<UFODivebomb>
which then attempted to install every package in nixos....
<catern>
:D
<catern>
yeah I like how "nix-env -i" tries to install everything
<catern>
I tried running that and then frantically C-c'd :)
johnramsden has joined #nixos
MP2E has quit [(Quit: reboot!)]
MP2E has joined #nixos
<joepie91>
man, packaging XNA stuff is a fucking nightmare
<joepie91>
Mono introduces a whole class of new ways for stuff to break with uninformative exceptions
roconnor has joined #nixos
pareidolia has quit [(Ping timeout: 240 seconds)]
pareidolia has joined #nixos
<justanotheruser>
How would I go about making pygame a python35 package?
filterfish has quit [(Ping timeout: 255 seconds)]
<justanotheruser>
it seems there might not be anything making it exclusively python27. So I am trying to define python35Packages.pygame
<justanotheruser>
Is there some recommended book that goes over the nix language?
<gchristensen>
justanotheruser: the nix pills series is incredible, by Lethalman
<justanotheruser>
checking it out
natas_ has joined #nixos
inflames has joined #nixos
<catern>
I have a shell script, with no deps outside of stdenv, and I want to put it into a Nix package so it shows up in my path. anyone know a good example of how to do this?
<justanotheruser>
In python35Packages I attempted pygame = pkgs.newScope ../development/python-modules/pygame { }; to no avail
drakonis has joined #nixos
<gchristensen>
catern: writeScriptBin
<catern>
oooh sounds like what I want :D
<joepie91>
I have successfully packaged Stardew Valley :D
vaibhavsagar has joined #nixos
lambdamu has joined #nixos
<catern>
gchristensen: oh hmm, this takes a string
<catern>
gchristensen: I don't want to embed my script into my nix expression, I'd rather it be separate
filterfish has joined #nixos
<gchristensen>
you could pass it (builtins.readFile ./your-file)
lambdamu_ has quit [(Ping timeout: 255 seconds)]
<catern>
oh
<catern>
true :D
vaibhavsagar_ has joined #nixos
<justanotheruser>
could someone please tell me where it is specified that pygame is only under python27Packages?
<justanotheruser>
I can't find that anywhere
vaibhavsagar has quit [(Ping timeout: 260 seconds)]
<joepie91>
aaaaaand the game crashed
<gchristensen>
justanotheruser: looks like it is for 17.03 but not in unstable
<justanotheruser>
gchristensen: is the master branch unstable?
mbrgm has quit [(Read error: Connection reset by peer)]
<catern>
gchristensen: thank you, that worked perfectly :)
<catern>
now I'm just slightly annoyed that "nix-env -f http://path/to/stuff.tar.gz -i" and "nix-env -f stuff.tar.gz -i" have different behavior
<catern>
that is kind of unintuitive imo
<dmj`>
this doesn’t actually show the ami manifest though
mizu_no_oto has quit [(Quit: ["Textual IRC Client: www.textualapp.com"])]
mbrgm has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] NeQuissimus pushed 1 new commit to master: https://git.io/vHSkb
<NixOS_GitHub>
nixpkgs/master 5fbab5d Tim Steinbach: linux: 4.12-rc4 -> 4.12-rc5
NixOS_GitHub has left #nixos []
<joepie91>
gchristensen: do you know of any prior art on bundling XNA games?
Infinisil has quit [(Quit: leaving)]
<joepie91>
er
<joepie91>
on packaging XNA games *
<joepie91>
I'm having considerable ongoing trouble with it :P
filterfish has quit [(Ping timeout: 255 seconds)]
<catern>
yeah, I wish that I could pass the http url of a git repo to nix-env -f
Infinisil has joined #nixos
Infinisil has quit [(Quit: leaving)]
<joepie91>
okay, fixed my issue. turns out that the bundled msc in the game didn't work for reasons, so it broke while compiling scripts in the game
<joepie91>
symlinked local msc to nixpkgs-installed msc in Mono
<joepie91>
crashed on corlib mismatch, twiddled around with exact Mono versions until it didn't
<joepie91>
(for future reference for people packaging this type of stuff)
<joepie91>
works with Mono 4.4 now
filterfish has joined #nixos
roconnor has quit [(Ping timeout: 240 seconds)]
akamaus has joined #nixos
spinus has quit [(Ping timeout: 246 seconds)]
roconnor has joined #nixos
vaibhavsagar_ has quit [(Remote host closed the connection)]
vaibhavsagar_ has joined #nixos
filterfish has quit [(Ping timeout: 255 seconds)]
eacameron has quit [(Ping timeout: 240 seconds)]
nckx has joined #nixos
eacameron has joined #nixos
<pikajude>
can someone who's not using WSL do a `man lndir`
<pikajude>
mine says "If the link already exists but doesn't point to the correct file, the program prints the link name and the location where it does point." at the bottom
<pikajude>
lndir is used to build symlinkJoin derivations
<pikajude>
i don't understand how it can possibly work, and it certainly doesn't on my machine
johnramsden has quit [(Quit: Leaving)]
drakonis has quit [(Remote host closed the connection)]
drakonis has joined #nixos
<c74d>
pikajude: "No manual entry for lndir"; from what package is it?
<pikajude>
c74d: i don't know actually
<pikajude>
i think it's from xorg
<pikajude>
xorg.lndir
<pikajude>
i had the store path handy and i did man /nix/store/....-lndir/share/man/...
<LnL>
my idea was that people should use that to ping the darwin people for relevant issues/prs
<ikwildrpepper>
ah, you can ping teams? cool
<ikwildrpepper>
LnL: thnx
<LnL>
pretty sure you can with @darwin-maintainers
<ikwildrpepper>
yeah, just my github-ignorance
mudri has joined #nixos
<Svarog>
when nixos-rebuild says something like these paths will be fetched: fontconfig-etc, mirrors-list, fc-cache, etc, where exactly are those things defined? they don't appear in the package list
<sphalerite>
Svarog: many of them are built locally as dependencies of the system
<sphalerite>
they'll often end up in /etc
<Svarog>
so if i'm building a custom live cd, can i just force it to use what's on my system instead of it redownloading all that stuff?
<sphalerite>
that depends. If it is the same stuff, it will use what's on your system
<sphalerite>
If it isn't, then it won't
<Svarog>
i keep trying to cache dependencies by adding them to system.extraDependencies, but there is always more it wants to download and some of it like that example with fontconfig-etc, i don't know how to cache
<sphalerite>
All derivations get "cached"
<sphalerite>
they live on your system anyway, they'll only get downloaded if you don't already have them
<Svarog>
what i mean is when i do nix-collect-garbage -d, it gets rid of things like fontconfig-etc, so next time i build the live cd it wants to download fontconfig-etc again
<sphalerite>
oh right
<Svarog>
and i don't know how to get it to download it once and then keep it
<sphalerite>
don't gc :)
<Svarog>
heh
<Svarog>
there's gotta be a way to add a root to it
<Svarog>
so gc doesn't collect it
<sphalerite>
yeah, you can build the system using nix-build and the result link will be a gc root
<Svarog>
adding things to system.extraDependencies has that effect too - someone here told me about that, possibly clever
<sphalerite>
I think if you build config.system.build.toplevel or something along those lines (I probably got the order wrong)
<Svarog>
but i can't add fontconfig-etc because the package doesn't exist
<sphalerite>
But your system isn't the same thing as the live system you're building
<Svarog>
and yet it still gets downloaded and built every time i build a live cd
<sphalerite>
so what you want to keep around is the system of the live CD
<Svarog>
yes, but that fontconfig-etc must be coming from somewhere
<Svarog>
and if it is, it's defined somewhere
<Svarog>
and i should be able to manually install it or download it or add a gc root to it
<sphalerite>
By the fontconfig module
<Svarog>
which is installed
<sphalerite>
I'm not sure it's accessible, but what you want to do isn't to add that specific file
<Svarog>
but it still keeps redownloading it
<sphalerite>
instead build config.system.build.toplevel, which has a transitive dependency on the fontconfig thing
<sphalerite>
That way you know *nothing* that's required for building the CD image will be GCed
<Svarog>
well it's not just fontconfig, there is about 20-30 dependencies that keep getting redownloaded
<sphalerite>
and you don't have to care specifically about the config thing
<Svarog>
pulseaudio is another one
<sphalerite>
yes
<sphalerite>
What's the expression you use to build your disk image?
<sphalerite>
somewhere in there you probably have config.system.build.isoImage, right?
<Svarog>
it doesn't matter that i have pulseaudioFull and pulseaudioLight cached locally, it seems to want puseaudio-10.0, which again doesn't appear on the list of packages
<Svarog>
just a sec
<Svarog>
hmm it's not a single expression as such - i've written a custom package set that contains iso = config.system.build.isoImage, where i took the definition of config from gnome
<Svarog>
pkgs/desktops/gnome-3/3.22/installer.nix
<Svarog>
so i then build it with nix-build '<custompkgs>' -A iso
<sphalerite>
Right
<Svarog>
i have the same problem when i try to build the default live cd without any customizations or my custom packages
<sphalerite>
just add somewhere something that lets you build config.system.build.toplevel instead
jensens has quit [(Remote host closed the connection)]
<Svarog>
readline, pulseaudio, fontconfig-etc opengl-drivers udev-rules, w3m and a bunch of other things are always redownloaded
<sphalerite>
Build that, and the result should have transitive dependencies on all that stuff
indi_ has joined #nixos
<Svarog>
where do i need to add that?
jensens has joined #nixos
pie_ has joined #nixos
<sphalerite>
I'm guessing that to build the image you invoke nix-build somehow, right?
<Svarog>
nix-build '<custompkgs>' -A iso
pietranera has joined #nixos
<sphalerite>
then add toplevel = config.system.build.toplevel alongside iso, and do nix-build '<custompkgs>' -A toplevel
<sphalerite>
then keep the result symlink around and it should stop GC from removing the dependencies
<Svarog>
oh
<Svarog>
hmm would that work if i added it to /etc/nixos/configuration.nix?
<sphalerite>
Possibly, but do you have a good reason to do that? Does the iso you're building have anything to do with your system configuration?
<sphalerite>
you could add config.system.build.toplevel to your extraDependencies of course, but it seems to me to make more sense to divorce it from the system config
<Svarog>
no but i'd like the system configuration to have everything i might need cached so it can be used in an environment without internet access
<Svarog>
i guess the effect is the same if i use nix-build, it's just that i like all dependencies to be in the main config file so i only ever need to worry about the main config file
<sphalerite>
I guess
roconnor has quit [(Ping timeout: 240 seconds)]
<sphalerite>
I prefer to keep the system config referring only to systemwide stuff, have a separate config for my user software, and for project-specific stuff it's a result symlink in the project dir
<Svarog>
i sort of have that in that i have a bunch of different modules with configuration.nix importing them all
reinzelmann has quit [(Ping timeout: 240 seconds)]
<Svarog>
i used to use nix-build to pull things i wanted cached locally but not installed, until i found out about system.extraDependencies which has the same effect
<Svarog>
keeps things in the store without having them installed
pie_ has quit [(Remote host closed the connection)]
pie_ has joined #nixos
orivej has joined #nixos
reinzelmann has joined #nixos
<sphalerite>
Svarog: you may also want to set nix's gc-keep-outputs and gc-keep-derivations options, that way the build dependencies for everything are kept
<NixOS_GitHub>
[nixpkgs] gebner pushed 1 new commit to master: https://git.io/vHSgB
<NixOS_GitHub>
nixpkgs/master 798e256 Gabriel Ebner: cups-filters: fix path to pdftops...
NixOS_GitHub has left #nixos []
pietranera has quit [(Ping timeout: 246 seconds)]
indi_ has quit [(Remote host closed the connection)]
<dhess`>
sphalerite: any idea why /nix/store might be read-only during nixos-install on this install image?
indi_ has joined #nixos
<dhess`>
sphalerite: or how I can fix it?
<Svarog>
thanks sphalerite, will try that as well
<dhess`>
very strange, never seen this issue before after many nixos-installs
<sphalerite>
dhess`: because of the configuration, I remember there was a commit a while back that mounted the nix store as read-write using unionfs to put a tmpfs on top of the read-only nix store from the image
<sphalerite>
dhess`: that remounting configuration might not be included in your image
<dhess`>
sphalerite: possible, this is an experimental aarch64 image from Dezgeg
<dhess`>
although I assume he's done tested some installs from it
<dhess`>
s/done//
pietranera has joined #nixos
indi_ has quit [(Ping timeout: 246 seconds)]
<Dezgeg>
I haven't run nixos-install from that one, just nixos-rebuild to the same sd card
justan0theruser has joined #nixos
<Dezgeg>
but I don't see how nixos-install should act any different
bennofs has joined #nixos
indi_ has joined #nixos
justanotheruser has quit [(Ping timeout: 240 seconds)]
<dhess`>
I can write /mnt/nix/store just fine, and /nix, but not /nix/store
<dhess`>
it's very odd
<Dezgeg>
no idea
<dhess`>
:(
zeus_ has quit [(Remote host closed the connection)]
<marvin3k>
i have a function which creates an a-priori unkown number of files via "runCommond" churning a binary input file. each created file in turn might result in an a-priori unkown number of derivations, determined via "builtins.readFile". a final "buildEnv" symlinks all resulting derivations. my problem is: when i build the first stage alone, parallelization is in effect. if i directly run the "buildEnv"
<marvin3k>
stage, all the (expensive) "runCommand" are executed in series and only the last stage of derivations in parallized. using "builtins.seq" or "builtins.deepSeq" didn't help, as well as "buildInputs". any tips on how to proceed? where is my mental model broken?
mudri has joined #nixos
faffolter has quit [(Ping timeout: 240 seconds)]
indi_ has quit [(Remote host closed the connection)]
indi_ has joined #nixos
bennofs has quit [(Ping timeout: 240 seconds)]
indi_ has quit [(Remote host closed the connection)]
indi_ has joined #nixos
mpcsh has quit [(Quit: THE NUMERICONS! THEY'RE ATTACKING!)]
mpcsh has joined #nixos
chpatrick has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] edolstra pushed 2 new commits to master: https://git.io/vHSVR
<NixOS_GitHub>
nixpkgs/master ca490a6 Daiderd Jordan: gdb: disable format warnings...
<NixOS_GitHub>
[nixpkgs] dguibert opened pull request #26535: gfortran7: added by wrapping gcc7 (master...dg/gfortran7) https://git.io/vH9ZI
NixOS_GitHub has left #nixos []
zeus_ has quit [(Ping timeout: 245 seconds)]
leat has quit [(Quit: leat)]
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
Amirkin_ has quit [(Quit: Page closed)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] globin pushed 6 new commits to master: https://git.io/vH9Zg
<NixOS_GitHub>
nixpkgs/master c5467a1 Robin Gloster: errbot module: needs network-online to connect properly
<NixOS_GitHub>
nixpkgs/master d138f9e Franz Pletz: gstreamer-plugins-bad: add deps for HLS & H.265
<NixOS_GitHub>
nixpkgs/master 934b723 Franz Pletz: libfann: init at 2.2.0
NixOS_GitHub has left #nixos []
Itkovian has joined #nixos
ertes-w has quit [(Ping timeout: 240 seconds)]
leat has joined #nixos
faffolter has quit [(Read error: Connection reset by peer)]
fabian_a has joined #nixos
mudri has joined #nixos
nithor has quit [(Ping timeout: 260 seconds)]
tmaekawa has joined #nixos
<jophish>
Why is the closure >41GB for a 28GB store path?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] gratuxri opened pull request #26536: Bump fvwm to 2.6.7 (master...master) https://git.io/vH9no
NixOS_GitHub has left #nixos []
endformationage has joined #nixos
FRidh has joined #nixos
FRidh has quit [(Remote host closed the connection)]
<jophish>
the store path requisites are <1GB
lhart has joined #nixos
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<lhart>
Hi everyone! I have just a little question. Is it expected that i need to sudo to be able to use nixos-rebuild? Thanks!
mudri has quit [(Ping timeout: 240 seconds)]
<marvin3k>
on nixos? yes, doing nixos-rebuild is going to change the operating system, like installed kernel or grub
<marvin3k>
when all you want is to install a piece of software in your local user profile "nix-env" might do the job
Itkovian has joined #nixos
freusque has quit [(Quit: WeeChat 1.7)]
Itkovian has quit [(Client Quit)]
<gchristensen>
nixos-rebuild build will not require sudo, switch and boot will
<gchristensen>
and maybe `test`
<marvin3k>
mh, ja, thats more true
<lhart>
ok, I wasn't sure and it's not used in the manual. hum ok...
Itkovian has joined #nixos
<gchristensen>
lhart: is that problematic?
fabian_a has quit [(Ping timeout: 255 seconds)]
Svarog has quit [(Quit: Ex-Chat)]
faffolter has joined #nixos
faffolter has quit [(Changing host)]
faffolter has joined #nixos
<lhart>
hum maybe because I can't nixos-rebuild without sudo and I think it screws permissions. Like asking password to suspend system, use an usb key etc...
<lhart>
I just made a fresh install and it's working properly
<lhart>
I ll need to try agoin to be sure, I think.
<lhart>
*again
nschoe has quit [(Quit: Program. Terminated.)]
nithor has joined #nixos
<lhart>
nixos-rebuild doesn<t seem to have permission to create symlinks... :/
<gchristensen>
that is the failure I would expect if you tried to do nixos-rebuild boot or nixos-rebuild switch without sudo
<lhart>
oh ok, my bad i read your answer too quickly... thann you!
<lhart>
th
<lhart>
*thank
<lhart>
Have a nice day!
lhart has quit [(Quit: Page closed)]
<gchristensen>
you're welcome, good luck and have fun...oh
xeviox has quit [(Ping timeout: 240 seconds)]
nithor has quit [(Ping timeout: 245 seconds)]
civodul has quit [(Quit: ERC (IRC client for Emacs 25.2.1))]
wmertens[m] has joined #nixos
jonte_ has joined #nixos
nevermind has joined #nixos
[0x4A6F] has joined #nixos
nithor has joined #nixos
logzet has joined #nixos
indi_ has joined #nixos
bennofs1 has joined #nixos
faffolter has quit [(Ping timeout: 255 seconds)]
Filystyn has joined #nixos
Filystyn has quit [(Changing host)]
Filystyn has joined #nixos
indi_ has quit [(Ping timeout: 240 seconds)]
digitus has joined #nixos
arianvp2 has quit [(Ping timeout: 258 seconds)]
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
mudri has joined #nixos
arianvp2 has joined #nixos
zeus_ has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nix] edolstra pushed 2 new commits to master: https://git.io/vH9EI
<NixOS_GitHub>
nix/master 25230a1 Eelco Dolstra: On macOS, don't use /var/folders for TMPDIR...
<NixOS_GitHub>
nix/master 177f399 Eelco Dolstra: Suppress spurious "killing process N: Operation not permitted" on macOS
NixOS_GitHub has left #nixos []
zeus_ has quit [(Ping timeout: 246 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nix] edolstra pushed 1 new commit to 1.11-maintenance: https://git.io/vH9EC
<NixOS_GitHub>
nix/1.11-maintenance 36f363b Eelco Dolstra: On macOS, don't use /var/folders for TMPDIR...
NixOS_GitHub has left #nixos []
iyzsong has quit [(Remote host closed the connection)]
danl1240 has joined #nixos
jgertm has joined #nixos
taktoa has joined #nixos
zeus_ has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nix] edolstra pushed 1 new commit to 1.11-maintenance: https://git.io/vH9zB
zeus_ has quit [(Read error: Connection reset by peer)]
zeus_ has joined #nixos
tvon has joined #nixos
Neo-- has quit [(Remote host closed the connection)]
Svarog has joined #nixos
<Svarog>
is there a way to generate a random number in nix?
<Svarog>
or more specifically, what i'd like to do is generate unique hostnames in a live cd, so if multiple instances are running on the same network they'll have unique hostnames
<dtzWill_>
that seems awfully non-deterministic :3
<Svarog>
i know :(
<Svarog>
i can't think of anything better
<Svarog>
open to suggestions
<dash>
Svarog: generate it from mac address
<makefu>
Svarog: how about not depending on a unique hostname?
<dtzWill_>
Svarog: do you need to verify the hostnames are unique across the set before deployment? If not, maybe just generate it on-the-fly at boot-time based on some physical attributes?
<Svarog>
hmm that could work
goibhniu has quit [(Ping timeout: 246 seconds)]
<Svarog>
how would i query something like a mac address from nix? or would i have to do it through a shell script and somehow call it from nix?
<makefu>
you could use an activation script which runs "ip addr" to set the hostname. activation scripts are run at the beginning of every boot. alternatively you could configure a systemd service which runs at startup
tvon has quit [()]
jensens has quit [(Ping timeout: 260 seconds)]
k2s has joined #nixos
tmaekawa has quit [(Remote host closed the connection)]
tmaekawa has joined #nixos
k2s has quit [(Client Quit)]
indi_ has joined #nixos
tmaekawa has quit [(Client Quit)]
tmaekawa has joined #nixos
tmaekawa has quit [(Client Quit)]
indi_ has quit [(Ping timeout: 240 seconds)]
arianvp2 has quit [(Quit: arianvp2)]
jonte_ has quit [(Ping timeout: 240 seconds)]
arianvp2 has joined #nixos
stallion has joined #nixos
jonte_ has joined #nixos
arianvp2 has quit [(Quit: arianvp2)]
mudri has quit [(Ping timeout: 260 seconds)]
<samueldr>
I'm having a hard time setting up nat networking with declarative containers
<samueldr>
well
<samueldr>
not that much
proteusguy has joined #nixos
<samueldr>
there's one issue left, where I'm unable to talk to the outside network from inside the container
<samueldr>
The manual, section 28.3 mentions using nat for this purpose
slabity has joined #nixos
jonte_ has quit [(Ping timeout: 240 seconds)]
faffolter has joined #nixos
faffolter has quit [(Changing host)]
faffolter has joined #nixos
<samueldr>
(making a gist of an elided configuration [I'll need to actually try that configuration later on])
<tilpner>
akamaus - Put nix.maxJobs = n; into your nixos-config
freezeboy has joined #nixos
<tilpner>
akamaus - Or maybe it was nix.buildCores, I think I set both :c
ixxie has quit [(Ping timeout: 255 seconds)]
freezeboy has quit [(Client Quit)]
freezeboy has joined #nixos
proteusguy has quit [(Ping timeout: 240 seconds)]
<freezeboy>
hi, is there somewhere on the nixos.org website or on github a raw diff (new packages/modules) between the 17.03 release and the state of the future 17.09 branch ?
jbgi has joined #nixos
<simpson>
freezeboy: You can ask for such a comparison on github using their tools.
<freezeboy>
simpson: you mean playing with git and sed ? or a better solution ?
<akamaus>
tilpner, thanks! I looked at nix-daemon.nix. Looks like buildCores controls parallelism inside a single build (aka make -j) and maxJobs sets the number of simultaneous builds.
<simpson>
freezeboy: I guess I don't understand what you want, sorry.
<dash>
freezeboy: you could look at the changes to the docs instead :)
<freezeboy>
dash: well, the release-notes.xml seems outdated in nixos-unstable
<dash>
they possibly have not been writtenyet
<freezeboy>
simpson: basically some kind a simplified output of some git diff release-17.03..master nixpkgs/top-level/all-packages.nix would be great
ris has joined #nixos
DutchWolfie has joined #nixos
DutchWolfie has quit [(Changing host)]
DutchWolfie has joined #nixos
<freezeboy>
dash: simpson: or just the content of the /nixos/options.html and /nixos/packages.html on the website but with the unstable branch would also be enough
riclima has joined #nixos
mudri has joined #nixos
sigmundv_ has joined #nixos
arianvp2 has joined #nixos
jeaye has quit [(Quit: WeeChat 1.7.1)]
Wizek_ has quit [(Ping timeout: 268 seconds)]
ixxie has joined #nixos
stallion has quit [(Quit: stallion)]
joehh has quit [(Ping timeout: 255 seconds)]
joehh has joined #nixos
arianvp2 has quit [(Client Quit)]
proteusguy has joined #nixos
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
arianvp2 has joined #nixos
Nixer has joined #nixos
Itkovian has joined #nixos
<Sonarpulse>
channels are screwed TT
<Sonarpulse>
more tears for CI
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] disassembler opened pull request #26537: lifelines: init at 2017-06-12-unstable (master...lifelines) https://git.io/vH9yC
NixOS_GitHub has left #nixos []
Guest42737 is now known as katyucha
<Nixer>
Could someone point me to where the nix cache downloader lives? I want to see the code that pulls down already built packages via nix-shell, nix-env, or nix-rebuild? Thanks
orivej has quit [(Ping timeout: 255 seconds)]
Sonarpulse has quit [(Ping timeout: 246 seconds)]
orivej has joined #nixos
arianvp2 has quit [(Quit: arianvp2)]
freezeboy has quit [(Quit: Konversation terminated!)]
loook has joined #nixos
<loook>
I'm getting a kernel panic on boot, after stage2, since systemd is not found. Any ideas what would cause this?
simukis has joined #nixos
<jophish>
peti: Recently (a couple of months ago probably) ghcWithPackages started to print a load of output when building, when previously it did not. Do you know what change this could have been, and if it's possible to reverse?
<loook>
I've had a working NixOS machine for over a year and I just rebooted into this issue.
<loook>
Mic92: Agreed. Must be a DigitalOcean issue then.
<Mic92>
loook: what is your boot loader?
<Mic92>
grub?
<loook>
Yes.
[0x4A6F] has quit [(Remote host closed the connection)]
<loook>
Trying to get in there so I can select the previous NixOS revision, but I haven't been able to get a DO console while GRUB is still up.
<loook>
If I can get into GRUB, this should be simple enough to fix.
<Mic92>
loook: can you get into the rescue shell?
<Mic92>
of the initrd
arianvp2 has joined #nixos
<loook>
DO has a rescue kernel which I can try booting. Currently, I'm making a snapshot of the machine before I fuck anything else up.
<Mic92>
loook: you could try to change your grub configuration from the rescue shell.
<loook>
Ok, nope, it won't slip in the rescue kernel.
<loook>
Mic92: How else were you thinking about getting into a rescue shell? By the time I get the DO console open, I'm booting NixOS stage1, I see stage2 fly by, then the panic.
[0x4A6F] has joined #nixos
<Mic92>
loook: the first question is if panicOnFail=1 is set in the digitalocean initrd.
arianvp2 has quit [(Quit: arianvp2)]
<Mic92>
loook: did you try to send Ctrl + C to the script?
<clever>
another thing i had thought of is a custom bootloader that supports network intervention
<Mic92>
We probably also need a rescue shell in stage-2
alx741 has quit [(Quit: alx741)]
<Nixer>
Can anyone point me to the nix cache downloading code? I know it lives in these repo: https://github.com/NixOS/nix/ somewhere, but don't know where yet.
logzet has quit [(Remote host closed the connection)]
<Mic92>
loook: you could ask the support to boot your vm with an nixos ISO.
<clever>
if you use another working nixos droplet, you can mount the old one to /mnt and then do "nixos-install --chroot" to chroot into it, and then do repairs, and restore the previous connections
<peti>
jophish: That was probably a side-effect of updating to the latest major release of hoogle.
<loook>
Mic92: Yeah, I've made a support ticket already; we'll see when they get back to me.
proteusguy has quit [(Ping timeout: 245 seconds)]
<loook>
clever: _Maybe_. I don't think they support mounting snapshots, or other droplet volumes, as generic volumes.
arianvp2 has quit [(Client Quit)]
<loook>
Will probably be switching to Linode, when this blows over.
<Mic92>
They have ec2 like block storage, but this is not the default.
<Mic92>
I have my stuff on netcup, which allows to upload custom iso as installation medium
tmaekawa has joined #nixos
indi_ has quit [(Ping timeout: 255 seconds)]
zetok has quit [(Ping timeout: 240 seconds)]
<clever>
Mic92: with the kexec stuff i linked previously, i can do an install without needing any custom iso to be mounted
<andrewrk>
I accidentally destroyed my boot partition
<andrewrk>
and then restarted my computer
<andrewrk>
:-(
<Mic92>
clever: does not help, if you have an unbootable system.
<Mic92>
andrewrk: dont panic
<clever>
andrewrk: you will need to boot from the install cd, mount the rootfs to /mnt, and put the new boot at /mnt/boot/, then run "nixos-install --chroot" and "nixos-rebuild boot"
<andrewrk>
that's not bad. I'll try that
<clever>
Mic92: yeah, the kexec requires a working linux kernel, of any distro
<Mic92>
fair
ertes has joined #nixos
<Mic92>
clever: I wait for your pull request to nixpkgs ;)
<Mic92>
clever: mhm, how about using the current routing table
zetok has joined #nixos
<clever>
slabity: mkBefore and mkAfter cant easily be overridden, you would need to checkout a clone of nixpkgs on the same revision, and just edit the file directly
<Nixer>
Does anyone know if the nix install script change recently (https://nixos.org/nix/install)? I can no longer install nix on centos.
<clever>
Mic92: in the case of digital ocean, they will just open your rootfs and tinker with /etc/network/interfaces any time the IP changes
<Nixer>
I'm seeing this error: error: removing extended attribute ‘security.selinux’ from ‘/nix/store/3rh146gsqybkc7br5lnlk53p1ll4lna7-env-manifest.nix’: Permission denied /tmp/nix-binary-tarball-unpack.yHy6iHuqu0/unpack/nix-1.11.10-x86_64-linux/install: unable to install Nix into your default profile
<clever>
Mic92: and a one-time inspection of the current IP cant deal with that
<loook>
Mic92: clever Now that I'm in grub, do I have any better option than to just select one of the older configurations?
<Mic92>
clever: creepy
<loook>
Anything I can add to the kernel line to help ensure I make it in?
vandenoever has joined #nixos
arianvp2 has joined #nixos
<slabity>
clever: That's disappointing. Do you know of a better way to make plymouth start before the LUKS password prompt appears?
<Mic92>
loook: press e
<loook>
Mic92: Sure, I'm familiar with its UI. What're you thinking?
<clever>
slabity: all i can think of is to modify nixpkgs directly, and maybe send a PR upstream so it always does it that way
<Mic92>
loook: then I do not understand your question.
<clever>
loook: setting debug1mounts forces it to "fail" after it has mounted everything
<clever>
so its not a true failure
<loook>
Hm. It also failed with boot.debug1 though, so I figured I'd let it go further.
<clever>
debug1 just forces it to fail sooner
<slabity>
clever: Can I override just that particular file? Or do I need to clone the entire nixpkgs repo?
<loook>
clever: Right.
<clever>
slabity: currently, there is no way to override modules, you need to clone the entire nixpkgs
civodul has joined #nixos
<loook>
clever: Unfortunately, when it fails, it prompts me for input but doesn't read any of my keystrokes.
<clever>
loook: ah, the keyboard drivers might not be in the initrd
<clever>
loook: try booting with init=/bin/sh then and see what that does
arianvp2 has quit [(Client Quit)]
<loook>
clever: Ok.
<clever>
that should replace stage2 with a shell, and it will have the full rootfs mounted, so it might have more drivers
<clever>
though systemd hasnt ran yet, so it might not have auto-loading
zeus_ has quit [(Read error: Connection reset by peer)]
zeus_ has joined #nixos
<loook>
Ah /bin/sh didn't exist.
<clever>
it will also falsely claim it doesnt exist
<clever>
tell it to continue anyways
<loook>
It continued onto the panic, without my input.
<clever>
the problem is that /bin/sh is a symlink with an absolute path, and its being checked from outside the chroot
<clever>
oh right, no input at the initrd
<clever>
loook: this one will take a bit more typing, but what about init=/nix/var/nix/profiles/system/sw/bin/bash ?
josePhoenix has joined #nixos
<loook>
O
<loook>
I'll try. It takes a lot of luck, refreshing the console at the right time while grub is still up.
<Nixer>
Is anyone familiar with the how the nix install (https://nixos.org/nix/install) interacts with SELinux? I'm running into an install failure suddenly: error: removing extended attribute ‘security.selinux’ from ‘/nix/store/3rh146gsqybkc7br5lnlk53p1ll4lna7-env-manifest.nix’
<Nixer>
Can't seem to track down the cod that is trying to modify the selinux attribute.
<dhess`>
Is there any way with NixOps to have the deployment host download from the binary cache on behalf of the target host?
<loook>
clever: Well, I have a bash shell now, but no userland.
<dhess`>
I've got a NixOps deployment target whose outbound Internet access is blocked
<clever>
loook: yeah, $PATH will be blank, try export PATH=/nix/var/nix/profiles/system/sw/bin
<dhess`>
what's really odd is that I previously deployed to this host just fine, but I had to re-install and now it's trying to download from the binary cache when I nixops deploy
<Nixer>
clever: Thanks. Hopefully, I'll be able to figure out why I can't install nix any longer.
<loook>
clever: So, we still don't know what's causing the boot to not find systemd, right?
<loook>
I could try to continue down stage2, but I'll likely end up at the same panic.
<clever>
loook: i would check this first, "ls -lh /nix/var/nix/profiles/system/sw/bin/systemd"
<clever>
loook: what does that say?
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<loook>
clever: It's not there.
<loook>
o.O
<clever>
loook: can you cd into /nix/var/nix/profiles and check all of the other system sylinks?
<clever>
loook: try looking in ls -l /nix/var/nix/profiles/system/systemd/lib/systemd/systemd
ryanartecona has joined #nixos
<loook>
That's there, though it's dated for 1969. It's a binary, not a link.
<clever>
yeah, that date is normal
<clever>
its jan 1st, 1970, minus your timezone offset
pie_ has quit [(Changing host)]
pie_ has joined #nixos
<clever>
loook: did you have the option to boot older generations in grub?
<loook>
Yes.
<clever>
have you tried booting those yet?
<loook>
Nope, once I got into grub I asked in here what the best move would be.
<clever>
ah
<clever>
try going back to grub and picking an older generation
<loook>
Didn't want to be hasty with my new grub access.
jgertm has joined #nixos
<srhb>
Is there a nicer way to check old versions of haskell deps than this? Get what you can with nix-shell, then use cabal install --dependencies-only for the remainder? I'm wondering if there's a way to ad-hoc run cabal2nix and put together everything in nix shell.
<srhb>
(Following what the dependency solver claims)
<srhb>
I suppose what I want is a recursive cabal2nix
<loook>
clever: The oldest config I have still does it.
<loook>
clever: This is almost certainly because it's the result of something DO has done.
<clever>
loook: can you get back into that bash shell and then run "nix-store --verify --check-contents" ?
<loook>
Yeah
<loook>
clever: Connection refused on the nix daemon
<clever>
loook: you may need to export NIX_REMOTE=local i think
k0001 has joined #nixos
<loook>
nix-store says local is an invalid setting for NIX_REMOTE
<clever>
just unset NIX_REMOTE?
<loook>
Nice.
ilyaigpetrov has quit [(Quit: Connection closed for inactivity)]
<loook>
I imagine this'll take a while.
<clever>
yeah
<loook>
clever: Thanks so much for the help thus far.
<clever>
its hashing everything in /nix/store to see if anything is currupt
<loook>
If I can get this running, even if I do it manually, with the grub boot still broken, it'll at least get my services back online and I can migrate to a hoster which actually allows custom ISOs.
<loook>
clever: /nix/store/*-unstable-16.09pre* was modified
<clever>
loook: in theory, once you can get it to boot fully, you can just "nixos-rebuild boot" and it will update the boot config
<loook>
That's the only error.
<clever>
loook: can you screenshot it?, the hashes can help in debugging the cause
<clever>
loook: also, do you know how to configure an ip with only "ip" or "ifconfig" ?
<loook>
It removed a lot of stale links, no it shows per-user/root/channels-11{1,2,3,4}-link
<clever>
yeah, that confirms its a channel, but that shouldnt have broken it
<clever>
id get the network up first
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dtzWill opened pull request #26539: elfkickers: init at 3.1 (master...feature/elfkickers) https://git.io/vH9h1
NixOS_GitHub has left #nixos []
<clever>
loook: "ip addr add 192.241.218.198/24 dev eth0; ip route add 192.241.218.0/24 dev eth0; ip route add via 192.241.218.1 dev eth0"
<clever>
loook: that will get you internet access, then you can manualy run sshd with an absolute path and &, which should get you a better terminal and copy/paste
<loook>
clever: The "via" in the second ip route command wasn't working.
<loook>
I removed it, while tinkering, to get this. That's likely the issue.
<clever>
you may need to delete the 192.241.218.1 route
<clever>
ip route del 192.241.218.1 dev eth0
<clever>
default via 192.168.2.1 dev eth0 metric 2
<clever>
it might also need the word default added, ip route add default via 192.241.218.1 dev eth0
Nixer has quit [(Quit: Page closed)]
<loook>
Nice, no more system error, but not yet a proper ping.
<loook>
Scratch that, system error, but promising hang beforehand.
<clever>
also keep in mind, dns may not work yet
<clever>
so you can only ping ip's
<loook>
Looks like I can ping the machine though, so we're probably good.
<loook>
Yep, agreed.
chthondweller has quit [(Ping timeout: 260 seconds)]
riclima has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<loook>
clever: Mimicking all the right sshd options will likely be a pain.
<clever>
loook: you can usualy just run sshd without any args and it will work
boomshroom has joined #nixos
<clever>
loook: but if you want the sshd-config to work, just read /nix/var/nix/profiles/system/etc/systemd/system/sshd.service
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<loook>
When I run sshd along, it says it doesn't find /etc/ssh/sshd_config and aborts.
<loook>
So I figure it needs a proper setup. When I provide it with the sshd_config, it wants the host files, etc.
<loook>
I'll take a look at the service.
<clever>
there should be a -c flag and an absolute path
pie_ has quit [(Remote host closed the connection)]
Filystyn has quit [(Quit: Konversation terminated!)]
<boomshroom>
Hello! I'm still trying to build the cross compiler. I sent a message to the mailing list, but I'm not sure if I need to subscribe before people will see it.
pie_ has joined #nixos
riclima has joined #nixos
<boomshroom>
I think I've made some progress, but I'm not sure if I really did.
DutchWolfie has quit [(Quit: Konversation terminated!)]
<slabity>
Is there anything earlier than boot.initrd.preDeviceCommands?
<disasm>
are they any known vulnerabilities for breaking out of a nixos container? I accidentally set immutableUsers to false in my config and now my sudo password no longer works, but I have a root shell open inside a container.
<clever>
disasm: do you have physical access or access to grub?
Itkovian has joined #nixos
<disasm>
clever: I do have physical access. I can always use a thumb drive, just was wondering if there was an easier way.
<clever>
disasm: if you have physical access, you can just pick an older generation, or add init=/bin/sh to the kernel commandline and get instant root
<clever>
no need for a thumbdrive
<disasm>
ah thx
indi_ has joined #nixos
<boomshroom>
Should I subscribe to the mailing list if I want to ask a question?
indi_ has quit [(Remote host closed the connection)]
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
faffolter has quit [(Ping timeout: 240 seconds)]
indi_ has joined #nixos
Itkovian has joined #nixos
indi_ has quit [(Ping timeout: 260 seconds)]
ericsagnes has quit [(Ping timeout: 255 seconds)]
<loook>
clever: Should I be able to spawn init from my shell and debug further?
<loook>
Since I've splipped bash in before init
<clever>
loook: yeah, if you exec stage2 from that shell, it will get pid1 and should try to boot fully
<loook>
The question is, I suppose, whether or not that will result in another panic.
<clever>
yeah
<loook>
Ideally not, unless bash itself fails.
<clever>
if you exec, it can result in another panic
<clever>
if you just run it normally without exec, it will drop back to the shell
<clever>
but systemd will get upset if its not pid 1
riclima has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
pie_ has quit [(Ping timeout: 260 seconds)]
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
michas has joined #nixos
johnsonav has quit [(Ping timeout: 246 seconds)]
Itkovian has joined #nixos
pie_ has joined #nixos
<loook>
clever: The end of stage-2-init.sh run systemd, but it sets PATH to /run/current-system/systemd/lib/systemd
<boomshroom>
I think I finally got a working cross compiler! :D And surprise, surprise, it looks nearly identical to gccCrossStageStatic.
<loook>
clever: "Trying to run as user instance, but the system has not been booted with systemd."
ericsagnes has joined #nixos
<clever>
so the elf file is at least valid
<loook>
I can try with --system
<clever>
it has to be ran as pid 1 via exec to make the system "booted with systemd"
<loook>
Yeah, it failed. :P
<loook>
It looks to me like activation is failing.
<clever>
what if you try running activate manualy?
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<loook>
Since, even after running it, I don't have a /var/run/current-system
<clever>
it should be just /run/current-system
sigmundv__ has joined #nixos
Itkovian has joined #nixos
<loook>
I think I see it.
<loook>
If so, this would be my fault.
<boomshroom>
Not that I'm a systemd hater, bu is it possible to build a nix system with alternate inits? I'd imagine the core of NixOS wouldn't be dependant on systemd, but there's probably a lot of scripts that are specific to a given init system. That said, it could still be possible to rewrite some of them.
<clever>
loook: ?
<loook>
clever: I have an activation script which is using the network to check for updates. That's failing the activation script here.
<simpson>
boomshroom: Sure. And several folks have done so on an ad-hoc basis, and many folks would welcome contributions to nixpkgs which help detangle this.
<clever>
boomshroom: there is an open issue about reviving the jobs framework, so services can be defined without using systemd terms
frio[m] has joined #nixos
<boomshroom>
Interesting. With how NixOS works, if you can make it boot with an alternate init, it would be possible to install multiple and select one at boot.
<simpson>
boomshroom: Maybe!
<boomshroom>
Again, I'm not a systemd hater, I just find it cool how with the right scripts, one could switch between completely different setups with minimal effort after the setup.
<boomshroom>
Go options!
<simpson>
I *am* a systemd hater, but I'm also tired and I have bigger fish to fry.
<clever>
i have managed to crash pid 1 by just spawning a systemd container on a malfunctioning fuse layer
<boomshroom>
Oh, no! My build failed! checking for shl_load... configure: error: Link tests are not allowed after GCC_NO_EXECUTABLES.
<clever>
andrewrk: because some distros dont make a combined iso that can also boot from usb
<simpson>
andrewrk: In general, when software exists, it's because some folks somewhere thought that there was a hole in the world and that it could be plugged by writing code.
hiratara has joined #nixos
<andrewrk>
clever, when you say "put the new boot at /mnt/boot/" what exactly does that mean?
<andrewrk>
mount the boot device there?
<clever>
yeah
<andrewrk>
clever, I mounted the root fs at /mnt, and there's already some stuf in /mnt/boot
<clever>
that stuff may be from when you accidentaly ran things without /boot mounted
<andrewrk>
so I should rm rf it
<clever>
andrewrk: what fs is /?
<andrewrk>
I think / is the live cd. but I have my ext4 fs mounted on /mnt
<clever>
ah, ext4 for the real /
<clever>
then you dont really even need a /boot fs
<clever>
if your doing legacy booting (non-efi)
<andrewrk>
I'm doing efi booting
<clever>
ah
<clever>
then youll want to make a fat32 and mount it to /mnt/boot
<clever>
as a safety, i would also "chmod 0 /mnt/boot" before mounting it
<andrewrk>
ok
<clever>
so you cant write to the directory under it
<clever>
you may also need to update /mnt/etc/nixos/hardware-configuration.nix to refer to the new fat32
spinus has quit [(Read error: Connection reset by peer)]
<andrewrk>
thanks
<andrewrk>
I'm basically following a section of the normal installation instructions at this point, right?
<clever>
almost, you can skip a full nixos-install and instead do nixos-install --chroot, followed by nixos-rebuild boot
spinus has joined #nixos
aanderse has quit [(Ping timeout: 240 seconds)]
<disasm>
andrewrk: I'd also use the flag --no-root-passwd so it doesn't ask you to reset the root pw
<clever>
if your using --chroot, that wont be an issue
<disasm>
oh, never mind :)
<clever>
since nixos-rebuild is doing the actual build
<Mic92>
3/go 8
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<disasm>
is there a way to do rotation with services.postgresqlBackup?
Wizek has quit [(Ping timeout: 246 seconds)]
qwebirc68691 has joined #nixos
slabity has joined #nixos
obadz has quit [(Ping timeout: 240 seconds)]
<slabity>
Is there someplace that lists valid packages for `listof package` option types? For example, what are valid packages for `hardware.opengl.extraPackages`?
obadz has joined #nixos
<clever>
slabity: everything put int opengl.extraPackages is added to the buildEnv that lands in /run/opengl-driver
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peti pushed 9 new commits to release-17.03: https://git.io/vHHnr
<NixOS_GitHub>
nixpkgs/release-17.03 6f295b8 Maximilian Bosch: oh-my-zsh: add module (#25140)...
<NixOS_GitHub>
nixpkgs/release-17.03 533f813 Maximilian Bosch: zsh-syntax-highlighting: Add more configuration options and move to module (#25153)...
<NixOS_GitHub>
nixpkgs/release-17.03 ffb4c1f Maximilian Bosch: programs.zsh.syntax-highlighting: refactor `highlighters` option for proper validation...
NixOS_GitHub has left #nixos []
<clever>
slabity: what are you trying to do?
<slabity>
clever: Just trying to learn how to use nixos and configure my system
<slabity>
I'm switching from gentoo, so it's very different
<clever>
ah
<slabity>
I'd like to try to enable opencl on intel though
<clever>
thats an option that you dont really need to ever modify
<clever>
ah
<slabity>
clever: Yea, that's why I'm trying to find the list of valid packages for that option
<clever>
it would probably have to be a package containing libraries related to intel's opencl
mudri has joined #nixos
johnsonav has joined #nixos
<catern>
gchristensen: Nix can't install fonts off of NixOS? what about for packages installed with Nix?
<clever>
catern: the font installing works by managing things in /etc/fonts/
markus1189 has joined #nixos
<clever>
catern: and the entire /etc framework is nixos only
<catern>
oh :(
markus1209 has joined #nixos
<catern>
it's not possible to use ~/.fonts?
<clever>
if you just nix-env a font, it will land somewhere in ~/.nix-profile/
<clever>
and you will have to manualy link ~/.fonts up with symlinks
<gchristensen>
I think artifact would be imprecise
<gchristensen>
but not invalid
<catern>
artifact is a term that people who aren't familiar with Nix will understand
<catern>
so that's why I would go with it
<gchristensen>
fair
<gchristensen>
though you may define it "a store path (or an artifact)" or reversed ("an artifact, (a store path in Nix's vocabulary)")
<catern>
hmm
<gchristensen>
but you don't have to of course
<catern>
that makes sense, I will do that :)
<catern>
thanks
<gchristensen>
you're welcome! :)
<catern>
further question (now more object-level than meta-level :D) does anyone know of a P2P content-addressable-store that could just be dropped in for Nix?
<catern>
content-addressability has the big benefit of not requiring trust, if I understand correctly
<clever>
catern: a common problem, is that the hash in /nix/store/<hash>-<name> is the hash of the build script, not the output
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] shlevy pushed 1 new commit to release-17.03: https://git.io/vHHWR