<aanderse>
oh /dev/nvme0n1... thats one of them fancy pants ssds isn't it
<clever>
aanderse: yep, ive got 3 of them
Dedalo has joined #nixos
<aanderse>
oh wow thats fantastic :D
endformationage has quit [Ping timeout: 250 seconds]
<aanderse>
we built a pc for my friend a few weekends ago and got one of those
jluttine has quit [Ping timeout: 272 seconds]
o1lo01ol1o has joined #nixos
<o1lo01ol1o>
Is it possible to pass the path of the nix-store for the output of a callCabal2nix derivation to another derivation? I have multiple binaries that are built in a dependency that my top-level derivation needs to be able to access.
<clever>
o1lo01ol1o: just treat its result as a string somewhere else, its just a normal derivation
alienpirate5 has joined #nixos
<o1lo01ol1o>
ah, ok, I see. thanks!
<{^_^}>
[nixpkgs] @griff opened pull request #49620 → nixos/rspamd: Add support for included files → https://git.io/fxxI5
hamishmack has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
andreabedini has joined #nixos
oida has quit [Ping timeout: 256 seconds]
oida has joined #nixos
r5d has joined #nixos
<{^_^}>
[nixpkgs] @pbogdan opened pull request #49622 → polybar: nuke reference to stdenv.cc, small clean up → https://git.io/fxxLN
<{^_^}>
[nixpkgs] @fpletz opened pull request #49623 → python27: add patch to fix CVE-2018-1000802 → https://git.io/fxxLx
<arq1>
i tried nix-env -iA userPackage -f '<nixos-unstable>' and nix-env -iA userPackages -f $HOME/.config/nixos-unstable/config.nix with this config file https://pastebin.com/mE2X4EP3
<arq1>
thanks it worked, then the nix-env file doesnt take any arguments, like the wiki one?
endformationage has joined #nixos
<lukego>
So I have a directory called /run/wrappers/bin containing juicy programs that I'd like to run inside my build sandboxes, but when I add this dir as a sandbox path the programs don't work because they can't find libcap-ng.so.0. Any idea what's up? Is it the way these programs are built (how?) or the way they are mapped into the sandbox?
<srhb>
lukego: suid wrappers are too juicy to be sandboxed! :-)
<srhb>
(I don't actually know, but it seems like it should be impossible to run those in a sandbox, since they are likely to be able to break out)
wazboy has quit [Ping timeout: 240 seconds]
ThatOtherGuy is now known as LogicAside
<lukego>
I want to be able to break out. My use case requires running stuff as root.
<srhb>
lukego: Sounds like you want sandbox off anyway then.
<lukego>
But I don't want to accidentally break out and reference untracked dependencies
<srhb>
That's what you're actively doing though (re libcap)
<lukego>
edolstra said this should work and I'm trying to reproduce his example from nix build '(with import <nixpkgs> {}; runCommand "foo" {} "/run/wrappers/bin/ping -c 1 8.8.8.8; exit 1")' --no-allow-new-privileges
<lukego>
Interesting. I suppose that depends on your <nixpkgs> matching *exactly* with the libcap_ng that was used to build /run/wrappers though?
<lukego>
hm, no, since it will be dynamic linking that finds it via LD_LIBRARY_PATH, not a mapped nix store path...
<lukego>
doesn't seem to work for me anyway though. I already had nix.conf mapping /run/wrappers into the sandbox and I still get the error finding libcap-ng.so.0 when I add it as a build input
<lukego>
weirdly if I add the --extra-sandbox-paths like you have then it can't find /run/wrappers/bin/ping anymore...
<srhb>
lukego: You can extract the exact path with ldd and use extra-sandbox-paths if you need.
<srhb>
Oh, what?
<lukego>
but if I take the nix store path that is going to break when I least expect it right?
<srhb>
Well, it is anyway when wrappers change out of sync with your expression :)
<srhb>
As you pointed out yourself.
<lukego>
I want to avoid that problem somehow :)
Synthetica has quit [Quit: Connection closed for inactivity]
<lukego>
since wrapper-building and expression-building are separate activities quite likely done using separate nixpkgs verisons at different times
<srhb>
Yeah..
<lukego>
(what's the point of the wrappers btw if not to use them like this?)
<srhb>
Solely system-purpose. They can't exist in the nix store, hence this problem..
o1lo01ol1o has quit [Ping timeout: 252 seconds]
ddellacosta has joined #nixos
<srhb>
lukego: Is there really no way you can bump this problem to not require root? Like, giving the builders access to do whatever it is you need without root..
<lukego>
nope. I'm using nix to benchmark system software that is doing stuff like taking over PCI devices from the kernel. This is already working with 16.03 and I'm trying to port our nix code to 18.09
<lukego>
and since it's benchmarking it's not okay to wrap it in a layer of virtualization
<srhb>
Ouch, okay.
<lukego>
Maybe disabling the sandbox is a necessary evil
ddellacosta has quit [Ping timeout: 252 seconds]
<srhb>
It seems simpler than keeping this in sync at least...
<lukego>
Often I feel like my life would be simpler if I just patched the kernel to give every user superuser privileges...
<lukego>
ok with the sandbox disabled I still can't run sudo :-( sudo: /nix/store/v6l2sacryfr88yqq0pq7sia8wfgm9q31-wrapper.c:203: main: Assertion `!(st.st_mode & S_ISUID) || (st.st_uid == geteuid())' failed
<yl[m]>
how to turn off distributed builds when invoking nix-build?
<yl[m]>
I can't figure out the `--option` syntax for it
romanofskiWork has quit [Ping timeout: 240 seconds]
<lukego>
Can I somehow disable all of this security bullshit in NixOS? :)
reinzelmann has joined #nixos
<yl[m]>
I'm trying to wrap a binary to add go to the path like this `wrapProgram $bin/bin/gometalinter --prefix PATH : "${makeBinPath [go]}"` but it's production the PATH `export PATH='/nix/store/eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee-go-1.11/bin'${PATH:+':'}$PATH`
<yl[m]>
where is the eeeeee... coming from?
<lukego>
Specifically I think that what I need is to pass the "real" sudo into the build instead of the wrapper that wants to use seccomp and enforce extra rules etc.
<drakonis1>
the hash you mean
<drakonis1>
you can fetch the hash
<drakonis1>
or let it be generated
<drakonis1>
its a sha256 hash
<yl[m]>
never mind, I had to pass `allowGoReference = true;` to the buildGoPackage.
<jayQ>
Hi there. In a local project, `nix-build override.nix -A package` works to build the source. How can I specify this override to run in stdenv.mkDerivation after fetching the source (containing override.nix) via fetchurl?
reinzelmann has quit [Quit: Leaving]
<lukego>
Could be that what I need is an option for security.wrappers to simply setuid root an executable instead of generating a wrapper with seccomp etc
<lukego>
then I could override sudo to use that option and do what I want
<ekleog>
yl[m]: the eeeeeeeeeeeeeeee comes from erased references
<ekleog>
I guess go by default includes references to it in all builds, but you usually don't want it in the closure of the packages, so it defaults to erasing references to itself out unless you allowGoReference
<lukego>
On further reflection I think that what I *really* want is to simply stash a setuid-root copy of bash somewhere that I can pass into the sandbox. I suppose I'd need to define a nixos module for this.
mvnetbiz has joined #nixos
<lukego>
Statically linked so that it can run sandboxed without worrying about dependencies.
sb0 has quit [Ping timeout: 268 seconds]
o1lo01ol1o has quit [Ping timeout: 250 seconds]
vk3wtf has quit [Ping timeout: 246 seconds]
jsgrant has quit [Remote host closed the connection]
<Izorkin>
How to fix error - running perl packages - ./pt-align - error - ...-coreutils-8.30/bin/env: ‘perl -I/nix/store/...-perl-5.’: No such file or directory
MichaelRaskin has joined #nixos
zack_moe has joined #nixos
<elvishjerricco>
lukego: Wait, is this for the Nix build sandbox?
andreabedini has joined #nixos
jackdk has quit [Ping timeout: 252 seconds]
astronavt has quit [Quit: Leaving]
hamishmack has quit [Ping timeout: 252 seconds]
sb0 has joined #nixos
<jayQ>
answer to question was to not use fetchurl, and instead use fetchTarball, e.g. `override = import "${fetchTarball { url = "..." }}/override.nix" { ... };`
jayQ has left #nixos ["ERC (IRC client for Emacs 26.1)"]
<CMCDragonkai>
It seems that `nix-shell` doesn't bring in `checkInputs`.
<CMCDragonkai>
Why is that?
<{^_^}>
[nixpkgs] @ilikeavocadoes opened pull request #49637 → pyditz: init at 0.10.2 → https://git.io/fxxzl
<CMCDragonkai>
I need to put the `checkinputs` into the `buildInputs` for `nix-shell` to bring in that dependency.
<imincik>
Hi guys, can anybody explain to me difference between github repo 'NixOS/nixpkgs' and 'NixOS/nixpkgs-channels' ? Thanks
<andi->
imincik: nixos/nixpkgs is the "development" repository where all changes will be merged into. nixpkgs-channels is populated with commits that hydra has finished building (on the respective branch)
thc202 has joined #nixos
<imincik>
great thanks andi. Does it means that no real people are pushing in to 'nixpkgs-channels' ?
<gchristensen>
my computer can play audio through the speakers, and when I plug in my headphones no sound plays. pulseaudio has shifted to "Headphones (plugged in)" automatically and nothing seems muted. any ideas?
<gchristensen>
I've tried restarting pulseaudio
<{^_^}>
[nixpkgs] @aszlig pushed to master « nixos/tests/chromium: Fix sandbox info matching »: https://git.io/fxxgR
<{^_^}>
[nixpkgs] @domenkozar pushed commit from @peti to release-18.09 « ghc: add new 8.4.4 version »: https://git.io/fxxwG
<srk>
hm, looks good
<{^_^}>
[nixpkgs] @domenkozar pushed to release-18.09 « chromium: use jumbo builds to speedup compilation »: https://git.io/fxxwl
orivej has joined #nixos
<srk>
I have auto-mute disabled, no idea what it does. I my case I have to use alsamixer due to jack running in front of pulse as pulse can't control levels/muting
<srk>
you can also run pulseaudio -vvv at command line to get tons of logs
<srk>
pulseaudio -k && pulseadio -vvv
<Taneb>
,help
<{^_^}>
Use `,` to list all commands, `,foo = Foo!` to define foo as "Foo!", `,foo =` to undefine it, `,foo` to output "Foo!", `,foo somebody` to send "Foo!" to the nick somebody
<Taneb>
,
<{^_^}>
Special commands: find locate tell - Commands sorted by use count, page 0 (use ,<n> to view page <n>): library tofu dnw pr -A imperative pills profiling escape'' paste ping pinning unfree unstable ask declarative error escape" help overlay stateVersion upgrade IFD NUR allah callPackage cloudfront dentalplan escape-special exec info loot nixlang++ runtimeDeps tias timer whomademe wololo wrapper arm bootfull channels context fancy-uninstall github
<Taneb>
,help find
<{^_^}>
find: Use `,` to list all commands, `,foo = Foo!` to define foo as "Foo!", `,foo =` to undefine it, `,foo` to output "Foo!", `,foo somebody` to send "Foo!" to the nick somebody
<Taneb>
,find cblas.h
<{^_^}>
Couldn't find any such files
<srk>
,locate cblas.h
<{^_^}>
Found in packages: atlas, caffe2, openblas
<srk>
try different headphones to rule out cable/jack issues
sg2002 has quit [Ping timeout: 252 seconds]
<gchristensen>
the headphones work on my phone just fine... anyway, thank you for the help -- I'll take a harder look later :)
<srk>
I have to use USB soundcard on my x230 as the jack socket is broken :|
<Taneb>
Is it OK to pass foo.src into a derivation if I want a specific file of something that isn't made visible in the output in nixpkgs?
<srhb>
Taneb: I'd say so, yeah.
<srhb>
Taneb: I think there's even precedent, if you grep a bit
<gchristensen>
srk: no better with other headphones either. hrm.
<gchristensen>
I'll try poking an old system version, and then poking the bio
<gchristensen>
s
<Taneb>
srhb: thank you
sg2002 has joined #nixos
johanot has quit [Ping timeout: 272 seconds]
<chris|>
has someone managed to get dfhack working for dwarf-fortress on 18.09? On my fresh install using the full pack, it just complains about not being able to detect the df version (xml version missmatch) and just dies
nschoe has quit [Quit: Program. Terminated.]
johanot has joined #nixos
<lassulus>
hmm, could be related to my PR to dwarf fortress, how do you test it chris| ?
<{^_^}>
[nixos-org-configurations] @srhb opened pull request #61 → Automatic GC if we drop low. Limits debatable... → https://git.io/fxxKE
mayhewluke has quit [Ping timeout: 268 seconds]
mayhewluke has joined #nixos
<lassulus>
alright, I will need some time to test this
<chris|>
while I have your attention, the switches on the package override do not seem to be passed to the callPackage ./wrapper in game.nix so they don't have any effect on the init config ;)
cinimod has joined #nixos
<cinimod>
Can I install nixpkgs without being root?
<Taneb>
What do you mean by "install nixpkgs"?
iyzsong has joined #nixos
<qyliss^work>
You can do a single-user install of Nix, if that helps
<symphorien>
then either you recompile nix to use another prefix (and can't use the binary cache) or use proot. I advise you to use the fork of proot by termux in this case, it is more stable.
<cinimod>
qyliss^work: the reason I am thinking of using single user mode is because I am trying to install a recent version of SUNDIALS for which I need a recent version of cmake
<cinimod>
symphorien: "fork of proot by termux" means nothing to me but I can google
<cinimod>
So it seems my system supports user namespaces
<symphorien>
if you are on debian or centos, you need to add a kernel parameter/sysctl to enable them
zduch4c has joined #nixos
<zduch4c>
hallo
<zduch4c>
How can I make `nix-build' use "{ allowUnfree = true; }" without adding it into "~/.config/nixpkgs/config.nix"? Is there some command-line switch that can do this?
<clever>
locallycompact: line 43 sets the size of the disk image, 41 makes it have a partition table, 42 makes it bootable, and 35 loads a configuration.nix value (held in dom0_config this time)
<clever>
locallycompact: youll need to cut lines 2-45 into its own file, and then refer to the rootDisk attr
<clever>
and then you can freely edit the config on lines 5-34
<cinimod>
ok so I have installed nix as single user and installed sundials
<cinimod>
Can I reference this (the library and the headers) from outside of the proot shell?
<clever>
locallycompact: it needs a `let` near the start, and a `in` between lines 44 and 45
<clever>
and the ; on 45 doesnt belong
<locallycompact>
then I get
<locallycompact>
error: anonymous function at /home/lc/.nix-defexpr/channels/nixpkgs/nixos/lib/make-disk-image.nix:1:1 called with unexpected argument 'partitioned', at /home/lc/disk.nix:36:12
<clever>
locallycompact: that seems to have been removed in july of 2018, as it didnt actually do anything by then
<locallycompact>
cool something is building
* locallycompact
sits :)
Guest16957 has joined #nixos
<cinimod>
So all I have to do now is tell haskell / cabal / stack not to look in the standard places but to look in .nix/store/ar9s3y7pzmg2dk5ww7d4c4rq58bjwrmx-sundials-3.2.1
Dedalo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<cinimod>
Off to #haskell or whatever
<cinimod>
Thanks all
<clever>
cinimod: use haskell.ghcWithPackages i believe
<eeva>
clever: yay that works really well :D
<{^_^}>
[docker] @domenkozar pushed to bash « Add bash to the image »: https://git.io/fxxMY
<{^_^}>
[docker] @domenkozar opened pull request #5 → Add bash to the image → https://git.io/fxxMZ
<clever>
locallycompact: mount /tmp -o remount,size=8g
<clever>
that will allow it to use up to 8g of ram, and the build should pass
o1lo01ol1o has joined #nixos
<locallycompact>
great, that completed
<locallycompact>
do I get rid of this xen stuff
<{^_^}>
[docker] @domenkozar pushed to bash « Add bash to the image »: https://git.io/fxxM5
<clever>
locallycompact: you can get rid of everything under virtualisation, kernelParams, systemd
<clever>
locallycompact: and depending on what you want, also get rid of the autologinUser
<clever>
and the /stuff isnt needed anymore
sb0 has quit [Quit: Leaving]
<nh2>
arianvp: I don't think so -- the promise of NixOS and nixops is that after you switch-configuration, the specified configuration is running, not some old software that isn't even in the closure of that configuration. That makes it a real issue for NixOS no matter if it's intended systemd behaviour or not. Also, from how I understood boucman, this behaviour goes _against_ his expectations that timer fires shouldn't be remembered
<cinimod>
clever: I don't think my colleagues are ready for nix just yet so I think I just need to provide them with an installation of sundials
<clever>
cinimod: installing libraries with nix wont make them findable by any compiler, and only the compilers within nix have been modified to be able to find them correctly
<etu>
note: currently hard linking saves -7.59 MiB
<etu>
Hmmmmm
<{^_^}>
[nixpkgs] @jyp closed pull request #49381 → Add cudnn7.3 over cuda9.0 and make tensorflow use it → https://git.io/fxQDY
jperras has joined #nixos
<etu>
A "nix-store --optimize" made me save space instead of loose space, that's nice :D
booglewoogle has quit [Remote host closed the connection]
booglewoogle has joined #nixos
silver has joined #nixos
lonokhov has quit [Quit: WeeChat 2.2]
<jonge>
etu: lol just did that on 2 machines, in both cases it was ~4GB saving
<jonge>
should do regularly
__Sander__ has joined #nixos
masterdonx has joined #nixos
<clever>
jonge: there is also a nix.conf flag to do it after every build/download
<clever>
auto-optimise-store = false
<jonge>
clever: will it be fast if it is done so often?
<jonge>
clever: any downsides to it?
<clever>
jonge: its limited to just the storepath thats being built/downloaded, so it can optimize faster
<jonge>
because if i run this immediately again, it takes some time..
<jonge>
oh, i see. this makes a lot of sense.
<clever>
jonge: but ive noticed that traversing the massive /nix/store/.links/ dir does get worse over time, and then even adding a new link becomes a cost
<jonge>
right
<locallycompact>
what's the way do fork nixpkgs and have a private cache of things and maintain it sensibly
<locallycompact>
if I just fork nixpkgs and add a thing and point at it I can still use the main nix cache and it will just build my little thing on whatever machine called it?
<clever>
locallycompact: you also dont have to fork nixpkgs if you just want to build one thing
<{^_^}>
[nixpkgs] @rickynils pushed to release-18.09 « nixos/google-compute-image: Fix network service »: https://git.io/fxxdB
<hyper_ch2_>
tilpner: no, I have not
<jonge>
locallycompact: no need to fork. i am just using pinned nixpkgs (so everyone has the same base) and then run `ourPackages = import ./pinnedNixPkgs { overlays = [ (import ./our_overlays {}) ....]; };` and that's it.
<locallycompact>
how do I pin?
<jonge>
locallycompact: i needed a few hours to understand how this works, but in the end it's really easy.
<locallycompact>
that sounds like what I want yeah
<tilpner>
hyper_ch2_ - It would allow you to change things whenever you want, without having to go through nixpkgs maintainers, and you still get some of the benefits of adding it to nixpkgs
<jonge>
locallycompact: i have an example project here: https://github.com/tfc/nix_cmake_example it uses pinned nixpkgs so it works for everyone. if you have questions about this, don't hesitate to ask
<locallycompact>
clever, in your example you have a loose patch file overriding a thingy, what's the way of overriding a fork of that package so the diff is tracked in git rather than loose patch files?
<clever>
locallycompact: just override src= to use fetchgit or fetchFromGitHub
<locallycompact>
ok cool
<locallycompact>
nic
<locallycompact>
e
periklis has quit [Remote host closed the connection]
<booglewoogle>
hey, I managed to change the version of my to-be-installed godot package, however the building of this older release has issues with GCC 7 ( https://pastebin.com/5ZZUr5Nm ). what override would I have to add to make it use GCC 5?
<booglewoogle>
tilpner: okay, I'll try this. what would be the best syntactical way to add this override? since the entry in systemPackages is already an overrideAttr, namely '(godot.overrideAttrs ( oldAttrs: { version = "2.1.5-stable"; src = fetchFromGitHub { owner = "godotengine"; repo = "godot"; rev = "2.1.5-stable"; sha256 = "1d77a7jgghykanclpssipshwak7c32flhylvqyy5i2bz9bq4gyry"; }; } ))'
<{^_^}>
[nixpkgs] @thoughtpolice pushed to master « nixos: fix #48917 by setting SYSTEMD_TIMEDATED_NTP_SERVICES »: https://git.io/fxxA8
<tilpner>
booglewoogle - That's generally a bad idea. You're using the godot 3 expression to build godot 2
<tilpner>
Which might work, but also might break in subtle ways
<{^_^}>
[nixpkgs] @peterhoeg opened pull request #49645 → pythonPackages.broadlink: init at 0.9 → https://git.io/fxxA1
<tilpner>
I did the equivalent of putting (callPackage ./godot {}) into my systemPackages
<booglewoogle>
tilpner: I see, so I'd better just place the fitting godot 2 expression somewhere local and call that?
<booglewoogle>
tilpner: out of curiosity, is there a way to call an expression from nixpkgs history, like the one above?
<{^_^}>
[nixops] @tomberek opened pull request #1037 → Only populate /etc/hosts using privateIPv4 if on the same zone or region → https://git.io/fxxxc
<tilpner>
Sure. You'd fetch an older copy of nixpkgs, import it and use the package from there
<tilpner>
But: 1. that will greatly increase the closure size 2. use old dependencies 3. be not maintained
<booglewoogle>
tilpner: ah, okay. makes sense. thanks a lot for all your help!
<tilpner>
You can try to call an old package expression with new dependencies, which would solve 1. and 2., but then you end up running a never-tested-before combination
mayhewluke has quit [Ping timeout: 245 seconds]
<{^_^}>
[docker] @peti pushed to master « Run "nix-store --optimise", too. »: https://git.io/fxxx5
<{^_^}>
locallycompact: If a Nix file foo.nix starts with something like `{ stdenv, cmake }:`, you can build it with `nix-build -E '(import <nixpkgs> {}).callPackage ./foo.nix {}'`
<infinisil>
But really, in nixpkgs, you can build e.g. the hello package via `nix-build -A hello` from the repo root
<locallycompact>
I don't follow sorry
<locallycompact>
why is it different
<locallycompact>
before I just ran nix-build
<locallycompact>
how do I change it so I just have to run nix-build
vidbina has quit [Ping timeout: 240 seconds]
drakonis has joined #nixos
turion has joined #nixos
drakonis_ has quit [Ping timeout: 245 seconds]
<cransom>
restructure the file so it doesn't take arguments. but the good way is as {^,^} said.
<infinisil>
Well for my own projects I prefer to have it not take any arguments
<infinisil>
But if it's a package in nixpkgs, the arguments are preferred, but then there's usually an attribute you can use for `nix-build -A`
<locallycompact>
I don't know all this vocabulary yet sorry
<locallycompact>
like I just want to define something in the let block that can be a package that I can put in environment.systemPackages
<MasseR>
When you're using S3 as a binary cache, how do you clean it?
<MasseR>
On a regular I would just call 'nix-collect-garbage'
dbmikus has quit [Quit: WeeChat 2.2]
<infinisil>
locallycompact: gist your file? It's hard to help if we can't see
<sphalerite>
MasseR: you can't really since you don't have gc roots for an S3 cache
<sphalerite>
MasseR: you could manually delete all files beyond a certain age or something, but there's no guaranteeing you won't delete something that someone might still want
<infinisil>
Oh didn't see that
<MasseR>
sphalerite: thanks
<sphalerite>
MasseR: you could also maybe do nix-store --gc --store s3://foo but I'm not sure that's implemented, and if it is it would probably delete everything from the cache
erasmas has joined #nixos
<MasseR>
Yeah that's what I was thinking as well
<MasseR>
Set up an EC2 instance, mount the S3 store, use that as nix store, install a derivation (gc root), and do a clean? :D
<locallycompact>
infinisil, hmm what do I put in environment.systemPackages?
<wirew0rm>
locallycompact: you could put a default.nix in your directory that does 'with import <nixpkgs> {}; pkgs.callPackage ./foo.nix {}' and call your package from
<infinisil>
locallycompact: I don't know, what do you want to put in it?
<locallycompact>
error: attribute 'a1' missing, at /home/lc/disk.nix:27:62
<infinisil>
Ah right
<infinisil>
Use nixpkgs.overlays = [ overlay1 ]; in the nixos config
<sphalerite>
wirew0rm: `pkgs.` is redundant there :)
<locallycompact>
in where?
<infinisil>
locallycompact: In the x86_config part
<infinisil>
That's the NixOS config
<locallycompact>
ok cool that works thanks
<locallycompact>
infinisil++
<{^_^}>
infinisil's karma got increased to 31
<wirew0rm>
sphalerite: thanks :)
<wirew0rm>
Is it possible to get nix-shell for an existing .drv in the store?
<wirew0rm>
ah that was a stupid question, it just works
infty has quit [Quit: Lost terminal]
dbmikus_ has joined #nixos
infty has joined #nixos
dbmikus has quit [Ping timeout: 252 seconds]
<turion>
I have a nextcloud running with services.nextcloud = { enable = true; nginx.enable = true; hostName = "owncloud.mydomain.com"; }. Now I'd like to forward www.mydomain.com/owncloud to the same instance as well
agander has quit [Ping timeout: 268 seconds]
<turion>
By adding some nginx config somehow, but how? The owncloud client doesn't like redirects, it seems
<turion>
I don't want it redirected, just rewritten internally
drakonis has quit [Ping timeout: 252 seconds]
<infinisil>
turion: Check out the nginx option globalRedirect
<sphalerite>
turion: it might be that you just need to add www.mydomain.com to the trusted domains for owncloud or something
tilpner has quit [Remote host closed the connection]
<turion>
infinisil, sphalerite, the problem seems to be that the client doesn't play well with a redirect. Instead we want some kind of reverse proxy
<infinisil>
turion: There's the proxyPass option
<turion>
proxy_redirect off; proxy_pass owncloud.mydomain.com; doesn't work, it still redirects
<infinisil>
(you are certain that these options were applied?)
<infinisil>
I'm using proxyPass myself and I'm pretty sure it works without redirecting (not for nextcloud though)
vidbina has joined #nixos
drakonis has joined #nixos
<turion>
infinisil, I'm running nextcloud :D
<turion>
The URL is only /owncloud for legacy purposes
<infinisil>
(I know?, proxyPass works for my websites and stuff is what I was trying to say)
endformationage has joined #nixos
<{^_^}>
[nixpkgs] @srghma opened pull request #49646 → WIP: pgFormatter: init at 3.2 → https://git.io/fxpkm
<b1000101>
anyone using Engimail (gnupg2) + Thunderbird on KDE? It keeps asking for passphrase all the time and I can't seem to disable it. I've tried to add use-standard-socket to $HOME/.gnupg/gpg-agent.conf but it didn't help.
sanscoeur has quit [Remote host closed the connection]
tilpner has joined #nixos
<srk>
top!
<eisbaer-north>
\o/
drakonis has quit [Ping timeout: 260 seconds]
drakonis_ has joined #nixos
<dtz>
\o/
<booglewoogle>
nice! well deserved
<avn>
domenkozar++ ;)
<avn>
domenkozar: btw, I'll sumbit few more patches to chromium today (some more repro and speedups)
<domenkozar>
that would be lovely :)
Ariakenom has quit [Ping timeout: 252 seconds]
<jonge>
hey there, i just installed nix on an ubuntu machine, made sure that qemu/kvm works and then ran "nix-build $nixpkgs/nixos/tests/login.nix" and ran into the following error: https://gist.github.com/tfc/5984853386e848ffcf8144717c9e9315
<jonge>
domenkozar: the whole laptop has 8 gigs of ram and nis is basically the only thing running here
<jonge>
domenkozar: there are a few gigabytes left at least. that's strange.
<jonge>
i also did not change qemu's memory limit for the VM. the same nixos test runs everywhere else obviously...
<domenkozar>
yeah that's strange :)
worldofpeace has quit [Read error: Connection reset by peer]
worldofpeace has joined #nixos
<avn>
domenkozar: I have two more ideas about it -- first one split out angle and skia, and second one -- drop gtk* dependencies from chromium (idk if they reasonable, and folks use gtk theming from chromium).
<avn>
btw, folks, are we have anyone, who use ffado in nixos?
vidbina has joined #nixos
<sphalerite>
b1000101: use-standard-socket is obsolete AFAIK and doesn't even do anything anymore
<sphalerite>
b1000101: that's not helpful though I guess >.<
<booglewoogle>
I lack the ability to make something of this. if anyone can, I'd appreciate it a lot!
* tilpner
:/
<booglewoogle>
was there maybe once a package 'osmesa' in nixpkgs, which was taken out for some reason? I mean, this patch has to have worked someday as it's in the godot history just like that
<tilpner>
No, this is something else
<booglewoogle>
hmmm
<tilpner>
And as I said before, I'm not sure this has ever worked
<avn>
booglewoogle: osmesa is part of mesa (should be)
<tilpner>
It was just meant as a starting point for you
<avn>
and probably should be part of GL_combined
<avn>
(need check, it already expired from my in-brain cache)
<booglewoogle>
tilpner: yeah, I know. if it doesn't, welp, can't help it. :) but for now I'm stuck with this error I guess
<booglewoogle>
avn: oh, okay. hum..
<avn>
Anyone familiar with java stuff in nixos? Would be nice to have pdftk replaced with pure java (non-gcj) version.
<tilpner>
booglewoogle - Why do you need this version of Godot built from source?
<booglewoogle>
tilpner: oof. wow. steam godot has a version-2.1 launch option which works flawlessly. I didn't realize it's on there, thanks a bunch again. all the hassle for nothing, haha
<tilpner>
:)
<jbgi>
currently the nixpkgs-unstable channel has binary caches for both darwin and x86, right? Would it make sense to have a stable channel that does the same?
exarkun2 has joined #nixos
<exarkun2>
is there a nix command to only download binaries from the cache that would be required to perform a certain build, but not actually do the build?
<exarkun2>
Or: Is there a good caching strategy I can use for Nix builds on CircleCI?
<worldofpeace>
joepie91: Yep I saw. It's always when you get the time to do things too :P
Theuni2 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<rhitakorrr>
Hey folks - I'm having some trouble getting WiFi working on my laptop. I have `networking.networkmanager.enable = true` in my configuration, but when I scan for WiFi with `nmcli dev wifi rescan`, I get "Error: No Wi-Fi device found." Any idea how to fix this?
<sphalerite>
I want to strace a systemd service. Anyone know a handy trick to attach to it right from the start?
<srk>
I wish! :D
<srk>
you can probably wrap ExecStart in strace -o
<sphalerite>
well that's the thing, I don't want to modify the nixos config for it :/
<sphalerite>
I suppose I could trace init
<sphalerite>
I'll get some noise, but it should work
agander_ has joined #nixos
<{^_^}>
[nixpkgs] @srghma opened pull request #49650 → amazon-image: fix typo in comment → https://git.io/fxpcB
mayhewluke has quit [Ping timeout: 240 seconds]
<exarkun2>
are python2.7-hyperlink and python2.7-spake2 mispackaged? my CI build just failed: collision between `/nix/store/3risdi5523sl863dqsi1q8b99nd407i4-python2.7-hyperlink-18.0.0/.pytest_cache/v/cache/nodeids' and `/nix/store/p6azz13zil902k83madp85jjgykkadjb-python2.7-spake2-0.8/.pytest_cache/v/cache/nodeids'
mayhewluke has joined #nixos
agander has quit [Killed (asimov.freenode.net (Nickname regained by services))]
agander_ is now known as agander
<exarkun2>
(unclear why my local builds don't fail that way)
<avn>
(semi-offtopic) are python gurus here? I need advice about gevent to asyncio migration. (I know that rewrite-all-to-haskell is an true answer ;))
<exarkun2>
avn (IRC): I am glad that you have anticipated me.
<simpson>
avn: What's your actual question?
nDuff has quit [Ping timeout: 260 seconds]
symphorien has quit [Ping timeout: 250 seconds]
<avn>
actual question is a how to not throw myself into rewrite-all-at-once. I bound to websocket and dbus implementations (latter is breaks even from recent gevent update).
<exarkun2>
Have all of your application logic factored such that it is independent of the choice between gevent or asyncio. Then you only have to rewrite your integration layer which is presumably very small.
<simpson>
Porting to or from gevent, and to or from asyncio, are kind of all-or-nothing propositions.
<simpson>
What exarkun2 said. Make as many units as possible network-agnostic and isolated.
ubert has quit [Quit: Leaving]
<avn>
so plan is bring in trollius and aiogevent, backport dbussy, upgrade (and isolate) logic, then move to 3.x dropping gevent and replacing websocket event.
<avn>
app is mainly proxy from websocket to dbus and tcp to services behind it
<avn>
mainly curious, if anyone have success story on migrating.
<simpson>
avn: Hm. We should move this to #nixos-chat. I think that you need a Python doctor.
<avn>
exarkun2: btw, I remember how you answer my silly twisted questions ages ago ;)
<simpson>
Actually, no, wait, I have stuff to do. I cannot be a Python doctor. But yeah, this doesn't sound like a good situation for incremental switchovers; this sounds like a rewrite.
<exarkun2>
:)
symphorien has joined #nixos
dbmikus_ has quit [Ping timeout: 245 seconds]
<avn>
Rebuilding chromium and gcc6 provoking me to offtopics
<hodapp>
rewrite it all in Haske- oh.
<hodapp>
but seriously, Python asyncio smells bad, along with literally everything in Python when concurrency is involved :-/
<avn>
simpson: btw, thank you for #nixos-chat hint
<hodapp>
I have been through some amount of pain trying to make asyncio play nice with certain things, like anything else that has an event loop
<avn>
hodapp: well, gevent is ok as long, as libraries worked as planned
<hodapp>
everything is okay as long as libraries work as planned
<hodapp>
sometimes the plan just sucks
juliendehos has quit [Quit: Leaving]
<avn>
hodapp: I do rewrite to haskell in small portions, where I can sell it to customers (for example where startup time matter)
jD91mZM2 has quit [Quit: WeeChat 2.2]
<exarkun2>
The real test is how things work when things go wrong.
<hodapp>
I moved a bunch of my Python stuff go Go.
<hodapp>
s/go Go/to Go/
<exarkun2>
hodapp (IRC): Speaking of bad smells! ;)
<avn>
exarkun2: yep. I sometimes doubt if dbus was good solution over zookeeper.
<hodapp>
exarkun2: it smells dusty and stale, but at least not like flaming machinery and burning duct tape.
buckley310 has joined #nixos
<nh2>
avn: dbus and zookeeper seem like some vastly different projects, one is distributed, the other isn't
<nh2>
or am I behind the times and dbus does Paxos now?
<simpson>
dbus is probably not a good solution, in general.
<exarkun2>
nh2 (IRC): I am sure someone is working very hard on making it so.
<exarkun2>
nh2 (IRC): Or, possibly, Blockchain.
<exarkun2>
brb need to run an ico
<avn>
nh2: I need rpc and something like process coordinator but on one host. To proxy from websockets to user sessions and back inside one machine.
sigmundv has quit [Ping timeout: 268 seconds]
<avn>
dbus generally works, and it already exists (due systemd and other session-related stuff)
ariutta_ has joined #nixos
<nh2>
ah, I see, one host
<joepie91>
anybody here know how to convince npm to look in a particular include path for native-module (ie. node-gyp) builds?
<joepie91>
I obviously don't have a magical global include folder
<joepie91>
and it seems to eat my env vars
<joepie91>
not passing them on to the build process
<exarkun2>
I switched my CI setup to the same version of nix (2.0.4, CI using it via docker image nixos/nix:2.0.4) as I have installed on my Ubuntu laptop but it still encounters build failures I don't encounter locally. What is going on?
<nh2>
exarkun2: simplest suspicion: The build isn't pure, and it reads some dirs from your Ubuntu that don't exist in your NixOS container
<exarkun2>
nh2 (IRC): Hm. Is there a `--pure` flag I can pass somewhere?
<nh2>
exarkun2: only for nix-shell, is your problem in nix-build?
<exarkun2>
nh2 (IRC): It is
<nh2>
then at least your env vars shouldn't have a chance to pollute your build
<{^_^}>
[nixpkgs] @peti pushed 4 commits to haskell-updates: https://git.io/fxpW7
<nh2>
exarkun2: things I'd try: 1. strace the build on ubuntu and see if it accesses some suspicous paths, like your home or /usr. 2. Try to enable the sandboxing feature of nix (don't remember if that's easy on Ubuntu), and see if that makes it fail on Ubuntu
<simpson>
I have a handful of branches. matador-next, matador-upstream, matador-broken-worked-on-friday, etc.
<rhitakorrr>
WiFi question. I enabled networkmanager in my config, but scanning for WiFi or trying to add a connection with nmcli prints "Error: No Wi-Fi device found." Ethernet works fine. Any idea how to fix this?
<srk>
rhitakorrr: rfkill list
Theuni2 has joined #nixos
Theuni2 has quit [Client Quit]
<rhitakorrr>
Thanks, srk. I'll try that out when I get home (killed my ssh connection messing with the network settings :/)
<srk>
rhitakorrr: there's a feature for that in nm!
<arianvp>
And I thought we were onto something! Darn
<joepie91>
final command in my case to get the `foundationdb` module on npm to install (assuming a `libfdb` folder in the root of the project containing libfdb_c.so and a foundationdb folder with the header files from the client .deb): CXX="$(readlink -f $(which g++)) -I../../../libfdb -L../../../libfdb" npm install
<joepie91>
works on non-NixOS also
dbmikus_ has joined #nixos
agander has quit [Read error: Connection reset by peer]
vidbina has joined #nixos
<rhitakorrr>
Oh, that's nice srk. Wish I knew about it earlier! haha
<arianvp>
idk, the module seems more complicated than needed
<Mic92>
arianvp: I wonder why it cannot read the ip address itself.
<Mic92>
This is something I would expect.
<arianvp>
OH I remember now
<arianvp>
consul refuses to start when there are multiple interfaces
<arianvp>
(this is because, you might want to listen to WAN gossip on interface 1, and LAN gossip on interface 2)
<arianvp>
for multi-dc support
<nh2>
arianvp: to me that looks like to bind to a specific interface and set that `advertise` option; I know consul has stuff like that for the case like AWS where the machine sees only its internal IP, not the external Internet-facing IP
pie_ has quit [Remote host closed the connection]
<nh2>
arianvp: but in my case, it just writes `{}` into the file (I used dedicated servers, and on my AWS staging deployment it also works without setting that stuff)
pie_ has joined #nixos
lopsided98 has quit [Quit: Disconnected]
<arianvp>
yeh I think it only is an issue when you have multiple network interfaces
<nh2>
arianvp: did you mean `network-online.target` instead of network-up.target`?
<arianvp>
yeh
<arianvp>
but I guess it needs "network.target" for that execPre script to work :P
<arianvp>
so makes sense
<arianvp>
(I think )
<arianvp>
So.. I saw you're sure the issue wasn't related to your timers?
<nh2>
arianvp: no, actually I'm not saying it isn't related to _timers_ just that the "canceling" part may be unrelated
<arianvp>
hmm maybe we do need "network-online.target"
<arianvp>
from the systemd docs: Whether any network interfaces are already configured when it is reached is undefined
lopsided98 has quit [Client Quit]
<arianvp>
not important for now, we can fix that later. First the reliable units part
<arianvp>
yeh so I've been trying to reproduce it locally, (by keep changing the ExecStart)
<arianvp>
but for now it has been working flawlessly
<arianvp>
:/
worldofpeace has quit [Ping timeout: 272 seconds]
<nh2>
arianvp: I suspect it is related to timers still; I changed my timer to 3s so that I can try and reproduce quicker
pie_ has quit [Excess Flood]
lopsided98 has joined #nixos
pie_ has joined #nixos
<gchristensen>
per fpletz's talk, network.target is basically only used at shutdown-time I think
<gchristensen>
and usually you want network-online.target
<arianvp>
then network.target makes sense
<arianvp>
I want consul to stop before the network shuts down
agander has quit [Ping timeout: 268 seconds]
Thra11 has quit [Ping timeout: 264 seconds]
<moredhel>
Hi, I'm trying to run coredns on my laptop, but it is failing sure to port 53 bring already taken. Is it dnsmasq that is causing this? I'm not sure how to disable it/use another resolver
<{^_^}>
[nixpkgs] @matthewbauer pushed 5 commits to master: https://git.io/fxpwI
<clever>
moredhel: netstat -anp | grep 53, as root, should tell you what program is to blame
pie_ has quit [Ping timeout: 245 seconds]
sanscoeur has joined #nixos
<v0|d>
gchristensen: Doesn't cover the case where you've got an eth and a wireless which is down.
<v0|d>
franz told me its possible to depend on interfaces though.
<gchristensen>
cool
<{^_^}>
[nixpkgs] @costrouc opened pull request #49657 → Add all python google-cloud-* modules → https://git.io/fxpwd
<arianvp>
nh2: easy way for me to verify a new version being deployed is doing "watch -d "systemctl show consul | grep ExecStart"
<arianvp>
you can easily spot if the ExecStart line actually changed during update
vk3wtf has joined #nixos
<nh2>
arianvp: so the `show` command would have the old line, while `cat` would have the new one?
<arianvp>
no, show should change
<arianvp>
and "Watch -d" will flash the parts of the lines that change
<nh2>
arianvp: I mean in the bad case
<arianvp>
yes
<arianvp>
in the bad case, "watch -d" wouldn't flash the ExecStart line white
vidbina has quit [Ping timeout: 252 seconds]
<arianvp>
nh2: another question. We're sure by now that the problem as nothing to do with the consul version bump right?
<nh2>
arianvp: ah yes that for sure. I would have merged it if I knew what my duties as a merger are
slyfox has joined #nixos
<nh2>
arianvp: I can't confirm your `show` vs `cat` understanding above though. I just tried it: `ps` showed the command line unchanged (bad), but both `show` and `cat` show the updated CLI
<nh2>
so I don't think we can use that as checking for the problem
<arianvp>
oh damnit
<arianvp>
okay then "watch -d "ps | grep consul"" it is :P
<arianvp>
that's probably more robust
cnidario has joined #nixos
<arianvp>
so when it happens, do you get like a warning in the systemd logs?
<neonfuz>
Any recommended way to get an environment variable into the user's env?
<neonfuz>
specifically trying to set $TERMINAL
<neonfuz>
not sure for example if anything will source .profile
ckauhaus has quit [Quit: WeeChat 2.2]
<neonfuz>
I'm using windowManager.i3
<neonfuz>
I guess I could add it to windowManager.i3.extraCommands?
<nh2>
arianvp: I seem to get "Current command vanished from the unit file" for both the situation where it updates correctly, and where it doesn't. But I suspect it's just a race. Also, I can't seem to reproduce it unless I put like a sleep 120 or something of that order in between the deploys, even though I've set my timer to 3 seconds
<neonfuz>
extraSessionCommands*
Guest16957 has quit [Remote host closed the connection]
<nh2>
neonfuz: maybe you could write it to a global config file (like a global .profile file), and condition it on the user name?
arianvp has quit [Quit: WeeChat 2.2]
<neonfuz>
hmm, I'm thinking now maybe just add "source ~/.profile" to extraSessionCommands
<nh2>
that's how I do things to colour the prompt of different users differently
arianvp has joined #nixos
<arianvp>
sorry nixos-upgrade :P
<arianvp>
what did you say? I lost logs
myguidingstar has quit [Ping timeout: 245 seconds]
<neonfuz>
I want to set this variable at the user level because that's the level I have my terminal emulator installed at
<neonfuz>
though maybe it'd be good to have a terminal emu installed on nixos lol
endformationage has quit [Ping timeout: 252 seconds]
<jasom>
shouldn't the terminal-emulator itself set that variable?
<neonfuz>
I want i3-sensible-terminal to find it
<neonfuz>
which is not ran from a terminal emulator
<neonfuz>
usually
<jasom>
oh $TERMINAL, not $TERM, sorry.
<neonfuz>
yep
halfbit has joined #nixos
<ddellacosta>
okay just want to express my joy to the world that after about three days I finally got NixOS installed and running successfully on my new Thinkpad X1 Extreme. Thanks to help from folks here and in #zfsonlinux I'll add
<neonfuz>
thanks though
<capisce>
ddellacosta: congratulations
<ddellacosta>
capisce: thanks!
acarrico has quit [Ping timeout: 272 seconds]
dbmikus_ has quit [Ping timeout: 250 seconds]
dbmikus_ has joined #nixos
<nh2>
ddellacosta: what were the things you struggled with -- from your message I conclude at least ZFS; other things?
<arianvp>
how do you do auto-upgrades with nixops?
<nh2>
arianvp: they don't
<arianvp>
:'(
<ddellacosta>
nh2 I confused myself trying to figure out how to get my already installed Windows 10 working in a dual-boot configuration
<arianvp>
yeh makes sense... because the nix files arent on the machine
<nh2>
arianvp: I think it's OK, if you want to automate it, you probably want to run a loop that fetches stable nixpkgs and deploys it with nixops
<arianvp>
problem is, I'm just managing one machine (my home server) and I cant always connect to it
<arianvp>
but meh guess it's not that much of a big deal
<arianvp>
other question, I know nix-build creates an implicit gcroot but why doesn't nix-shell?
<nh2>
arianvp: wouldn't be too surprised if you can deploy with nixops from that machine itself, so you can put the loop there
mayhewluke has quit [Ping timeout: 272 seconds]
<arianvp>
I hate it that nix-collect-garbage throws away my dev environments
<nh2>
arianvp: I suspect the `nix-build` gc root is the `result` symlink dropped in the current directory, nix-shell doesn't drop files
<__monty__>
Wasn't nix-shell originally only meant as a development tool for nix expressions?
<nh2>
arianvp: what dev environment, Haskell?
<arianvp>
__monty__: yes but i've come to abuse it for actual development
<__monty__>
Also, arianvp, sounds like you might like johnw's approach.
<nh2>
IIRC for Haskell the nix-shell dev entironment is .env on the package or so, so maybe if you just `nix-build -A yourpackage.env`, it'll drop a `result` symlink for that?
mayhewluke has joined #nixos
<__monty__>
nix-shell as dev environment has some annoying shortcomings imo.
feep has joined #nixos
<nh2>
arianvp: I think I'll have to write a script now that runs nixops in a loop, keeps appending `-ui` to the consul config, and checks the `ps` output, as I can reproduce this stuff one out of 10 times, and apparently only if I sleep 2 minutes in between, I'm sick of doing this all day long :/
<arianvp>
__monty__: yes maybe, but nix-build is not granular enough to be useful for development
<arianvp>
nh2: woohoo automation
<arianvp>
watching the talk now
<__monty__>
arianvp: Sure, that's not what I'm suggesting though.
<__monty__>
It's basically to use myEnvFuns (and buildEnvs) but for dev the former.
<arianvp>
is it mentioned in that talk you linked?
<arianvp>
i'm watching it now,
<__monty__>
Disadvantage is you have to manually specify all the packages you want available in your dev env and adding one requires rebuilding but I've found that I usually add packages in bursts and then stick with what I have for a while so for me the advantages far outweight the disadvantages.
<arianvp>
because i dont understand (yet) I think.
<__monty__>
arianvp: Yes, I got it from that talk. Every other guide I've seen talks about using nix-shell and I just don't like it.
<arianvp>
ooh myEnvFun is actaully a function
<{^_^}>
[nixpkgs] @globin opened pull request #49658 → gitlab: refactor and fix test → https://git.io/fxp6z
<arianvp>
oh that's really cool
<__monty__>
It also makes ghcWithHoogle more useful afaics.
<{^_^}>
#18315 (by domenkozar, 2 years ago, open): Remove myEnvFun
<__monty__>
: )
ottidmes has joined #nixos
<avn>
__monty__: I upvote it once again ;) we need revamp it ;)
<johnw>
__monty__: nowadays I use direnv
jtojnar has quit [Ping timeout: 268 seconds]
<ottidmes>
I am trying to package bitwarden_rs, which is, as the name implies, a rust package, I tried following the manual, but am now confronted with the following error: Permission denied (os error 13), any idea how to fix this? Maybe I need to use rust nightly which is what the AUR package does, but I doubt that affects this particular error
<__monty__>
johnw: How do you generate the .envrc's?
jbgi has quit [Ping timeout: 252 seconds]
<__monty__>
johnw: And how does it address the shortcomings of nix-shell?
<johnw>
I copy the .envrc from other projects
<johnw>
it uses nix-shell, but caches the result, then thereafter uses its own cache
<__monty__>
johnw: Did it lead to a QoL improvement though? Or was it just an experiment to see *if* it's possible to work around nix-shell's shortcomings?
<johnw>
very much QoL
<johnw>
because of emacs-direnv
<johnw>
now every buffer is naturally project-relative
<johnw>
including things like M-!, which nix-buffer didn't solve
<symphorien>
hakujin: macos is (usually) case insensitive
<symphorien>
ottidmes: is the cargoSha256 right ?
<hakujin>
symphorien: yep. are filenames included in the calculation? I guess what I'm really asking is: how is the sha256 field calculated?
<clever>
hakujin: depends on if its recursive or flat hashing
<ottidmes>
symphorien: I guess, it first failed on that of the fetchFromGithub sha256, then on that of the cargo sha256 and I modified it to the one reported at failure
<clever>
oh, but i think the filename is included either way
<symphorien>
ottidmes: then it can be that the cargo.lock and the cargo.toml are not in sync
erasmas has quit [Quit: leaving]
<symphorien>
oh, also try on nixos-unstable, not on stable
<ottidmes>
symphorien: I will check unstable, or should I just use the mozilla overlay?
<symphorien>
does aur build in a sandbox ?
<clever>
hakujin: if recursive == false, then the path is a file, and does not support directories (fetchurl for ex)
<symphorien>
you are using buildRuntPackage right ?
<symphorien>
buildRustPackage got improvements in master
<__monty__>
johnw: Still trying to grok nixos, don't make me learn a whole new OS : > Thanks anyway though. Stuff for thought.
<dmc>
symphorien, no
<johnw>
haha
<johnw>
if only it had a decent editor
<ottidmes>
symphorien: yes, I am using rustPlatform.buildRustPackage, but ok, then I will use unstable and see if that helps
<symphorien>
ottidmes: @AUR: well if these files are not in sync, cargo will connect to the internet to update them so it will work on aur but not in nix
<__monty__>
666 \,,/
Mr_Keyser_Soze has joined #nixos
__monty__ has quit [Quit: leaving]
jbgi has joined #nixos
<ottidmes>
symphorien: this is the tree at 1.3.0, the latest release: https://github.com/dani-garcia/bitwarden_rs/tree/e0614620efdd170a0e9b406695314fb005298597 not sure if there is an easy way to check if they match, checking by eye is too much, but I will first try unstable, just have to update it, because my stable is more recent than my unstable checkout :P
<symphorien>
no it is not visible
* colemickens
is going in on NixOS + HyperV, wish me luck
<ottidmes>
symphorien: I tried with unstable, still failed with the same error :(
<drakonis>
there's a hundred new pull requests wow
<hyperfekt>
i was hoping someone could help me fix an expression i've been banging my head against several hours yesterday. when i add something with boot.initrd.postDeviceCommands from configuration.nix, it shows up fine in the init script, but if i try from bcachefs.nix, nothing happens. here it is: https://github.com/hyperfekt/nixpkgs/blob/bcachefs_cryptroot/nixos/modules/tasks/filesystems/bcachefs.nix compare btrfs.nix and nixos/modules/sy
<hyperfekt>
the kernel modules are added just fine, and it doesnt matter what i put behind boot.initrd.postDeviceCommands, even simple strings don't work
booglewoogle has quit [Remote host closed the connection]
<colemickens>
ottidmes: thanks! I have some of that already from the Azure image stuff in nixos/modules. I think this will not to be so bad ;)
<ottidmes>
colemickens: I have had it working fine with this, but never really used it, since I like my multi monitor pure NixOS setup too much, and this way I am not tempted to start playing some game ;)
<colemickens>
Ah yeah, see, I have my gaming PC that sits idle except for 4 hours a month maybe, and jobs to be built! Plus it's another cool example to talk about when I start writing up all the cool things I can build from my nixcfg+nixpkgs.
<colemickens>
Of course, if I do this, it will be harder to justify spinning up 48 core EPYC packet.net machines though, which is still really fun to me :)
<ottidmes>
colemickens: Also nice if you combine it with one of those X server implementations on Windows, then you can intergrate NixOS nicely in Windows
<ottidmes>
symphorien: I gotten further by changing the git version to the latest rather than the release commit, now I get this: https://pastebin.com/wBRUguC8 I am going to try adding openssl as a dependency somehow, since it seems that is the cause of the error
<ottidmes>
I love this CPU fan, i7 920 overclocked from 2.66Ghz to 3.8Ghz at 100% CPU load and max temp of 55 degrees Celcius, while the previous fan could get close to 80 without overclocking
<MichaelRaskin>
ottidmes: SGT Puzzles! Also TORCS.
<ottidmes>
MichaelRaskin: sorry? I only understand 2 words from that sentence :P so I am puzzled :P
<MichaelRaskin>
You said that running NixOS prevents temptation to start some game.
<ottidmes>
symphorien: hopefully the last error, now I get: error[E0554]: #![feature] may not be used on the stable release channel from pear_codegen
<nh2>
arianvp: my deploy-loop script is runnning now, of the 3 times it has reproduced the errors, it took 32, 38 and 80 deploys respectively -- annoyingly long
<ottidmes>
symphorien: I will use nightly then, thanks for the help :)
<aanderse>
ottidmes: man i'm loving steam proton. got a whole bunch of windows games running perfectly
<symphorien>
there is no nightly in nixpkgs
<ottidmes>
symphorien: mozilla overlay right?
<symphorien>
right
<colemickens>
I'd be very curious to peek at someone's nixos configuration if they're using NetworkD with Libvirt and/or Docker...
<drakonis>
ottidmes, wine is grand.
<colemickens>
I may have been a bit hasty to switch away from NM...
<ottidmes>
colemickens: Every time I tried moving away from Network Manager, I ran into network problems with my desktop, so I have just given up and stick with Network Manager
<Zajcev_>
guys, anybody with working pam_pgsql?
<ottidmes>
Going to Windows is not just for the games, it's remote desktop is also still one of the best, just to name one, but honestly I like not having games on my Linux machine, even though I know it is possible given some effort
<steveeJ>
should setting { nixpkgs.config.allowUnfree = true; } take effect for the current or the next build?
<nDuff>
"Current" meaning a build that's already running? Since evaluation happens (ie. the derivations are made concrete) before actual building starts at all, it wouldn't take effect until a new evaluation takes place with that change in place.
<ottidmes>
steveeJ: if you set it, and then rebuild, it should take effect, like any other config does, so directly
<steveeJ>
seems like even after switching to a config that has it set it's not effective though
<ottidmes>
steveeJ: how are you building? nix-env/nix-shell/nix-build or nixos-rebuild?
<steveeJ>
ottidmes: nixos-rebuild with a config inserted via -I
bfrog has quit [Ping timeout: 264 seconds]
<steveeJ>
I think it has to do with an import. I'm importing <nixos-unstable> in my config to pull in some packages from there
xourt has joined #nixos
<ottidmes>
steveeJ: correct, that would cause it
<steveeJ>
somehow this worked before I refactored my config repo
<ottidmes>
steveeJ: unstable = import /wheel/fork/nixpkgs-unstable { config = { allowUnfree = true; }; overlays = []; }; is what I use in my overlay
<xourt>
When running nix-store --gc --print-roots , what are the {memory:number} garbage roots such as {memory:0} -> /nix/store/12zhmzzhrwszdc8q3fwgifpwjkwi3mzc-gcc-7.3.0-lib ?
<xourt>
I am trying to completely empty my /nix/store/ not sure how to do this ...
<steveeJ>
ottidmes: that works, thanks! why wouldn't `{ inherit config; }` work if it's set there?
<ottidmes>
symphorien: why is it recompiling all its dependencies every build? so it is not using nix for its dependencies?
<symphorien>
ottidmes: this is a limitation of buildRustPackage.
<symphorien>
use buildRustCrate (if I remember the name correctly) instead if you want a finer grained thing
<steveeJ>
ottidmes: it uses cargo internally. there's carnix which can bypass cargo
<ottidmes>
steveeJ: the config is the nixpkgs config, not that of NixOS, so you would need { config = config.nixpkgs; } probably
dbmikus_ has quit [Ping timeout: 252 seconds]
<steveeJ>
ottidmes: that works
oida has quit [Remote host closed the connection]
<steveeJ>
cool. I'm still wondering why it worked before :D
oida has joined #nixos
pie_ has joined #nixos
<ottidmes>
symphorien: If I need to recompile a few more times, I guess I will try that
<ottidmes>
symphorien: cool, its running tests now :)
stepcut has joined #nixos
yosslid_ has quit [Ping timeout: 246 seconds]
<hyperfekt>
can anyone give me a pointer on what would be the workflow for debugging a nixos module?
<clever>
hyperfekt: `nix repl '<nixpkgs/nixos>'` and then eval `config` and `options` to inspect the current state of all modules in your current config
<clever>
imports = [ ./foo.nix ]; to load a custom module in your configuration.nix
<clever>
`-I nixos-config=./configuration.nix` on most nix commands, to make them use a different configuration.nix
<xourt>
*bump* does anyone know what {memory:0} -> /nix/store/12zhmzzhrwszdc8q3fwgifpwjkwi3mzc-gcc-7.3.0-lib means in nix-store --gc --print-roots?
pie_ has quit [Ping timeout: 268 seconds]
<clever>
xourt: one of the /proc/*/maps files refers to that path, check them all with a grep
<xourt>
ok thanks
<ottidmes>
symphorien++ thanks for the help! the rust package seems to be working fine now. I am just going to clean things up and see if I can use buildRustCrate instead
acarrico has joined #nixos
<hyperfekt>
clever: oh sweet. i've been working from a fork, disabledModules makes that unnecessary. let's see if i can find the bug with the repl
<hyperfekt>
clever: btw, i made that PR for the option you inspired (#49506)
<ottidmes>
clever: nice tip! I did not know about disabledModules, I have just been using different names to work around it
<hyperfekt>
currently i'm working on getting bcachefs encrypted root to work, and i figured out how, it's just that my code doesn't do what i think it does :b
<clever>
ottidmes: its a somewhat recent addition, ive also been messing with service names in the copy to make them not conflict, prior to discovering that
<ottidmes>
clever: just this week I had a friend ask me how to replace a module without modifying nixpkgs, so just messaged him ;)
<Mic92>
let's make developing nixops more fun again
<clever>
Mic92: heh, i'm currently adding a new backend to nixops
<Mic92>
clever: consider adding types :)
pie_ has joined #nixos
<Mic92>
there are also editor integrations that show type errors inline
<clever>
half of my problem is just a lack of documentation with this cloud provider
<clever>
i assumed that the ssh keys your authorizing, are a list of unique id's for the ssh keys, from the api to create them
<maurer>
When I first used the GCE backend, the docs were a nightmare
<maurer>
I'll be they're better now
<clever>
it then gave an error, saying .... is not a colaborator
<clever>
i assumed it was my userid, but no, it was the unique id i put in to the ssh keys list
<maurer>
(it used to be that if you followed the docs, you'd get errors in the nix evaluation, because they referred to a mix of features from unstable, and the way things were done in stable)
<maurer>
(and these were very much Not Compatible)
<clever>
Mic92: so what looked like a permission error, was actually a 404 for the ssh keys listing! lol
<Mic92>
clever: I also had to debug azure ones, it was a nightmare. I had instant flashbacks to windows error messages.
cnidario has quit [Remote host closed the connection]
<Mic92>
clever: maybe mitm proxy helps, assuming you are using some python library?
<clever>
Mic92: i already patched the python library to print all request and response bodies
<clever>
that didnt help with the api giving vague errors
<clever>
Error 422: users: "7e24caef-6317-4b2b-8830-d5e9b77f27d2" are not collaborators on this project
<clever>
Mic92: how would you parse this error?
<Mic92>
clever: no idea.
<Mic92>
I also prefer often just to use the REST api directly instead on relying on badly written libraries.
<clever>
my first idea, was that i lacked permission to create devices
<clever>
Mic92: that string is directly from the json that the REST api spit out
Drakonis__ has quit [Quit: Leaving]
<clever>
and the library is just a thin wrapper around the REST api
<clever>
with zero type checking
<clever>
the docs also dont tell you what it expects in any fields
<clever>
"string", lol
<nh2>
arianvp: I now ran with the timer disabled, it deployed 200 times without issue, so the timer definitely seems at fault
<clever>
Mic92: so i can now create and destroy sshkeys, and machines, but the machine doesnt have an ip immediate after creating, causing nixops to break
<clever>
now i need to loop and check the state until it gains an ip
<{^_^}>
[nixpkgs] @matthewbauer pushed to master « release-corss.nix: fix ppc-embedded name »: https://git.io/fxpSA
ilikeheaps has joined #nixos
nDuff has quit [Quit: zzz]
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Mic92>
clever: Today somebody told me that amazon took like 1/4 hour until there metadata services picked up a machine. That would be a long loop.
<hyperfekt>
what the hell. nix repl simply ignores my config i gave it with -I? is there something obvious i'm missing?
<clever>
Mic92: :O
copumpkin has quit [Read error: Connection reset by peer]
<clever>
hyperfekt: what args did you run it with?
<azazel>
hey guys, I'm using requiredBy in a service definition. I was expecting that if that service failed to activate, those specified by requiredBy would be as well, but it's not
azdle has joined #nixos
<hyperfekt>
ooh, that was the obvious thing :j thanks once again. i'll have the nix argument scheme figured out soon enough
schopp0r has left #nixos ["Error from remote client"]
schopp0r has joined #nixos
<azazel>
the "uwsgi" service I specified in requiredBy started (and activated) _after_ my service, but was started nonetheless
<clever>
azazel: you probably also want to set after