushed to master « gns3Packages.{server,gui}{Stable,Preview}: 2.1.7 -> 2.1.9 »: https://git.io/fAeap
<lorilan> __monty__, when you say stable branches, you're talking about ghc itself ?
<lorilan> __monty__, because the version seems to all have a x.y.z schema
<__monty__> lorilan: I think the extensions relevant to completion are neocomplete.vim and neco-ghc.vim, you'll also want either hasktags or fast-tags for navigation and then ghcid (this is pretty minimal, a couple niceties are vim-indent-objects, vim-easy-align, haskell-vim, vim-commentary, vim-bracketed-paste, vim-haskellConcealPlus, hlint-refactor-vim and vim-hoogle and I've been looking into vim-brittany, ale
<__monty__> and the like.)
<ongy> ok, disregard me. I derped and put "virtualization." before my `containers =` and stuff wasn't found
<ongy> so different question: is there a "strict" mode that validates all keys against matching something?
<__monty__> lorilan: Yes, it's always x.y.z but y is always even for releases, because they happen from a stable branch.
<__monty__> lorilan: ghc861 is just a name for the package without dots because those mean something in nix.
<lorilan> __monty__, so ghc7103 is the exception :p
<__monty__> No, that's 7.10.3 : )
<lorilan> __monty__, that's a big list of plugin ! thank you !
<lorilan> __monty__, erf … the absence of dots is misleading
<ongy> because that's not a decimal . ;)
<__monty__> In principle 861 should be 806, I guess nix conventions don't entirely align with ghc conventions.
<__monty__> 7.10.3 would be 710, ghc convention is xyy with 0 for the first y if it's single digit and just dropping the patch version.
<inquisitiv3> Is there a reason why it's called "Package names" when they're called "derivates" in Nix?
<lorilan> __monty__, well versioning scheme can become a hell as I recently discovered at work so I won't mind a little misalignment between the two :p
<__monty__> Do you mean derivations, inquisitiv3?
<__monty__> lorilan: Read up on the controversies around the PVP if you want versioning drama : )
<monad_cat> I'm trying to override a single package in haskellPackages, but trying to build `my-packages` errors out with infinite recursion. What am I doing wrong? https://www.irccloud.com/pastebin/Jp5f4j50/
<inquisitiv3> __monty__: Yep
<michalrus> Dezgeg: it works consistently, thank you. ♥
<michalrus> Dezgeg: perhaps then https://github.com/NixOS/nixpkgs/commit/dda74d9e50dbd8a412de743a53e9cfd585407342 should be cherrypicked onto 18.03?
<__monty__> monad_cat: Try super.haskell.packages... instead.
duncan^ has joined #nixos
<foldingcookie> what does it mean that when I do `nixos-rebuild switch` I get errors about the option `system.nixos.stateVersion` set in configuration.nix not existing?
mayhewluke has quit [Ping timeout: 256 seconds]
<__monty__> inquisitiv3: So the question was directed at my use of package name? I guess I could've said derivation name, just wanted to distinguish it from attribute name.
<foldingcookie> also, unrelatedly, how can I resolve name collisions that cause channels to be skipped?
<monad_cat> __monty__: in which place?
<foldingcookie> I think I must have misnamed something and now nix-channel --update does nothing
<Dezgeg> ok, I'll pick it
<__monty__> monad_cat: In your definition of haskellPackages.
mayhewluke has joined #nixos
<monad_cat> __monty__: so you mean `super.haskell.packages` instead of `super.haskellPackages`?
<inquisitiv3> __monty__: You used the term, but it isn't only you. E.g. if you search channels with `nix search [pattern]` you will get search matches that list "Package name" in addition to attribute name.
<{^_^}> [nixpkgs] @dezgeg pushed to release-18.03 « nixos/qemu-vm: Always add a virtio RNG device to the quest »: https://git.io/fAeVl
<__monty__> inquisitiv3: Ah ok. So the attribute name is basically the name in nix the derivation is assigned to, while the package name is an attribute in the record that describes the package. This is supposed to be more human readable but it turns out to be muy inconvenient because it means nix has to evaluate all the derivations just to list their names.
Fare has quit [Ping timeout: 272 seconds]
revtintin has quit [Quit: WeeChat 1.9.1]
<inquisitiv3> __monty__: And it's very confusing, because 1) according to Nix documentation there's no such thing as "package" in the Nix, and 2) new users is confused which name they should use and what they means.
Fare has joined #nixos
<__monty__> inquisitiv3: Yes, in nix you write expressions that describe packages and are evaluated into derivations which can be evaluated to build the package. 2) is an unfortunate side effect and harms nix's UX>
tychog has joined #nixos
<monad_cat> __monty__: `super.haskell.packages` ends up not having `override` on it
<__monty__> Yeah, you need to specify the ghc version so super.haskell.packages.ghcXYZ.override.
johanot has joined #nixos
<__monty__> haskellPackages is an alias for some haskell.packages.ghcXYZ, currently 8.2.2 is my guess but I'm not actually sure.
patrl has joined #nixos
<adisbladis> firefox
adisbladis has left #nixos ["ERC (IRC client for Emacs 26.1)"]
kolb has quit [Quit: ZNC 1.6.3+deb1ubuntu0.1 - http://znc.in]
<{^_^}> [nixpkgs] @wucke13 opened pull request #45305 → apmplanner2: init at 2.0.26 → https://git.io/fAewa
patrl has quit [Ping timeout: 272 seconds]
<lorilan> I have a following question on my haskell/cabal/nix setup. whene I try to run `ghcid` I run into a "<command line>: can't load .so/.DLL for: libz.so" error
<lorilan> now if I edit the LD_LIBRARY_PATH env var to include the path to said lib, it works fine
<ongy> so, I have xpra in my packages=[] for my user. But after `nixos-rebuild switch` it's not available to the user. Any idea how to debug? (also not as winswitch which afaik is the old name of the project)
<lorilan> but isn't the nix-shell supposed to handle this for me ?
<__monty__> The current haskellPackages is haskell.packages.ghc843 btw.
Anton-Latukha has joined #nixos
<__monty__> Oh, and 8.6.1 hasn't been released yet. Not sure why it's included in nixpkgs already.
<__monty__> Maybe for ghc devs using nix?
<monad_cat> __monty__: I have this now, but it complains that an attribute `compiler` is missing. I tried adding `compiler = super.haskell.packages.ghc822.compiler` at the same level as `overrides`, to no avail https://www.irccloud.com/pastebin/3KecuAKB/
<ongy> hmm, how can I get the user env up, after I su into my user?
<__monty__> monad_cat: When does it complain?
<__monty__> monad_cat: Oh, wait. it should be haskellPackages = super.haskell.packages...
<monad_cat> then it just goes to infinite recursion :(
<__monty__> You need to use the long form for the time override too.
<__monty__> And you might want to use ghc843, it's the current default in nixpkgs.
<monad_cat> I've just added `haskell.compiler = super.haskell.compiler` but it compiles about something else
<monad_cat> is there maybe a well-documented way of overriding a specific haskell package somewhere?
<__monty__> monad_cat: I have some overrides here: https://github.com/toonn/nix-config/blob/master/config.nix
<monad_cat> with `ghc843` it complains that `ghc822` is missing even though it's nowhere to be found in that file
<__monty__> Using overrides isn't the current best practice though. You might want to read up or look for examples of overlays.
<monad_cat> thanks __monty__, I'll adapt this and report with results
<monad_cat> oh? checking...
<Thra11> What's the significance of gcc-wrapper (as opposed to just gcc)?
goibhniu has quit [Ping timeout: 268 seconds]
MinceR has quit [Ping timeout: 255 seconds]
MinceR has joined #nixos
iyzsong has joined #nixos
<{^_^}> [nixpkgs] @7c6f434c pushed to master « mono514: init at 5.14.0.177 »: https://git.io/fAerH
oida has quit [Quit: byez]
Fare has quit [Ping timeout: 272 seconds]
<hyper_ch> can I give two users the same userid?
orivej has joined #nixos
<infinisil> hyper_ch: Nope
<hyper_ch> it works when I edited the passwd directly in debian/ubuntu... I could just give them the same user and group id and both could login with different passwords
<monad_cat> Hey again __monty__ :) I tried destilling your nix config down, but I'm still getting an infinite recursion :( https://www.irccloud.com/pastebin/uSuQYJHD/
<__monty__> monad_cat: Oh, you can drop the myHaskellPackages. That's for packages I'm developing. The overrides go in haskell843...{ overrides = <HERE> };
<{^_^}> [nixpkgs] @NeQuissimus pushed 4 commits to release-18.03: https://git.io/fAeol
<__monty__> monad_cat: I noticed you switched to myEnvFun instead of buildEnv. Did you read up on the difference? With a nix-shell I think you'll want to use buildEnv. I use the myEnvFuns so I can load my haskell env when I develop and don't have multiple haskell envs causing collisions.
<monad_cat> still infinite recursion :D https://www.irccloud.com/pastebin/21zKcuRz/
<{^_^}> [nixpkgs] @NeQuissimus merged pull request #45304 → linux: 4.14.64 -> 4.14.65 linux: 4.17.16 -> 4.17.17 linux: 4.18.2 -> 4.18.3 linux: 4.4.149 -> 4.4.150 linux: 4.9.121 -> … → https://git.io/fAeaA
<{^_^}> [nixpkgs] @NeQuissimus pushed 5 commits to master: https://git.io/fAeoE
<monad_cat> __monty__: I'll make it perfect once it builds
<monad_cat> still infinite recursion with `buildEnv` too
<{^_^}> [nixpkgs] @NeQuissimus pushed 4 commits to release-18.03: https://git.io/fAeoz
<__monty__> monad_cat: Try time = super.callHackage
<monad_cat> `error: attribute 'callHackage' missing`
iyzsong has quit [Quit: ZNC 1.7.0 - https://znc.in]
<__monty__> Oh, replace the second _ with super.
<monad_cat> if only there were typings on the nix lang :P
<monad_cat> infinite recursion then
iyzsong has joined #nixos
endformationage has joined #nixos
<{^_^}> [nixpkgs] @worldofpeace opened pull request #45306 → nixos/geoclue2: add ability to whitelist agents and append to config → https://git.io/fAeoV
<__monty__> Ok, let's do time = super.time, forget about callHackage for a second.
<monad_cat> it's building there
<monad_cat> then*
<__monty__> Can you do just callHackage or replace the first _ with self and do self.callHackage?
<monad_cat> undefined variable in the first case, infinite recursion in the second
graphene has quit [Read error: Connection reset by peer]
graphene has joined #nixos
<__monty__> I'm hoping someone more knowledgeable takes over : ) No idea why the callHackage call causes infinite recursion tbh.
iyzsong has quit [Quit: ZNC 1.7.0 - https://znc.in]
orivej has quit [Ping timeout: 260 seconds]
iyzsong has joined #nixos
<monad_cat> yeah, do you know where I could find it's source? Google doesn't help
<__monty__> pkgs/development/haskell-modules/make-package-set.nix
sm0rux has quit [Ping timeout: 256 seconds]
inquisitiv3 has quit [Ping timeout: 240 seconds]
asymmetric has joined #nixos
Fare has joined #nixos
<asymmetric> hi folks, i'm not sure i understand the difference between override and overrideAttrs
<Myrl-saki> asymmetric: .override overrides the arguments
<Myrl-saki> asymmetric: .overrideAttrs overrides the derivation
<asymmetric> i see
johanot has quit [Quit: leaving]
<Myrl-saki> I'm now curious where overrideAttrs is defined
<asymmetric> so basically, if i want to modify the derivation produced by build-maven.nix, to add buildInputs, i'd use overrideAttrs
<Myrl-saki> Ah, knew it.
<Myrl-saki> overrideAttrs is by mkDerivation itself
<Myrl-saki> While override is by callPackage
<Myrl-saki> Cool stuff. :)
<Myrl-saki> asymmetric: And yes, yes you would.
<asymmetric> where's mkDerivation defined btw? tried grepping but couldn't find it
<Myrl-saki> asymmetric: Just like how if you have local source code, you'd do `foo.overrideAttrs (oldAttrs: { src = ./.; })`
<Myrl-saki> I'm pretty sure I saw mkDerivation somewhere in stdenv.
alex`` has quit [Remote host closed the connection]
<Myrl-saki> ,locate stdenv/generic/make-derivation.nix
<{^_^}> Couldn't find any packages
<Myrl-saki> I have no idea how to use the bot.
<Myrl-saki> oldandwise: Yeah, x86-64 has permissions to build x86.
thekolb has joined #nixos
alex`` has joined #nixos
<infinisil> ,find make-derivation.nix
<asymmetric> thanks :)
<asymmetric> Myrl-saki: is the explanation in https://nixos.org/nixpkgs/manual/#sec-overrides unclear or am i thick?
<asymmetric> [override] is used to override the arguments passed to a function.
<asymmetric> The function overrideAttrs allows overriding the attribute set passed to a stdenv.mkDerivation call, producing a new derivation based on the original one.
sir_guy_carleton has quit [Quit: WeeChat 2.0]
<asymmetric> ah i think i see.. overrideAttrs overrides the argument of mkDerivation, which is an attribute set
<oldandwise> tried, `nix build --builders ssh://192.168.2.10 nixpkgs.cscope --max-jobs 0` won't connect
<{^_^}> [nixpkgs] @Infinisil merged pull request #44880 → nixos/nix-daemon: build-max-jobs, add support to auto → https://git.io/fN7Hy
<{^_^}> [nixpkgs] @Infinisil pushed 2 commits to master: https://git.io/fAe6L
<infinisil> oldandwise: I've never seen ssh:// work with Nix 2.0 :/
<infinisil> sphalerite also had some flaky experience with it
<oldandwise> error says: `error: unable to start any build; either increase '--max-jobs' or enable remote builds`
<infinisil> Oh yeah no that's a different erro
<infinisil> Oh or do you expect it to be built on the ssh:// host?
<infinisil> If it couldn't connect then it should throw an error I think
<oldandwise> yes, coz client is 32-bit slow machine
<clever> oldandwise: try ssh-ng:// ?
<oldandwise> 32-bit clien also had error saying `__divmoddi4` so i thought remote build could fix it?
<clever> oldandwise: also, try with sudo on the nix-build
<symphorien> oldandwise: see https://nixos.wiki/wiki/Distributed_build#Prerequisites to check if ssh works
<Dezgeg> the __divmoddi4 problem is a known one in 32-bit x86
<Dezgeg> it needs fixing in nixpkgs
<symphorien> also --builders only works with trusted users
<oldandwise> now, i get different error: error: getting status of '/home/john/nixpkgs.cscope': No such file or directory
<clever> oldandwise: nix build is a bit random with its arg parsing, try nix-build
<oldandwise> trie that.. still errors
<monad_cat> __monty__: I managed to fix it with above, and then using `hp.hackage-server`. Weird :P https://www.irccloud.com/pastebin/GOwAXuxh/
<clever> oldandwise: what exact command did you try with nix-build?
<oldandwise> tried, `nix-build --builders ssh://192.168.2.10 nixpkgs.cscope --max-jobs 0`
<clever> oldandwise: the syntax is different there
<clever> oldandwise: sudo nix-build --builders ssh://192.168.2.10 '<nixpkgs>' -A cscope --max-jobs 0
<infinisil> (Ohh, --builders != --option substituters, I was having trouble with the latter one)
<oldandwise> tried that... error is unable to open SSH connection to 'ssh://192.168.2.164': cannot connect to '192.168.2.164'; trying other available machines...
jtojnar has quit [Ping timeout: 240 seconds]
<clever> oldandwise: `ssh 192.168.2.164 nix-store --version`
oida has joined #nixos
<oldandwise> gave: nix-store (Nix) 2.0
<oldandwise> no password asked...
<clever> oldandwise: oh, but what about sudo ssh 192.168.2.164 nix-store --version
<clever> you may need to set the user in that cmd
<oldandwise> now, it keep asking for password, and password does not work.. won't let me in...
<clever> oldandwise: try with user@ip in the cmd
<clever> sudo changes the default user for ssh
<oldandwise> 'sudo ssh johnny@192.168.2.164 nix-store --version' asked for password, is accepted, and prints result 'nix-store (Nix) 2.0'
jtojnar has joined #nixos
<clever> ah, then your not using an ssh agent, so it looked in /root/.ssh/id_rsa instead of /home/johnny/.ssh/id_rsa
<clever> sudo nix-build --builders ssh://johnny@192.168.2.10 '<nixpkgs>' -A cscope --max-jobs 0
hiroshi- has joined #nixos
<oldandwise> unable to open SSH connection to 'ssh://johnny@192.168.2.120': cannot connect to 'johnny@192.168.2.120'; trying other available machines...
<clever> oldandwise: try starting ssh-agent, set the vars it showed, then ssh-add, then do the above cmd again
hiroshi has quit [Ping timeout: 248 seconds]
hiroshi- is now known as hiroshi
joebobjoe has joined #nixos
graphene has quit [Remote host closed the connection]
graphene has joined #nixos
<oldandwise> error has change. It's http://termbin.com/jrpg
iyzsong has quit [Ping timeout: 260 seconds]
joebobjoe has quit [Ping timeout: 240 seconds]
<clever> so the remote nix builder is now "working"
<clever> but the remote machine wont accept the storepath
<clever> a: add johnny to the trusted users on the remote machine
<clever> b: add your ssh key to root on the remote machine and switch to root
<clever> c: setup nix keypairs to sign builds, and configure the 2 machines properly
<clever> do one of the above to fix it
<oldandwise> k... doing option#1 in /etc/nixos/configuration.nix nix.trustedUsers = ["johnny"];
<tychog> how does one put a nix setup in git? I tried a git init in /etc/nixos/, but found it not to persist across rebuilds. use-case: annotating/viewing history, backing up / sharing my setup.
<clever> adding yourself to the trustedUsers on the local machine would also get rid of the need to use sudo
* oldandwise crossing fingers...
wrl has quit [Read error: Connection reset by peer]
JonReed has joined #nixos
magnetophon has joined #nixos
<magnetophon> I'm having trouble compiling some, but not all pkgs. I get: "error: while setting up the build environment: getting attributes of path '/etc/hosts': Permission denied". It's not a config issue, since I've build a really minimalistic config and tried from there, with the same result. the permissions of the file in question look ok. What can I try next?
aexoxea has quit [Quit: Goodbye!]
arjen-jonathan has joined #nixos
wrl has joined #nixos
<tychog> @magnetophon: try sudo?
<magnetophon> tychog: thanks, but I already tried using both my user and root acount. or am I misunderstanding you?
periklis has joined #nixos
<tychog> magnetophon: no, in that case I'm not sure either. fwiw I can stat /etc/hosts just find without sudo.
<tychog> btw /etc/hosts should symlink to /etc/static/hosts then to some /nix/store entry.
<JonReed> magnetophon: You can also reference your Github issue here ( https://github.com/NixOS/nixpkgs/issues/41651 ) so that people who can help you (I don't really know what's going on) have more info to deal with
<{^_^}> #41651 (by magnetophon, 10 weeks ago, open): error: while setting up the build environment: getting attributes of path '/etc/hosts': Permission denied
<gchristensen> inside a build, /etc/hosts isn't the same as your machine's /etc/hosts
bennofs has joined #nixos
<magnetophon> JonReed: thanks. I avoided that, since I made that issue into a bit of a mess: hard to get info out of
<LnL> gchristensen: yep :D
tychog has quit [Quit: WeeChat 2.1]
<magnetophon> tychog: mine goes: /etc/hosts -> /etc/static/hosts /home/me/myNixosConfig/hosts and that file has: Access: (0644/-rw-r--r--) Uid: ( 1000/ bart)
<magnetophon> so it looks like that's my issue, right?
<gchristensen> it won't be inside a nix build
joebobjoe has joined #nixos
<magnetophon> gchristensen: so you are sure that the solution is not to change the permissions of the hosts file in my home dir?
<gchristensen> inside a build, /etc/hosts isn't the same as your machine's /etc/hosts
<magnetophon> gchristensen: thanks. any suggestions on how to troubleshoot further?
Fare has quit [Ping timeout: 272 seconds]
<gchristensen> magnetophon: what is failing to build?
<magnetophon> gchristensen: for example when I try to update the version of ir.lv2
<LnL> ah, it's the fetchurl build that fails?
<LnL> that does use the actual /etc/hosts
<magnetophon> LnL: idk. how do I find out?
<LnL> the error log, is it hello-2.10 that failed or hello-2.10.tar.gz
roconnor has quit [Quit: Konversation terminated!]
<LnL> yeah, source in this case
<LnL> so your symlink to /home is the problem
<magnetophon> LnL: I did "chown root /home/me/nixosConfig/hosts" and also "chmod -R 777 /home/me/nixosConfig/hosts", but get the same result
<LnL> doesn't matter, /home/me isn't readable for other users
<magnetophon> LnL: 2 questions: how come this setup worked for so long for me then? how do I fix it? (I want hosts to point to a blacklist I can easily download and keep up to date, preferably inside my nixos config git repo)
<magnetophon> LnL: for now I'll rebuild with a default hosts file, and see if it fixes the issue. thanks!
<LnL> I don't know, I'd expect lots of things to break if you do that
<magnetophon> LnL: it works now!!!! great! I still don't understand why it didn't work after rebuilding with a minimal config though... must have been PEBCAC....
<oldandwise> thanks clever
<magnetophon> Many thanks LnL. Ive been strugling with this for over a month. Now I can go back to contributing to NixOS! :D
<magnetophon> LnL: are you LnL7 on github? I want to credit you in my issue closing remark.
Fare has joined #nixos
<LnL> yeah, that's me
<magnetophon> :)
bennofs has quit [Ping timeout: 276 seconds]
<{^_^}> [nixpkgs] @LnL7 merged pull request #45157 → oracle-instantclient: fixes dylib rpath on macos → https://git.io/fNh0n
<{^_^}> [nixpkgs] @LnL7 pushed 2 commits to master: https://git.io/fAeXf
endformationage has quit [Ping timeout: 240 seconds]
<LnL> magnetophon: fyi, you can do something similar to what you had before by making nix copy the file instead
<LnL> only difference is that you need to nixos-rebuild switch to update it
<magnetophon> LnL: that would make it even better imo. thanks!
rprije has quit [Read error: Connection reset by peer]
<LnL> just remove the quotes :)
rprije has joined #nixos
joebobjoe has quit [Ping timeout: 272 seconds]
<{^_^}> Channel nixpkgs-18.03-darwin advanced to https://github.com/NixOS/nixpkgs/commit/47b68dead3d (from 2 hours ago, history: https://channels.nix.gsc.io/nixpkgs-18.03-darwin)
philippD has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]
alexteves has joined #nixos
Havvy has quit [Ping timeout: 240 seconds]
Havvy has joined #nixos
mayhewluke has quit [Ping timeout: 260 seconds]
mayhewluke has joined #nixos
semilattice has joined #nixos
revtintin has joined #nixos
hax404 has joined #nixos
vmandela_ has joined #nixos
johnw has joined #nixos
<{^_^}> [nixpkgs] @Infinisil merged pull request #45292 → ecos: init at 2.0.6 → https://git.io/fAelR
<{^_^}> [nixpkgs] @Infinisil pushed 3 commits to master: https://git.io/fAe1Y
johnw has quit [Ping timeout: 276 seconds]
periklis has quit [Remote host closed the connection]
<{^_^}> [nixos-artwork] @samueldr merged pull request #33 → Adds a revision of the logo SVG file with matching font → https://git.io/fNhMh
<{^_^}> [nixos-artwork] @samueldr pushed 3 commits to master: https://git.io/fAe1l
mayhewluke has quit [Ping timeout: 272 seconds]
Wizek has joined #nixos
<Wizek> hello
<Wizek> anyone knows how to use `nix-shell -p 'haskell.packages.ghc843.developPackage { root = ./.; src = ./.; }' -j4 --run 'cabal new-repl'`?
<mpickering> I think the `new-repl` here is the error
<Wizek> It seems to complain if I remove src attr or keep it
<mpickering> what's the error exactly?
<Wizek> 'developPackage' at /nix/store/cc5i4y0hv1y8igqscy568sxxc0s4zlyg-nixpkgs-18.09pre149415.8395f9aa85e/nixpkgs/pkgs/development/haskell-modules/make-package-set.nix:207:7 called with unexpected argument 'src', at (string):1:94
<mpickering> and if you remove it?
<Wizek> sec and copying
<Wizek> I remember having used this some months ago, not sure if it was with or without src
<clever> Wizek: if you read the file listed in the error, it only accepts root, source-overrides, overrides, modifier, and returnShellEnv
<Wizek> clever: yup, I got as far, but something somewhere then is missing src
<clever> can you paste that error?
<mpickering> It's above
<mpickering> "variable $src or $srcs should point to the source"
<Wizek> clever: yup, it's above
<clever> oh
<clever> use -E not -p
<clever> -p will build that (shell only) derivation, and then put it into PATH
<Wizek> trying
<Wizek> clever: that seems to have done the trick! Thanks!
<{^_^}> [nixpkgs] @Mic92 merged pull request #45289 → vscode: Wrap with GAppsHook → https://git.io/fAeZf
<{^_^}> [nixpkgs] @Mic92 pushed 2 commits to master: https://git.io/fAeMI
<Wizek> This is what ended up working `nix-shell -E '(import <nixpkgs> {}).haskell.packages.ghc843.developPackage { root = ./.; }' -j4 --run 'ghcid -c "cabal new-repl" -T main'`
<{^_^}> [nixpkgs] @Mic92 merged pull request #45283 → libmsgpack: 2.1.5 -> 3.0.1 → https://git.io/fAe3O
<{^_^}> [nixpkgs] @Mic92 pushed 2 commits to master: https://git.io/fAeMz
<clever> what does -T do?
<clever> ah, just runs main in ghcid
<clever> ghci*
magnetophon has quit [Remote host closed the connection]
mayhewluke has joined #nixos
aanderse has joined #nixos
<{^_^}> [nixpkgs] @Mic92 merged pull request #45293 → libqmatrixclient: enable Darwin platform → https://git.io/fAelp
<{^_^}> [nixpkgs] @Mic92 pushed 2 commits to master: https://git.io/fAeMy
<Wizek> clever: yup. Btw, the woes seem not to be over. As it turns out developPackage doesn't give me cabal on path that lead to an abi error likely due to version mismatch. I tried asking nix for cabal with `-p cabal` but apparently that's mutex with `-E`. Any ideas?
<aanderse> anyone know bundix well? i'm looking at some ruby stuff that wants me to call bundle like so:
<aanderse> bundle install --without development test
<gchristensen> manveru does
<clever> Wizek: nix gives you cabal as a library, not a binary
<aanderse> if i pass the --without development test then the Gemfile skips over some deps
<clever> Wizek: so you have to compile and run Setup.hs and use that as your cabal binary
<aanderse> i'm trying to figure out how to get bundix to skip those deps as well
<clever> Wizek: you can also just runhaskell Setup.hs
<aanderse> it appears bundix doesn't pass args onto bundler
joebobjoe has joined #nixos
<Wizek> clever: I'd rather use ghcid somehow, or is that not very possible?
<clever> Wizek: i use ghcid -c "runhaskell Setup.hs repl"
<Wizek> checking
<clever> new-repl may also work
<{^_^}> [nixpkgs] @Mic92 merged pull request #45277 → hugo: 0.46 -> 0.47 → https://git.io/fNjH1
<{^_^}> [nixpkgs] @Mic92 pushed 2 commits to master: https://git.io/fAeDf
ryanartecona has joined #nixos
<Thra11> Does nixpkgs have any rules or guidelines for naming patch files?
aanderse has quit [Ping timeout: 276 seconds]
hotfuzz has joined #nixos
<manveru> oh, he already left?
Havvy has quit [Ping timeout: 260 seconds]
<symphorien> Thra11: if it is a patch for a CVE, it should be named after the CVE
<symphorien> otherwise, I don't know
<samueldr> Thra11: not sure, but if I were to write one up, it would be "make them using git format-patch" as you can also add metadata (commit message) to them that way https://git-scm.com/docs/git-format-patch
<samueldr> * when possible
<clever> i just do `git diff > reason.patch
<clever> or `diff -ru` if i dont have a .git
<gchristensen> Anton-Latukha: hey
Drakonis has joined #nixos
endformationage has joined #nixos
ryanartecona has quit [Quit: ryanartecona]
revtintin has quit [Quit: WeeChat 1.9.1]
<Thra11> In this case, I've just copied a patch from archlinuxarm's build sources, so it does already have a name. Not sure if I should get nix to fetch it from archlinuxarm instead.
<clever> Thra11: there is a pkgs.fetchpatch function
<clever> works a lot like fetchurl
<Thra11> Yes, I think I've seen it in use.
<Thra11> I guess it depends whether there's a stable url to download the patch from
graphene has quit [Remote host closed the connection]
<clever> yeah
<clever> and if its a nix only patch
graphene has joined #nixos
b has quit [Quit: Lost terminal]
Havvy has joined #nixos
vmandela_ has quit [Quit: Leaving]
kisik21 has joined #nixos
<kisik21> An interesting idea struck me - is it possible to bootstrap LFS (in broad sense - Linux with all packages built from scratch) from NixOS? Assuming FHS hierarchy in the resultant system.
aarvar has joined #nixos
<clever> kisik21: it might be possible to create the /tools/ set from nixos, just follow the usual directions and make sure cc-wrapper doesnt inject too much junk
<kisik21> Maybe just, y'know, somehow bootstrap /tools/-like hierarchy with Nix?
hyper_ch has quit [Remote host closed the connection]
<kisik21> I'm too lazy to go through all of instructions of building the intermediate toolchain
<kisik21> It
<kisik21> it's too long
<kisik21> and too manual
<kisik21> oh, my musl cross-compiler is ready
<clever> kisik21: you can probably just bind-mount /nix into the chroot and use nix-shell -p as a fake /tools
<clever> /tools is more about having a full toolchain that doesnt live in /usr/
<clever> and nix already counts as that
<kisik21> That would be interesting
<kisik21> and my musl cross-compiler doesn't work...
<kisik21> it isn't in the path of nix-shell
<kisik21> seems like I'm doing it wrong
<kisik21> nix-shell --arg crossSystem '{ config = "x86_64-unknown-linux-musl"; }' '<nixpkgs>' -p stdenv
<{^_^}> [nixpkgs] @primeos opened pull request #45309 → iputils: 20161105 -> 20180629 → https://git.io/fAe93
<kisik21> I want to modify this command so x86_64-unknown-linux-musl toolchain will magically appear in the path, what did I do wrong?
Drakonis has quit [Remote host closed the connection]
<clever> (import <nixpkgs> { crossSystem = (import <nixpkgs/lib>).systems.examples.musl64; }).hello
<clever> kisik21: this is what i have tested in the past
<clever> try just putting that directly into nix-shell -E '....' ?
Drakonis has joined #nixos
hyper_ch has joined #nixos
<oldandwise> i'm keep on getting this `__divmoddi4` error even if i do remote build. I'm only interested in a command line cli of octave. How can I do this?
<kisik21> clever: works.
<symphorien> oldandwise: if you eval on 32 bit, then the derivation is 32 bits and the remote builder will build 32bit even if it is 64bits capable
<kisik21> Would be nice to get rid of hello part though.
<clever> kisik21: you can also do .mkDerivation { name = "name"; }
<oldandwise> if the `__divmoddi4` is in GUI, ... i only want cli
graphene has quit [Remote host closed the connection]
<symphorien> oldandwise: (octave.override{qt=null;})
graphene has joined #nixos
<symphorien> I guess the configure script will give up on building the CLI in this case
<symphorien> the GUI*
<oldandwise> if this `divmoddi4` is going to be fixed soon..i'll just wait
<symphorien> this is a 32bits problem right ?
<oldandwise> seems so...
<symphorien> so I am not sure it will e ever fixed
<symphorien> there are less and less 32 bits users and 32 bits is increasingly broken
acarrico has quit [Ping timeout: 272 seconds]
<symphorien> unless you are motivated enough to fix it, of course :)
acarrico has joined #nixos
Fare has quit [Ping timeout: 272 seconds]
<kisik21> If I haven't enabled nixpkgs.allowUnfree, would my system be perfectly free software, including every part that is coming from NixOS? (kernel and firmware included)
<emily> depends how high your standards are
<emily> the stock kernel isn't linux-libre, for instance
<emily> you might be interested in GuixSD, which is based on Nix but much more GNU-y?
<kisik21> emily: are there linux-libre packages?
<kisik21> emily: also GuixSD is not nix-based, despite sharing similar concepts
<gchristensen> GuixSD is guaranteed to be all-free, since they don't even allow unfree software to be available
<emily> I don't think so, but I don't know for sure.
<emily> kisik21: Guix uses Nix under the hood, from my understanding.
<emily> not Nix-the-language.
<emily> Nix-the-software/store-model.
<gchristensen> I wonder why we don't have linux-libre available
<gchristensen> or if the reason is just, nobody interested in doing so has done it
<kisik21> Probably no one ever bothered to write a derivation
<kisik21> I've never used linux-libre before...
<jeaye> emily: Guix uses all of its own code.
<emily> jeaye: oh, okay. sorry for my error.
<emily> it still uses the Nix store model, though, right?
<gchristensen> yeah, and they used to use nix-daemon
<kisik21> Yeah, but it's /gnu/store
<jeaye> It uses the same principles, but that's all, emily.
<kisik21> They like the brand
<emily> gchristensen: I guess the only real reason to care about linux-libre is ideological, and so anyone who would package it would probably be dissatisfied with nixpkgs' attitude to licensing in general?
silver_ has joined #nixos
rprije has quit [Remote host closed the connection]
rev_strangehope has quit [Ping timeout: 272 seconds]
<emily> it tends to only show up in the kind of distros that get FSF approval, I feel
<Dezgeg> what does linux-libre even provide if you simply don't install any firmware packages?
<kisik21> did I install any firmware packages?
<kisik21> I dunno if I am!
<{^_^}> [nixpkgs] @joachifm merged pull request #45197 → dolphinEmuMaster: 2018-07-22 -> 2018-08-17 → https://git.io/fNhbJ
<{^_^}> [nixpkgs] @joachifm pushed 2 commits to master: https://git.io/fAeHT
rprije has joined #nixos
<symphorien> even without firmware, the linux source contains some blobs whose license is at best unclear
<jeaye> Dezgeg: Linux itself comes with binary blobs.
silver has quit [Read error: Connection reset by peer]
<clever> Dezgeg: ive noticed that some of my wifi drivers have support for open-source firmware, at least in terms of loading it, but ive never actually found said firmware
<gchristensen> you have to explicitly turn on unfree firmware (hardware.enableRedistributableFirmware)
<Dezgeg> you can do ls -la /run/current-system/firmware/
rev_strangehope has joined #nixos
<emily> Dezgeg: Linux itself has some blobs inside
<Dezgeg> didn't all the firmware blobs get removed from latest versions?
<emily> I think the argument is that they're tiny enough that they're legally unproblematic or something
lorilan has quit [Quit: Leaving]
<emily> it's blobs that aren't part of the firmware stuff
<emily> that are the problem
<kisik21> Dezgeg: a bunch of files from /nix/store/5wfhbvai8qnrirf1vgxqks7vxg8qpj5i-firmware-linux-nonfree-2018-03-20...
<emily> aiui
<gchristensen> great!
acarrico has quit [Ping timeout: 272 seconds]
<gchristensen> kisik21: set: hardware.enableRedistributableFirmware = false;
<kisik21> I wonder if it will break my GPU or WiFi...
<kisik21> would be funny to recover!
<kisik21> I know, we have rollback
<clever> kisik21: select an older generation, yeah
<kisik21> the best part about NixOS? if you haven't broke the bootloader, everything else is recoverable.
<Dezgeg> right, there still seem to be some sort of microcode files like drivers/gpu/drm/nouveau/nvkm/subdev/pmu/fuc/gf100.fuc3
Dagger has joined #nixos
<kisik21> 01:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] RV730/M96 [Mobility Radeon HD 4650/5165]
<kisik21> Does anyone have this thing running without unfree firmware?
<kisik21> it uses radeon kernel module, FYI
<kisik21> (is it ok to ask these things here though?)
<gchristensen> I don't see why not
<Dezgeg> based on http://linux-libre.fsfla.org/pub/linux-libre/releases/4.6.1-gnu/deblob-4.6 they drop a bunch of radeon stuff
asymmetric has quit [Ping timeout: 244 seconds]
<Dezgeg> I don't know for how recent cards/chips that is
<Dezgeg> I guess for both old and new, since both 'radeon' and 'amdgpu' are mentioned
<kisik21> means I have good chances of not breaking my system then
<{^_^}> [nixpkgs] @Infinisil pushed 2 commits to master: https://git.io/fAeH0
<{^_^}> [nixpkgs] @Infinisil merged pull request #45149 → riot-web: 0.15.7 -> 0.16.0 → https://git.io/fNhCD
sm0rux has joined #nixos
<kisik21> how does the kernel build infrastructure works even? where are the packages, where is the source? it doesn't look like a normal derivation to me!
<kisik21> `nix edit nixpkgs.linuxPackages.kernel`
<Dezgeg> yes it's split into quite many pieces
<Dezgeg> sources are in e.g. pkgs/os-specific/linux/kernel/linux-4.14.nix
<kisik21> Oh, that looks like something one can build on
<kisik21> Let's say I've modified the derivation to download sources from other place. How to try to build the kernel with default config?
<kisik21> You can assume I have an overlay, if it's needed.
<kisik21> (and how to write it to an overlay anyway?)
<kisik21> s/write/add/
<symphorien> easiest way is nixos-rebuild -I nixpkgs=/path/to/clone
<kisik21> symphorie: I don't have a checkout of nixpkgs with my kernel added in.
<kisik21> I have a file and I have an overlay and I don't know how to integrate it with an overlay...
<symphorien> well this I don't know
<kisik21> where to inject the super.callPackage os-specific/linux/kernel/linux-libre-4.14.nix? and yes, I have my overlay hierarchy based on nixpkgs' one... if one day I'll decide to push my changes for the whole world to see, it will be easier to integrate.
<kisik21> should I even use callPackage?
<kisik21> Dezgeg: how the source package you mentioned gets called? nixpkgs.kernelPackages.linux_something?
<Dezgeg> it is linux_4_14 in all-packages.nix IIRC
<kisik21> just linux_4_14?
<kisik21> no kernelPackages?
<Dezgeg> and linuxPackages_4_14 is built from the linux_4_14
<kisik21> so I would have to copy half of nixpkgs to my overlay?
<kisik21> let me see all-packages.nix
<kisik21> okay, so it has the package in top-level
<kisik21> and then...
<kisik21> mmhm
<kisik21> I got it
<kisik21> What does the myriad of patches for linux-4.14 in nixpkgs do? Are these critical to have a stable system?
neonfuz has quit [Quit: WeeChat 1.9]
aarvar has quit [Quit: Leaving.]
sm0rux has quit [Ping timeout: 256 seconds]
worldofpeace_ has joined #nixos
<{^_^}> [nixpkgs] @Thra11 opened pull request #45310 → vim-qml: init at 2018-07-22 → https://git.io/fAeQg
worldofpeace has quit [Ping timeout: 260 seconds]
vcunat has quit [Ping timeout: 256 seconds]
<{^_^}> [nixpkgs] @Thra11 opened pull request #45311 → vim-lastplace: init at 2017-06-13 → https://git.io/fAeQw
ambro718 has joined #nixos
<ambro718> Hi. I am trying to move a nixos installation to a new disk. I set up partitions and copied boot and root filesystems (with rsync -aHAX), then used nixos-enter to get into a chroot, update the configuration file with new drive paths, and tried nixos-rebuild boot. But it gives an obscure error: https://paste.ubuntu.com/p/rwdW3fWtDQ/
Anton-Latukha has quit [Quit: Leaving.]
slyfox has quit [Ping timeout: 240 seconds]
<ambro718> Am I doing something wrong? How is this supposed to be one? Note that just using nixos-install from the host system would not be good as it would use the wrong nixpkgs and it would download lots of unnecessary things because it would use the host's nix store.
<symphorien> try disabling sandbox
<ambro718> how?
<kisik21> something like --option no-sandbox? i'm not sure
<symphorien> edit /etc/nix/nix.conf and put here sandbox=false
<Dezgeg> nixos-install would be the solution, and it has been fixed to not use the host store
<Dezgeg> I don't know if it handles NIX_PATH or -I nixpkgs sanely, though
<kisik21> should
<kisik21> nothing prevents it from handling it
<Dezgeg> yep, it should be fixed of course if it doesn't
<ambro718> Dezgeg: so how do I do it? Do I need to specify NIX_PATH relative to the mouted root?
init_6 has quit [Ping timeout: 240 seconds]
<ambro718> BTW I can't edit /etc/nix/nix.conf, it's unwritable in nix store. Is there a command line option?
arjen-jonathan has quit [Ping timeout: 255 seconds]
<Dezgeg> I suppose if you try it and get it wrong it will warn and complain
<Dezgeg> maybe ideally it'd use the host store for instantiation so all paths were as if they were outside the chroot
<ambro718> How to I pass nix-build options to nixos-rebuild?
joebobjoe has quit [Ping timeout: 268 seconds]
<kisik21> ambro718: the same way? IIRC almost all nix and NixOS tools share their command line options
<ambro718> Dezgeg: how is using the host store ideal? Using the existing store in the installed system seems much more appropriate because it happens to have almost everything needed.
Fare has joined #nixos
<Dezgeg> instantiation would only create the small .drv files
<kisik21> ambro718: last time I was forced to reinstall my system, it used EVERYTHING from target store. No downloads.
<Dezgeg> but yes actual copying would happen there
semilattice has quit [Ping timeout: 244 seconds]
slyfox has joined #nixos
abueide has joined #nixos
<ambro718> okay "--option build-use-sandbox false" solved the error
drakonis__ has joined #nixos
jtojnar has quit [Ping timeout: 244 seconds]
jD91mZM2 has quit [Quit: WeeChat 2.0]
<ambro718> seems like the install went fine with nixos-rebuild from chroot now, hope it actually works else I'll have to reboot and setup all the cryptsetup and mounts again
Drakonis has quit [Ping timeout: 256 seconds]
<kisik21> ambro718: If you write a good configuration.nix, nixos-install is the only thing you'll need to do. And most configuration.nix files are good by default.
<kisik21> (for now, definition is simple - the bad one won't even be built)
<{^_^}> Channel nixpkgs-unstable advanced to https://github.com/NixOS/nixpkgs/commit/b9fafcfc302 (from 12 hours ago, history: https://channels.nix.gsc.io/nixpkgs-unstable)
<ambro718> kisik21: what about fixups of /dev paths in hardware-configuration.nix? And I still need to fixup the resume_device and resume_offset stuff for hibernation.
<ambro718> the bootloader install device had to be updated to the new /dev/disk/by-id/.. path and the filesystems to new /dev/disk/by-uuid/ paths
abueide has quit [Read error: Connection reset by peer]
<kisik21> ambro718: No hibernation stuff whatsoever, swap on LVM on LUKS, surprisingly worked out of the box.
<kisik21> oh, now, resumeDevice=/dev/mapper/nixos-swap
<kisik21> yeah
thetet has joined #nixos
<kisik21> wait, there's an option for that?
<kisik21> I've configured it via boot.kernelParams!
<kisik21> did I do it right?
<ambro718> kisik21: your resume device is pointing to what, a cryptsetup thing?
sm0rux has joined #nixos
<ambro718> kisik21: I have my swap inside the root filesystem as a file, for the simple reason that I only have one password to enter at boot.
<ambro718> though are you solving that by putting LVM inside cryptsetup?
<kisik21> Yeah. One flash drive with a second keyfile partition to rule them all.
<kisik21> And a passphrase - just in case.
<ambro718> ah ok, good to know. I'm sticking with no lvm and swap file for now since I know this works for me :)
trevthedev has joined #nixos
<kisik21> My setup waits 10 seconds for a special flash drive to be connected to the right side of the computer (for unknown reasons, this is important!), and fall backs to a passphrase if a keyflash is not available. (Usually it is, because it's inserted at boot)
<{^_^}> [nixpkgs] @Thra11 opened pull request #45312 → qtcreator: enable on aarch64 and armv7 → https://git.io/fAe7N
foldingcookie has quit [Ping timeout: 260 seconds]
troydm has quit [Ping timeout: 244 seconds]
<ambro718> I think the flash drive is just convenience, it doesn't solve the "evil maid" issue AFAIK.
<kisik21> I have no maids at home, sadly.
<trevthedev> same
<kisik21> And my relatives are too incompetent to perform a real evil maid attack...
<ambro718> i.e. someone may modify the boot code on your system without you knowing and then when you boot up and insert key / passphrase they have it
<kisik21> I know!
<ambro718> :)
<{^_^}> [nixpkgs] @TravisWhitaker opened pull request #45313 → ghc: Use 8.2.2 binary for boot, no parallel builds on aarch64. → https://git.io/fAe7p
<kisik21> But really, no one at my home even knows what is an evil maid attack
<{^_^}> [nixpkgs] @TravisWhitaker closed pull request #45166 → ghc: Use 8.2.2 binary for boot, no parallel builds on aarch64. → https://git.io/fNh1S
<ambro718> ok time to reboot :)
<kisik21> good luck!
ambro718 has quit [Quit: Page closed]
ambro718 has joined #nixos
<kisik21> ambro718: did it work?
<ambro718> Hah it booted perfetly :) Doesn't usually work on the first try.
<kisik21> That's NixOS, it's normal
aarvar has joined #nixos
<kisik21> It always works on a first try
<tilpner> Unless you're debugging bootloaders or drivers
<kisik21> I'm talking about end-user experience
<tilpner> s/debuggin/configuring/
<kisik21> if you're working with bootloaders, it always fails at least one time
<kisik21> tilpner: well, maybe
<kisik21> never tinkered with it beyond setting a few boot options
jtojnar has joined #nixos
<ambro718> oh one thing I've learned is to always boot in EFI mode, or you can't install EFI bootloader. Seems so backward.
<kisik21> no EFI computers except an Intel Atom tablet.
<kisik21> And y'all know how broken Intel Atom is
<ambro718> all new computers have EFI
<kisik21> I know.
<kisik21> Wanna have one. EFI seems cool.
<kisik21> No need for bootloaders at least...
<ambro718> I don't think you would see any difference.
mayhewluke has quit [Ping timeout: 276 seconds]
<kisik21> EFI Shell makes a difference.
<kisik21> Seems like it can even boot Linux without GRUB, if you remember the commands.
joebobjoe has joined #nixos
<kisik21> I'm starting to think this is my last hope to boot that Atom tablet
<ambro718> if you say so :) I just need it to load my grub.
mayhewluke has joined #nixos
trevthedev has quit [Quit: WeeChat 2.0]
<kisik21> GRUB never got past "Loading linux and initrd..." for me on EFI...
thetet has left #nixos [#nixos]
<kisik21> I'm starting to think something is wrong with the handover to Linux kernel
aminb has quit [Quit: WeeChat 2.2]
<kisik21> oh, what bootloader does NixOS install USB uses on EFI/
<ambro718> grub2
<ambro718> oh sorry no idea
<ambro718> but it probably is grub2 IIRC
<kisik21> No wonder it doesn't work then. GRUB sadly never worked for me, being a good tool everywhere else though.
aminb has joined #nixos
<kisik21> oh, my new linux-libre derivation is starting to build
<kisik21> nice! expecting something awesome to land in the $out dir
<{^_^}> [nixpkgs] @LnL7 merged pull request #44630 → gtk3: add x11Support for darwin → https://git.io/fNDFg
<{^_^}> [nixpkgs] @LnL7 pushed 4 commits to master: https://git.io/fAe5c
<ambro718> I learnt all I need to get the efi grub working is 1) make a partition with bios_grub flag (that is actually required because for some reason the nixos setup also wants to install bios stuff) 2) boot.loader.grub.efiInstallAsRemovable = true; which is good if you want to move disk to another computer and expect it to boot
<kisik21> what does the second option do exactly to grub? looks promising
<ambro718> kisik21: I think the default setup without this "just" adds a boot entry to the EFI configuration telling it what to run from which partition etc. That option also does some installation on the disk itself allowing an EFI firmware to detect your bootable thing automatically and let you boot it even if it has no advance configuration of that thing.
aarvar has quit [Ping timeout: 265 seconds]
<kisik21> oh I see
<ambro718> so if I move disk to a different computer, I will supposedly be able to boot from it (if not otherwise then via boot override)
<kisik21> BOOTX64.EFI
<kisik21> check for file named like that in your ESP, I bet it would be present there
<ambro718> yes
<kisik21> ok
<kisik21> cool!
rtjure has quit [Ping timeout: 272 seconds]
<clever> ambro718: bios_grub is only needed if you forgot to set boot.loader.grub.device = "nodev";
<clever> ambro718: youve accidentaly configured nixos to both legacy and efi at the same time (a common mistake)
<kisik21> but doesn't it allow for hybrid drive?
<kisik21> something bootable on both BIOS and UEFI?
<ambro718> clever: ah, good to know. I don't mind though.
<clever> kisik21: you can also just do that on purpose, to get exactly that
<kisik21> Would be a nice trick to show on parties.
<kisik21> And make a recovery USB drive.
<kisik21> Or both at the same time.
<clever> install as removable is also good for usb drives
<clever> boot.loader.grub.efiInstallAsRemovable
<kisik21> kinda obvious, *removable*!
<clever> this configures grub to use the path that is used on removable drives
rtjure has joined #nixos
<clever> which the uefi firmware will load without having the efivars setup
<kisik21> also Windows likes to clobber it
<kisik21> so no Windows on that drive
<kisik21> Windows protection
<kisik21> sounds like it's against a rootkit
<kisik21> is Windows hating allowed here?
arjen-jonathan has joined #nixos
<clever> kisik21: as long as you also bash darwin/apple? :P
<kisik21> Ave NixOS! who needs darwin anyway when we have NixOS
<clever> :D
<kisik21> seriously, I can't imagine my life without Nix now
<kisik21> It's too cool to stick it on a shelf
<kisik21> It almost became my religion
<kisik21> but I think it's more of a topic for #nixos-chat, isn't it?
<clever> probably
<kisik21> ok, going on-topic - I have some progress with linux-libre. It's building right now, 9 workspaces away from irssi window.
<ambro718> just yesterday I set up a coworker with vim+youcompleteme in centos via nix in 10 minutes, after he has already been compiling various llvm stuff for hours :(
Anton-Latukha has joined #nixos
<clever> ambro718: i often scp my vim.nix file to another machine, nix-build it, and then just manually jam it into $PATH
<ambro718> people just don't realize how useful nix is, you can get almost any piece of SW running on almost any linux distro.
<kisik21> clever, ambro718: I just use home-manager.
<clever> ambro718: the biggest problem ive had, is that f7/f8 dont work, because that machine had keybinds in screen for f7/f8, so vim never got them, lol
<clever> kisik21: in the above case, i wanted to use my vim, without changing the defaults for other users on the machine
<ambro718> I don't use vim myself so I wouldn't know the details, I know they were happy with it :)
<kisik21> home-manager doesn't change defaults for other users on the machine. If you don't have your users sharing one account, obviously
<ambro718> home manager seems interesting but I see a giant warning about experimentalness :)
<clever> kisik21: shared user on a shared machine
<kisik21> ambro718: oops! seems like I haven't noticed it
<kisik21> no, really, it's working too good to be beta
<kisik21> or alpha
<kisik21> Oh, I've had a question for a long time! How to add more perl packages to irssi closure? Need some things for trigger.pl to work.
<JonReed> Is `nix repl` available by default on NixOS or it comes from some package?
<kisik21> nix-repl uses Perl, so it's a separate package.
<kisik21> or wait
<clever> JonReed: nix repl is part of nix 2.0
<kisik21> oh, nice, what's nix-repl package is then?
<JonReed> I used "nix-repl" before, now switched to unstable and it says that "nix-repl" package no longer exists
<kisik21> so it got integrated!
<kisik21> That's good
bennofs has joined #nixos
<kisik21> One of my most beloved tools is nix-shell though. Trying out apps (even complex ones, like nixpkgs.gnome3.evolution) and not clobbering the environment.
drakonis__ has quit [Read error: Connection reset by peer]
kisik21 has quit [Quit: leaving]
<ambro718> mate desktop is packaged now in nixos-unstable right?
joebobjoe has quit [Ping timeout: 276 seconds]
<ambro718> is nixos-unstable reasonably stable now? I'd like to give it a try before the 18.09 release.
goibhniu has joined #nixos
Drakonis has joined #nixos
<gchristensen> no, nixos-unstable will be branched in to release-18.09 and then will be stabilized there
<ambro718> ah ok, I'm trying it own anyway
<gchristensen> ok
<ambro718> otherwise I always end up with issues with the new release and then fixes need to be backported :)
rprije has quit [Remote host closed the connection]
<gchristensen> that is the point of the stabilization process! :)
Thra11 has quit [Ping timeout: 240 seconds]
rprije has joined #nixos
<JonReed> `nix-prefetch-url` now also comes by default with NixOS on unstable, right?
acarrico has joined #nixos
<clever> JonReed: its part of the nix package
<clever> so its available on anything that has nix
<JonReed> clever: Ok. Thanks. That's great.
<infinisil> ,locate bin nix-prefetch-url
<{^_^}> Found in packages: nix1
__monty__ has quit [Quit: leaving]
<infinisil> Ah, it doesn't display nixStable, because the package name is the same, and it prefers shorter attribute names
<ambro718> I figured that from some point on, "command-not-found" works also if you don't use channels. How does that work?
aarvar has joined #nixos
<JonReed> locate bin google-chrome-stable
<joepie91> JonReed: you're missing a comma at the start
<JonReed> , locate bin google-chrome-stable
<{^_^}> Couldn't find any packages
<joepie91> huh? that should not have worked
<joepie91> ... infinisil?
<joepie91> the bot is space-tolerant?
<LnL> unfree
<infinisil> Yeah, nix-index can't know about unfree stuff because hydra doesn't build it
ambro718 has quit [Quit: Konversation terminated!]
<JonReed> Is there something on nix that has the same functionality as `locate` (finding a package for command that you haven't yet installed)?
<joepie91> infinisil: yes, but they did `, locate` instead of `,locate`
<infinisil> Ah heh
<infinisil> The way it works is: Throw away the ",", separate into space-separated words, match commands on the list of words
<infinisil> Notably the word-splitting is done with Haskell's words function, which drops all spacing
justanotheruser has quit [Ping timeout: 272 seconds]
graphene has quit [Read error: Connection reset by peer]
graphene has joined #nixos
arjen-jonathan has quit [Ping timeout: 276 seconds]
asymmetric has joined #nixos
thetet has joined #nixos
thetet has left #nixos [#nixos]
kisik21 has joined #nixos
justanotheruser has joined #nixos
justanotheruser has quit [Read error: Connection reset by peer]
Thra11 has joined #nixos
graphene has quit [Read error: Connection reset by peer]
graphene has joined #nixos
<Lisanna> > let a = { b = 0.5; }; in a.b * 2
<{^_^}> value is a float while an integer was expected, at (string):169:26
<Lisanna> :/
<Lisanna> > 0.5 * 2
<{^_^}> 1
<JonReed> Is there a way to specify another `/boot` location for `nixos-rebuild switch`? Something ("systemd-1 on /boot type autofs") has mounted itself to /boot, so I can't mount my usb device there and want to mount it to another path and supply that path to `nixos-rebuild`.
<mpickering> How does the `inNixShell` attribute work
<Lisanna> welp, floats are broken
<mpickering> It seems that if I use a derivation which mentions it as a build input to `mkShell` then it is set to `True`
<mpickering> which seems wrong
<Lisanna> > let a = 0.5; in a * 2
<{^_^}> 1
Guanin has quit [Read error: Connection reset by peer]
Guanin has joined #nixos
justanotheruser has joined #nixos
jtojnar has quit [Read error: Connection reset by peer]
<mpickering> Seems related to this - https://github.com/NixOS/nixpkgs/issues/9438
<{^_^}> #9438 (by domenkozar, 2 years ago, closed): pkgs.lib.inNixShell giving false positives
IRCsum has quit [Remote host closed the connection]
jtojnar has joined #nixos
LnL has quit [Excess Flood]
IRCsum has joined #nixos
LnL has joined #nixos
patrl has joined #nixos
<clever> JonReed: have you tried `umount /boot` ?
JonReed_ has joined #nixos
<JonReed_> clever: Yes. It's some "systemd-1" is there even after "sudo umount /boot".
<clever> JonReed_: does anything appear in `journalctl -f` when you try to umount it?
<clever> JonReed_: and what does configuration.nix say about /boot/ ?
JonReed has quit [Ping timeout: 252 seconds]
<clever> JonReed_: systemctl stop boot.automount
troydm has joined #nixos
<JonReed_> clever: Thanks! You have no idea how much you're helping sometimes. Some of this stuff could take me a day to figure out on my own.
<clever> JonReed_: ive also had the fun of my /boot drive becoming corrupt and having to swap it out
<clever> JonReed_: and one of my systems uses a 64mb usb stick for /boot lol
<JonReed_> clever: :D
<Lisanna> > (0.5 + 0.5) * (2.0 + 0.5)
<{^_^}> value is a float while an integer was expected, at (string):169:1
<{^_^}> Channel nixos-18.03-small advanced to https://github.com/NixOS/nixpkgs/commit/47b68dead3d (from 8 hours ago, history: https://channels.nix.gsc.io/nixos-18.03-small)
<LnL> heh, I'm really not convinced introducing floats was a good idea
<emily> Lisanna: oh dear
<Lisanna> can't seem to find a workaround :(
<emily> maybe it should have been +. *. instead >.>
<emily> Lisanna: can you just convert both arguments to floats before multiplying?
<emily> in a similarly horrible way as converting floats to integers
<emily> it's like shelling out to bc, but worse
<Lisanna> emily but, there are no ints involved
<emily> there are
orivej has joined #nixos
<emily> or uh
<clever> Lisanna: internally, + and * just get translated into calls of the __add and __mul functions
<emily> hmm
<Lisanna> > (0.1 + 0.1) * (0.1 + 0.1)
<{^_^}> value is a float while an integer was expected, at (string):169:1
<emily> > 1.0*(2.0+0.5)
<{^_^}> 2.5
<emily> > (0.5+0.6)*(2.0+0.5)
<{^_^}> value is a float while an integer was expected, at (string):169:1
<emily> ...well, okay
<clever> Lisanna: which are then implemented here: https://github.com/NixOS/nix/blob/master/src/libexpr/primops.cc#L2281-L2285
<clever> if both arguments are floats, it will oh, i see the problem
<Taneb> The attribute for chromium in nixpkgs has a parameter "gnomeSupport", any idea what benefits as a user (using gnome) that would give me?
<clever> if both arguments are floats, AND NOT THUNKS, it will add the floats, and make a new float
<clever> but if either argument is a thunk (another expression), it will force them to be ints
<Lisanna> oh lol
<clever> prim_add needs to be fixed
<clever> it needs to forceValue arg0 and arg1, THEN check its types
<emily> amazing
<emily> > let x = 0.5 + 0.5; y = 2.0 + 0.5; in builtins.seq x (builtins.seq y (x*y))
<{^_^}> 2.5
<emily> Lisanna: you can just define your own arithmetic that forces its arguments first
<emily> what a perfectly normal problem to have
<clever> let __add = a: b: "boo!"; in 1+2
<clever> > let __add = a: b: "boo!"; in 1+2
<{^_^}> 3
<clever> dang!
<Lisanna> bah
<emily> is it builtins.__add?
<emily> > let builtins = builtins // { __add = a:b:"boo!"; }; in 1+2
<{^_^}> 3
<clever> emily: nope, its using the global __add
<clever> > __add == builtins.add
<{^_^}> false
<clever> seperate primops for the same function i believe
<clever> and __ is a special trick to add a non __ variant to builtins
<emily> maybe making forcing thunks explicit is just part of the plot to turn nix into a call-by-push-value language
<emily> clever: ah, thanks
<clever> emily: another fun thing, is that `true` and `false` are just global variables, that contain the expected values: https://github.com/NixOS/nix/blob/master/src/libexpr/primops.cc#L2155-L2159
Thra11 has quit [Ping timeout: 260 seconds]
<emily> well, it's okay because they're immutable, right?
<clever> scopedImport is the only way to change that
<emily> IIRC in R5RS Scheme it's legal to just (set! + ...)
<emily> much to the pain of compiler writers everywhere
<Lisanna> okay, yeah, builtins.seq can be used to work around it
<Lisanna> > let a = { b = 0.5; }; c = a.b; in builtins.seq c (c * 2)
asymmetric has quit [Ping timeout: 244 seconds]
<{^_^}> 1
patrl has quit [Ping timeout: 272 seconds]
silver__ has joined #nixos
<emily> yeah, I'd define add = a: b: builtins.seq a (builtins.seq b (a + b)) and just... give up on infix?
<clever> emily: scopedImport can be abused to overwrite __add
<emily> maybe you can write an evaluator that turns [a + [b*d]] into let bd = b*d; builtins.seq bd (a+bd)
<emily> clever: oh, I see, cool
<emily> nix has so many magic features :o
<emily> to go with the magic bugs I suppose
<clever> hmmm, not actually working...
<clever> { $$ = new ExprConcatStrings(CUR_POS, false, new vector<Expr *>({$1, $3})); }
<clever> oh, + is a little different from - * and /
silver_ has quit [Ping timeout: 276 seconds]
<clever> wait, thats string +....
grumble has quit [Remote host closed the connection]
<mpickering> Where are the attributes of `buildInputs` defined?
<clever> due to strings, float + float is a lot more complex
<mpickering> sorry, `buildPackages`.
alex`` has quit [Ping timeout: 276 seconds]
<clever> > builtins.unsafeGetAttrPos "buildPackages" pkgs
<{^_^}> { column = 5; file = "/var/lib/nixbot/state/nixpkgs/pkgs/top-level/stage.nix"; line = 73; }
<clever> mpickering: that is where it is defined
<mpickering> It seems to be something to do with cross compiling
<clever> mpickering: there is a comment on line 24 of that file explaining it more
aarvar has quit [Ping timeout: 268 seconds]
grumble has joined #nixos
<clever> emily: try running `nix-instantiate --eval` on the 3 files in here: https://github.com/cleverca22/nix-tests/tree/master/scopedImport
<mpickering> ok thanks
<mpickering> So if I'm not cross compiling it will just be likes `pkgs`
<clever> mpickering: yeah
<mpickering> I'm having a look at this issue - https://github.com/NixOS/nixpkgs/issues/45307
<{^_^}> #45307 (by mkawalec, 7 hours ago, open): callHackage creates infinite recursion in package overrides
<mpickering> Perhaps it's nothing to do with the difference between them then
<clever> mpickering: and if you are cross-compiling, buildPackages will ignore the cross-compile flags
<clever> mpickering: you shouldnt use rec in overrrides, thats what self is for
<mpickering> I know :) ?
<clever> just mentioning it incase you didnt
<clever> i see your reply at the bottom now, and that would explain things
<mpickering> Well it's a useful fact but not related at all to the problem
thetet has joined #nixos
joebobjoe has joined #nixos
IRCsum has quit [Remote host closed the connection]
IRCsum has joined #nixos
joebobjoe has quit [Ping timeout: 272 seconds]
<infinisil> > lib.id == lib.id
<{^_^}> false
<clever> infinisil: did you see the scopedImport example above?
<mpickering> > { x = lib.id; } == { x = lib.id; }
<{^_^}> false
drbrule has quit [Ping timeout: 240 seconds]
<infinisil> Pretty sure functions are always non-equal
<mpickering> { x = y:y; } == { x = y:y; }
<mpickering> > { x = y:y; } == { x = y:y; }
<{^_^}> true
<clever> > let __mul = a: b: "boo!"; in 1 * 2
<{^_^}> "boo!"
<clever> :D
<infinisil> Nasty!
<kisik21> what are you doing?
<clever> abusing nix internals!
<infinisil> mpickering: Huh, why did that work :/
<infinisil> > y:y
<{^_^}> "y:y"
<infinisil> AH lol
<infinisil> Right
<infinisil> clever: Oh also, I found out why (import <nixpkgs> { config.allowUnfree = true; }).hello.name gave infinite recursion
<clever> infinisil: why?
<infinisil> > true = !true
<{^_^}> true defined
<infinisil> > (import <nixpkgs> {config.allowUnfree = true; }).hello.name
<{^_^}> hnix#340 (by jwiegley, 1 week ago, open): Correctly implement function equality testing within structures
<{^_^}> infinite recursion encountered, at (string):145:10
<infinisil> Somebody defined true=!true lol
<clever> ooooo
<kisik21> > (import <nixpkgs> {}).hello
<infinisil> Seriously, assigning key words should be disallowed
<{^_^}> "<derivation /nix/store/m5chcjbj7p08sy8adqpkr2rzf6rmdfh6-hello-2.10.drv>"
<kisik21> nice
<kisik21> > true
<{^_^}> infinite recursion encountered, at (string):145:10
<infinisil> > true = builtins.true
<{^_^}> true defined
<kisik21> > true
<{^_^}> true
<kisik21> > true = !true
<{^_^}> true defined
<kisik21> > true
<clever> > let __div = a: b: a+b; in 1 / 100
<{^_^}> infinite recursion encountered, at (string):145:10
<{^_^}> 101
<kisik21> I like it
<clever> now, try and explain why 1/100 == 101 :P
<kisik21> > false = !false
<{^_^}> false defined
<infinisil> clever: Ugh..
<kisik21> > 1/100
<{^_^}> /var/lib/nixbot/state/nixpkgs/1/100
<kisik21> what?
<kisik21> > 1 / 100
<{^_^}> 0
<rotaerk> > let { true = !false; false = !true } in true
<{^_^}> error: syntax error, unexpected '}', expecting ';', at (string):169:36
<kisik21> > builtins.readFile /var/lib/nixbot/state/nixpkgs/1/100
<{^_^}> getting status of '/var/lib/nixbot/state/nixpkgs/1': No such file or directory
<kisik21> > builtins
<{^_^}> { abort = <PRIMOP>; add = <PRIMOP>; addErrorContext = <PRIMOP>; all = <PRIMOP>; any = <PRIMOP>; attrNames = <PRIMOP>; attrValues = <PRIMOP>; baseNameOf = <PRIMOP>; builtins = { abort = <PRIMOP>; add =...
<gchristensen> hey clever, kisik21, maybe take this {^_^} chatter elsewhere, like #bottest
<kisik21> ok
<infinisil> Or #nix-lang would be better
<gchristensen> sure!
alex`` has joined #nixos
<{^_^}> [nixpkgs] @benley opened pull request #45314 → vdiff-magit: Include git in nativeBuildInputs → https://git.io/fAepL
rfold has joined #nixos
simukis has quit [Quit: simukis]
vidbina has quit [Ping timeout: 268 seconds]
Anton-Latukha has quit [Ping timeout: 268 seconds]
<{^_^}> [nixpkgs] @JonathanReeve opened pull request #45315 → dragon: init at 18.04.3 → https://git.io/fAepC
acarrico has quit [Ping timeout: 272 seconds]
slack1256 has joined #nixos
rfold has quit [Ping timeout: 276 seconds]
acarrico has joined #nixos
alex`` has quit [Quit: WeeChat 2.2]
Mateon3 has joined #nixos
Mateon1 has quit [Ping timeout: 240 seconds]
Mateon3 is now known as Mateon1
jdnavarro has quit [Remote host closed the connection]
jonreeve has joined #nixos
thc202 has quit [Ping timeout: 268 seconds]
<jonreeve> Does anyone know the proper way to declare a variable in Nix, just so that I can reuse it elsewhere in the file?
<jonreeve> I'm trying to do something like: `dots = "/home/jon/dotfiles/dotfiles";`
<jonreeve> And then: ` home.file = {".spacemacs".source = "${dots}/spacemacs";};`
Fare has quit [Ping timeout: 240 seconds]
<jonreeve> in my home.nix (using home-manager)
<ldlework> jonreeve: let BINDINGS in EXPR
<jonreeve> Aha cool
<jonreeve> Where should I put that?
<jonreeve> I'm getting "unexpected let"
<ldlework> you can structure your module this way
<jonreeve> So, put it at the very top of the file? It's just my home.nix
<ldlework> { pkgs, ...}: with pkgs; let foo = "bar"; in { config = { home.file.foo.text = foo; }; }
<ldlework> etc
Fare has joined #nixos
<jonreeve> Awesome, thanks!
jluttine has quit [Ping timeout: 240 seconds]
jluttine has joined #nixos
mayhewluke has quit [Ping timeout: 276 seconds]
mayhewluke has joined #nixos
joebobjoe has joined #nixos
<{^_^}> Channel nixos-18.03 advanced to https://github.com/NixOS/nixpkgs/commit/47b68dead3d (from 10 hours ago, history: https://channels.nix.gsc.io/nixos-18.03)
slack1256 has quit [Ping timeout: 256 seconds]
Fare has quit [Ping timeout: 272 seconds]
justanotheruser has quit [Ping timeout: 268 seconds]
IRCsum has quit [Remote host closed the connection]
<{^_^}> [nixpkgs] @timokau merged pull request #45301 → ntl: 11.2.1 -> 11.3.0 → https://git.io/fAe2c
<{^_^}> [nixpkgs] @timokau pushed 2 commits to master: https://git.io/fAehM
IRCsum has joined #nixos
justanotheruser has joined #nixos