gchristensen changed the topic of #nixos to: NixOS 18.03 and Nix 2.0 are released! || 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
<zybell_> one *line* for agetty (I think agetty turns to line read after baudrate is detected)
<zybell_> one passwd for login
<zybell_> user 'oo' incorrect
pingu_ has joined #nixos
<zybell_> the f was read by agetty
day|flip has joined #nixos
<{^_^}> [nixpkgs] @yegortimoshenko pushed to yegortimoshenko-patch-4 « kernel: drop tuxOnIce patch »: https://git.io/vpSQx
<{^_^}> [nixpkgs] @yegortimoshenko opened pull request #40411 → kernel: drop tuxOnIce patch → https://git.io/vpSQp
day|flip has quit [Remote host closed the connection]
day|flip has joined #nixos
day|flip has quit [Client Quit]
day|flip has joined #nixos
<{^_^}> [nixpkgs] @xeji merged pull request #40411 → kernel: drop tuxOnIce patch → https://git.io/vpSQp
<{^_^}> [nixpkgs] @xeji pushed commit from @yegortimoshenko to master « kernel: drop tuxOnIce patch (#40411) »: https://git.io/vpS7O
<{^_^}> [nixpkgs] @anderslundstedt opened pull request #40412 → plex: 1.9.6 -> 1.13.0 → https://git.io/vpS73
sary has joined #nixos
<{^_^}> Channel nixos-unstable-small advanced to https://github.com/NixOS/nixpkgs/commit/8e7f0bc8623 (from 5 hours ago, history: https://channels.nix.gsc.io/nixos-unstable-small)
day|flip has quit [Remote host closed the connection]
day|flip has joined #nixos
phreedom has quit [Remote host closed the connection]
phreedom has joined #nixos
<boomshroom> I may have no idea what's wrong with cross-compiling, but I'm making progress on my custom init.
<boomshroom> Only on NixOS can I build an entire linux system from scratch every time I make a small modification to my program.
<boomshroom> It's like Make, but better.
tmaekawa has joined #nixos
<drakonis> make is terrible ftfy
<drakonis> wait, that's autotools i'm thinking of
<zybell_> you can make the same with make
pingu_ has quit [Ping timeout: 276 seconds]
<{^_^}> [nixpkgs] @yegortimoshenko opened pull request #40413 → stage-1: remove tuxonice handling → https://git.io/vpS7z
<boomshroom> That's one of the reasons I don't use C and C++.
<{^_^}> [nixpkgs] @yegortimoshenko pushed 0 commits to yegortimoshenko-patch-4: https://git.io/vpS7a
<boomshroom> The problem is my init is doing weird things: Send the shutdown signal, it prints "s". Do it again, prints "h". It's supposed to print "shutting down" and then shut down.
<boomshroom> The fact that it's in haskell would naturally make things more complicated.
aarvar has quit [Ping timeout: 246 seconds]
<Lisanna> My /nix/store/trash has a lot of things in it, and nix-collect-garbage is taking a REALLY long time. Is there a faster way to just purge everything in there?
<clever> Lisanna: the GC is taking a long time because its running `rm -rf /nix/store/trash`
<boomshroom> Lisanna: I often have the same problem.
<Lisanna> clever are you sure? It's unlinking a bunch of stuff in /nix/store/trash, but so far all it's actually said is "finding garbage collector roots..."
<clever> Lisanna: it may be trash left-over from a previous gc?
<Lisanna> okay
<Lisanna> so, the first thing it does is purge /nix/store/trash, then it does a regular GC cycle, and then purges /nix/store/trash again?
<clever> double-check, do you happen to have 2 GC's running at once?
<Lisanna> no, just one
<Lisanna> I've interrupted previous runs though
<Lisanna> so, that would amke sense if that's how it behaves
<Lisanna> ugh at this point it seriously might be faster to just reformat the whole disk
<clever> Lisanna: thats why i tend to use nix-collect-garbage --max-freed=1g
<clever> so it deletes 1gig at most, rather then everything
<Lisanna> I really probably just need to run it daily
<Lisanna> I had it set to weekly
<Lisanna> what's the value of --max-freed?
<Lisanna> just prevents the GC from running for too long and interrupting your work?
<clever> yeah
<clever> after deleting that much, it will stop
<Lisanna> well, my work is interrupted with the fact that most of my disk is gone, so I'm desperate to get rid of all this garbage
<Lisanna> I'll leave it sit for a day, if it hasn't made progress I'll just blast the whole disk and redeploy
<clever> the more you try to delete at once, the longer it will take to even give you 1 byte
<boomshroom> I find if you get to that point, a purge might be a good idea anyways. I cleared, I was SEVENTY GIG earlier today.
<Lisanna> clever really?
<clever> Lisanna: yeah, it will MOVE all the garbage into /nix/store/trash, as it collects garbage
<clever> Lisanna: and you wont gain any space until it has either found all the garbage, or hit the max-freed limit
<clever> then it starts to actuall delete
<boomshroom> clever: In that case, incremental collections might be a good idea.
<Lisanna> boomshroom I estimate that I have about 1.5TB of garbage
<drakonis> geeeeeeeeeeeez
<drakonis> hmm
<drakonis> i have a problem
<clever> Lisanna: welp, there goes your entire cache of downloaded things :D
<drakonis> i need to garbage collect my efi entries
<{^_^}> [nixpkgs] @dezgeg pushed 2 commits to master: https://git.io/vpS7A
<{^_^}> → 261bc1dc by @fahadsadah: sunxi-tools: master20151216 -> master20171130
<{^_^}> → 714aff40 by @dezgeg: sunxi-tools: Use upstream install rule
<boomshroom> Lisanna: My store is only on a 300GB partition. My laptop only has around 16GB for the store and home.
<Lisanna> clever I have a fast network, redownloading isn't a big deal
<boomshroom> drakonis: That was why I did a GC, resulting in 70GB disapearing.
<clever> Lisanna: about your only option is to wait for GC to finish
<drakonis> i need to resize
<{^_^}> [nixpkgs] @dezgeg closed pull request #40030 → sunxi-tools: master20151216 -> master20171130 → https://git.io/vprED
sary has quit [Remote host closed the connection]
<Lisanna> thank god I don't have any stateful work that can't be rebuilt... I write *everything* as nix expressions that I keep on my laptop, so I can rebuild everything I need
<drakonis> 119 generations removed
<drakonis> hmmm
<drakonis> i can see why i ran out of space
<boomshroom> Nix and NixOS are amazing aren't they that you end up with 100 different versions of the same system all usable.
<boomshroom> correction: most of which are usable.
<Lisanna> ...just as long as you remember to run garbage collection regularly ^^
<cindylind> why isn't auto gc a default?
<Lisanna> cindylind I had it set to one week, I need more regular garbage collection because my workflow generates a lot of garbage
<boomshroom> I'm usualy afraid of GCing since I tend to build a lot of custom compilers and those take FOREVER!
goibhniu has quit [Ping timeout: 260 seconds]
<Lisanna> anything I build that I care about I have GC roots automatically registered for
<cindylind> ^^
<Lisanna> well, it's not going to purge any faster if I sit and stare at df output all day, might as well go do something else
<boomshroom> Lisanna: No, but it can be fun watching the number tick down.
<Lisanna> boomshroom mine goes in a very slow mountain pattern
<Lisanna> goes up a bit, down a bit more, up a bit, down a bit more
<Lisanna> no idea wtf else is causing data to be created... probably some journal
<Lisanna> I killed all my automatic build processes
<clever> Lisanna: what filesystem are you using?
work_ has joined #nixos
<Lisanna> btrfs
<boomshroom> Lisanna: I did spend a GC on my laptop watching the disk usage, it did peak a few times.
thc202 has quit [Quit: thc202]
<Lisanna> iotop shows that btrfs-transcti... is doing something
<cindylind> btrfs can be problematic on a device with a slow hdd
<Lisanna> hmm, okay
<boomshroom> Has anyone tried tried building from source a non-linux kernel as a Nix derivation.
<Lisanna> :p
silver has quit [Read error: Connection reset by peer]
<boomshroom> Lisanna: So that means that it's theoretically possible for a minimalist Nix-based distro to use the Hurd kernel. :)
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<boomshroom> My quest to Nix the world is progressing steadilly.
<boomshroom> For some reason, weechat doesn't seem to like emoji.
<drakonis> you probably want a terminal emulator that does
Supersonic112 has joined #nixos
Supersonic has quit [Disconnected by services]
<boomshroom> drakonis: My terminal will display them (though not perfectly). Pasting them into Weechat replaces them with a series of question marks.
<drakonis> i need to garbage collect
<drakonis> i think i'm doing it wrong
Supersonic112 is now known as Supersonic
<cindylind> I gc'd the entire os accidently. :(
oltoAltn has joined #nixos
<boomshroom> cindylind: wat?
olto has quit [Ping timeout: 265 seconds]
<drakonis> ah crap
ericsagnes has quit [Ping timeout: 240 seconds]
<drakonis> i need to expand my efi partition but windows
<drakonis> is a huge piece of shit
<drakonis> because my kernels are stored on the efi partition
Lisanna has quit [Quit: Lisanna]
mbrgm has quit [Ping timeout: 264 seconds]
Have-Quick has joined #nixos
<{^_^}> [nixpkgs] @oxij opened pull request #40414 → rspamd: 1.6.6 -> 1.7.3 → https://git.io/vpS5r
mbrgm has joined #nixos
sbach has left #nixos ["Leaving"]
<{^_^}> Channel nixos-18.03 advanced to https://github.com/NixOS/nixpkgs/commit/ef74cafd3e5 (from 10 hours ago, history: https://channels.nix.gsc.io/nixos-18.03)
<drakonis> i have to garbage collect my kernels
<{^_^}> [nixpkgs] @oxij opened pull request #40415 → uim: 1.8.6 -> 1.8.6-20180501-git, switch to fetchFromGitHub → https://git.io/vpS5o
cindylind has quit [Ping timeout: 260 seconds]
<{^_^}> [nixpkgs] @oxij opened pull request #40416 → nixos: fix some types → https://git.io/vpS56
<{^_^}> [nixpkgs] @oxij opened pull request #40417 → python: introduce `config.pythonPackageOverrides` for proper overrides → https://git.io/vpS5i
jtojnar has quit [Read error: Connection reset by peer]
jtojnar has joined #nixos
<{^_^}> [nixpkgs] @oxij opened pull request #40418 → lib, nixos: fix module aliases in presence of defaults → https://git.io/vpS5X
<{^_^}> [nixpkgs] @oxij opened pull request #40419 → linuxPackages.kernel: config: fix DEBUG_INFO setting → https://git.io/vpS5M
ericsagnes has joined #nixos
<{^_^}> [nixpkgs] @ttuegel opened pull request #40420 → kde-gtk-config: Set path for GSettings schemas → https://git.io/vpS5S
drakonis has quit [Remote host closed the connection]
drakonis has joined #nixos
d4 has quit [Ping timeout: 276 seconds]
oida has quit [Remote host closed the connection]
oida has joined #nixos
broccoli has quit [Ping timeout: 240 seconds]
nD5Xjz_ has quit [Ping timeout: 265 seconds]
drummeur has joined #nixos
<drummeur> i wanted to stop by and thank everyone who helped me yesterday---I've got everything up and running now!
Have-Quick has quit [Quit: Have-Quick]
Have-Quick has joined #nixos
asuryawanshi has joined #nixos
broccoli has joined #nixos
nD5Xjz has joined #nixos
broccoli has quit [Ping timeout: 240 seconds]
drummeur has quit [Quit: Page closed]
<{^_^}> [nixpkgs] @moaxcp opened pull request #40421 → nixos/tests: add jbake → https://git.io/vpSdh
Have-Quick has quit [Quit: Have-Quick]
Have-Quick has joined #nixos
nD5Xjz has quit [Ping timeout: 276 seconds]
day|flip has quit [Remote host closed the connection]
justan0theruser has quit [Quit: WeeChat 2.0]
justanotheruser has joined #nixos
nD5Xjz has joined #nixos
griff_ has quit [Quit: griff_]
acarrico has quit [Ping timeout: 255 seconds]
<{^_^}> Channel nixpkgs-18.03-darwin advanced to https://github.com/NixOS/nixpkgs/commit/2d867df5b4c (from 8 hours ago, history: https://channels.nix.gsc.io/nixpkgs-18.03-darwin)
asuryawanshi has quit [Remote host closed the connection]
nD5Xjz has quit [Ping timeout: 264 seconds]
nuncanada has quit [Quit: Leaving]
nD5Xjz has joined #nixos
<{^_^}> [nixpkgs] @yegortimoshenko merged pull request #40413 → stage-1: remove tuxonice handling → https://git.io/vpS7z
<{^_^}> [nixpkgs] @yegortimoshenko pushed to master « stage-1: remove tuxonice handling »: https://git.io/vpSb9
freeman42x]NixOS has quit [Ping timeout: 260 seconds]
<gchristensen> samueldr: stupid mistake ...
<drakonis> rookie... mistake...
<jcrben> so I basically rebuilt my system from scratch since I upgraded libinput and everything depended on it - then I found that i forgot to do sudo so it failed - and I didn't use --keep-failed so I guess I'll be rebuilding again from scratch...
<drakonis> shouldn't there be a kernel package that always points to the latest?
<samueldr> gchristensen: for how long?
<jcrben> dropped into a nix-shell for libinput and briefly flirted with the idea of trying to test it that way, but that's a bit too deep into C for me right now
<gchristensen> samueldr: I took a couple hours break buut ... a while.
<boomshroom> jcrben: If you try to rebuild as sudo, most of the pieces should still be in the store, so it shouldn't take as long as the initial failed build.
sary has joined #nixos
<jcrben> boomshroom: huh, seems like it's rebuilding most all of this stuff from scratch. but I suppose it's hard to tell - it's just a long scrollback of stuff
<jcrben> sounds good tho - heading out for a political fundraiser dance party so maybe it'll be done when I'm back
<gchristensen> jcrben: that is my strategy any time I'm doing a big rebuild... find something else to do :)
beaky has joined #nixos
griff_ has joined #nixos
marusich has quit [Ping timeout: 260 seconds]
Have-Quick has quit [Quit: Have-Quick]
spear2 has quit [Remote host closed the connection]
<drakonis> hmm
smallville7123 has joined #nixos
Ross has quit [Ping timeout: 240 seconds]
Have-Quick has joined #nixos
Ross has joined #nixos
smallville7123 has quit [Quit: rosa]
giaco has joined #nixos
<giaco> hello
fatsau has quit [Ping timeout: 256 seconds]
<boomshroom> giaco: Hi!
<giaco> Helo!
<boomshroom> giaco: Looking into trying NixOS, or here to chat about or get help with an existing installation?
<giaco> I have finally succeeded in compiling Nix from source code on android. Now I have it make-installed into a custom path and I have just sourced source <localprefix>/etc/profile.d/nix.sh. Problem: error: undefined variable 'exec_prefix' at /data/data/com.termux/files/nix/local/share/nix/corepkgs/config.nix:14:40
<giaco> Am I missing something?
<boomshroom> giaco: Before starting, are you using the standard store path of /nix/store?
<drakonis> he's trying to run it inside termux
asuryawanshi has joined #nixos
<drakonis> termux doesn't provide a full environment
<giaco> boomshroom: no, I have compiled it with custom paths obviously
<boomshroom> giaco: Ok, that does mean more complications than usual, but it's doable.
<giaco> here's how I compiled it
<giaco> ./configure --prefix=/data/data/com.termux/files/nix/local --with-store-dir=/data/data/com.termux/files/nix/store --localstatedir=/data/data/com.termux/files/nix/var --disable-doc-gen
<giaco> make & make install went ok. I had to recompile libboost but that was a termux package problem. Everything else went kind of smooth
Have-Quick has quit [Quit: Have-Quick]
<giaco> I just want to try too bootstrap it and see what happens and report it to the community
<giaco> (btw I am already using nix on android under proot environment)
cindylind has joined #nixos
<giaco> do you have any idea why I am receiving undefined variable 'exec_prefix'? The line 14 says nixBinDir = fromEnv "NIX_BIN_DIR" "${exec_prefix}/bin";
<giaco> I do not have set any NIX_BIN_DIR environment variable in my terminal session, but I have never set one. Do I have to set it?
<cindylind> Anyone know how to make a package overlay for a python package? I did it the way the manual says and it really messes things up. This pr may be relevent https://github.com/NixOS/nixpkgs/pull/40417
<drakonis> hmm
<drakonis> it is a full environment
<drakonis> neat.
<giaco> drakonis: :)
smallville7123 has joined #nixos
<giaco> I am going to release a nix package on termux repository if I end up with a working setup into my custom prefix
<giaco> any idea why 'exec_prefix' is not set after a "make install"?
broccoli has joined #nixos
<drakonis> awesome
<giaco> I just tried to export NIX_BIN_DIR but I received the very same error undefined variable 'exec_prefix'
<cindylind> Here's what I what sorta doing https://cryptbin.com/SWHwjH#567acf6b468d113fef1d5fdd3331507e
broccoli has quit [Ping timeout: 256 seconds]
Have-Quick has joined #nixos
Maxdamantus has quit [Ping timeout: 260 seconds]
<giaco> recompiling adding option exec_prefix to ./configure
Have-Quick has quit [Quit: Have-Quick]
Have-Quick has joined #nixos
Rusty1_ has quit [Quit: Konversation terminated!]
smallville7123 has quit [Quit: rosa]
Guanin_ has quit [Ping timeout: 256 seconds]
<{^_^}> [nixpkgs] @Ericson2314 pushed 94 commits to staging: https://git.io/vpSxC
<{^_^}> → dc524390 by R. RyanTM: rocksndiamonds: 4.0.1.4 -> 4.1.0.0
<{^_^}> → 4d0ab5d6 by @ayyjayess: polybar: add i3 to path
<{^_^}> → d0cfb273 by @vbgl: ocamlPackages.ocaml_pcre: cleanup
Have-Quick has quit [Quit: Have-Quick]
boomshroom has quit [Quit: WeeChat 2.0]
Guanin_ has joined #nixos
jD91mZM2 has joined #nixos
orivej has quit [Ping timeout: 246 seconds]
cindylind has quit [Ping timeout: 260 seconds]
griff_ has quit [Quit: griff_]
drakonis has quit [Read error: Connection reset by peer]
alex`` has quit [Quit: WeeChat 2.1]
Maxdamantus has joined #nixos
Mateon3 has joined #nixos
Mateon1 has quit [Ping timeout: 240 seconds]
Mateon3 is now known as Mateon1
alex has joined #nixos
<giaco> recompiling with exec_prefix does not solve the problem
<giaco> installing with make install exec_prefix=/data/data/com.termux/files/nix/local it not solving the problem
<giaco> what the hell is "exec_prefix"?! I am gonna give up. error: undefined variable 'exec_prefix' at /data/data/com.termux/files/nix/local/share/nix/corepkgs/config.nix:14:40
simukis has joined #nixos
stumble has joined #nixos
griff_ has joined #nixos
iqubic has joined #nixos
zybell_ has quit [Read error: Connection reset by peer]
Ariakenom has joined #nixos
brandon__ has joined #nixos
brandon_ has quit [Read error: Connection reset by peer]
broccoli has joined #nixos
smichel has joined #nixos
broccoli has quit [Ping timeout: 255 seconds]
griff_ has quit [Quit: griff_]
<{^_^}> Channel nixos-unstable-small advanced to https://github.com/NixOS/nixpkgs/commit/a2981671523 (from 3 hours ago, history: https://channels.nix.gsc.io/nixos-unstable-small)
Guest81 has joined #nixos
Guest81 has quit [Remote host closed the connection]
kiwi_0 has joined #nixos
kiwi_0 has quit [Remote host closed the connection]
blatherstop has joined #nixos
blatherstop has quit [Remote host closed the connection]
marusich has joined #nixos
work_ has quit [Quit: Connection closed for inactivity]
raynold has joined #nixos
i-am-the-slime has joined #nixos
<i-am-the-slime> Hi there, I would like to replace some files in /nix/store/...-user-environment/share/icons/hicolor ...
<i-am-the-slime> How should I do that?
phdoerfler has joined #nixos
<jD91mZM2> i-am-the-slime: You probably want to create a package override that replaces the files in a post-install hook
<i-am-the-slime> Do I "want" or "have" to? :) So that's for each application individually? Do you have an example?
tertle||eltret has quit [Quit: Connection closed for inactivity]
marusich has quit [Ping timeout: 240 seconds]
juhe has joined #nixos
coot has quit [Remote host closed the connection]
jacob has joined #nixos
jacob is now known as Guest84635
coot has joined #nixos
kreetx has joined #nixos
user__ has joined #nixos
hotfuzz has quit [Read error: Connection reset by peer]
hotfuzz has joined #nixos
udono has joined #nixos
<{^_^}> [nixpkgs] @LnL7 pushed to staging « Revert "darwin.libSystem: multiple outputs" »: https://git.io/vpSjP
<{^_^}> [nixpkgs] @rycee pushed to release-18.03 « lib.types: fix loaOf behavior for long lists »: https://git.io/vpSj1
<jD91mZM2> Just how horrible of an idea would it be to make a setuid wrapper for nixos-rebuild?
<jD91mZM2> Or maybe make it still require sudo but passwordless
d4 has joined #nixos
<judson> Is it always this hard to get a python app added?
<jD91mZM2> Judson: What's up?
phdoerfler has quit [Quit: Leaving.]
<{^_^}> [nixpkgs] @uri-canva opened pull request #40424 → bazel: 0.12.0 -> 0.13.0 → https://git.io/vpSjx
TonyTheLion has joined #nixos
smallville7123 has joined #nixos
broccoli has joined #nixos
broccoli has quit [Ping timeout: 240 seconds]
tmaekawa has quit [Quit: tmaekawa]
orivej has joined #nixos
<user__> Hey, guys! I build nixpkg, and it try to create file in ${nix store path of that pkg}/var directory. How can i redirect it to current user home dir, using libredirect? Or something else?
<user__> And one more question: if my pc's hardware changes, should I run nixos-generate-config, or nixos + linux kernel can detect it automatically?
<user__> and probably systemd can do something
<jD91mZM2> user__: You want to re-run nixos-generate-config. But don't worry - it won't overwrite your configuration.nix :)
phdoerfler has joined #nixos
<user__> jD91mZM2: Thank you! :)
FRidh has joined #nixos
pSub has quit [Remote host closed the connection]
<{^_^}> Channel nixos-17.09 advanced to https://github.com/NixOS/nixpkgs/commit/967e40787c5 (from 2 days ago, history: https://channels.nix.gsc.io/nixos-17.09)
user__ has quit [Quit: Leaving]
<{^_^}> [nixpkgs] @megheaiulian opened pull request #40425 → avocode: init at 2.26.0 → https://git.io/vp9vt
stumbleplus has joined #nixos
Guest84635 has quit [Quit: Lost terminal]
stumbleplus has quit [Remote host closed the connection]
<{^_^}> Channel nixpkgs-18.03-darwin advanced to https://github.com/NixOS/nixpkgs/commit/c04fbec77cc (from 63 minutes ago, history: https://channels.nix.gsc.io/nixpkgs-18.03-darwin)
stumble has quit [Quit: Leaving]
alex has quit [Ping timeout: 255 seconds]
stumble has joined #nixos
udono has quit [Ping timeout: 256 seconds]
broccoli has joined #nixos
stumble has quit [Remote host closed the connection]
stumble has joined #nixos
stumble has quit [Client Quit]
alex has joined #nixos
jason has joined #nixos
Guest85183 has joined #nixos
jason is now known as Guest49585
alex has quit [Read error: Connection reset by peer]
stumble has joined #nixos
aarvar has joined #nixos
Guest85183 has quit [Client Quit]
alex has joined #nixos
smallville7123 has quit [Quit: rosa]
b has quit [Quit: Lost terminal]
<{^_^}> [nixpkgs] @Mic92 pushed commit from @Fresheyeball to release-18.03 « pywal: init at 2.0.5 (#38443) »: https://git.io/vp9vo
TonyTheLion has quit [Quit: leaving]
swflint has quit [Ping timeout: 265 seconds]
<Myrl-saki> Is nixpkgs also a module?
swflint has joined #nixos
periklis has joined #nixos
<electrocat> is there a decent way of creating nixos containers on non-nixos systems?
smichel has quit [Quit: smichel]
<Guest49585> electrocat: Containers specifically? You can run Nix on non-NixOS systems without a container.
ambro718 has joined #nixos
<electrocat> i know, yes, just nix is not sufficient
<Guest49585> Fair enough. Then you need VirtualBox or something similar.
<electrocat> i dont want vms though
<Guest49585> I thought you did. OK. What do you want?
<stumble> You want a container that isn't running an OS but is running NixOS?
<Guest49585> Because if it is running an OS then it's a VM.
<electrocat> i just want to install nixos to some root so i can run it as a container, but nixos-install demands filesystems and bootloaders and stuff
<Guest49585> So you want a VM. Which is cool. There are instructions for running NixOS in VirtualBox in the manual.
<electrocat> i mean for running it in things like docker or systemd-nspawn
<Guest49585> It seems that you don't know that NixOS is an operating system. Is that right? If you run two operating systems at once, one of them has to be inside a VM, because the whole point of an operating system is that it believes it has direct access to system resources.
periklis has quit [Ping timeout: 240 seconds]
orivej has quit [Ping timeout: 264 seconds]
<stumble> You can have Nix, which is a package manager, or NixOS, which is an operating system.
<Guest49585> If you want an operating system, go for it. Then you need to make it think it has direct access to system resources. You do that either by installing it so that it HAS access to system resources, or by running it in a VM.
<{^_^}> [nixpkgs] @Mic92 opened pull request #40426 → linuxPackages.bcc: make it useable as a python library → https://git.io/vp9fg
<MichaelRaskin> You might want to search for nixos-container
<electrocat> MichaelRaskin: i want something like that, but it has to work on non-nixos systems
<Myrl-saki> Rule 34p: If it exists, there's a Docker variant of it?
<Myrl-saki> electrocat: nixops?
<Guest49585> electrocat: Could you just confirm that you know that NixOS is an operating system? Then I'll be happy! So you want Docker or something similar, running a separate OS (not the native OS)?
<electrocat> nixops only seems to be able to deploy to vms
<MichaelRaskin> Did you try just nix-build-ing a nixos system with /dev/null as root FS etc., inspecting its path and running nixos-container from there
<Myrl-saki> electrocat: Yeah, but it's pretty close to what you want.
<Myrl-saki> Just name dropping it in case you do not find what you want.
<electrocat> MichaelRaskin: i was going to try that, but first i wanted to confirm there was a better option
<Dezgeg> nixops can do libvirtd containers, I've heard
<electrocat> Dezgeg: really? that would be perfect
<Guest49585> I thought you DIDN'T want VMs!
<MichaelRaskin> electrocat: the better option is doing this, then finding out what attribute you need to extract to build only the container stuff
<electrocat> Guest49585: yes, i really know what that is, i just want a nixos installation that doesn't care about linux kernels or filesystems or bootloaders
<Guest49585> So VirtualBox. See the manual.
<stumble> I'm sure there are other ways too, but VirtualBox has the advantage of being documented.
<sphalerite> so I've put boot.debug1devices on my kernel commandline, but it doesn't seem to be having any effect… I'm getting the message "/nix/var/nix/profiles/system/init: line 9: /proc/cmdline: No such file or directory" right before it prints <<< NixOS Stage 2 >>>, maybe that's why?
<sphalerite> /proc/cmdline does exist in the fully booted system
Guest49585 has quit [Quit: restarting client]
stumble has quit [Remote host closed the connection]
stumble has joined #nixos
<sphalerite> Myrl-saki: no, nixpkgs is not a module
<Dezgeg> your initrd didn't get executed at all maybe?
<sphalerite> electrocat: you should be able to make a container yourself as well, iirc boot.isContainer disables the bootloader and filesystem requirements.
<electrocat> sphalerite: ah nice
<sphalerite> Dezgeg: oooh yes, that's a possibility!
<electrocat> for now it demanded a root filesytem
<electrocat> which i pointed to /dev/null as MichaelRaskin suggested
<sphalerite> yeah now that you mention it… it's not printing the stage 1 banner
<Myrl-saki> sphalerite: What's wrong with declaring `options` and `config` in the same module, again?
<sphalerite> Myrl-saki: nothing.
<Myrl-saki> sphalerite: I gave the module system a closer look.
alex has quit [Ping timeout: 255 seconds]
<Myrl-saki> sphalerite: Do you remember that systemPackages thing that I asked help with?
<sphalerite> no
<Myrl-saki> Let me find the gist.
<{^_^}> [nixpkgs] @knedlsepp closed pull request #36825 → WIP: pythonPackages.macfsevents: init at 0.8.1 → https://git.io/vxeMk
<Myrl-saki> sphalerite: What's wrong with declaring options in the root set here?
<MichaelRaskin> Yeah, I am better at brute-force extraction of stuff from NixOS quasi-builds than at keeping track of proper application of NixOS, sorry
coot has quit [Quit: coot]
<sphalerite> Myrl-saki: nothing AFAIK, but what's probably problematic is redeclaring the environment.systemPackages option.
<sphalerite> I'm not an expert on the system either though.
<{^_^}> [nixpkgs] @FRidh pushed 32 commits to staging: https://git.io/vp9JI
<{^_^}> → 408563d0 by @markuskowa: rtl-sdr: 0.5.3 -> 0.5.4
<{^_^}> → fbe7ebca by @akazakov: Update pants to 1.6.0
<{^_^}> → 095fe5b4 by @oxij: nixos: rename system.{stateVersion,defaultChannel} -> system.nixos.\1
<{^_^}> [nixpkgs] @FRidh closed pull request #25985 → Python: add sitecustomize.py, listen to NIX_PYTHONPATH → https://git.io/vHT7r
<Myrl-saki> sphalerite: No worries. I guess there's also merge strategies for `option` declarations?
<sphalerite> My guess would be that that's just a fixed recursive set merge which is either biased on one side or just complains about collisions. I don't really know though.
<electrocat> ah nice
<electrocat> got my container working
<electrocat> the boot.isContainer is very much required
<LnL> FRidh: around?
<sphalerite> Myrl-saki: would it maybe be easier to achieve what you want to do using an overlay that replaces nixos-icons with null?
<MichaelRaskin> Yeah, nixos-container probably sets it too, but directly building with boot.isContainer sounds saner
<{^_^}> [nixpkgs] @FRidh reopened pull request #25985 → Python: add sitecustomize.py, listen to NIX_PYTHONPATH → https://git.io/vHT7r
<FRidh> LnL: yep
<electrocat> yeah i just did nixos-install with the option, and did 'systemd-nspawn /path/to/init'
<Myrl-saki> sphalerite: That's also what was suggested last time. And yeah, I agree. I think it's even better to override the specific module that I required this hack for.
<electrocat> without boot.isContainer it gets stuck somewhere during systemd boot
<Myrl-saki> sphalerite: It was the NixOS module manual that required nixos-icons, which indirectly requires rust. I should have "patched" that instead of removing nixos-icons from systemPackages.
<Myrl-saki> NixOS manual module*
<LnL> FRidh: I've been working on this recently https://github.com/LnL7/nixpkgs-python/tree/updates It's still pretty wip, but perhaps you could take a look at it sometime
<{^_^}> Channel nixos-18.03-small advanced to https://github.com/NixOS/nixpkgs/commit/c04fbec77cc (from 2 hours ago, history: https://channels.nix.gsc.io/nixos-18.03-small)
smichel has joined #nixos
<sphalerite> Myrl-saki: oh yeah, I've had issues with that too
<Myrl-saki> I should really put that as an issue.
<Myrl-saki> Rust is not a trivial build.
<Myrl-saki> And if you're running an i686 machine, then chances are, your machine won't even be able to build rust.
<Myrl-saki> Sanely, I mean.
<electrocat> i feel like i should document this somewhere, this might be useful for a lot of people
<FRidh> LnL: "avoid PYTHONPATH" that I like to see happening as well.
<FRidh> I just started looking into some old work abbradar and I did with using a sitecustomize.py
<FRidh> also because of https://github.com/NixOS/nixpkgs/issues/40427 so we might be able to get rid of the magical sed
<sphalerite> Myrl-saki: I'm guessing the definition of sanely has shifted a bit in the past, imagine running gcc bootstraps back in 2002 or something D:
<Myrl-saki> sphalerite: I used to run Gentoo back in 2015 on a 2008 machine.
<Myrl-saki> sphalerite: I think I switched to Debian 3 days later.
<sphalerite> haha
<Myrl-saki> Though, TBF, Gentoo was faster than I expected.
<Myrl-saki> Err
<Myrl-saki> faster to build*
<sphalerite> Must have been all the -march=native
<sphalerite> :p
<Myrl-saki> sphalerite: :P
<Myrl-saki> Gotta have them funroll-loops dude.
<Myrl-saki> Really though. I think I was able to build X11 and Firefox in 2 days at most.
<Myrl-saki> Of course, it at most saved me 1 second of my life, so 0/10, would not do again.
<Myrl-saki> Turns out that in most systems, the slowest thing is the person using it.
<sphalerite> :)
<{^_^}> [nixpkgs] @jfrankenau opened pull request #40429 → sass: 3.4.22 -> 3.5.6 → https://git.io/vp9JH
<sphalerite> building everything on my chromebook is also more reasonable than I would have thought
<sphalerite> a full system rebuild takes maybe 2 or 3 days
<sphalerite> I still haven't got a full-featured browser though :/
<Myrl-saki> sphalerite: Ikr. I expected weeks.
<Myrl-saki> sphalerite: armv7l?
<sphalerite> yep
<Myrl-saki> sphalerite: I have access to a build farm and qemu if you'd like.
<sphalerite> I was running fedora's firefox build for a while, but eventually that stopped working mysteriously, with segfaults
<sphalerite> Myrl-saki: I think firefox literally can't build natively on armv7 because of address space limitations.
<sphalerite> i.e. it involves a process that needs more than 3GB of RAM
<LnL> Fridh: currently it kind of buildenv's the dependencies together but I want to do something somilar with pth files
<Myrl-saki> sphalerite: Linking, I guess?
<sphalerite> I'm not sure. Some rust stufff
<Myrl-saki> sphalerite: I was able to build Rust in i686 FWIW.
<rszibele[m]> is rebuilding all packages for nixos with -march=native easily doable? I remember one could do something like that on debian..
<FRidh> LnL: so you will use site.addsitedir then
<Myrl-saki> rszibele[m]: "Not really"
<sphalerite> Myrl-saki: yeah rust itself works, but the firefox build fails
<Myrl-saki> rszibele[m]: Nixpkgs specifically disables -march=native.
<Myrl-saki> rszibele[m]: So you'd have to patch that script first.
<Myrl-saki> rszibele[m]: Then after that, you'll have to pray that any nix copying will work.
<LnL> FRidh: yeah, probably
<sphalerite> rszibele[m]: the trouble with that is it goes against one of the core features of nix — build reproducibility
<FRidh> LnL: you're on Darwin, right? Could you test the following script for me https://gist.github.com/FRidh/c70d67e5ccbeb0b19e0c0d1eb59627dc
xy3_ is now known as xy2_
<FRidh> the python docs don't state that `del os.environ` works on OSX
<Myrl-saki> myrl@c:~/nixpkgs$ git stash pop
<Myrl-saki> Segmentation fault
<Myrl-saki> OH NO
<sphalerite> Myrl-saki: nice.
<rszibele[m]> sphalerite: the default flags for gcc must be defined somewhere
<LnL> Fridh: pretty sure it does
<sphalerite> rszibele[m]: yes, it is *possible*, it's just not a good idea.
<rszibele[m]> I mean it would be cool, but I don't think my system would be that much more performant.
<FRidh> LnL: using this in a `sitecustomize.py` allows us to prevent leaking of env vars
<Myrl-saki> sphalerite: I'm trying to build firefox.
<Myrl-saki> myrl@c:~$ nix build $(nix-instantiate ./nixpkgs --arg system '"armv7l-linux"' -A firefox) --cores 48 --max-jobs 16
<Myrl-saki> It says 225 packages to build.
<sphalerite> yeah there's plenty of deps :p but it did get up to building firefox itself for me
<sphalerite> and then that ended up failing with https://gist.github.com/lheckemann/07f5f0e1998514846509db057a6eecdc
<Myrl-saki> sphalerite: Then 3GB limit?
<sphalerite> yeah pretty sure that was the reason, since I added plenty of swap and it didn't get filled
<Myrl-saki> Yeah, that does seem scary.
<Myrl-saki> sphalerite: IIRC, that's an ongoing issue.
<LnL> FRidh: main problem I've had with PYTHONPATH are projects with mixed python2/python3 tools
<Myrl-saki> sphalerite: And even PAE can't fix it.
<sphalerite> Yeah
<Myrl-saki> sphalerite: Anyways, I remember that issue, but i thought it was with Rust itself. "Can't compile Firefox because Rust on i686 is uncompilable." Might have remembered wrong.
<sphalerite> I suspect this is a "won't fix" issue, because there's always cross compilation, and the majority of new consumer devices you'd want t ocompile anything on will probably be 64-bit sooner or later
<Myrl-saki> sphalerite: I'll report back if anything.
<sphalerite> cool, thanks :)
<Myrl-saki> sphalerite: No problem. I'm literally the only one using the build machine, so I thought it'd be a waste to not use it for Nix stuff.
<sphalerite> :D
<sphalerite> what kind of machine is it?
<Myrl-saki> Also, I encountered a promblem where `nix build` hangs but `nix-build` doesn't. Was weird.
<Myrl-saki> sphalerite: EPYC! :D
<sphalerite> oh right and you're building with qemu-user?
<flokli> Myrl-saki: nice :-)
<sphalerite> That might well also result in some difficulties.
<Myrl-saki> Yep.
<Myrl-saki> Yeah, I agree.
<pstn> How is the cache for `nix search` actually meant to work? I can only call nix search with an -u to get results even after just having called `nix-search -u` on it's own, it still complains when I just want to search for something.
<Myrl-saki> sphalerite: Back then, I used to use my RasPi for some qemu problems.
<Myrl-saki> sphalerite: But I decided to pull the plug on it. :(
<LnL> Fridh: adding a python3 based tool to a python2 project changes PYTHONPATH and depending on the buildInputs/propagatedBuildInputs ordering is might end up before other dependencies
<Myrl-saki> sphalerite: qemu-user really dies on Haskell compilation.
<sphalerite> I mostly use scaleway C1 servers for offloading builds
<sphalerite> €3/month, they're not very powerful machines, but they're ARMv7 and they do work
<Myrl-saki> sphalerite: Seems like they're better than DO though.
<Myrl-saki> I mean, sure, it's ARM.
<sphalerite> oh?
<Myrl-saki> DO is like $5/mo for 1C/512MB.
<Myrl-saki> IIRC.
<sphalerite> oh then yeah, these are 4 cores and 2GB
<sphalerite> And bare metal
<sphalerite> Hm, does qemu-user use KVM when available?
<FRidh> LnL: correct, that still happens. We need to get rid of `propagatedBuildInputs` in `buildPython*`, and do the propagating ourselves. See the `requiredPythonModules` function. It should be a small change now. Add a new parameter, `runtimeInputs`, and apply `requiredPythonModules` to it.
phdoerfler has quit [Quit: Leaving.]
<Dezgeg> it's probably quite hard/impossible to use KVM while not emulating a full kernel
<sphalerite> aw ok, was hoping I'd have an easy way to build 32-bit ARM stuff on a 64-bit ARM server that supports 32-bit as you suggested https://github.com/NixOS/nix/pull/1916#issuecomment-387543970
<{^_^}> [nixpkgs] @timokau opened pull request #40430 → giac: remove liblapackWithAtlas dependency → https://git.io/vp9UV
<Dezgeg> could be worth trying
<sphalerite> Dezgeg: do you know an easy way to get a 32-bit ARM KVM machine set up?
<Dezgeg> I have something like this: qemu-system-aarch64 -enable-kvm -machine virt,gic-version=host -cpu host,aarch64=off -bios /home/tmtynkky/u-boot.bin -s -nographic -drive if=none,format=raw,file=/home/tmtynkky/hydra-arm-vm.img,id=root -device ich9-ahci,id=ahci -device ide-drive,drive=root,bus=ahci.0 -m 3072 -smp 8 -device virtio-net,netdev=netdev0 -netdev
<Dezgeg> user,id=netdev0,hostfwd=tcp::3000-:3000,hostfwd=tcp::2222-:22
<Dezgeg> I guess "easy" can be relative
<{^_^}> [nixpkgs] @Synthetica9 opened pull request #40431 → nasm: set -Wno-error=attributes for gcc8 → https://git.io/vp9U9
<Myrl-saki> flokli: Indeed very nice.
smallville7123 has joined #nixos
Ariakenom has quit [Read error: Connection reset by peer]
<sphalerite> Dezgeg: if that "just works" I'll be happy :p how do I build the u-boot image?
<sphalerite> and the HD image
<Dezgeg> that's just ubootQemuArm
<Dezgeg> and the v7 install image (resized up)
<Myrl-saki> I think NixOS is the only OS I used that doesn't have vi pre-installed.
<sphalerite> cool, thanks!
<sphalerite> Myrl-saki: but nano! :p
<Myrl-saki> sphalerite: I cringe physically every time I have to open nano. :P
<Myrl-saki> sphalerite: But hey, at least it taught me to use `less` for viewing stuff. :P
<flokli> Myrl-saki: it's also possible to "cross-boot" just the kernel in qemu without uboot
<flokli> Didn't try to boot PAE kernel, though
<flokli> Uboot is limited to highmem=off currently
i-am-the-slime has quit [Ping timeout: 256 seconds]
<Dezgeg> yeah -kernel -initrd -cmdline works but it's a pain to dig those out
<{^_^}> [nixpkgs] @Synthetica9 opened pull request #40432 → nasm: set -Wno-error=array-bounds for libglvnd → https://git.io/vp9Te
<flokli> Dezgeg: thought about just using a dumb kernel, then kexec'ing from initrd into the 'right' one
<flokli> But #nixos-aarch64 should be the more appropriate channel for that kind of stuff ;-)
<sphalerite> Oh yeah, I know it's possible to embed an initrd into a kernel at build time. Is it also possible to "retrofit" a kernel image (with no special config) with an initrd?
<Dezgeg> ohh, but then you have to rely on kexec working properly :P
<flokli> Dezgeg: I hope it does on armv7l and aarch64 :-P
<Dezgeg> btw eventually I finally found out a change to u-boot that makes highmem
<Dezgeg> =on work on both arm and aarch64
<ben> How do I install nixpkgs.urxvt_font_size in my user environment so that urxvt can find it?
<ben> do i need a wrapper for urxvt?
civodul has joined #nixos
<ben> oh, there's a wrapper, rxvt_unicode-with-plugins, cool
<ij> Is it the case that the store path's hash is made from the inputs only? How do you then find out whether a build is undeterministic?
<ben> ij: optimism, mostly
<ben> some sandboxing
<ij> ah, ok, then I've got that right
<ij> you could also check that something's invalidating cache all the time
periklis has joined #nixos
seanparsons has quit [Read error: Connection reset by peer]
<ben> I think most builds aren't 100% bit-for-bit deterministic, but I'm not that well-informed
<Myrl-saki> Set `-j1` and redefine `random()` to `return 4;`
<Myrl-saki> :P
<sphalerite> Myrl-saki: I believe that was Sonarpulse
<Myrl-saki> sphalerite: How about the formal proof?
<sphalerite> yeah Sonarpulse I think
griff_ has joined #nixos
nD5Xjz has quit [Quit: ZNC 1.6.5 - http://znc.in]
periklis has quit [Ping timeout: 246 seconds]
<Myrl-saki> Sonarpulse: Ping. Is there a "raw" of the inference rules? I'm trying to understand it, but proof b has `t0` unbounded and I'm confused as to what each offset means, nor what is it 'offseting'
<Myrl-saki> The bash script filled in some info, but I'm confused as what "target" means.
<Myrl-saki> https://nixos.org/nixpkgs/manual/#idm140737317773760 has the definition of targetPlatform, FWIW.
thc202 has joined #nixos
<sphalerite> build = where are we building; host = where will the code we're building run; target = what will the code we're building build for
novum has quit [Quit: leaving]
<Myrl-saki> sphalerite: Oh okay.
nD5Xjz has joined #nixos
<Myrl-saki> sphalerite: Why is the dependency code specified in bash and not in Nix? Or is it a combination of both?
<sphalerite> because the rest of stdenv is mostly bash as well
<Myrl-saki> This is the bash script that's cited by the formal proof. https://github.com/NixOS/nixpkgs/blob/master/pkgs/stdenv/generic/setup.sh#L358
<sphalerite> AFAIK
seanparsons has joined #nixos
<Myrl-saki> There's also this line.
<Myrl-saki> # Platform offsets: build = -1, host = 0, target = 1
<infinisil> I'm getting a bit anxious just hearing "formal proof" and "bash" together in a sentence
novum has joined #nixos
<Myrl-saki> infinisil: Haha. :P
<Myrl-saki> Formally verified shell scripts seems pretty cool.
<Myrl-saki> infinisil: It's not a formal proof but just inference rules, FWIW.
<Myrl-saki> infinisil: You might enjoy this comment though. :P https://github.com/NixOS/nixpkgs/blob/master/pkgs/stdenv/generic/setup.sh#L389-L391
<infinisil> I'm glad to say that I know what a homomorphism is
nD5Xjz has quit [Ping timeout: 265 seconds]
<Myrl-saki> Is it different from a cismorphism?
* Myrl-saki groans
<infinisil> Probably, no idea :P
<Myrl-saki> infinisil: FWIW, this is the specific excerpt.
<Myrl-saki> We can define the process precisely with Natural Deduction using the inference rules. This probably seems a bit obtuse, but so is the bash code that actually implements it!
<infinisil> Hah
<Myrl-saki> > define ... using the inference rules > bash code that implements it
<{^_^}> error: syntax error, unexpected ELLIPSIS, expecting ')', at (string):34:8
<infinisil> Hah
<Myrl-saki> > 1 + 1
<{^_^}> 2
<Myrl-saki> What's > supposed to be? Nix?
<stumble> Yes.
<infinisil> You mean why the ">" character and not something else to trigger nix eval?
<Myrl-saki> > Oh okay.
<{^_^}> error: syntax error, unexpected ')', expecting ID or OR_KW or DOLLAR_CURLY or '"', at (string):35:1
<Myrl-saki> Ugh, sorry, I wanted to try out something then forgot that I was in the middle of trying out something.
<Myrl-saki> infinisil: Nah, just that I'm surprised that there is a Nix repl.
<sphalerite> There's #bottest if you want to spam it
<infinisil> Myrl-saki I just coded it recently :)
<Myrl-saki> sphalerite: Nah, except for the `> 1 + 1`, everything was unintentional. :P
pingu_ has joined #nixos
<Myrl-saki> I'm confused as to what "offset" means.
<ij> ben, Wait, no, the hash should change when the content changes.
<Myrl-saki> ij: Source code is input.
<Myrl-saki> ij: I think?
<Myrl-saki> I'm actually not sure anymore.
<sphalerite> Dezgeg: it's not liking the net device specification: qemu-system-aarch64: -netdev user,id=netdev0,hostfwd=tcp::3000-:3000,hostfwd=tcp::2222-:22: Device needs media, but drive is empty
<sphalerite> but I don't know what that means :p
<ij> But the store should be content addressable, so hashing the content should give you the store path.
<sphalerite> ij: it isn't, except for fixed-output derivations.
<sphalerite> ben: yeah a lot of builds aren't deterministic but it is a goal.
<Dezgeg> maybe you forgot the -device virtio-net,netdev=netdev0 part?
<Dezgeg> I guess you can drop the hostfwd parts if you don't need them
<Myrl-saki> `TEMP=$(tempf); gcc <(long int random() { return 0; }) -o $TEMP; export LD_PRELOAD=${TEMP}:${LD_PRELOAD}`
<Myrl-saki> sphalerite: :D
<Myrl-saki> Or maybe `return 4;` since we know that's from a fair dice roll.
nD5Xjz has joined #nixos
<sphalerite> roll 1d4294967296
b has joined #nixos
<sphalerite> Myrl-saki: do you mean rand as opposed to random?
<sphalerite> nvm me
<sphalerite> Dezgeg: nope, I have that…
pingu_ has quit [Ping timeout: 240 seconds]
<Myrl-saki> builder for '/nix/store/b3a10dsjwzn4gfmi8g384w3v6d4iyhp4-orc-0.4.28.drv' failed with exit code 2
<Myrl-saki> Sad.
<Myrl-saki> orc_add2_rshift_add_s16_22:
<Myrl-saki> backup function : PASSED
<Myrl-saki> qemu: uncaught target signal 11 (Segmentation fault) - core dumped
<Myrl-saki> FAIL orc_test (exit status: 139)
<Myrl-saki> I don't have an ARMv7 machine in hand, so. :(
<sphalerite> oh pfft I was doing stuff wrong
<sphalerite> it was just a bad script
hiratara_ has joined #nixos
tmaekawa has joined #nixos
hiratara has quit [Ping timeout: 255 seconds]
hiratara_ is now known as hiratara
<sphalerite> Dezgeg: now I'm getting controller reset failed (0x1) and Device 0: unknown device when it tries to boot
<Dezgeg> oops
<Dezgeg> s/-machine virt,gic-version=host/-machine virt,gic-version=host,highmem=off/ for unpatched u-boot I guess
<gchristensen> any ideas on what is causing this kernel to fail to load the root fs from iPXE? it is the only server hardware which doesn't handle it okay out of about nine: https://gist.github.com/grahamc/979a7b09625c84f833dd5f9ea3e0cecc
tmaekawa has quit [Client Quit]
phdoerfler has joined #nixos
<Dezgeg> sounds like your initrd doesn't get loaded/used at all
<Dezgeg> maybe it's too big or something
<sphalerite> Dezgeg: now it says "undefined instruction", gives some register info, and resets
<Dezgeg> huh
<gchristensen> seems ... unlikely I think? it is pretty new hardware, and the same size initrds / bzImages work well elsewhere. one thing is it is new: AMD EPYC 7401P
<gchristensen> Dell R6415
<Dezgeg> I guess you can boost the kernel loglevel to hopefully see more info
<Dezgeg> sphalerite: any full log? where it crashes like that
<Dezgeg> maybe try giving it less ram as well
<sphalerite> same with 1G of RAM
<gchristensen> ah! yes! thank you
<pstn> My mysql server isn't starting any more because the service wants to connect as root but is denied: https://share.xndr.de/iZae2sae9kai0sei8Ohquilee0aola/mysql.log Trying to bring up mysqld_safe is also not succesfull.
stumble has left #nixos ["ERC (IRC client for Emacs 25.3.1)"]
<pstn> https://share.xndr.de/zievaC4Avi1thahwohbumuquur2oof/mysql.nix Here is the config of the container.
<{^_^}> [nixpkgs] @xeji pushed commit from @Synthetica9 to gcc8 « nasm: set -Wno-error=attributes for gcc8 (#40431) »: https://git.io/vp9IZ
<{^_^}> [nixpkgs] @xeji merged pull request #40431 → nasm: set -Wno-error=attributes for gcc8 → https://git.io/vp9U9
<Dezgeg> sphalerite: huh, what happens with this same u-boot binary I have: https://www.cs.helsinki.fi/u/tmtynkky/u-boot.bin
<sphalerite> Dezgeg: same hting
<pstn> I could just reset the server and add my db dump, but I'd rather recover.
<Dezgeg> well, then I'm out of ideas
stumble has joined #nixos
<sphalerite> :/
<Unode> does anyone here use stack installed via nix (nix-env -iA stack) and does it read your configuration ? I can't get stack to load my settings on ~/.ghc/ghci.conf which seems to have been fixed back in 2017 (https://github.com/commercialhaskell/stack/issues/2937)
<Myrl-saki> If I understand correctly
<sphalerite> https://www.coreboot.org/Initramfs how would I adapt these instructions for a zImage and ARM?
<Dezgeg> I guess time to extract the kernel and do the -kernel zImage -initrd initrd.img ...
<Myrl-saki> Actually, I don't understand correctly.
<Myrl-saki> I'm confused. If I need a program in runtime, then use buildInputs?
<sphalerite> whoops, just realised I'm running late and need to leave.
<sphalerite> Thanks for the help Dezgeg, I'll try that later, tomorrow probably :)
<Myrl-saki> Err.
<Myrl-saki> Wait.
<Myrl-saki> Why shouldn't I use `depsTargetTarget` for things that have to run at runtime?
<gchristensen> Dezgeg: I wonder if this machine doesn't support bzImage? https://gist.github.com/grahamc/979a7b09625c84f833dd5f9ea3e0cecc has the debug log
<Dezgeg> non-bzImages got dropped like 15 years ago
<Myrl-saki> For example, if my package outputs a bash script, wouldn't I use `depsTargeTarget`?
<Dezgeg> but it's clearly just ignoring your initrd
<gchristensen> ahhh
stumble has quit [Remote host closed the connection]
stumble has joined #nixos
<gchristensen> maybe I should try passing initrd=initrd like I do for the ARM boxes
<Dezgeg> I've never heard of that, is that something ipxe-specific?
<gchristensen> I don't know, but the ARM boxes had a similar problem and this was the advice I got
<gchristensen> gsc.io/nixos-packet-aarch64-2018-04-12v1/netboot.ipxe
<Dezgeg> ok, strange
logzet has joined #nixos
stumble has quit [Remote host closed the connection]
<Dezgeg> ah, it's some EFI-specific option, no wonder I haven't run to it :P
<{^_^}> [nixpkgs] @Ma27 opened pull request #40433 → sasview: fix build → https://git.io/vp9Lf
stumble has joined #nixos
rihards has joined #nixos
<{^_^}> [nixpkgs] @xeji closed pull request #37097 → rspamd: 1.6.6 -> 1.7.0 → https://git.io/vxLDq
<Dezgeg> are the rest of the servers not doing EFI boot then?
<gchristensen> ahhh yeah thats it
Neo-- has joined #nixos
<Myrl-saki> error: Nix database directory ‘/nix/var/nix/db’ is not writable: Permission denied
<Myrl-saki> Oh wait.
<Myrl-saki> Is nix-repl only usable on Nix1?
<{^_^}> [nixpkgs] @FRidh merged pull request #38851 → Add nvidia_x11 as a dependency in Theano → https://git.io/vxxyw
<{^_^}> [nixpkgs] @FRidh pushed 3 commits to master: https://git.io/vp9LW
<{^_^}> → 6d8bf113 by @smatting: libgpuarray: add nvidia_x11 as dependency when cudaSupport
<{^_^}> → e0cf5824 by @smatting: Theano: fix broken compiler; Add nvidia_x11 as dependency when cudaSupport
<{^_^}> → 91e9f1d7 by @FRidh: Merge pull request #38851 from smatting/fixtheano
<sphalerit> Myrl-saki: it should work with 2.0 as well, but you might need to set NIX_REMOTE=daemon
<sphalerit> Or you could use `nix repl` instead
xcmw has joined #nixos
Leo` has joined #nixos
schoppenhauer has quit [Ping timeout: 240 seconds]
<{^_^}> [nixpkgs] @sshisk opened pull request #40434 → Update rspamd →
xcmw has quit [Client Quit]
<Myrl-saki> Err
<Myrl-saki> ix-build -E '(import <nixpkgs> {crossSystem = (import <nixpkgs/lib>).systems.examples.aarch64-multiplatform;}).callPackage ./tools/nix {}'
<Myrl-saki> no configure script, doing nothing
<Myrl-saki> installing
<Myrl-saki> /nix/store/0yp2vyfsq689srdv50rrmxzd5ls664vs-stdenv/setup: line 1224: /nix/store/nwjvnq6r482869g1j0fn48fby27y41dx-coreutils-8.29-aarch64-unknown-linux-gnu/bin/mkdir: cannot execute binary file: Exec format error
<{^_^}> [nixpkgs] @sshisk opened pull request #40435 → doRename: use visible attribute → https://git.io/vp9Lp
<jD91mZM2> sphalerit: Thanks for telling me about scaleway! Looks really nice :)
<Myrl-saki> Hmm, was my fault, aparently.
<jD91mZM2> sphalerite* (is this your alt?)
schoppenhauer has joined #nixos
xcmw has joined #nixos
<Myrl-saki> Wtf.
logzet_ has joined #nixos
logzet has quit [Ping timeout: 265 seconds]
<Myrl-saki> So, cross compiling.
<Myrl-saki> A certain string in `checkPhase` makes `configurePhase` fail.
<Myrl-saki> Even with `doCheck = false;`
xcmw has quit [Client Quit]
<Myrl-saki> I "get" why.
<Myrl-saki> Okay, I don't. I have a guess as to why.
<Myrl-saki> phases = [ "unpackPhase" "patchPhase" "checkPhase" "configurePhase" "installPhase" ];
<Myrl-saki> (The order is intentional)
<Myrl-saki> checkPhase has this.
<Myrl-saki> echo ./tools/build.sh ${compilerString}test ${gcc}/bin/$gcc
<Myrl-saki> Specifically, `${gcc}` is making configurePhase not run.
<Myrl-saki> configuring
<Myrl-saki> no configure script, doing nothing
<{^_^}> [nixpkgs] @volth opened pull request #40436 → perlPackages.DBFile: db -> db.lib → https://git.io/vp9tT
<Myrl-saki> This happens even if I remove checkPhase from `phases`.
<jD91mZM2> I'd be OK with this if you said it detected the gcc call inside buildPhase... but wtf
<Myrl-saki> I have a feeling what's happening is that including gcc changes PATH.
<jD91mZM2> Does it work correctly if you run checkPhase afterwards instead?
<Myrl-saki> jD91mZM2: Bash doesn't even seem to work.
<Myrl-saki> jD91mZM2: I placed `exit 1;` in patchPhase. Didn't run.
<Myrl-saki> With the checkPhase.
<Myrl-saki> patching sources
<Myrl-saki> configuring
<Myrl-saki> no configure script, doing nothing
<Myrl-saki> installing
<Myrl-saki> Without checkPhase.
<Myrl-saki> patching sources
<Myrl-saki> builder for '/nix/store/4kfjwb6nw1dcrwdf0rvraipql5pyz38g-hlibc-aarch64-unknown-linux-gnu.drv' failed with exit code 1
xcmw has joined #nixos
<Myrl-saki> Oh right.
<Myrl-saki> The most telling thing.
<Myrl-saki> /nix/store/0yp2vyfsq689srdv50rrmxzd5ls664vs-stdenv/setup: line 1224: /nix/store/nwjvnq6r482869g1j0fn48fby27y41dx-coreutils-8.29-aarch64-unknown-linux-gnu/bin/mkdir: cannot execute binary file: Exec format error
<Myrl-saki> Ofc, that disappears if I remove ${gcc}
<Myrl-saki> So, the fix is...
<Myrl-saki> `checkPhase = if doCheck then ...`
<Myrl-saki> The best part is that this only happens in checkPhase.
<Myrl-saki> Err, no.
<Myrl-saki> It actually happens on other locations.
<jD91mZM2> What happens if instead of ${gcc} you just use "gcc" like you assume it's in your PATH?
d4 has quit [Ping timeout: 256 seconds]
<jD91mZM2> I can usually run `cmake` from various phases without any issues
<Myrl-saki> jD91mZM2: I used to do that, but I had to use clang-unwrapped and not-clang-wrapped in the same project.
<Myrl-saki> Err
<Myrl-saki> not-clang-unwrapped.. so clang-wrapped.
<Myrl-saki> Basically, our clang wrapped script is very incomplete compared to gcc, which we had a .spec for. In order to reduce the errors, I used clang-unwrapped for building things "raw" and clang-wrapped for testing phase.
* Myrl-saki is working on a libc.
<Myrl-saki> jD91mZM2: But yes, before I had to incorporate clang, we used `gcc` in PATH. And it worked.
<Myrl-saki> jD91mZM2: So, apparently, I'm stupid. What was happening was that I was building gcc.
<Myrl-saki> building '/nix/store/fazf3bmymlxdqwm96i4cj8w44wn49vkv-gcc-wrapper-7.3.0-aarch64-unknown-linux-gnu-aarch64-unknown-linux-gnu.drv'...
<Myrl-saki> Oh okay.
logzet_ has quit [Remote host closed the connection]
<Myrl-saki> It's the gcc-wrapper that fails to build.
nD5Xjz has quit [Quit: ZNC 1.6.5 - http://znc.in]
nD5Xjz has joined #nixos
<Myrl-saki> The wrapper itself. GCC itself is fine.
<{^_^}> [nixpkgs] @FRidh pushed 2 commits to staging: https://git.io/vp9tA
<{^_^}> → a4b7be9e by @volth: perlPackages.DBFile: db -> db.lib
<{^_^}> → a96f176f by @FRidh: Merge pull request #40436 from volth/patch-126
<{^_^}> [nixpkgs] @FRidh merged pull request #40436 → perlPackages.DBFile: db -> db.lib → https://git.io/vp9tT
<Myrl-saki> IOTW, I'll go send this closure to a v8 machine.
logzet has joined #nixos
jacob has joined #nixos
jacob is now known as Guest78178
d4 has joined #nixos
d4 has quit [Client Quit]
<giaco> hello! Having problem in running nix compiled from source. error: undefined variable 'exec_prefix' at <my prefix>/share/nix/corepkgs/config.nix:14:40
smallville7123 has quit [Quit: rosa]
<{^_^}> [nixpkgs] @Mic92 opened pull request #40438 → xorgserver: add mesa_noglu for dri package → https://git.io/vp9qD
<{^_^}> [nixpkgs] @florianjacob opened pull request #40439 → matomo: 3.4.0 -> 3.5.0 → https://git.io/vp9qS
rokka has joined #nixos
<Myrl-saki> Well.
<Myrl-saki> This is just confusing me.
<Myrl-saki> hardeningDisable=all exec /nix/store/kqlbhd3jhjbz2d9ir1nbbdlbbfayyaqa-aarch64-unknown-linux-gnu-gcc-cross-wrapper-7.3.0-aarch64-unknown-linux-gnu-stage-final/bin/gcc
<Myrl-saki> myrl@myrl:~/Dev/hlibc[6▲ clang-nix••]$ ls /nix/store/kqlbhd3jhjbz2d9ir1nbbdlbbfayyaqa-aarch64-unknown-linux-gnu-gcc-cross-wrapper-7.3.0-aarch64-unknown-linux-gnu-stage-final/bin/gcc
<Myrl-saki> ls: cannot access '/nix/store/kqlbhd3jhjbz2d9ir1nbbdlbbfayyaqa-aarch64-unknown-linux-gnu-gcc-cross-wrapper-7.3.0-aarch64-unknown-linux-gnu-stage-final/bin/gcc': No such file or directory
<Myrl-saki> Oh, I'm stupid.
<{^_^}> [nixpkgs] @ruuda opened pull request #40440 → libressl: init at 2.7, change url to https → https://git.io/vp9me
phdoerfler has quit [Quit: Leaving.]
<sphalerit> jD91mZM2: yeah sphalerit is matrix, sphalerite is weechat
<Myrl-saki> Ughh
<Myrl-saki> `buildPackages.buildutils` gets me `ar: Command not found`
<{^_^}> [nixpkgs] @FRidh merged pull request #40425 → avocode: init at 2.26.0 → https://git.io/vp9vt
<{^_^}> [nixpkgs] @FRidh pushed commit from @megheaiulian to master « avocode: init at 2.26.0 (#40425) »: https://git.io/vp9mV
<Myrl-saki> I giff up.
acarrico has joined #nixos
broccoli has quit [Ping timeout: 264 seconds]
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
joehh has joined #nixos
rokka has quit [Quit: rip]
realrokka has joined #nixos
phdoerfler has joined #nixos
<Mic92_> Myrl-saki: ar is in binutils
<Myrl-saki> Mic92_: Yeah, my mistake, that was supposed to be binutils.
<Myrl-saki> But yeah, `ar: Command not found` this is with cross compilation.
ambro718 has quit [Quit: Konversation terminated!]
<{^_^}> [nixpkgs] @volth opened pull request #40441 → cpan2nix: update perl packages → https://git.io/vp9Yb
goibhniu has joined #nixos
juhe has quit [Quit: Leaving.]
<giaco> Do you know how to tell the builder where to find libs? It seems not considering LD_LIBRARY_PATH
<Sonarpulse> Myrl-saki: hi
griff_ has quit [Quit: griff_]
<Sonarpulse> elvishjerricco: ping if you are up
<elvishjerricco> Sonarpulse: hi
<Sonarpulse> elvishjerricco: morning!
<Sonarpulse> yeah I left you those comments and just want to sleep
acarrico has quit [Ping timeout: 264 seconds]
<Sonarpulse> elvishjerricco: just got on my computer now, so I
<Sonarpulse> 'm going to start with the --extra-framework one
<elvishjerricco> Sonarpulse: Going out to breakfast with the family now. Will take a look at your responses in a few hours
<Sonarpulse> elvishjerricco: ok sounds good
joehh has quit [Ping timeout: 255 seconds]
<Myrl-saki> Sonarpulse: What's the use of mapOffset?
<Myrl-saki> Sonarpulse: I've read it a lot over, and I think I understand most of it now.
<Sonarpulse> Myrl-saki: you'd be one of the few, congrats :)
<Myrl-saki> Sonarpulse: :P
<Myrl-saki> Sonarpulse: Can I ask you a few stuff?
<Sonarpulse> yes
<Sonarpulse> the point of map offset is to "squeeze" or "stretch" the offsets
<Sonarpulse> were mapping the offsets from the child's perspect onto the parents perspective
<Sonarpulse> Myrl-saki: are you interested in reimplementing findInputs in nix by any chance?
<Sonarpulse> I want to do that for testing and teaching
<Sonarpulse> there is an old function that did the old propagation in lib/deprecated.nix
<Sonarpulse> would need to massively overhaul it
<Myrl-saki> Sonarpulse: That sounds like it'd be cool to do, but I don't understand how there's an interaction between bash and nix.
<Sonarpulse> Myrl-saki: well findInputs is bash
<Myrl-saki> Sonarpulse: Right. I mean I don't understand how findInputs can be in bash.
<Sonarpulse> Myrl-saki: oh look at the bottom of stdenv setup
<Sonarpulse> the progpagated inputs are put into bash
<Sonarpulse> *put into files in nix-support
<Sonarpulse> Myrl-saki: search for the variable propagatedInputsFile in particular
<Sonarpulse> in fixup phpase
<Myrl-saki> Oh wait, I think I understand it now.
<Myrl-saki> Sonarpulse: Is this used to map the propagated inputs' build and host offsets?
<Myrl-saki> s/build and host/host and target/
<Sonarpulse> Myrl-saki: those files?
<Sonarpulse> sort of?
<Sonarpulse> we go through the offsets, and then see if there is any file of deps matching those offsets
<Myrl-saki> Wait, brb, I have to leave for an hour.
<Sonarpulse> ok
<Myrl-saki> I'll just message you when I'm back, I'm interested in understand this.
<Sonarpulse> Myrl-saki: ok
<Sonarpulse> Myrl-saki: I have to go out about then too :(, but I'll keep laptop on so you can see ping
rihards has quit [Quit: rihards]
<{^_^}> [nixpkgs] @Ericson2314 opened pull request #40442 → haskell-generic-builder: Add extra framework dirs → https://git.io/vp93H
nuncanada has joined #nixos
<{^_^}> [nixpkgs] @Ericson2314 pushed 55 commits to staging-18.03: https://git.io/vp9ss
<{^_^}> → 302c4c5f by @Ericson2314: stdenv: Put back isArm, with deprecation notice.
<{^_^}> → 042cd6ec by @taku0: oraclejdk: add version 10
<{^_^}> → 5d7d0842 by @matthewbauer: openjdk: get rid of warning
<{^_^}> [nixpkgs] @Ericson2314 opened pull request #40443 → haskell-generic-builder: Add extra framework dirs → https://git.io/vp9sn
xcmw has joined #nixos
asuryawanshi has quit [Remote host closed the connection]
asuryawanshi has joined #nixos
asuryawanshi has quit [Remote host closed the connection]
asuryawanshi has joined #nixos
devhell has joined #nixos
devhell has quit [Client Quit]
Ariakenom has joined #nixos
blum has joined #nixos
RyanGlScott has joined #nixos
<blum> Hello. I just noticed my laptop loses its wired network connection when not charging. It's probably some wierd power management option, but I am not sure where could that be coming from, and Google isn't being useful. Any suggestions?
<RyanGlScott> A question about how the nixpkgs repo is organized. If I see something like:
<RyanGlScott> "let pkgs = import <nixpkgs> { }; in pkgs.haskellPackages.callPackage ..."
<manveru> blum: powertop maybe
<RyanGlScott> Does pkgs.haskellPackages.callPackage correspond to a particular subdirectory of nixpkgs?
<FRidh> RyanGlScott: no, its an attribute
<blum> manveru: Will try to install it and mess around for a bit. Thanks.
<manveru> RyanGlScott: not really, you'll have to check all-packages.nix for the haskellPackages attribute and follow that path
<RyanGlScott> Perhaps a better way to phrase my question is: how /do/ folks generally look up the source code for things they get from nixpkgs?
<RyanGlScott> I mean, it clearly lives somewhere
<RyanGlScott> I'm just not sure how the mapping works from a nixpkgs attribute to its home on the GitHub repo
pingu_ has joined #nixos
<{^_^}> [nixpkgs] @xeji merged pull request #40433 → sasview: fix build → https://git.io/vp9Lf
<{^_^}> [nixpkgs] @xeji pushed commit from @Ma27 to master « sasview: fix build (#40433) »: https://git.io/vp9G4
<{^_^}> [nixpkgs] @xeji merged pull request #40432 → libglvnd: set -Wno-error=array-bounds → https://git.io/vp9Te
<{^_^}> [nixpkgs] @xeji pushed commit from @Synthetica9 to gcc8 « libglvnd: set -Wno-error=array-bounds (#40432) »: https://git.io/vp9GE
<RyanGlScott> manveru: Thanks, it looks like pkgs/top-level is what I was looking for.
<manveru> RyanGlScott: you can also do something like `(import <nixpkgs> {}).haskellPackages.callPackage` in `nix repl` and see what the location is
<samueldr> RyanGlScott: it won't be able to be used for attribute sets (like haskellPackages) without a package name
<samueldr> but for a package, e.g. ACME: nix edit nixpkgs.haskellPackages.ACME
<{^_^}> [nixpkgs] @xeji pushed commit from @vbgl to master « ocamlPackages.bitstring: 2.1.1 -> 3.0.0 (#40034) »: https://git.io/vp9G2
<{^_^}> [nixpkgs] @xeji merged pull request #40034 → ocamlPackages.bitstring: 2.1.1 -> 3.0.0 → https://git.io/vprwF
<RyanGlScott> Oh wow, `nix edit` is nice
Myrl-saki has quit [Ping timeout: 256 seconds]
Have-Quick has joined #nixos
<giaco> do you have any idea why nix ignores LD_LIBRARY_PATH?
phdoerfler has quit [Quit: Leaving.]
Myrl-saki has joined #nixos
<{^_^}> [nixpkgs] @jfrankenau opened pull request #40444 → nginxModules.ipscrub: init at 99230f6 → https://git.io/vp9Zk
RyanGlScott has left #nixos [#nixos]
<{^_^}> [nixpkgs] @Ericson2314 opened pull request #40445 → Guile: Backport cross fixes → https://git.io/vp9ZY
<infinisil> > builtins.unsafeGetAttrPos "xmonad" haskellPackages # This is how nix edit does it
<{^_^}> { column = 3; file = "/var/lib/nixbot/state/nixpkgs/pkgs/development/haskell-modules/configuration-nix.nix"; line = 217; }
<{^_^}> [nixpkgs] @ttuegel merged pull request #40420 → kde-gtk-config: Set path for GSettings schemas → https://git.io/vpS5S
<{^_^}> [nixpkgs] @ttuegel pushed 2 commits to master: https://git.io/vp9Zc
<{^_^}> → 1c3492f3 by @ttuegel: kde-gtk-config: Set path for GSettings schemas
<{^_^}> → 1df114a8 by @ttuegel: Merge pull request #40420 from ttuegel/bugfix/kde-gtk-config/gsettings
asuryawanshi has quit [Ping timeout: 240 seconds]
<gchristensen> anyone have NixOS's GRUB working over a serial console? I've added serial --unit=0 --speed=115200 --word=8 --parity=no --stop=1; terminal --timeout=10 serial console; copying from a known-good config (Ubuntu's) and it isn't rendering properly
blum has quit [Quit: blum]
<{^_^}> [nixpkgs] @xeji merged pull request #40439 → matomo: 3.4.0 -> 3.5.0 → https://git.io/vp9qS
<{^_^}> [nixpkgs] @xeji pushed commit from @florianjacob to master « matomo: 3.4.0 -> 3.5.0 (#40439) »: https://git.io/vp9Zz
<{^_^}> [nixpkgs] @Ericson2314 opened pull request #40446 → guile: Do nothing! → https://git.io/vp9Zg
<Dezgeg> I think I have seen uefi grub do the right thing on serial
<Dezgeg> also this: nixos/tests/installer.nix: boot.loader.grub.extraConfig = "serial; terminal_output.serial";
<gchristensen> yeah, I tried those lines as well
<Dezgeg> aybe it only works on a 80x24? :P
<gchristensen> :D
<gchristensen> https://gist.github.com/grahamc/6ee18ab09846e97e4a29f3d16675be1f here is the working config from Ubuntu, on the same class hardware
<{^_^}> [nixpkgs] @ttuegel opened pull request #40447 → kde-gtk-config: Set path for GSettings schemas → https://git.io/vp9nJ
<gchristensen> updated it to include the broken config from NixOS
<Dezgeg> IIRC there was something with the GRUB on the ThunderX where it would work if most of the NixOS boilerplate was commented out...
<gchristensen> I wonder if the insmods are breaking it?
<gchristensen> yeah! that is what I'm wondering :D
<gchristensen> I'm going AFK for an hour or so, and I'll have more fun with this when I get back.
<samueldr> I'm curious if it's the font and/or gfx payload
FRidh has quit [Quit: Konversation terminated!]
jD91mZM2 has quit [Quit: WeeChat 2.0]
<gchristensen> I could rmmod those things... but I'll try that latere
leat has quit [Ping timeout: 240 seconds]
pingu_ has quit [Ping timeout: 240 seconds]
RyanGlScott has joined #nixos
<RyanGlScott> How do I combine the -A and -p flags with nix-shell?
<jcrben> giaco: could you share a link to the code you're dealing with? not that I'm going to be much help
<RyanGlScott> If I run `nix-shell release.nix -A env`, it works
<RyanGlScott> If I run `nix-shell release.nix -p haskell.compiler.ghc822 -A env`, it fails with:
<RyanGlScott> `error: attribute 'env' in selection path 'env' not found`
<{^_^}> [nixpkgs] @Ericson2314 pushed 6 commits to staging-18.03: https://git.io/vp9nM
<{^_^}> → 0fcad047 by @shlevy: Guile: Fix cross-compilation
<{^_^}> → 84156826 by @Ericson2314: guile: Fix cross more, including 1.8
<{^_^}> [nixpkgs] @Ericson2314 merged pull request #40445 → guile: Backport cross fixes → https://git.io/vp9ZY
<{^_^}> → 9a4cc935 by @mdorman: guile-1.8: Fix and clean up `libtool` propagation fixes
<{^_^}> [nixpkgs] @Ericson2314 merged pull request #40446 → guile: Do nothing! → https://git.io/vp9Zg
<{^_^}> [nixpkgs] @Ericson2314 pushed 6 commits to master: https://git.io/vp9ny
<{^_^}> → 0fcad047 by @shlevy: Guile: Fix cross-compilation
<{^_^}> → 84156826 by @Ericson2314: guile: Fix cross more, including 1.8
<{^_^}> → 9a4cc935 by @mdorman: guile-1.8: Fix and clean up `libtool` propagation fixes
<giaco> jcrben: sure. Take a look at this strace
<{^_^}> [nixpkgs] @xeji merged pull request #40416 → nixos: fix some types → https://git.io/vpS56
<{^_^}> [nixpkgs] @xeji pushed 3 commits to master: https://git.io/vp9n9
<{^_^}> → 44d612d6 by @oxij: nixos: services.xserver.desktopManager.xterm.enable: fix type
<{^_^}> → fb2a8d66 by @oxij: nixos: systemd.services.*.environment: fix type
<{^_^}> → 8c78d3a5 by @xeji: Merge pull request #40416 from oxij/os/fix-some-types
<jcrben> RyanGIScott: man nix-shell shows "{{--packages | -p} packages... | [path]}" - written like that it kinda seems like it's one or the other write?
<giaco> jcrben: I am trying to run nix from a custom prefix environment. Make and make install goes well, but nix seems to spawn subprocesses that does not consider parent LD_LIBRARY_PATH
<RyanGlScott> OK. So how do I evaluate default.nix using a particular package, then?
<jcrben> write -> right
leat has joined #nixos
<jcrben> using -E gives you more flexibility. for example this is a nix-shell command I used recently: nix-shell -E 'with (import <nixpkgs> {}); callPackage /home/ben/nix-group/nixpkgs/pkgs/development/libraries/libinput/default.nix {}'
<RyanGlScott> I'm not sure how that solves the problem...
<MichaelRaskin> drakonis: by the way, I am not enthusiastic about shipping SVG icons for LO just separately, but that means I don't want this in the main build, and I don't want to test it myself with environments I try to avoid. If you want to submit a patch that fetches these icons and installs them in the wrapper, I will merge it.
Sonarpulse has quit [Ping timeout: 276 seconds]
<RyanGlScott> I suppose I'd want to say `nix-shell -p haskell.compiler.ghc822 default.nix`
<RyanGlScott> But it doesn't let me do that
<RyanGlScott> So I can either get a shell with a particular version of GHC, or I can build default.nix. But I can't do both.
<infinisil> RyanGlScott: Evaluate a default.nix with a specific package? Do you mean a default.nix that has a bunch of { stdenv, foo, bar }: at the top?
<RyanGlScott> Yes
rihards has joined #nixos
<RyanGlScott> I want to evaluate that, but with a particular version of GHC
<infinisil> ,callPackage RyanGlScott
<{^_^}> RyanGlScott: If a Nix file ./foo.nix starts with something like `{ bar, baz }:`, you can build it with `nix-build -E '(import <nixpkgs> {}).callPackage ./foo.nix {}'`
<jcrben> giaco: I see you opened https://github.com/NixOS/nix/issues/2160 - I'm not sure but it seems like you might need to build a custom nix-env which looks for libandroid or has that environment variable passed in
<infinisil> RyanGlScott: That, but place the arguments you want to override in the last {}
<RyanGlScott> Surely there's a more succint way to do that?
<RyanGlScott> I'd like to quickly switch between GHC versions for testing purposes
<RyanGlScott> Preferably with nix-shell
<jcrben> RyanGIScott: put it in a bash alias or function? it seems reasonably succinct
<giaco> jcrben: yes. but such lib is already in LD_LIBRARY_PATH
<RyanGlScott> I mean, `nix-shell -p haskell.compiler.ghc822` is nice and short
<jcrben> RyganGIScott: you should see some of my docker commands with a half-dozen switches...
<RyanGlScott> I don't see why that can't be extended to envelop a .nix file
pacman99 has joined #nixos
<infinisil> RyanGlScott: I mean, if you have it as an attrset, you can do `the.attrset.override { ghc = <other version>; }`
<RyanGlScott> infinisil: Apologies, I'm not quite sure where that it supposed to go.
<giaco> jcrben: problem seems that nix-channel (and probably other nix executables) spawns different sub pids and LD_LIBRARY_PATH is not passed
<infinisil> And if you had a file without such arguments but instead something like `{ pkgs ? import <nixpkgs> {}, compiler ? "ghc822" }:` at the top, you could call that with `nix-shell --argstr compiler ghc842`
<rihards> how can I get something like nix-channel --list-generations? i want to use nix-channel --rollback, but I want to know the dates when nix-channel --update was used
silver has joined #nixos
<RyanGlScott> Hm. Is there a way I can get cabal2nix to do this?
<hodapp> What's the "right" way to deal with having one custom Python dependency that needs a custom version of something (say, numpy), and then needing several libraries alongside it that all also require numpy, and thus need that same custom version?
<hodapp> right now I've been copying the entire derivation, changing numpy to the custom version, and then doing this recursively, manually, for everything that depends on those
<{^_^}> [nixpkgs] @dotlambda pushed to master « home-assistant: 0.69.0 -> 0.69.1 »: https://git.io/vp9cg
<jcrben> RyanGIScott: nix run --help - it seems like that has better UX
<{^_^}> [nixpkgs] @xeji merged pull request #40430 → giac: remove liblapackWithAtlas dependency → https://git.io/vp9UV
<{^_^}> [nixpkgs] @xeji pushed commit from @timokau to master « giac: remove liblapackWithAtlas dependency (#40430) »: https://git.io/vp9co
Have-Quick has quit [Quit: Have-Quick]
<RyanGlScott> jcrben: It's not clear to me after reading that how to accomplish what I need with `nix run`
<RyanGlScott> For one thing, I'm not sure what the equivalent of `-A env` is
<jcrben> RyanGIScott: yep, doesn't really have good docs. anyhow, if I were you I'd just write a bash function and do something like np <versionnumber> which uses -E and callPackage
<RyanGlScott> Well, I give up for now. I'll hold off on using Nix for my project until I figure this out.
<hodapp> hm... is there basically a local rather than global packageOverrides?
<jcrben> RyanGIScott: I hardly ever write out long commands more than once. not good for sanity
<infinisil> RyanGlScott: Can you give a bit more context? I don't quiet understand your problem
<RyanGlScott> infinisil: I have a Haskell library that I want to build with different versions of GHC.
<RyanGlScott> cabal2nix gives me a nice way to come up with a default.nix file
<infinisil> hodapp: overlays should work
<RyanGlScott> And then I can run it with nix-env -A env default.nix
<RyanGlScott> The problem is that locks me in to a particular GHC version
Have-Quick has joined #nixos
cnidario has joined #nixos
<RyanGlScott> I need a simple way to switch between GHC versions to ensure that my project builds with all of them
<RyanGlScott> Surely I'm not the only person that's desired this :)
<jcrben> RyanGIScott: you do realize that passing an expression with -E works perfectly, right? but you're saying it's too verbose and you don't want to use an alias or bash function?
<infinisil> RyanGlScott: You're using the default cabal2nix . > default.nix and then build it with a different file that contains `pkgs.haskellPackages.callPackage ./default.nix {}`?
<RyanGlScott> infinisil: Right, I have a separate release.nix file which does the latter.
<infinisil> And you build it with `nix-build release.nix`?
<RyanGlScott> Currently it's:
<RyanGlScott> `let pkgs = import <nixpkgs> { }; in pkgs.haskellPackages.callPackage ./default.nix { }`
<RyanGlScott> Which is simple, but it works
<infinisil> Alright, then add an argument at the top of that file: `{ compiler ? "ghc822" }:`
<infinisil> Use `pkgs.haskell.packages.${compiler}.callPackage` instead
<infinisil> And build it with `nix-build release.nix --argstr compiler ghc842` for a different compiler
<RyanGlScott> infinisil: Wow, that's nice!
<RyanGlScott> Thanks.
<RyanGlScott> And apologies for drowning the channel with noob questions :)
<infinisil> The channel is there for that, np :)
<RyanGlScott> Another question I've wondered for a while
deepfire` has quit [Ping timeout: 240 seconds]
<jcrben> I'm a noob too. keep in mind tho that nix-shell does not accept the same params as nix-build
<RyanGlScott> Why is it that when building Haskell packages from source, it bothers to run all of the tests?
<infinisil> jcrben: Such as?
<RyanGlScott> Even if the package is only used as a dependency?
<infinisil> Dependencies should be tested too, why wouldn't they have to be?
akapav has quit [Quit: ZNC 1.6.5 - http://znc.in]
jD91mZM2 has joined #nixos
<jcrben> infinisil: for example --arg - manpage says "This option is accepted by nix-env, nix-instantiate and nix-build"
<RyanGlScott> Right, but even if you're only building a library component?
<infinisil> jcrben: Nah nix-shell also accepts --arg, see its manpage
<infinisil> Bbl, gotta eat
<jcrben> infinisil: that quote comes from the nix-shell manpage. but perhaps a docs bug. I wish that nix-shell accepted the -f flag that nix-env uses, too
akapav has joined #nixos
deepfire` has joined #nixos
<hodapp> for the example in https://nixos.org/nixpkgs/manual/#how-to-override-a-python-package-using-overlays how does one use this? is this for config.nix?
RyanGlScott has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]
Have-Quick has quit [Quit: Have-Quick]
<jcrben> RyanGIScott: as far as running tests, that's controlled by doCheck - you could probably override it if you'd like? https://github.com/NixOS/nixpkgs/blob/master/pkgs/development/haskell-modules/generic-builder.nix
<jcrben> RyanGIScott: if you've got the time, I found that the nix pills series helped a lot...
knupfer has joined #nixos
Have-Quick has joined #nixos
maingo has joined #nixos
ryantrinkle has joined #nixos
Have-Quick has quit [Quit: Have-Quick]
ilyaigpetrov has joined #nixos
Have-Quick has joined #nixos
broccoli has joined #nixos
orivej has joined #nixos
<mkaito> I'm writing a package for a java app. The derivation is basically just "download jar and run it with jre". How do I do the "run with jre" part in the most straightforward way?
<{^_^}> [nixpkgs] @volth opened pull request #40449 → perlPackages.GD: fix build on staging → https://git.io/vp9W8
<{^_^}> [nixpkgs] @dezgeg merged pull request #40362 → libb2: 0.97 -> 0.98 and fix build on ARM → https://git.io/vpyk5
<{^_^}> [nixpkgs] @dezgeg pushed 2 commits to master: https://git.io/vp9Wu
<{^_^}> → 03318efe by @dotlambda: libb2: 0.97 -> 0.98
<{^_^}> → 409b97c9 by @dotlambda: libb2: fix build on ARM
Have-Quick has quit [Quit: Have-Quick]
alex`` has joined #nixos
sary has quit [Ping timeout: 260 seconds]
Have-Quick has joined #nixos
<jD91mZM2> Wow ok my /boot just filled up lol
<jD91mZM2> How do I clean it up?
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
camsbury has joined #nixos
<hodapp> okay, I think I grok how to use an overlay within a shell.nix but still have a hard time understanding the example at https://nixos.org/nixpkgs/manual/#how-to-override-a-python-package-using-overlays
<hodapp> https://github.com/NixOS/nixpkgs/issues/6756#issuecomment-288005837 and why isn't that doing this with an overlay? or is it?
ThatPako has joined #nixos
<ThatPako> Any idea why compiling with `openjdk8` and `gradle` as buildinputs results in the compiler complaining about JavaFX stuff missing?
Have-Quick has quit [Quit: Have-Quick]
xcmw has joined #nixos
xcmw has quit [Client Quit]
griff_ has joined #nixos
<infinisil> mkaito: Should be just using the openjdk package
<infinisil> mkaito: ${openjdk}/bin/java -jar /path/to.jar
<infinisil> > src = a: if a ? src then let pos = builtins.unsafeGetAttrPos "src" a; in "https://github.com/NixOS/nixpkgs/blob/master${lib.removePrefix (toString <nixpkgs>) pos.file}#L${toString pos.line}" else "Can't get source of this attribute set"
<{^_^}> src defined
<infinisil> > src hello
<infinisil> Very nice
jD91mZM2 has quit [Quit: WeeChat 2.0]
<{^_^}> [nixpkgs] @xeji pushed commit from @florianjacob to release-18.03 « matomo: 3.4.0 -> 3.5.0 (#40439) »: https://git.io/vp98v
jD91mZM2 has joined #nixos
xcmw has joined #nixos
pingu_ has joined #nixos
marusich has joined #nixos
deepfire` has quit [Ping timeout: 265 seconds]
nick_l has joined #nixos
<nick_l> How do I convert this into something that I can put in configuration.nix? nix-env -iA hies -f https://github.com/domenkozar/hie-nix/tarball/master
<giaco> is it possible to set to nix-env a custom LD_LIBRARY_PATH or a custom LD_PRELOAD when it launches subprocesses? I have a working nix-env that fails in launching bash as my bash needs a LD_LIBRARY_PATH that nix-env seems dropping
rauno has quit [Ping timeout: 240 seconds]
deepfire` has joined #nixos
ThatPako has quit [Quit: Lost terminal]
<nick_l> Was process accounting recently fixed in NixOS?
<nick_l> I remember atop to always show that it "was not enabled", but somewhere around 18.03 it started working.
cnidario has quit [Read error: Connection reset by peer]
<{^_^}> [nixpkgs] @xeji merged pull request #40373 → nixos/mesos: fix non-deterministic service and test failure → https://git.io/vpy4L
<{^_^}> [nixpkgs] @xeji pushed 3 commits to master: https://git.io/vp94f
<{^_^}> → 0d2a4ed4 by @xeji: nixos/tests/mesos: wait for all services and ports
<{^_^}> → ec00b6fb by @xeji: nixos/mesos-slave: start after docker
<{^_^}> → c482b41f by @xeji: Merge pull request #40373 from xeji/fix-mesos-test
<nick_l> infinisil: that's the kind of stuff I want to see in NixOS.
<{^_^}> [nixpkgs] @xeji merged pull request #40380 → virtmanager: add missing virt-install runtime dependencies → https://git.io/vpyXJ
<{^_^}> [nixpkgs] @xeji pushed to master « virtmanager: add missing virt-install runtime deps (#40380) »: https://git.io/vp94T
<infinisil> Heh :P
<nick_l> infinisil: compiling Haskell code gets old *so* fast.
jD91mZM2 has quit [Quit: WeeChat 2.0]
<infinisil> You mean manually without hie?
<{^_^}> [nixpkgs] @xeji merged pull request #40410 → nixos/tests/acme: fix on i686, improve timing → https://git.io/vpSQe
<{^_^}> [nixpkgs] @xeji pushed to master « nixos/tests/acme: fix on i686, improve timing (#40410) »: https://git.io/vp94q
<nick_l> infinisil: no, I haven't used hie before, but hie is now compiling a pile of code.
<infinisil> Ah yeah, no binary caches for that, it takes a while
<nick_l> infinisil: before Nix I actually had to compile everything manually, but now I just don't know how much I am going to torture my CPU.
<nick_l> infinisil: I have an i7, but it still feels as if it's not doing anything :P
marusich has quit [Ping timeout: 260 seconds]
<nick_l> infinisil: I did configure my machine to run Nix processes with low I/O and CPU priority, because otherwise I wouldn't be able to type anything anymore.
<nick_l> What's the most cores one can get in a laptop? 8?
<{^_^}> [nixpkgs] @xeji pushed to release-18.03 « virtmanager: add missing virt-install runtime deps (#40380) »: https://git.io/vp944
rauno has joined #nixos
jD91mZM2 has joined #nixos
ryantrinkle has quit [Ping timeout: 240 seconds]
griff_ has quit [Quit: griff_]
<Izorkin> how to send folders to remote pc with nixops?
freeman42x]NixOS has joined #nixos
raynold has quit [Quit: Connection closed for inactivity]
don_quijote[m] has joined #nixos
knupfer has quit [Remote host closed the connection]
<nick_l> Izorkin: there is nixops scp, if that's what you mean.
<nick_l> cat /proc/cpuinfo even lists bugs now. We are truly living in the future.
knupfer has joined #nixos
udono has joined #nixos
orivej has quit [Ping timeout: 264 seconds]
<Izorkin> nick_l: need automatic copying config files to remote pc with nixops deploy config
pako has joined #nixos
<nick_l> Izorkin: in that case, I would just build a nix expression. As such there won't really be a "copying concept".
<nick_l> Izorkin: That is, act if your config files are just another package.
<nick_l> Izorkin: you could first compress all the files you want to send, for example.
<nick_l> Also depends on the amount of files.
camsbury has quit [Ping timeout: 268 seconds]
pako has quit [Client Quit]
<nick_l> Load average: 12.88 12.49 10.62 o.O
<giaco> is it possible to set to nix-env a custom LD_LIBRARY_PATH or a custom LD_PRELOAD when it launches subprocesses? I have a working nix-env that fails in launching bash as my bash needs a LD_LIBRARY_PATH that nix-env seems dropping
<Izorkin> nick_l: thanks
<nick_l> Izorkin: are you using it in combination with a cloud provider?
<Izorkin> nick_l: no, home server
rihards has quit [Quit: rihards]
<{^_^}> [nixpkgs] @fadenb opened pull request #40450 → [18.03] graylog: 2.4.3 -> 2.4.4 → https://git.io/vp9Bt
<nick_l> Izorkin: I think using NixOS reduced maintenance costs for me on such home servers.
klntsky has quit [Ping timeout: 255 seconds]
<nick_l> Izorkin: nixops is perhaps overkill for home servers, but if you are using Nix anyway, you can just as well do it.
<Izorkin> me use nixops to create libvirtd virtual pc
jD91mZM2 has quit [Quit: WeeChat 2.0]
klntsky has joined #nixos
<{^_^}> [nixpkgs] @jcrben opened pull request #40451 → libinput: 1.10.0 -> 1.10.6 → https://git.io/vp9BB
ThatPako has joined #nixos
<{^_^}> [nixpkgs] @Mic92 merged pull request #40438 → xorg.xorgserver: fix build by adding mesa_noglu for dri package → https://git.io/vp9qD
<{^_^}> [nixpkgs] @Mic92 pushed 2 commits to master: https://git.io/vp9By
<{^_^}> → 991f8db5 by @Mic92: xorgserver: add mesa_noglu for dri package
<{^_^}> → 8cdbdfb3 by @Mic92: Merge pull request #40438 from Mic92/xorgserver
<infinisil> Izorkin: What folder do you need to have on the remote machine? Where should it be?
broccoli has quit [Ping timeout: 256 seconds]
<{^_^}> [nixpkgs] @xeji closed pull request #40451 → libinput: 1.10.0 -> 1.10.6 → https://git.io/vp9BB
jD91mZM2 has joined #nixos
<Izorkin> infinisil: need send duplicate config (need to working command nixos-rebuild) and custom zsh themes to /etc/nixos folders
<infinisil> Huh? You wanna use nixops or not?
Have-Quick has joined #nixos
<infinisil> It messes stuff up if you use nixops but use nixos-rebuild
aszlig has quit [Quit: Kerneling down for reboot NOW!]
foldingcookie has quit [Ping timeout: 255 seconds]
<Izorkin> infinisil: there were cases when nixops stopped working (not workin sshd, not assigned Ip, etc) Needed as a reserve
<srk> happens to me from time to time as well
aszlig has joined #nixos
<infinisil> Note that if you use nixos-rebuild on your machine you can't use nixops again until you restore the keys
<srk> looks like if the deployment suceeds but fails during service restart I can't nixops ssh to target sometimes
<infinisil> Izorkin: You can do the configuration.nix thing with `nix.nixPath = [ "nixos-config=${/path/to/config.nix}" "nixpkgs=${/path/to/nixpkgs}" ];`
<infinisil> And why would you put custom zsh themes in /etc/nixos? You can just install them as a normal package with systemPackages and refer to it via `/run/current-system/sw/<theme path in theme package>`
<{^_^}> [nixpkgs] @andersk opened pull request #40453 → pytorch: 0.3.1 -> 0.4.0 → https://git.io/vp9Rw
camsbury has joined #nixos
<nick_l> infinisil: How can I use hie without it building the world?
<nick_l> infinisil: it's still building.
<infinisil> Well what is it building?
<{^_^}> [nixpkgs] @dotlambda merged pull request #40450 → [18.03] graylog: 2.4.3 -> 2.4.4 → https://git.io/vp9Bt
<{^_^}> [nixpkgs] @dotlambda pushed 3 commits to release-18.03: https://git.io/vp9RM
<{^_^}> → 8252dee6 by @fadenb: graylog: wrap graylogctl
<{^_^}> → 7d8dbc5f by @fadenb: graylog: 2.4.3 -> 2.4.4
<{^_^}> → 465de598 by @dotlambda: Merge pull request #40450 from fadenb/graylog_2.4.4_on_release-18.03
<infinisil> Because you will need to compile all haskell dependencies, no way around that
<nick_l> infinisil: I saw Codec.Binary.UTF8.Generic
<infinisil> Yeah, no way around that
<nick_l> infinisil: is there nothing cached anywhere?
<infinisil> Nope
<{^_^}> Channel nixos-18.03-small advanced to https://github.com/NixOS/nixpkgs/commit/efd2addff65 (from 82 minutes ago, history: https://channels.nix.gsc.io/nixos-18.03-small)
<nick_l> infinisil: from my experience of using the C tools on NixOS, it sort of seems that Haskell then is completely useless?
<infinisil> Huh?
<nick_l> infinisil: why is the situation so bad?
<infinisil> I don't get it, what's so bad about what?
<nick_l> infinisil: I don't have to compile all the C tools all the time.
<nick_l> infinisil: it looks like I need to setup a build server in order to use Haskell.
<infinisil> hie-nix is using stack2nix, to make it work at all
<infinisil> If hie were packaged in nixpkgs that wouldn't be necessary and you could have it from the cache
<nick_l> infinisil: OK, so why does it do that?
<infinisil> But it turned out to be rather hard to package that one correctly
<infinisil> So domenkozar kindly created hie-nix for it to be usable on nix at all
<nick_l> OK, so it's not Haskell that's broken on NixOS, it's just hie-nix.
<infinisil> Why would you need to set up a build server just to compile a thing?
<infinisil> No
<infinisil> Nothing is broken
<nick_l> infinisil: I consider it broken.
<nick_l> infinisil: you do not consider it broken. I understand.
<infinisil> Then you need to reconsider your notion of broken
winem_ has joined #nixos
<{^_^}> [nixpkgs] @jfrankenau opened pull request #40454 → taskwarrior: install zsh completion → https://git.io/vp9Rp
<Izorkin> infinisil: I want to do without creating a custom package
<nick_l> infinisil: If compiling a piece of software is *mandatory* for it to be used, it has limited value.
<infinisil> nick_l: Sorry to disappoint, but every software you ever used had to be compiled at some point
<nick_l> infinisil: for example, I also kicked out Virtualbox extensions, because it took too much time to compile.
<{^_^}> [nixpkgs] @ArdaXi opened pull request #40455 → nixos: security.pam: Rename enableU2F and add opts → https://git.io/vp90U
<jcrben> nick_l: perhaps the haskell programmers should organize to come up with a shared binary cache?
<infinisil> nick_l: Be glad that nixos has such a huge build server and cache at all, there are distros which need to compile *everything*, as in actually everything
<nick_l> infinisil: sure, I have run Gentoo too.
<nick_l> infinisil: but in a world where the choice is between Gentoo and Debian, I pick Debian.
<nick_l> infinisil: I am also not sure why you are arguing.
deepfire` has quit [Ping timeout: 255 seconds]
<nick_l> jcrben: the solution seems to be pretty clear to me: it should be packaged using native nixpkgs dependencies.
<infinisil> You seem to have the notion of "I can't use packages that I have to compile myself"
<nick_l> infinisil: yes, because I don't have the time to do that.
<jcrben> nick_l: seems reasonable - probably just requires some motivation for someone to make that happen
<infinisil> nick_l: There was an attempt for that, but it failed because hie dependencies changed so often
camsbury has quit [Quit: Leaving...]
camsbury has joined #nixos
<joko> What would be the best way of upgrading nix in mac os?
<nick_l> joko: probably depends on the method of installation.
camsbury has quit [Client Quit]
<infinisil> nick_l: See #32947 and #34447
<{^_^}> https://github.com/NixOS/nixpkgs/pull/32947 (by shlevy, closed): Add haskell-ide-engine.
<{^_^}> https://github.com/NixOS/nixpkgs/issues/34447 (by Infinisil, closed): haskell-ide-engine broken
camsbury has joined #nixos
<joko> nick_l: I have an old install using the curl command
winem_ has quit [Ping timeout: 240 seconds]
<jcrben> joko: nix upgrade-nix altho that's pretty new...
<joko> jcrben: I'm on 1.11.16
jD91mZM2 has quit [Quit: WeeChat 2.0]
<joko> so no nix command available
<nick_l> joko: one way to "upgrade", would be to remove it completely and install it again.
<joko> nick_l: about to do so, is there anyone else?
<infinisil> Have you tried just running the curl thing again?
<nick_l> joko: you might also be able to just nix-env -iA <right package>.
<nick_l> joko: but if you want to make sure this doesn't happen again in the future, you could ask whoever wrote the curl script to make it such that upgrades work.
<nick_l> I think if you make the decision to implement something, you should also implement the things that come with it, which are "uninstallation", and version management.
hph^ has joined #nixos
<nick_l> Upgrading is a specific kind of version management (one that goes one version in the future).
<infinisil> Izorkin: If you don't want to create a custom package (which I would do tbh), then your next best bet is `environment.etc."some-name".source = ./path/to/folder;`
hph^ has quit [Max SendQ exceeded]
<{^_^}> Channel nixpkgs-18.03-darwin advanced to https://github.com/NixOS/nixpkgs/commit/465de5981f3 (from 21 minutes ago, history: https://channels.nix.gsc.io/nixpkgs-18.03-darwin)
jtojnar has quit [Quit: jtojnar]
<{^_^}> [nixpkgs] @xeji pushed to release-18.03 « nixos/tests/acme: fix on i686, improve timing (#40410) »: https://git.io/vp90o
jtojnar has joined #nixos
hph^ has joined #nixos
Neo-- has quit [Ping timeout: 240 seconds]
<nick_l> infinisil: I installed hie, tried it on a an extremely simple file, but basic completion doesn't even seem to work.
vaninwagen has joined #nixos
<nick_l> infinisil: I even loaded the xmonad project, but inferior haskell mode doesn't even load that project.
<nick_l> So, while it could be the case that xmonad was not setup professionally, it does sort of show that the tooling is not ready.
broccoli has joined #nixos
camsbury has quit [Quit: Leaving...]
<infinisil> Needs some debugging, it took me a bit to get it to work
<xnaveira[m]> Hi, I finally made a python script work in a shell created like this: nix-shell -p 'python27.withPackages(ps: with ps; [ numpy scipy opencv pillow ])'
<nick_l> infinisil: sure, but when it is in that state I would never advertise it to others.
<xnaveira[m]> How can I add those same pytho packages in my configuration.nix?
<xnaveira[m]> i don't undetstand the with ps part
sigmundv_ has joined #nixos
<jcrben> nick_l: agreed on version management
<infinisil> nick_l: Sorry but that's mean, lots of people have spent lots of time to work on hie and nixos, for free for that matter, and you're gonna complain that they shouldn't have advertised it because you couldn't get it to work within a minute?
<nick_l> infinisil: that's exactly right.
<nick_l> infinisil: where I am from that's not mean, it's called realistic.
<nick_l> infinisil: you have your point of view, I have mine. That's fine.
<infinisil> If you want a flawless experience, use Ubuntu and VSCode
<infinisil> Not NixOS with haskell and LSP
<nick_l> infinisil: I tried VSCode and it was not flawless.
<nick_l> Ubuntu certainly is not.
tmaekawa has joined #nixos
drakonis has joined #nixos
tmaekawa has quit [Client Quit]
<jcrben> nick_l: would kinda like to help out with the core CLI, but tough to get into C++. whereas it was easy to drop into something like homebrew and patch a quick fix despite not really using ruby
<nick_l> jcrben: modern C++ is like Ruby.
<infinisil> nick_l: So, how about improving the situation for it? What can *you* do to make it better?
<nick_l> infinisil: I could fix the whole thing to meet my standards, if I had the time.
<nick_l> infinisil: Haskell has had a long history of building broken IDEs.
<nick_l> infinisil: the only thing that has ever really worked is inferior haskell mode.
<Izorkin> infinisil: thanks
Have-Quick has quit [Quit: Have-Quick]
<nick_l> Eclipse in the Java world was also a failure. IntelliJ is the only small company that produced something useful in this area, I think.
<nick_l> JetBrains*
alex`` has quit [Quit: WeeChat 2.1]
<infinisil> nick_l: Well I won't discuss with you anymore, if you want to bash all projects that don't meet "your standards", then go talk to it in #nixos-chat. Ask here if you have a question regarding nix/nixos
<jcrben> nick_l: huh, ruby <--> c++ is an interesting view - you sound like quite the coding wizard :) altho someone did point me to https://github.com/inspector-repl/inspector the other day which is apparently something like pry for C++
<nick_l> jcrben: whatever doesn't kill you, simply makes you stranger.
rauno has quit [Ping timeout: 256 seconds]
<nick_l> jcrben: cool, didn't know about that, but I'd just use a proper IDE for C++ development.
<{^_^}> Channel nixos-18.03-small advanced to https://github.com/NixOS/nixpkgs/commit/453498677ab (from 2 hours ago, history: https://channels.nix.gsc.io/nixos-18.03-small)
<nick_l> I think there are people who have setup Visual Studio on Linux, but QtCreator is also pretty good.
<jcrben> nick_l: I set up vscode.cpptools on nixos (couldn't get nix debug symbols to work right on darwin) which is my preferred C/C++ editor, but it's weird - lines are skipped randomly, local variables don't show up, sometimes I'll see <value optimized out> despite loading symbols
hph^ has quit [Ping timeout: 240 seconds]
<jcrben> since I don't do it professionally, hard for me to know what weirdness is just C things and what might be a bug in the debugger
<jcrben> but apparently inspector-repl is better about that
<nick_l> jcrben: I don't use a Mac, because I don't want to mess with broken stuff essentially.
<jcrben> haven't had the chance to try it
<nick_l> jcrben: I have used a Mac a couple of times, but I have never had a bug free experience, which I do expect for a premium device.
<nick_l> jcrben: by "a couple of times", I mean as a workstation and by "bugs", I mean things that have been acknowledged by Apple.
<nick_l> It would be smart for them to give me their most expensive toys for free. Not sure if I would still want to run a Mac on it then.
<jcrben> nick_l: my experience with a mac is that it's been far more reliable/smooth than linux (which I ran off and on from 2006 - 2015), but I'm trying to migrate back to linux . what hardware do you run? I use a laptop a lot
<nick_l> jcrben: I think a Dell XPS 9560 is the best that money can buy now.
<nick_l> jcrben: you do have to ask for a machine without Intel wireless.
<nick_l> jcrben: I own Intel stock, but I think they just suck for that.
<jcrben> nick_l: this is getting a bit off-topic, but that's interesting since the general consensus has always been that Intel has the best hardware for linux compatibility
<nick_l> jcrben: no, that's not the consensus. Intel graphics is "compatible" and "easy to use". Intel wireless has never been good, AFAIK.
<nick_l> jcrben: on NixOS, nvidia is easy to use.
<joepie91> shrug, it works quite well on my laptop
<joepie91> (intel wireless)
<nick_l> joepie91: and on tons of others people's devices it doesn't (see bug reports).
<nick_l> joepie91: microcode crashes on Intel for example.
<joepie91> I mean, you haven't linked any bug reports, so :P
<nick_l> joepie91: Intel even wrote an advisory and basically has shown that their hardware cannot cope with a nearby USB3 port being used.
<joepie91> what specific hardware though? because there's quite a lot of different models
rauno has joined #nixos
<nick_l> joepie91: in layman's terms (which you are not), that means they created shit.
<joepie91> er, no, not really
<joepie91> "it has a problem" != "it's shit"
<nick_l> joepie91: since their competitors do not have this problem, how else would you call it?
<nick_l> broadcom is shit too, btw.
<joepie91> thing is, you seem very determined to call everything "shit", seemingly without making any careful considerations on exactly *how* bad something is, what might cause it, and so on
<joepie91> this isn't a constructive attitude
<infinisil> Please take this to #nixos-chat, this is offtopic
<{^_^}> [nixpkgs] @vcunat pushed to master « xorg.xorgserver: drop makeWrapper »: https://git.io/vp9uy
<joepie91> and it's not exactly healthy for a channel like this either
broccoli has quit [Ping timeout: 256 seconds]
drakonis has quit [Read error: Connection reset by peer]
bpa has joined #nixos
sigmundv_ has quit [Ping timeout: 240 seconds]
udono has quit [Ping timeout: 240 seconds]
joehh has joined #nixos
nick_l was banned on #nixos by gchristensen [*!52a9aaf1@*]
nick_l was kicked from #nixos by gchristensen [nick_l]
<kreetx> I'm trying to create a lamp in nixos in virtualbox in macos, but get a very weird "string index out of range" error
<kreetx> the debug output for nixops is here https://pastebin.com/HhEUXe7T
<kreetx> I think it's not a python 2 vs 3 issue as nixops (I think!) depends on the 2
rain1 has quit [Quit: Leaving]
rain1 has joined #nixos
<{^_^}> [nixpkgs] @xeji merged pull request #40449 → perlPackages.GD: fix build on staging → https://git.io/vp9W8
<{^_^}> [nixpkgs] @xeji pushed commit from @volth to staging « perlPackages.GD: fix build on staging (#40449) »: https://git.io/vp9zP
<{^_^}> [nixpkgs] @Ericson2314 pushed 10 commits to master: https://git.io/vp9zD
<{^_^}> → 2a450cec by @lopsided98: uboot: add patch to increase maximum path length in extlinux.conf
<{^_^}> → d29202bc by @lopsided98: uboot: 2017.11 -> 2018.03, cross fixes
<{^_^}> → d50bcaa8 by @lopsided98: uboot: sheevplug and guruplug: use mainline U-Boot
<{^_^}> [nixpkgs] @Ericson2314 pushed 10 commits to release-18.03: https://git.io/vp9zH
<{^_^}> → d29202bc by @lopsided98: uboot: 2017.11 -> 2018.03, cross fixes
<{^_^}> → 2a450cec by @lopsided98: uboot: add patch to increase maximum path length in extlinux.conf
<{^_^}> → d50bcaa8 by @lopsided98: uboot: sheevplug and guruplug: use mainline U-Boot
<{^_^}> [nixpkgs] @jfrankenau opened pull request #40457 → Install shell completions for beets, borg, exa, youtube-dl → https://git.io/vp9z7
<{^_^}> [nixpkgs] @Ericson2314 pushed 2 commits to master: https://git.io/vp9gv
<{^_^}> → 29ef982a by @Ericson2314: Merge commit '130a0cbde6feae6b51f35a007c78cc8a25a6c3c9' into strict-deps-base
<{^_^}> → 3401db62 by @Ericson2314: Merge branch 'strict-deps-base'
<{^_^}> [nixpkgs] @Ericson2314 pushed 11 commits to staging-18.03: https://git.io/vp9gY
<{^_^}> → d29202bc by @lopsided98: uboot: 2017.11 -> 2018.03, cross fixes
<{^_^}> → 2a450cec by @lopsided98: uboot: add patch to increase maximum path length in extlinux.conf
<{^_^}> → d50bcaa8 by @lopsided98: uboot: sheevplug and guruplug: use mainline U-Boot
marusich has joined #nixos
<kreetx> so I've found the what/why: nixops runs "VBoxManage showvminfo --machinereadable <ID>" and gets a line "videocapopts=" which doesn't have a value
joehh has quit [Ping timeout: 246 seconds]
moet has joined #nixos
<kreetx> .. and of course there is a ticket and a fix for this (https://github.com/NixOS/nixops/pull/773/commits/e704ffedb9a4433775930e901952a22f6844ea06). Now need to figure out why I don't have that fix.. Sorry for the noise :)
<moet> hellooo.. this seems like a fairly simple question, but i'm unsure of how to proceed.. i run nix on osx, and was wondering how to update
<infinisil> Glad we could help :P
<moet> `nix-env -u` doesn't seem to do much and `nix-channel --list` shows no subscribed channel
<moet> i'm wondering if i should `nix-channel --add` one of the nixpkgs from here https://nixos.org/channels/
<moet> (or, if maybe osx is not supposed to be subscribed for some reason?
<moet> )
<kreetx> moet: I think every user of nix has a channel and is subscribed
<symphorien> moet: if you have a multi user install, when the user has no channel, it uses root's channels by default
<moet> i'm not sure if it's a multi-user install, but `sudo nix-channel --list` also doesn't show any channels :)
joehh has joined #nixos
<symphorien> nix-info will tell you
<symphorien> and iirc on mac, use sudo -i to run nix-channel as root
<moet> ooh, that's interesting
<moet> and it worked
<symphorien> also, again I am not sure, but nix-env -u will upgrade to the current state of the channel
<MichaelRaskin> I use Nix and I don't use a channel. But I am a committer and I need my checkout anyway
<moet> `sudo -i` followed by `nix-channel --list` ... also, nix-info is pretty nice!
tmp711[m] has left #nixos ["User left"]
<symphorien> to "really" upgrade, you need to nix-channel --upgrade first
<kreetx> symphorien: what does "really" mean?
<moet> i think upgrade the list of packages
<moet> vs upgrading the installed packages
<symphorien> kreetx: like the difference between apt upgrade and apt update
<moet> thanks symphorien
<kreetx> ah, so the nix-channel --upgrade is the apt update part -- thanks!
<symphorien> and regarding channels, on vanilla NixOS, root is always subscribed to one channel iirc
<kreetx> was wondering why I get no updates with nix-env -u..
<symphorien> other users default to root's channel
<symphorien> but nixos-rebuild --update will call nix-channel --upgrade
<moet> yep, and after doing `sudo -i` the `nix-channel --update` command is giving me my first updated package list! thanks!
Arcaelyx has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<cizra> How do I list reverse dependency list of a package? ie. I'm right now wondering why I have strace installed.
<symphorien> nix-store --query --refererers
<symphorien> or --referrers-closure
<symphorien> iirc
<cizra> Whoa, cool
<symphorien> with nix 2 you have nix why-depends
<infinisil> cizra: Or better yet for this specific case: nix-store -q --tree /run/current-system
<infinisil> Shows you all dependencies in a tree, so you know where what comes from
<infinisil> (But nix why-depends is even better)
ilyaigpetrov has quit [Quit: Connection closed for inactivity]
coot has joined #nixos
<cizra> Thanks!
<kreetx> so there was a bug in nixops and it's fixed in the 1.6 version, but I'm running 1.5.2. I think this must be related to the channel I'm on (on my macos), so I should switch to something more bleeding edge -- how do I do this? Or, how to I find out which channels are even available to me?
<jcrben> I would like to see something like https://pkgstats.archlinux.de/fun for nixos - has anyone ever talked about that?
sergej has joined #nixos
<symphorien> kreetx: upgrade you user's and root's channels, and then nix-env -u
sergej has quit [Quit: Bye Bye]
<ison[m]> jcrben: How is that information generated? Just from downloads? Or do users volunteer that information?
<symphorien> if you follow the 17.09 channel, you may never get nixops 1.6, though
<cizra> infinisil: How do I do nix-store -q --tree /run/current-system for nix-shell with extra packages installed? (if it's easy, I don't need it very badly)
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<symphorien> cizra: you mean nix-shell -p ?
<giaco> it is possible in config.nix to setup a LD_LIBRARY_PATH?
<{^_^}> [nixpkgs] @thoughtpolice merged pull request #40412 → plex: 1.9.6 -> 1.13.0 → https://git.io/vpS73
<{^_^}> [nixpkgs] @thoughtpolice pushed commit from @anderslundstedt to master « plex: 1.9.6 -> 1.13.0 »: https://git.io/vp92p
<symphorien> giaco: do you mean configuration.nix on NixOS ?
<cizra> symphorien: yessir
alex`` has joined #nixos
<infinisil> kreetx: The nixosVersion thing?
<giaco> symphorien: no config.nix in a version of nix compiled from source
<infinisil> giaco: What do you need this for?
<symphorien> nix without NixOS cannot control services, environment and other global system level things
<jcrben> ison[m]: it's not based on downloads - users opt-in
<Unode> any clues as to what might be the reason for http://dpaste.com/3GA0G7S ? Using stack with "nix: enable". Also tried different lts releases but optparse-applicative-0.14.2.0 fails in all of them.
<symphorien> so if you software need LD_LIBRARY_PATH, wrap it
<kreetx> infinisil: sorry, what are you asking? :) The issue I had was this here https://github.com/NixOS/nixops/pull/773, and it's fixed in the commit referenced there
<symphorien> *your
<kreetx> but the fix is in nixops 1.6, but I have 1.5 -- so I'm now trying to switch channel to unstable
<infinisil> kreetx: Ah I see, well I had a problem with nixosVersion that was fixed with 1.6, so I thought maybe you had that one too :)
<infinisil> kreetx: Yeah you probably need nixos-unstable/nixpkgs-unstable
<kreetx> yes
<kreetx> if not running nixos it's probably nixpkgs-* I want?
<infinisil> Indeed
<symphorien> cizra: no idea for nix-shell -p. For nix-shell somefile.nix I would say nix-instantiate the file to get a .drv and query the dependencies of the .drv
<cizra> symphorien: OK, no worries.
Rusty1_ has joined #nixos
<giaco> infinisil: I am porting nix to Android under Termux linux environment. I am almost there. It compiles, it runs, but I need to customize where the libs are using LD_LIBRARY_PATH
mekeor has joined #nixos
lopsided98 has quit [Quit: Disconnected]
<{^_^}> [nixpkgs] @timokau opened pull request #40458 → giac: Remove lapack configFlag (#40430) → https://git.io/vp9a3
<giaco> sadly nix seems to drop the LD_LIBRARY_PATH from env and I need to find another way to boostrap nix, so I am trying to static compile the executables referred in config.nix
vaninwagen has quit [Quit: WeeChat 2.1]
lopsided98 has joined #nixos
<giaco> do you know if there's an option that behaves just like LD_LIBRARY_PATH?
<symphorien> what do you mean "nix drops LD_LIBRARY_PATH" ?
joehh has quit [Ping timeout: 256 seconds]
<giaco> symphorien: yes. When nix-env calls the shell in config.nix (shell = "/data/data/com.termux/files/usr/bin/bash") it drops the LD_LIBRARY_PATH. My bash needs LD_LIBRARY_PATH=/data/data/com.termux/files/usr/lib
<symphorien> then wrap bash
<giaco> symphorien: do you mean statically compile?
<symphorien> hum
<symphorien> this may be dumb
lopsided98 has quit [Client Quit]
<symphorien> bash would be a shell script adding the environment variable and then execing bash
<symphorien> but if bash is your shell...
<symphorien> then you have to do the wrapper in C
<kreetx> another error building my LAMP from macos on virtualbox: nixops deploy tells me I'm on x86_64-darwin, but x86_64-linux is needed to build append-initrd-secrets. I guess this is not in the binary cache and it wants to build it locally, but couldn't it build it in the virtualbox/nixos somehow?
<moet> ooh.. um is it not possible to deploy to virtualbox with nixops from an osx host?
<giaco> I do have a backup shell that does not need the misplaced lib: /data/data/com.termux/files/usr/bin/sh
<kreetx> moet: I am deploying to virtualbox
lopsided98 has joined #nixos
<symphorien> then fine
<kreetx> not sure why it wants to build outside of it
broccoli has joined #nixos
<moet> kreetx: i got an error about some assertion failing, which appears to be `assert stdenv.isLinux;`
<infinisil> I bet clever could help for the nix installation if he were here
<kreetx> moet: yes, I got the same one!
<symphorien> giaco: to have an example, if you have firefox installed via nix, do cat $(which firefox)
<giaco> symphorien: so you are saying to set shell = "/data/data/com.termux/files/usr/bin/nixbash" and create nixbash to run #!//data/data/com.termux/files/usr/bin/sh export and bla blah?
<moet> kreetx: i'm curious if maybe this isn't a supported use case .. yet
<symphorien> and the wrapProgram function of nixpkgs can create such wrappers automatically
knupfer has quit [Ping timeout: 240 seconds]
Synthetica has quit [Quit: Connection closed for inactivity]
<symphorien> exactly
<symphorien> nixpkgs has tons of such wrapped programs. most notably binutils-wrapped vs binutils-unwrapped, gcc, clang and so on
<moet> kreetx: seems relevant https://github.com/NixOS/nixops/issues/560
<kreetx> moet: I got rid of that error by destroying and deleting my vbox and recreating it. Since the erroring one was built with nixops 1.5 and I was now using 1.6 then I thought it was related to that. And after the recreation that certain error went away
<kreetx> moet: haha
<kreetx> nice title
<{^_^}> [nixpkgs] @dotlambda merged pull request #40457 → Install shell completions for beets, borg, exa, youtube-dl → https://git.io/vp9z7
<{^_^}> [nixpkgs] @dotlambda pushed 5 commits to master: https://git.io/vp9aP
<{^_^}> → 836be602 by @jfrankenau: beets: install zsh completion
<{^_^}> → bb07e890 by @jfrankenau: borgbackup: install shell completions
<{^_^}> → 5948024c by @jfrankenau: exa: install man page and shell completions
<moet> kreetx: you were able to get past the assertion with a later version of nixops?
<kreetx> yes
<moet> huh..
<kreetx> but I only got it in between switching
<giaco> symphorien: thank you very very much. This is a great idea
<kreetx> i.e it wasn't there when I did all with 1.5
<kreetx> and it is not here (yet?) when doing all with 1.6
raynold has joined #nixos
<kreetx> but the issue you linked is directly relevant, thanks!
<moet> kreetx: i'm on virtualbox 5.2.12 and nixops 1.6 and it seems nixops wants to build locally (on host, not guest)
<kreetx> same here
<moet> i wish there was a `nixops deploy` suboption to just "build on the target"
<moet> :heh: but maybe that's difficult to achieve
<kreetx> I guess it makes sense to have deploy mean "deploy", but that flag would still be great to have in our situations
<infinisil> I think this might be possible by setting up a remote builder
<moet> yeah, that's what the nixops issue #560 above is about infinisil
<kreetx> it definitely is, one just needs to keep walking the path some more :)
pingu_ has quit [Ping timeout: 240 seconds]
<infinisil> You mean nixops#560 :P
<{^_^}> https://github.com/NixOS/nixops/issues/560 (by 3noch, open): Help poor macOS users who can't build many closures on their host
<kreetx> yes..
<infinisil> (I implemented the bot linking to it)
<moet> it seems like that issues has several workarounds outlinked
<kreetx> moet was my bot (no offence!), but how would one trigger the real bot?
<infinisil> kreetx: Huh? What real bot?
<{^_^}> [nixpkgs] @andersk opened pull request #40459 → pypy: 5.10.0 -> 6.0.0 → https://git.io/vp9aj
Guest78178 has quit [Quit: Lost terminal]
<kreetx> infinisil: I thought you meant you implemented a bot that links to the github issue when someone mentions macos or something
marusich has quit [Quit: Leaving]
<kreetx> in here or on github on the nixops repo perhaps
pingu_ has joined #nixos
<moet> kreetx: it sounds like if you add your ssh key to the guest-os after nixops fails to build on your host-os and then re-run the deploy command, nixops will build on the guest-os ... nixops#260
<{^_^}> https://github.com/NixOS/nixops/issues/260 (by wmertens, open): Support building fully remotely
<infinisil> kreetx: Not sure I understand, I just implemented this ^^
<infinisil> Linking to it when somebody mentions <project>#<issue number> (and if project is empty use nixpkgs)
<kreetx> ah, yes now I see!
<{^_^}> Channel nixpkgs-18.03-darwin advanced to https://github.com/NixOS/nixpkgs/commit/b6aa5923854 (from 61 minutes ago, history: https://channels.nix.gsc.io/nixpkgs-18.03-darwin)
<kreetx> moet: what's your use case, what are you trying to do?
<symphorien> giaco: btw, are user namespaces enables on android ?
<symphorien> *enabled
<giaco> symphorien: no
<moet> kreetx: learn nixops :)
<symphorien> sad
<kreetx> moet: ah, yes. So I'm trying to get a local dev environment running for work, and we all have macos. So my current target is getting a basic lamp stack running into which I would mount some git repo where I do my dev work
<{^_^}> [nixpkgs] @hedning opened pull request #40460 → gnome3.gnome-shell: 3.28.1 -> 3.28.2 → https://git.io/vp9VZ
<kreetx> was thinking that maybe you have some similar tangible goal
<infinisil> giaco: btw, gist allows you to put multiple files into one gist
smichel has quit [Quit: smichel]
<kreetx> the bigger goal is to run nix, nixops etc to deploy haskell to prod servers
leo60228 has joined #nixos
<giaco> infinisil: I've just download a ruby shell thing to upload file. Excuse me, I'm still getting used to it
<symphorien> giaco: does you shell support [[ ?
<infinisil> Yeah np, just thought I'd mention it :)
<symphorien> usually it is a shell builtin
hamishmack has joined #nixos
<leo60228> i'm reading a distro-universal (except for nixos, because of configuration.nix) guide to make a pci passthrough vm, and it says to download and install the GuC, DMC and HuC firmware
<moet> kreetx: yes, that's my bigger goal
<leo60228> does nixos do that automatically, is there an option for it, or should i do something else?
<moet> i'm planning to run my website on aws, but i want to define it with nixops first
<giaco> symphorien: I am running sh (dash symlink actually) instead of bash because bash needs LD_LIBRARY_PATH. I can: A find a way to push LD_LIBRARY_PATH into nix, B compile bash statically
<moet> so i'm starting out locally, where i hope to define a service to run on the nixops deployed machine
<leo60228> i can't tell if it's in nixpkgs, though
<kreetx> moet: one day we will get there! :p
<leo60228> and none of those options explicitly mention that they get firmware from outside nixpkgs
<kreetx> anyway, I'm off to sleep, bye
<giaco> obviously option A is preferred
<symphorien> giaco: dash does not support [[
<leo60228> the firmware is part of the i915 driver, but i'm not sure if it's required or nix uses it
<symphorien> this is too advanced shell syntax
<symphorien> write a more simple shell script
<moet> kreetx: :wave:
<leo60228> use [ "x" -eq "y" ] instead
<leo60228> of [[ "x" == "y" ]]
<symphorien> giaco: I think a case statement works here
<giaco> that is a nix file. I am not changing that file, I am changing shell
<giaco> is /data/data/com.termux/files/nix/store/f7d630zvrixjh9v9iydvfza632mwk22c-unpack-channel.sh
<leo60228> symphorien: what package are these drivers in https://01.org/linuxgraphics/downloads/firmware
<symphorien> leo60228: no idea
<symphorien> maybe nix-locate can tell you
<symphorien> giaco: ah. can you explain me where this file comes from ?
<symphorien> and why you can't modify it ?
kreetx has quit [Ping timeout: 240 seconds]
<leo60228> it's in firmware-linux-nonfree
<leo60228> hardware.enableRedistributableFirmware = true;, then
<giaco> symphorien: this files comes from nix make && make install. I am just trying to launch my first command: nix-channel --update
<symphorien> ok then nix needs bash to be bash
<symphorien> so:
Arcaelyx has joined #nixos
Arcaelyx has quit [Client Quit]
<symphorien> don't call you wrapper for bash nixbash but bash and put it earlier on the PATH
<leo60228> how do i append to an existing array? pretty new to nix
<leo60228> *prepend
<symphorien> leo60228: options of type list are automatically merged
<symphorien> to prepending there is lib.mkBefore iirc
<symphorien> *to prepend
<leo60228> thanks
<giaco> symphorien: thank you, it worked. I have successfully nix-channel --update
<symphorien> once you have bootstrapped, nix should provide its own bash I think
<moet> hrm..
<symphorien> you just have to ensure /bin/sh works good enough
<leo60228> option = pkgs.lib.mkBefore [ "whatever" ]; gives me error: infinite recursion encountered, at undefined position
<giaco> symphorien: thank, it is trying to boostrap. Have some errors be finally we are unstuck https://gist.github.com/33b96d851fd5dbdf5b3503633566b58c
<symphorien> can you paste: /data/data/com.termux/files/nix/store/ac1pzls6p1s2z10svz5ah8wzz4z0qqfj-unpack-bootstrap-tools.sh ?
acarrico has joined #nixos
<infinisil> logzet: Add "lib" to the argument list at the top of the file and use that instead
<infinisil> leo60228: ^^
<leo60228> thanks
<leo60228> ended up just using a different option
<leo60228> not sure which is the right one, but i'll find out
seanparsons has quit [Read error: Connection reset by peer]
leo60228_ has joined #nixos
leo60228 has quit [Disconnected by services]
leo60228_ is now known as leo60228
<symphorien> giaco: it looks like the bootstrap tools have a wrong content
<symphorien> nothing matches the glob pattern /data/data/com.termux/files/nix/store/dnzmv85z4ycy0dq58lyvp4vsp81ybz5k-bootstrap-tools.tar.xz.drv
alex`` has quit [Quit: WeeChat 2.1]
<symphorien> btw, which linker does android use ?
<giaco> I see
<leo60228> so uh
<leo60228> why is nix recompiling gtk
<giaco> symphorien: the content of /data/data/com.termux/files/nix/store/9741plqqjkjnaxm7yzw5jh7x63kz9sn6-bootstrap-tools/lib is https://gist.github.com/b0a65fbfc9835e6f8804101af0decbc8
<leo60228> wait
<leo60228> is it compiling gtk or nvidia_x11
<leo60228> it seems to be doing both
<leo60228> but nvidia_x11 is propietary
<leo60228> isn't it?
<moet> is anyone here familiar with NIX_REMOTE? the only mention i can find of it is https://github.com/NixOS/nixops/blob/549ce4399f5975765fb22c31d2c7dd9233187608/nixops/deployment.py#L624
<moet> i've tried setting it to things other than "daemon" and i just get errors saying "invalid setting for NIX_REMOTE"
<symphorien> giaco: can you check that your shell is able to expand the corresponding glob expression ?
seanparsons has joined #nixos
<symphorien> moet: NIX_REMOTE has only effect on nix1
<infinisil> moet: Afaik the valid settings are: "" for not using the daemon, "daemon" for using the daemon, and "local?root=/path/to/root" for using a different root
<symphorien> it tells wether nix should open /nix/store itself or via the daemon
leo60228 has quit [Ping timeout: 260 seconds]
hiratara has quit [Ping timeout: 276 seconds]
xy2_ has quit [Ping timeout: 240 seconds]
<symphorien> giaco: actually, can you add -x as argument to bash in your wrapper ? it will be ultra noisy, but might help
<giaco> symphorien: are you going to be here in an hour or so? I have an emergency here, have to go. But hey, thanks a lot, really near to have nix on android no-root-required
hiratara has joined #nixos
<symphorien> no sorry
<symphorien> good luck
<giaco> symphorien: I will be happy to keep up on this, but later.
boxscape has joined #nixos
Ross has quit [Quit: ZNC - http://znc.in]
boxscape has quit [Client Quit]
Ross has joined #nixos
civodul has quit [Quit: ERC (IRC client for Emacs 25.3.1)]
<moet> symphorien: so based on that i should set NIX_REMOTE= to enter that branch and then use NIX_REMOTE_SYSTEMS= to ... a file name formatted somehow
Sonarpulse has joined #nixos
<{^_^}> [nixpkgs] @xeji pushed to master « nixos/tests: set default stateVersion in test VMs »: https://git.io/vp9wP
<infinisil> Is there *any* way to find out what "warning: dumping very large path (> 256 MiB); this may run out of memory" actually dumps?
<infinisil> My config is rather big, and this message only recently started popping up without changing anything afaik
<infinisil> So I really don't have a clue
<johnw> infinisil: I'd love to know the answer too
infinisil has quit [Quit: Configuring ZNC, sorry for the join/quits!]
infinisil has joined #nixos
Ariakenom has quit [Read error: Connection reset by peer]
iqubic has quit [Ping timeout: 246 seconds]
<tilpner> infinisil - -vv 2>&1 | less, then /dumping
<infinisil> I just thought of adding -v's
<infinisil> But, how can I do this with nixops??
<tilpner> For me, that shows "copied source '...' -> '/nix/store/...'", which I know is the culprit
Sonarpulse has quit [Ping timeout: 256 seconds]
simukis has quit [Ping timeout: 248 seconds]
moet has quit [Quit: leaving]
<infinisil> Ohhhh found it, thanks tilpner
<infinisil> Now it seems obvious, it was the nixbot project directory with it's huge .stack-work folder I was using without lib.cleanSource
orivej has joined #nixos
<judson> Oddball nixops question: I'm setting up HTTP with ACME. There's kind of a chicken-and-egg problem with that, because until ACME gets the certificate in place, Apache (iirc) won't start. But ACME needs Apache to start in order to get the certificate.
<judson> I've been using a toggle locally to do one deploy, get the first certificate, and then enable TLS with a subsequent deploy.
<judson> Is that what other people do?
<judson> Or is there a straightforward configuration that'll work without a redeploy?
<gchristensen> tbh I use nginx and it creates an invalid cert to allow startup
<infinisil> Judson: The nginx service in nixos does it flawlessly automatically, it first starts a server the certificate, then when it has one starts it with the certificate
<infinisil> English where
<infinisil> Yeah what gchristensen said
<judson> And it does HTTP->HTTPS redirects somehow with that?
Have-Quick has joined #nixos
<gchristensen> hmm what do you mean?
hiratara has quit [Quit: ZNC - http://znc.in]
<judson> For years now I've configured my domains to redirect HTTP traffic to the HTTPS service.
<gchristensen> you can forceSSL to turn it on
<judson> That may be the stumbling block actually - not that Apache won't boot.
<infinisil> > option "services.nginx.virtualHosts.<name>.forceSSL"
<{^_^}> "Whether to add a separate nginx server block that permanently redirects (301)\nall plain HTTP traffic to HTTPS. This will set defaults for\n<literal>listen</literal> to listen on all interfaces on th...
<judson> Does that work with ACME as well?
hiratara has joined #nixos
<gchristensen> yeah, it works fine, LE doesn't validate the cert on setup
<judson> Huh. So a placeholder self-signed or something might be enough.
<samueldr> that's what nixos does
<judson> Honestly, more of this may be already solved - some of this is practicing modules on a problem I understand pretty well.
<judson> samueldr, nixos does that with Apache?
<samueldr> ah, with the previously mentioned nginx
<samueldr> so yes, if you use a self-signed cert, with http→https redir, it should work
<judson> Okay. I should look at the nginx module.
<infinisil> Btw, for anybody using nixops that doesn't want to deal with nixops horrible CLI but still be able to eval options and generally use nix-* commands with the config, you can do this: https://github.com/Infinisil/system/tree/master/deploy/eval
pingu_ has quit [Remote host closed the connection]
<judson> Nix doesn't have the equivalent of Haskell's $ does it?
<sphalerite> Judson: no
<infinisil> Judson: Not really, but I recently came up with this:
<infinisil> > compose (x: x + 1) (x: x * 2) 5
<{^_^}> 11
<infinisil> > compose (x: x + 1) (x: x * 2) (x: x - 3) 5
<{^_^}> 5
<judson> > compose (x: x + 1) (x: x * 2) (x: y: x + y) 2 3
<{^_^}> value is a function while an integer was expected, at (string):35:24
<judson> My use case ends with mapAttrsToList
<judson> I guess I could (compose ....) mapAttrs ...?
<infinisil> Maybe
<infinisil> compose isn't in the lib or anything btw though
<infinisil> > :v compose
<{^_^}> compose = let c = f: v: if builtins.isFunction v then c (x: f (v x)) else f v; in c lib.id
<Ralith> rip parametricity
manu_ has joined #nixos
<Ralith> I remember when people talked about nix getting a type system someday
<manu_> Hi, I need your help please. How can I change /etc/hosts? The file is read-only which is okay. But I need to use this file now
<infinisil> Ralith: Hmm what do you mean with parametricity? where does it not work?
<infinisil> > option "networking.extraHosts" # manu_
<{^_^}> "Additional verbatim entries to be appended to <filename>/etc/hosts</filename>.\n"
<mudri[m]> Ralith: I think Profpatsch is still on with the type system stuff.
<infinisil> (okay that was unnecessary use of > :P)
<mudri[m]> Or at least, was fairly recently.
<manu_> Then I need to perform nixos-rebuild switch?
<manu_> and after my test I do nixos-rebuild switch again? Right?
deepfire` has joined #nixos
<manu_> infinisil: How is the complete line? I'm new to nixos
<Ralith> infinisil: any time you branch on type, parametricity dies
<manu_> or where I can find the manual to this option? :)
<judson> manu_: yes, do a nixos-rebuild switch after updating that entry.
<mudri[m]> infinisil: if parametricity were true, compose would act the same for functions and non-functions.
<infinisil> I should make > option link to the manual too
<infinisil> mudri[m]: Ah I see, yeah
<infinisil> But this compose function looks really nice :)
<mudri[m]> There's probably something you can do with higher-order functions that makes compose behave weirdly.
<infinisil> In usage at least
<judson> I'll admit, as I'm looking at it, "just rub some parens on it" is working pretty well :)
<infinisil> Oh, I could change compose such that it's usable like `compose f g h end x`
logzet has quit [Remote host closed the connection]
<infinisil> Which wouldn't have the downsides
Have-Quick has quit [Quit: Have-Quick]
<mudri[m]> Taking a list of functions is basically the same thing, and also should preserve parametricity (in some imaginary type system).
<infinisil> True
<infinisil> And it wouldn't be as hacky
<sphalerite> oh hi mudri[m]
<mudri[m]> 410 polished off? ;-)
<sphalerite> oh yeah I need to check my mark for that. I don't think I'll be pleased with it :/
Jackneilll has joined #nixos
<infinisil> > composeAlt = fs: if fs == [] then lib.id else x: (lib.head fs) (composeAlt (lib.tail fs) x)
<{^_^}> composeAlt defined
<infinisil> > composeAlt [ (x: x + 1) (x: x * 2) ] 10
<{^_^}> 21
<infinisil> There we go
<mudri[m]> (don't wanna get too off-topic here...)
<judson> > composeAlt (x: x + 1) (x: x * 2) (x: y: x + y) 2 3
<manu_> Judson/Infisil: Okay, I got it :)
<{^_^}> value is a function while a list was expected, at (string):6:52
Jackneill has quit [Read error: Connection reset by peer]
<judson> manu_, right on :)
<infinisil> Judson: I'm not sure why you're expecting this to work, it doesn't work in haskell either
<sphalerite> mudri[m]: there's always #nixos-chat :p
tmplt has quit [Ping timeout: 256 seconds]
<infinisil> (and composeAlt takes a list)
<manu_> I've another question: my normal user account has no home and no login after each reboot. How do I change this in the configuration.nix? I don't know why this was lost
<infinisil> Wait or does it
<judson> > composeAlt [(x: x + 1) (x: x * 2) (x: y: x + y) ] 2 3
<{^_^}> value is a function while an integer was expected, at (string):36:28
<infinisil> Nah, the compose thing only works with a single argument, as with haskell
<judson> I'm not sure what I'm expecting :)
<sphalerite> manu_: no home and no login..? whaaa?
tmplt has joined #nixos
jackdk has joined #nixos
<manu_> yes, I need to edit /etc/passwd each time and then I can login
<sphalerite> manu_: you need to define your user in your configuration.nix
<sphalerite> /etc/passwd is generated by nixos and shouldn't be edited directly
<manu_> I've tried this by changing the preconfigured guest account. But I cannot build it unfortunately. Maybe an error in my configuration?
<manu_> changing the name, the uid etc
<manu_> where I can find an example?
jtojnar has quit [Read error: Connection reset by peer]
jtojnar has joined #nixos
<manu_> I've found one.
<infinisil> preconfigured guest account? I don't think there should be one by default
aarvar has quit [Ping timeout: 240 seconds]
<infinisil> manu_: But if you have an error you should post it here
<sphalerite> manu_: just users.manu.isNormalUser = true; should be enough to have the user created
<manu_> I've defined the user's group elsewhere in the configuration and wasn''t aware of this. Solved. :)
<manu_> How can I update the installed packages?
<manu_> The whole system if needed?
<manu_> to recent software versions?
jbboehr has quit [Ping timeout: 276 seconds]
<infinisil> manu_: nixos-rebuild switch --upgrade
<{^_^}> [nixpkgs] @matthewbauer pushed 4 commits to master: https://git.io/vp9oj
<{^_^}> → 3e293b2d by @ruuda: libressl: use https url in metadata
<{^_^}> → 4371e5be by @ruuda: libressl_2_7: init at 2.7.3
<{^_^}> → 5bd19cfc by @ruuda: libressl: 2.5.5 -> 2.7.3
<{^_^}> [nixpkgs] @matthewbauer merged pull request #40440 → libressl: init at 2.7, change url to https → https://git.io/vp9me
<manu_> infinisil_: would you suggest it to do this often? Can my system be unbootable with this?
<manu_> maybe not,
<sphalerite> manu_: yes, it does make sense to do it often
seanparsons has quit [Read error: Connection reset by peer]
<manu_> With often I mean some times a week/month
<sphalerite> and your system may be made unbootable, but it's A, very rare, and B, you can just boot an older generation1
<sphalerite> s/1/!/
<manu_> Can I read elsewhere which modules/software has been updated?
spear2 has joined #nixos
<sphalerite> you can look at the git log, but that might be a bit like drinking from a fire hose :')
<sphalerite> maybe not quite for stable I guess
<manu_> haha
<manu_> do I need then nixos-rebuild boot --upgrade instead of switch?
<judson> Aha! security.acme explicitly adds nginx and lighttpd as dependencies on the selfsigned target but not apache.
<judson> :/
<manu_> switch only changes for the running system I guess...
<samueldr> switch activates and makes it the boot default
<samueldr> test activates without making a boot entry
<samueldr> (activates meaning it uses what's changed, "switches" to it)
<judson> manu_, there'll be a few things that get "stuck" after a switch. Sometimes rebooting makes sense as a result.
<judson> e.g. it took me a while to work out how to clear the dmenu cache, which'll hold onto the old paths.
<judson> And e.g. running Chrome will be the old version, so links from other programs will open the new one.
<judson> And they will be sad.
thc202 has quit [Ping timeout: 256 seconds]
<judson> Is there a Vim plugin for Nix that'll match let and in?
<judson> (i.e. % jumps from let to matching in and v/v?)
pako_ has joined #nixos
ThatPako has quit [Ping timeout: 256 seconds]
<giaco> symphorien: I'm back
seanparsons has joined #nixos
broccoli1 has joined #nixos
broccoli has quit [Ping timeout: 240 seconds]
ison[m] has left #nixos ["User left"]
<judson> I know I'm preaching to the choir here, but boy howdy does Nixops make this straightforward.
broccoli has joined #nixos
broccoli1 has quit [Ping timeout: 248 seconds]
ericsagnes has quit [Ping timeout: 246 seconds]
<{^_^}> Channel nixos-18.03-small advanced to https://github.com/NixOS/nixpkgs/commit/b6aa5923854 (from 4 hours ago, history: https://channels.nix.gsc.io/nixos-18.03-small)
Have-Quick has joined #nixos