<ornxka>
actually how do i just skip tests entirely
vonfry has quit [Ping timeout: 240 seconds]
werner291 has joined #nixos
Rusty1 has quit [Remote host closed the connection]
<clever>
ornxka: i would set HOME=$NIX_BUILD_TOP
<clever>
ornxka: thats a temp dir that nix creates, and will clean up when the build is done
<ornxka>
ahh great thanks
wildtrees has quit [Quit: Leaving]
drakonis has joined #nixos
vonfry has joined #nixos
werner291 has quit [Client Quit]
o1lo01ol1o has joined #nixos
werner291 has joined #nixos
drakonis_ has quit [Ping timeout: 250 seconds]
<ornxka>
hm this package seems to run its tests in buildPhase (which then fail and nix thinks the whole phase failed and stops)
zqxw has joined #nixos
sigmundv has quit [Remote host closed the connection]
zqxw has quit [Client Quit]
drakonis_ has joined #nixos
xkapastel has joined #nixos
Rusty1 has joined #nixos
nekroze has joined #nixos
drakonis1 has joined #nixos
drakonis has quit [Ping timeout: 240 seconds]
werner291 has quit [Quit: werner291]
kleisli has quit [Ping timeout: 252 seconds]
kleisli has joined #nixos
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/Je6vG
Okinan has joined #nixos
<ornxka>
is there a way to let package building continue even if buildPhase fails?
<ornxka>
it builds the binary but then tests fail
ambro718 has quit [Ping timeout: 265 seconds]
<ornxka>
so it could easily just install it afterwards
<clever>
ornxka: read the source to see how you should disable tests
pareidolia has quit [Ping timeout: 240 seconds]
<nekroze>
Has anyone had any success using the `networking.vswitches` options lately? I have been using them for a while to create a bonded bridge for VM's out of two nic's but I did an update today and now it creates the nic with the service as per usual but then the service instantly starts stopping and then removes it again but I can see nothing specific causing a restart. If I run the ExecStart and ExecStop it
<nekroze>
works and I get no errors, for now I am just having to manually run the ExecStart of this service after a boot otherwise I have no networking. Any suggestions?
peanutbutter144 has quit [Quit: No Ping reply in 180 seconds.]
<evils>
kicad is build from source, mattermost-desktop uses a pre-build binary, mattermost-desktop uses rpath and it's the only thing i've seen do that, so i'm associating rpath with binary distributed packages
<{^_^}>
[nixpkgs] @jtojnar pushed to staging-next « guile: fix unbound variable in setup-hook »: https://git.io/Je6JW
<evils>
wrong channel xD
endformationage has quit [Ping timeout: 265 seconds]
xkapastel has quit [Quit: Connection closed for inactivity]
<clever>
jackdk: it will generate a url like this, based on the revision (which must be a string)
<clever>
124 sha256 = editedCabalFile;
<clever>
jackdk: and editedCabalFile must also be a string, containing the sha256 of the cabal file
<clever>
jackdk: so you want foo = self.callHackageDirect "foo" "1.2.3" { revision = "42"; editedCabalFile = "hash"; }
<jackdk>
clever: wow, cool! I couldn't see that machinery (I was poking at make-package-set.nix)
<jackdk>
thanks clever++
<{^_^}>
clever's karma got increased to 249
reallymemorable has quit [Quit: reallymemorable]
<clever>
jackdk: all of that magic is in pkgs/development/haskell-modules/generic-builder.nix
reallymemorable has joined #nixos
mexisme has quit [Ping timeout: 250 seconds]
mexisme has joined #nixos
<jackdk>
clever: I think your sample isn't quite right: callHackageDirect takes a { pkg, ver, sha256 } and trying to use callHackage is giving me unexpected argument errors
<clever>
oh
<clever>
i was mis-remembering the args for callHackageDirect
<jackdk>
This is the comment above it in make-package-set.nix, which seems wrong: # callHackageDirect :: Text -> Text -> AttrSet -> HaskellPackage
<clever>
it might have changed after the comment was made
domogled has quit [Remote host closed the connection]
domogled has joined #nixos
justan0theruser has quit [Ping timeout: 246 seconds]
<jackdk>
hum. I can't see how to feed that attrset with revision and editedCabalFile in
<pie__>
so i guessss nixos enables cgroups v1 and v2 but all the controllers end up in the v1 hierarchy because you can only mount a controller once?
hyper_ch has joined #nixos
Rusty1 has quit [Quit: Konversation terminated!]
domogled has quit [Quit: domogled]
domogled has joined #nixos
vonfry has joined #nixos
chocopuff has joined #nixos
<vonfry>
I find nixpkgs use `placeholder "out"` instead of `$out`. What are difference between them? I cannot find a description about `placeholder` on the manual page.
<{^_^}>
vonfry: 2 weeks, 1 day ago <tilpner> Oh, and check if rclone does what you want
<{^_^}>
vonfry: 2 weeks, 1 day ago <tilpner> Run appimage-run with APPIMAGE_DEBUG_EXEC=bash , then set XDG_DATA_DIRS inside and exec into AppRun
<jackdk>
not sure if this is the best way to do it but oh well
<clever>
jackdk: the revision goes into the 2nd {}, on line 7
<cdepillabout>
When trying to build my nixos system from the current nixpkgs master, I get the following error when the manual is being built: `ln: failed to create symbolic link './generated/2g0nicc0xrlk8lah3crd6hcn5dk8kf04-generated-docbook': Permission denied`. This happens when building `gk55y87b59i87cx3ngsmjlf2ib4ca6aa-nixos-manual-combined.drv`.
<cdepillabout>
Could someone else try to rebuild their system from nixpkgs master and let me know if you also get this error? If so, I'll look into it and try to send a fix.
<clever>
jackdk: yeah, it wont accept a revision easily
<jackdk>
clever: is there a doc anywhere that describes the pseudotypesignatures used in make-package-set.nix? I intend to PR a correction to the docstring for callHackageDirect
<lovesegfault>
clever: Yeah, I used bfg to clean all secrets
cyraxjoe has quit [Ping timeout: 276 seconds]
<lovesegfault>
do git grep REMOVED $(git rev-list --all)
<kandinski>
how can I know which generation was booted?
<kandinski>
which system build, I mean
CMCDragonkai has joined #nixos
<CMCDragonkai>
Hi, I've found that on NixOS using NVIDIA 435.21 I only see 3/4 GPUs. However when I then used Ubuntu with the same 435.21, I ended up getting 4 GPUs.
<kandinski>
I'm having a repeated issue with sound after I upgraded to 19.09 on an Intel thinkpad, but I can't well reproduce it. At a first approximation, it seems to me that there's sound on boot, but not on coming back from suspend.
<lovesegfault>
Goddamit andi- I think brandon broke my server
<lovesegfault>
No, `boot.extraModulePackages = [ linuxPackages_4_19.nvidia_x11 ];`
<lovesegfault>
also, `man configuration.nix` can help
<clever>
lovesegfault: you must fetch the modules from config.boot.kernelPackages
<clever>
lovesegfault: if you get it directly from pkgs.linuxPackages, you risk getting modules for another kernel
<CMCDragonkai>
yea i know
<CMCDragonkai>
not sure how this solves the 3/4 GPUs?
<CMCDragonkai>
why would it not detect one of them?
<clever>
does lspci detect it?
<lovesegfault>
CMCDragonkai: There's some weird setting thing that causes a bizarre unknown to me interaction that makes one of the GPUs disappear
asxa has joined #nixos
<lovesegfault>
in my experience changing that setting helps
<lovesegfault>
clever: Yes, it does
zupo has joined #nixos
asxa has left #nixos [#nixos]
<lovesegfault>
at least IME
asxa has joined #nixos
asxa has left #nixos [#nixos]
<lovesegfault>
and it's always the card on the first slow (lowest PCI ID) that doesn't show
<lovesegfault>
brb
<CMCDragonkai>
yea i noticed that too
<gyroninja>
Is slack plain white for anyone else?
<lovesegfault>
back
<lovesegfault>
gyroninja: Did you just install it for the first time
<CMCDragonkai>
lovesegfault: but my `xserver.enable` was already false
<lovesegfault>
CMCDragonkai: just try it
<gyroninja>
lovesegfault: No, it's been installed for a long time
<lovesegfault>
worst case scenario you can change it back
<CMCDragonkai>
lovesegfault: yea... it will take some time
<Dandellion[m]>
I'm attempting to package https://github.com/janusvr/janus. I get as far as compiling but fail when it tries to import `#include <btBulletDynamicsCommon.h>` even though I have `bullet` in BuildInputs, and that header file is present: `include/bullet/btBulletDynamicsCommon.h `
<Dandellion[m]>
So i'm a bit confused as to what I need to do make it recongnize it's there
<lovesegfault>
CMCDragonkai: Why?
<gyroninja>
actually I think I might know the issue
<CMCDragonkai>
need to unload the ubuntu
<lovesegfault>
CMCDragonkai: Ah :)
<gyroninja>
What's the name of the opengl cube test application?
<lovesegfault>
gyroninja: idk about cube, I know of gears
<lovesegfault>
which comes from the glxinfo pkg, IIRC
<clever>
,locate bin glxgears
<{^_^}>
Found in packages: glxinfo, driversi686Linux.glxinfo
<gyroninja>
that was what I was thinking of thanks
<gyroninja>
that loads fine, so my theory is wrong
<gyroninja>
I can't seem to figure out how to get slack to log and I'm not sure where the base directory is
<CMCDragonkai>
lovesegfault: does your server also use onboard intel graphics?
<CMCDragonkai>
lovesegfault: the server i have has onboard intel graphics and 4x rtx 2070
<kandinski>
clever: do you have a Twitter handle?
<CMCDragonkai>
so im wondering if there's an intel problem?
<{^_^}>
[nixpkgs] @Lassulus pushed commit from @Moredread to master « mediathekview: 13.2.1 -> 13.5.0 »: https://git.io/Je6Ud
<gyroninja>
ah yes logging stuff in the config folder. I wonder what genius decided to do that.
<gyroninja>
(.config/Slack/logs)
<gyroninja>
nothing interesting in there though
<lovesegfault>
CMCDragonkai: I doubt it, I don't think Xeon Platinum CPUs have onboard GFX?
<CMCDragonkai>
this one does
<CMCDragonkai>
cause im plugged into the VGA
<gyroninja>
going to just try and rollback slack and see if it starts working agian
<gyroninja>
*again
vonfry1 has joined #nixos
vonfry1 has quit [Client Quit]
Ariakenom has joined #nixos
<lovesegfault>
I mean, the server has IPMI and that's VGA
<lovesegfault>
so idk
<gyroninja>
nope rolling it back did not fix it
vonfry has quit [Ping timeout: 276 seconds]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<gyroninja>
turns out I just needed to control+R
<Ariakenom>
the description for hashed password has "mkpasswd -m sha-512" which afaict is 1 round. which to me seems strange https://nixos.org/nixos/options.html#users.users.%3Cname%3F%3E.hashedpassword
<clever>
Ariakenom: i tend to just run passwd, then steal a line from /etc/shadow
cinimod` has joined #nixos
<andi->
Ariakenom: feel free to propose a better default in the description :)
<andi->
I think that benefits everyone
zupo has joined #nixos
<Ariakenom>
clever: I think that's how I set it. That seems to be the same method
<DigitalKiwi>
that's also useful if for some reason you need to set a password on a raspberry pi that you can get the sd card but don't have hdmi/existing ssh/other issues
<Ariakenom>
from checking my /etc/shadow
aveltras has joined #nixos
<Ariakenom>
andi-: I haven't figured that out yet :p. just started looking into what methods to use when I saw that and it seemed strange
<andi->
interestingly crypt(3) says: the SHA-256 and SHA-512 implementations support a user-supplied number of hashing rounds, defaulting to 5000. If the "$id$"
<andi->
characters in the salt are followed by "rounds=xxx$", where xxx is an integer
<andi->
so that means 5000 would be the default not 1
<Ariakenom>
ah, I should have checked crypt's docs, thanks.
hlolli_ has quit [Remote host closed the connection]
<gwen>
Hi, could it be that since 19.09 the elpy emacs package is missing?
<andi->
Ariakenom: more strangly mkpasswd does not accept the rounds parameter with sha-512
<andi->
says: Method not supported by crypt(3).
<Ariakenom>
andi-: mkpasswd -m sha-512 -R 100000 works for me
<etu>
gwen: Before 19.09 was branched off a lot of work to the emacs ecosystem was done. And an emacs overlay was formed to provide up-to-date emacs packages. And also #nixos-emacs was formed for emacs questions :)
<andi->
Ariakenom: strange... I am on 19.09 with this machine. Did nix-shell -p mkpasswd
<gwen>
etu: Thank you very much, I will look into that :)
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
rembo10 has joined #nixos
<lovesegfault>
How can I interpolate an integer into a string
rauno has quit [Quit: Leaving]
<lovesegfault>
foo is an int, I want "foo=${foo}"
<clever>
> "${toString 42}"
<{^_^}>
"42"
<lovesegfault>
Magic
Heirlung has joined #nixos
<Ariakenom>
> toString 42
<{^_^}>
"42"
<Ariakenom>
I agree with your result, bot. Well calculated.
Makaveli7 has joined #nixos
Soo_Slow has joined #nixos
philr has joined #nixos
<Dandellion[m]>
I'm attempting to package https://github.com/janusvr/janus. I get as far as compiling but fail when it tries to import `#include <btBulletDynamicsCommon.h>` even though I have `bullet` in BuildInputs, and that header file is present: `include/bullet/btBulletDynamicsCommon.h `
<Dandellion[m]>
So i'm a bit confused as to what I need to do make it recognize it's there
<clever>
Dandellion[m]: try adding pkgconfig to nativeBuildInputs
lovesegfault has quit [Quit: WeeChat 2.6]
dansho has quit [Quit: Leaving]
magneticduck has joined #nixos
zupo has joined #nixos
<darthdeus>
why are things from my nix-env avaialble when I run `nix-shell --pure`? I thought that it's meant to isolate the outside environment for building
<darthdeus>
but instead I can say use `make` even though it's just installed via the env
knupfer has joined #nixos
<clever>
darthdeus: nix-shell always provides make
<darthdeus>
clever: that's not true, if I uninstall it from my env then running `make` inside nix-shell says "make is not installed"
<clever>
darthdeus: what does the prompt say when you run nix-shell?
<{^_^}>
[nixpkgs] @globin pushed to structured-attrs « treewide: configureFlags is a flat list »: https://git.io/Je6Td
<darthdeus>
just `[nix-shell:~/projects/pills/hello]$`
<darthdeus>
or what do you mean?
<clever>
darthdeus: and what does `type make` say?
<darthdeus>
bash: type: make: not found
<clever>
darthdeus: what are the contents of shell.nix?
<niksnut>
gchristensen: we can't build blender with cuda because of the cuda license
<darthdeus>
clever: uhm where do I find that?
<niksnut>
the blender-bin flake is cuda-enabled though
<clever>
darthdeus: what are the exact args you ran nix-shell with?
<darthdeus>
clever: `nix-shell something.nix`
bahamas has joined #nixos
<clever>
darthdeus: then what is the contents of something.nix ?
<clever>
darthdeus: thats not using the stdenv or buildInputs, so that will just completely break everything nix-shell does
<clever>
darthdeus: it only works with nix-build
<darthdeus>
ah lol, the pills don't mention that at all :\
<darthdeus>
damn
<clever>
darthdeus: nix-shell assumes your using stdenv.mkDerivation, and the stdenv provides make/gcc
<clever>
darthdeus: `nix-shell always provides make` also assumes your using the stdenv
<darthdeus>
clever: I haven't reached stdenv yet, so uhm yeah ... thanks tho
<clever>
darthdeus: one min...
<darthdeus>
the pills are good, but there are so many things I've run into like this one, kinda annoying :P
oxford has quit [Ping timeout: 260 seconds]
gxt has quit [Ping timeout: 260 seconds]
<clever>
darthdeus: they start with the internals of nix, that the stdenv is built around
<clever>
darthdeus: and i find a lot of people stop reading half way, and then try building a house with only the bricks from a foundation, and they havent even figured out where 2x4's and nails are yet
<{^_^}>
[nixpkgs] @cdepillabout pushed 2 commits to master: https://git.io/Je6TA
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<clever>
darthdeus: line 428 will just `source $stdenv/setup`, which assumes the derivation has a `stdenv = something;` and that `$stdenv/setup` will put things into PATH
<clever>
darthdeus: so if your thing wants to work with nix-shell (a fun excercise!), you must have a `stdenv = something`, and that something must contain a `$out/setup`, which adds all of $baseInputs to $PATH
knupfer has quit [Ping timeout: 245 seconds]
<{^_^}>
[nixpkgs] @xfix opened pull request #73805 → qt512.qtbase: add patch for CVE-2019-18281 → https://git.io/Je6kv
<blitzclone_>
I'm trying to use nixops from a fedora system, but the ssh it uses seems to barf on the host's ssh config. Does anyone know how to ignore the local configuration or use the system ssh? In particular I get:
<blitzclone_>
/etc/crypto-policies/back-ends/openssh.config: line 3: Bad configuration option: gssapikexalgorithms
<blitzclone_>
ma
<blitzclone_>
/etc/crypto-policies/back-ends/openssh.config: terminating, 1 bad configuration options
chloekek has joined #nixos
<clever>
blitzclone_: you may need to comment out those lines in your ssh host config
<darthdeus>
clever: hmm, so in the standard stdenv that doesn't just add things to PATH, but it has to replace it completely, right?
<clever>
darthdeus: $stdenv/setup will add all of $buildInputs to PATH
<clever>
darthdeus: nix-shell --pure, will just start that script with an empty PATH, so setup is adding to an empty list
<blitzclone_>
clever: mmh...
<darthdeus>
ah cool
<darthdeus>
clever: ty :)
<darthdeus>
nix really isn't easy to start with lol
<darthdeus>
and I've used haskell for years
<clever>
darthdeus: i learned nix first, and then i learned ghc internals, lol
<clever>
darthdeus: haskell itself came a while after that
<darthdeus>
lol, I was really only doing haskell when nix started coming around
<{^_^}>
[nixpkgs] @JohnAZoidberg pushed 2 commits to master: https://git.io/Je6kL
<CMCDragonkai>
xerver tries to start, then goes to stop, then starts then stop and then the screen goes blank
<darthdeus>
I remember one guy on our team was showing it to us and was like "here's this amazing new linux distribution based on functional concepts" and I'm like "uhm if it's so great, why are you running it in a virtualbox" ... and that was the end of my interest :|
<darthdeus>
(was like 7 years ago)
<clever>
darthdeus: if you want to see the setup that stdenv uses, nix-build '<nixpkgs>' -A stdenv -o stdenv && ls -lh stdenv/setup
<clever>
darthdeus: you could use that as a base for your own stdenv thing, if you really want to try making nix-shell work
<Yaniel>
reminds me of the story I read yesterday about someone debugging why their emails would not make it farther than 500 miles physically
<clever>
Yaniel: oh yeah, that was a fun story
<darthdeus>
wow nice upload
<ezemtsov>
Hi people. I try to install telega.el with melpaPackages and it constantly gives me error of wrong TDLib version. I can see that in .nix definition of telega they have a dependency on 1.4 while nixpkgs has already 1.5. How does that work? How can I ensure that I'm using 1.5 version which is already there?
<clever>
Yaniel: the original statement made no sense at all, but turned out to be exactly right
<clever>
Dandellion[m]: the issue, is that nix is only adding the include/ dir to the -I paths, so every package that tries to prevent the same issues nix solves (by using include/bullet) breaks nix
<clever>
Dandellion[m]: you need to do something like: NIX_CFLAGS_COMPILE = "-I${bullet.dev}/include/bullet";
<ezemtsov>
are there any telega.el users?
<darthdeus>
1.5Gbps damnnnn ... our "best" ISP only recently bumped their max from 300Mbps to 500
<Ariakenom>
rather, how do I check Qt version used by KDE Plasma?
<clever>
Ariakenom: find a binary within kde, and run `nix-store -qR` on that binary
<clever>
Ariakenom: was it fixed in 5.12 or a patch added after 5.12 was released?
<clever>
Ariakenom: take that path to qtbase, and run `nix-store --query --deriver`, then take the new .drv path and run it thru `nix-store --query --binding src`
<{^_^}>
[nixpkgs] @FRidh pushed 169 commits to staging-next: https://git.io/Je6Ie
<clever>
Ariakenom: ah, that may have been GC'd already
<Ariakenom>
not sure I've ever run gc on this install. but maybe
chocopuff has quit [Remote host closed the connection]
<darthdeus>
btw this might be too early in my learning to ask, but say that I have a python/js project I want to work on ... would it make sense to do something like create foo.nix with all the pip/npm packages in it, and then develop using nix-shell?
<darthdeus>
or is there some other way to do things?
ezemtsov has quit [Remote host closed the connection]
<Dandellion[m]>
darthdeus: that's a fine way to develop using nix! check out direnv or lorry to automatically spawn the environment
<clever>
Ariakenom: yeah, that is confusing
<darthdeus>
Dandellion[m]: nice :) how would that work with editors though? do I then need to say run my vim inside the same nix-shell, or just configure it somehow to look at the right paths?
<symphorien>
running vim inside the nix-shell works very well
<clever>
darthdeus: the python could then pass those args on you ycm, which then passes it on to a gcc/clang (not sure what it uses internally)
<clever>
in my example, its meant to be ran inside a nix-shell (line 4 reads the var that nix sets)
smatting has joined #nixos
ezemtsov has joined #nixos
<tobim[m]>
I'm working on a nixos module, and need to create a directory in ExecStartPre, but I run into an weird problem with that: `systemd[1549]: zeek.service: Failed at step CHDIR spawning /nix/store/gnw6yrqy249n62r4q8vy12ispviv3dav-coreutils-8.31/bin/mkdir: No such file or directory`. Did anyone here ever run into this or a similar issue?
<Dandellion[m]>
clever: thanks, it builds now, well, at least it doesnt error now...
<{^_^}>
[nixpkgs] @globin pushed to structured-attrs « treewide: configureFlags is a flat list »: https://git.io/Je6I4
* Dandellion[m]
sent a long message: < >
<clever>
tobim[m]: the WorkingDirectory= may apply to the `ExecStartPre` also, i typically just make that dir the home of a user, and set createHome=true;
<clever>
Dandellion[m]: your message is unreadable! 2019-11-20 05:19:02 * Dandellion[m] sent a long message: < >
<tobim[m]>
@clever: i do use `${pkgs.coreutils}/bin/mkdir` in the module, so I don't see how the WorkingDirectory has an effect here. Note that the path it complains about is absolute. I'm thinking that maybe this runs before `/nix/store` is mounted.
<clever>
tobim[m]: systemd itself is in /nix/store, so thats imposible
<clever>
tobim[m]: can you pastebin your nixos module?
<clever>
tobim[m]: line 47, you must quote that directory, or it wont work right
FRidh has quit [Quit: Konversation terminated!]
thc202 has joined #nixos
<clever>
tobim[m]: also, try using a multi-line string in systemd.services.<name>.preStart instead of ExecStartPre, then try doing `pwd` and `ls -l /var/log/zeek` from that multi-line string
<tobim[m]>
Thanks, but I still get the same error.
<{^_^}>
[nixpkgs] @bendlas pushed commit from @ivan to release-19.09 « chromium: 78.0.3904.97 -> 78.0.3904.108 »: https://git.io/Je6Lt
<turion>
I'm currently maintaining CI'ing a Haskell project with nix & Travis (https://github.com/sebastiaanvisser/clay/). I want to support multiple versions of GHC. My approach was defining a list of all versions and then looping over it and build the package for every GHC version. Unfortunately, this often errors with "Out Of Memory" on Travis when it tries to load 4 GHCs into ~3GB memory. How do I improve that?
knupfer has quit [Remote host closed the connection]
<DigitalKiwi>
and (may be a style thing but it's a widely adopted style...) use more new lines in default.nix
ezemtsov has quit [Remote host closed the connection]
knupfer has joined #nixos
<turion>
The two things I need:
<turion>
* Build with all GHCs locally, ideally using only nix
<turion>
* Only a single place where I configure the GHC versions (a file, or if need be, a directory)
<turion>
Nice to have:
<turion>
* CI with all GHCs without OOM
<darthdeus>
clever: hmmmm, that might even be easier to setup than regular ycm setup where you have to rely on generated configs :O ... this could even work for stuff like jedi in python where usually it's painful with virtualenvs mixed with global stuff, but with nix all of these issues could go away :O
<darthdeus>
I'm so intrigued now :P
<darthdeus>
really hoping I can get a working setup before I have to go back to work and be productive lol
<{^_^}>
[nixpkgs] @Mic92 merged pull request #73691 → tasksh: enable on Darwin → https://git.io/JeKnL
<turion>
The bottleneck in my nix setup seems to be that I'm trying to build everything with one `nix-build` call. In an ideal world, this would be fine, but unfortunately nix is not very efficient when it comes to memory. It should be able to build everything step-by-step, but it isn't.
<turion>
Should I make a text file where each line is a GHC version, and both nix and a travis build script can loop over it?
<clever>
tobim[m]: `-j2` can limit how many things nix builds at once
<turion>
clever: Is there anything similar for memory usage?
<turion>
Ideally I'd like to have the different GHC versions to be different travis jobs. This can be achieved with setting different environment variables in the travis script. But if I do that, how does nix get notified of the different variables? Is there a travis2nix script?
<symphorien>
you can pass --arg $ENVVAR
<symphorien>
--argstr foo $ENVVAR rather
<symphorien>
then have default.nix be a function with an argument foo
<turion>
symphorien: Ok, but then I'd need to build each version individually locally, right?
<turion>
Or collect all versions both in a .nix file and a .travis.yml file?
<symphorien>
you can have a default value of "all" for foo, and add logic in the default.nix which builds everything if foo is "all" ,
liff has joined #nixos
<manveru>
adisbladis: seems like the tools/go/vcs used by vgo2nix can't resolve some stuff because it doesn't implement the same behaviour as go mods on purpose :|
<dredozubov>
What's the most convenient way of setting nixpkgs to a local directory/repo instead of a channel? I guess it should be done in .nix-defexpr somehow, but my first few attempt failed.
<manveru>
i'm trying to update the terraform-providers-gandi package, but as usual google is using some silly domain shenanigans that can't be resolved by vgo2nix
tmaekawa has joined #nixos
ezemtsov has joined #nixos
<turion>
symphorien: Ok, that's quite nice, thanks. Still I'd need to maintain the supported versions in two different places then
erictapen has joined #nixos
<tilpner>
dredozubov: Set NIX_PATH=nixpkgs=$HOME/dev/nixpkgs and teach .nix-defexpr to respect NIX_PATH
<DigitalKiwi>
dredozubov: temporarily or permanently? -I nixpkgs=/path or set nixPath in your config are ways
<symphorien>
well, travis can only read statically generated configuration, right ? So you can't extract the list of ghc versions from nixpkgs
<ezemtsov>
hi. Is there a way configure emacs-overlay and emacsGit specifically to be used with default windowManager.exwm ?
<tilpner>
dredozubov: Not necessarily convenient to set up, but convenient to use afterwards, as all tools will use the same source of nixpkgs
knupfer has quit [Quit: knupfer]
<Dandellion[m]>
oh wow lmao, I added a prefonconfigure phase with export NIX_CFLAGS_COMPILE="$NIX_CFLAGS_COMPILE -I${bullet}/include/bullet"
knupfer has joined #nixos
<Dandellion[m]>
clever: ^
<immae>
tilpner: how do you "teach .nix-defexpr to respect NIX_PATH" ?
<tilpner>
dredozubov: If you want to do that, remove all links in ~/.nix-defexpr, and add a file ~/.nix-defexpr/default.nix with the content "import <nixpkgs> {}"
<dredozubov>
I'm looking for something that will be uniform for all nix tools, so tilpner's suggestion hits closer to home.
<tilpner>
Now you can do nix-env -iA hello without specifying a channel name
<turion>
symphorien: Yes, that's right. I wonder whether I can extract the configuration from travis to nix. Or extract it from the same place (textfile) into both. (But then I don't know how to split into jobs)
<dredozubov>
I'll try that, thanks
<symphorien>
turion: travis is yaml, right ? write it in json (which is valid yaml) and then read it with builtins.fromJson
<darthdeus>
is there some way to get nix-shell to print the offending expression? `error: expression does not evaluate to a derivation (or a set or list of those)`
<etu>
ezemtsov: So it basically creates a session which launches the command defined in there.
<symphorien>
the offending expression is "the whole file", darthdeus
cosimone has joined #nixos
<darthdeus>
symphorien: ah right, but still, can I somehow print what the file "evaluates to"?
<symphorien>
probably with nix-repl
<darthdeus>
I know what the error is in this case, I'm just curious if I can find some diagnostics that confirm it other than looking at it
<symphorien>
nix repl* sorry
<ezemtsov>
@etu windowManager.exwm should already do all that. Isn't that possible just to substitute the default package with emacsGit?
<symphorien>
`:p import ./foo.nix`
<turion>
symphorien: That's a gread idea! I'll try! Thanks!
<DigitalKiwi>
tilpner: does this not work for what they want nix.nixPath = [ "nixpkgs=/home/kiwi/nixpkgs" "nixpkgs=/home/kiwi/nixpkgs-1" ] ++ options.nix.nixPath.default;
<darthdeus>
symphorien: ah right, ty :)
<tilpner>
DigitalKiwi: No, nix-env ignores NIX_PATH by default
<tilpner>
DigitalKiwi: Why do you set nixpkgs twice?
<etu>
ezemtsov: Doesn't look like it. I've just read the module
<DigitalKiwi>
idk
mexisme has quit [Ping timeout: 246 seconds]
Chiliparrot has joined #nixos
<DigitalKiwi>
those two have different things in them
<ezemtsov>
etu: I know for sure that one can override exwm emacs with super.emacs26.override {} in overlays. That can be used for imagemagick support for example
<etu>
ezemtsov: yeah, that would work
<ezemtsov>
can the same approach be used for just using a different package
erictapen has quit [Ping timeout: 240 seconds]
<ezemtsov>
I guess I don't understand how overrides work
<etu>
ezemtsov: So, you want to override emacsWithPackages in an overlay to change the exwm module
Hedgework has quit [Ping timeout: 276 seconds]
ggp0647 has joined #nixos
erictapen has joined #nixos
tmaekawa has quit [Quit: tmaekawa]
cyraxjoe has joined #nixos
MightyJoe has quit [Ping timeout: 276 seconds]
Hedgework has joined #nixos
<MasseR>
Hey, I'm getting a "error: 'https://xxx.s3.eu-central-1.amazonaws.com' does not appear to be a binary cache" error when running nix copy from a docker image. Locally I can execute it fine
<MasseR>
I have tried curling the nix-cache-info from the s3 store, both from my local machine and from the docker machine, both are 200 ok
knupfer has quit [Ping timeout: 250 seconds]
hmpffff has joined #nixos
hmpffff_ has quit [Ping timeout: 240 seconds]
cosimone has quit [Remote host closed the connection]
cosimone has joined #nixos
Neo-- has joined #nixos
steevveenn has joined #nixos
drakonis_ has quit [Ping timeout: 250 seconds]
drakonis_ has joined #nixos
<steevveenn>
Hi! I was willing to use a specified package to be used for virtualizatiln , using virtualbox. I changed my configuration for now using the dedicated option : virtualisation.virtualbox.host.package = [n1903.virtualbox]; where n1903 is importing the channel for nixos 19.03. However I cannot managet to perform nixos-rebuild switch. It
<steevveenn>
complains: error: attribute 'override' missing at nix/var/..... myconfig.nix
<steevveenn>
I have not succeeded in understanding the error message, any hint ?
hyper_ch2 has joined #nixos
<tilpner>
steevveenn: nix/var? Where did you put your config?
ezemtsov has quit [Remote host closed the connection]
<steevveenn>
tilpner always at the same place: /etc/nixos
<tilpner>
steevveenn: You should create a past with the full error, and perhaps your config
domogled has quit [Ping timeout: 240 seconds]
LysergicDreams has quit [Ping timeout: 240 seconds]
iyzsong has joined #nixos
Soo_Slow has quit [Quit: Soo_Slow]
<steevveenn>
I have put the dedicated config file content and the error using --show-trace: https://pastebin.com/43qSJKXw
<steevveenn>
at the end the error is: while evaluating the attribute 'package.override' at undefined position:attribute 'override' missing, at /nix/var/nix/profiles/per-user/root/channels/nixos/nixos/modules/virtualisation/virtualbox-host.nix:8:16
LysergicDreams has joined #nixos
<lukego>
How can I insert some custom settings into /etc/sshd/sshd_config? I see that services.openssh.extraConfig is defined with "mkOrder 0" but I don't really understand if that means I can somehow append a string to that value via my nixos configuration.nix?
domogled has joined #nixos
equivrel has quit [Remote host closed the connection]
equivrel has joined #nixos
<steevveenn>
I feel a bit dumb: I modified the line into : virtualisation.virtualbox.host.package.override = [n1903.virtualbox]; and it compiles. I had not seen the type of the package option.
hmpffff has quit [Quit: nchrrrr…]
<steevveenn>
I have not run to vbox to check if that works, but I am confident. nixos compiles the configuration ok. sorry for the noise.
orivej has quit [Ping timeout: 265 seconds]
hmpffff has joined #nixos
steevveenn has quit [Remote host closed the connection]
alexherbo2 has joined #nixos
FRidh has joined #nixos
__monty__ has joined #nixos
LysergicDreams has quit [Ping timeout: 265 seconds]
LysergicDreams has joined #nixos
<Izorkin>
etu: rechecked php 7.4? )
ThatDocsLady has joined #nixos
<tilpner>
,tell steevveenn Sorry, didn't see your reply (no ping). You want virtualisation.virtualbox.host.package = n1903.virtualbox;
<{^_^}>
tilpner: I'll pass that on to steevveenn
chreekat has quit [Quit: quitting]
zupo has quit [Ping timeout: 245 seconds]
<tilpner>
,tell steevveenn And it seems odd for the same machine to be virtualbox host and guest at once, you probably only want host or guest, not both
<{^_^}>
tilpner: I'll pass that on to steevveenn
stigo has joined #nixos
zupo has joined #nixos
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
<lukego>
Oh seems I can just set services.openssh.extraConfig to a string and it magically merges at the end.
<lukego>
That makes sense from an interface perspective but it's not the impression I got from my naive reading of the nixos module definition for openssh.
<{^_^}>
[nixpkgs] @FRidh pushed 76 commits to python-unstable: https://git.io/Je6Yf
<bdju>
I keep experiencing mouse lag/freezing but with no obvious cause. temperatures aren't high enough for throttling, cpu isn't maxed, ram isn't maxed. I'm using Sway
Unode has quit [Read error: Connection reset by peer]
Unode has joined #nixos
ent has joined #nixos
Unode has quit [Read error: Connection reset by peer]
domogled has quit [Quit: domogled]
Unode has joined #nixos
bahamas has joined #nixos
coderobe has quit [Read error: Connection reset by peer]
dingenskirchen has quit [Quit: dingenskirchen]
dingenskirchen has joined #nixos
m0rphism1 has joined #nixos
simpson has left #nixos ["WeeChat 1.0.1"]
<FRidh>
adisbladis: just had a look at poetry2nix and looks really good. I suppose extending the default overrides.nix for nixpkgs unstable/stable would be very useful. With a more extended example (including how to add overrides yourself or use nixpkgs python packages like those that are bindings) I think this could be a recommended way for creating python envs.
<adisbladis>
FRidh: Yeah that's very much my intent :) So far overrides.nix has been me dogfooding it. Definitely needs more docs.
<turion>
error: don't know what to do with argument 'builtins.attrNames((import <nixpkgs> {}).haskell.compiler)'
<turion>
What else would nix eval expect other than a nix expression?
<adisbladis>
"I think this could be a recommended way for creating python envs." <3
<{^_^}>
[nixpkgs] @alexbakker opened pull request #73814 → zenpower: init at 0.1.5 → https://git.io/Je6OM
seanparsons has joined #nixos
nexgen has joined #nixos
<hyper_ch2>
hi gchristensen
mariel__ has quit [Ping timeout: 240 seconds]
m0rphism1 has quit [Ping timeout: 245 seconds]
Ariakenom has quit [Ping timeout: 246 seconds]
<FRidh>
poetry is getting a lot of traction. You've shown that essentially all the info needed is there, and its really easy to use without generating all kinds of files. It's important to document the algorithm that is used to choose a certain source of a pkg so people know what files its going to grab. Not sure though whether these overrides should be in nixpkgs or not. With cabal2nix they are not, which at times can be a bit annoying. At the same
<FRidh>
time, it may be good to keep things out when possible. In nixpkgs I suppose we could just fetchgit it as that's allowed with pure evaluation.
<adisbladis>
FRidh: Btw, I meant to ask you something. I have a package where the dist-info ends up not recognising the version number: /nix/store/mc0xpkjzhmrnpl4qyljbmcyvxsygncf0-python3.7-django-scopes-1.1.0/lib/python3.7/site-packages/django_scopes-_.dist-info/METADATA -> http://dpaste.com/0K1ZBPP
<adisbladis>
Do you have any ideas?
<{^_^}>
[nixpkgs] @alexbakker opened pull request #73815 → zenmonitor: init at 1.2 → https://git.io/Je6Oh
m0rphism1 has joined #nixos
<FRidh>
adisbladis: was this from a wheel?
Ariakenom has joined #nixos
<adisbladis>
FRidh: Nope, it's an sdist
<FRidh>
adisbladis: do you have a link to the sdist? I've seen it sometimes when people use something like setuptools_scm to inject a version number, and they've uploaded an sdist where the number was not injected
<deni>
When doing local haskell development with nix if I do nix-build everything works fine. And then I use nix-shell and inside use `cabal new-build` so that I can have incremental compilation (good for dev loop speed). Now...that creates a symlink in my local folder called `.ghc.environment.x86_64-linux-8.6.5`. And when I exit nix-shell and try to run a "full" nix-build again I get the following error:
<deni>
`ghc: can't find a package database at /home/deni/.cabal/store/ghc-8.6.5/package.db
<deni>
deleting the .ghc.environment.x86_64-linux-8.6.5 symlink fixes the issue
<adisbladis>
FRidh: My long-term hope is that poetry2nix can live as a flake, and that overrides go in the flake.
<FRidh>
adisbladis: ha, have a look at the setup.py. They want to import the value from another file...but the file does not exist so thye use the default ? which cannot be transformed into something valif
<adisbladis>
But maybe it's more sensible to embed overrides into nixpkgs?
o1lo01ol1o has quit [Remote host closed the connection]
<FRidh>
I thought for now having inside poetry2nix repo or in a separate repo a community-maintained overrides.nix with branches for nixpkgs-unstable/nixos-19.09/...
<FRidh>
outside nixpkgs makes it easier to add other maintainers
<FRidh>
actually, may as well be in the same repo as poetry2nix
coderobe has joined #nixos
<adisbladis>
FRidh: Indeed, patching the version to `version = '${version}'` fixed it :)
<asymmetric>
huh, hunspell marks "didn't" as incorrect - this doesn't happen in thunderbird (uses hunspell too) so i wonder if it's a problem with the nixpkg
<asymmetric>
has anyone else seen this?
<FRidh>
Bit off-topic but this is by the way one thing I am concerned about with flakes, patching. While we can patch source in a derivation, we cannot patch flakes.
<{^_^}>
[nixpkgs] @FRidh merged pull request #72370 → Ansible: Fix connection plugins and add support for NETCONF → https://git.io/Jez1M
<{^_^}>
[nixpkgs] @FRidh pushed commit from @mweinelt to master « Ansible: Fix connection plugins and add support for NETCONF (#72370) »: https://git.io/Je63K
<hexa->
thx
psyanticy has joined #nixos
<FRidh>
hexa-: did you happen to bring up the issue on the ansible tracker?
<hexa->
FRidh: I did not as I felt their reason for trying to reuse sys.executable for the connection looked legit.
<FRidh>
yea they probably won't like the change
iyzsong has quit [Ping timeout: 252 seconds]
<hexa->
regarding "Instead they should rely on their executables already pointing to the right Python.", that would require them to patch their shebangs dynamically on most systems, no?
<hexa->
and reusing from sys.executable is also a reasonable choice for venvs
<hexa->
s/from //
gwen has quit [Quit: Leaving]
<mpickering>
Anyone know which package `wasm-ld-9` is in?
<qyliss>
,locate wasm-ld-9
<{^_^}>
Couldn't find in any packages
<mpickering>
,locate wasm-ld
<{^_^}>
Found in packages: lld_6, lld_7, lld_8
<adisbladis>
mpickering: Looks like lld_9
<mpickering>
ah I see
<infinisil>
I should update the ,locate database, or make it automated at some point
<mpickering>
hmm it seems lld_9 doesn't provide wasm-ld-9 still explicitly but it does have the right version
reallymemorable has joined #nixos
DavidLeung[m] has joined #nixos
kisonecat has joined #nixos
o1lo01ol1o has joined #nixos
o1lo01ol1o has quit [Ping timeout: 246 seconds]
<infinisil>
,locate wasm-ld
<{^_^}>
Found in packages: lld, lld_6, lld_8, llvmPackages.bintools
<infinisil>
The nix-index database has been updated
bahamas has quit [Read error: Connection reset by peer]
<infinisil>
(to latest stable)
Rusty1 has joined #nixos
Izorkin has joined #nixos
Ariakenom has quit [Ping timeout: 265 seconds]
Makaveli7 has quit [Ping timeout: 240 seconds]
domogled has joined #nixos
genesis has quit [Ping timeout: 240 seconds]
<asymmetric>
is nix-index what's used under the hood to find the missing package when i run e.g. cowsay?
<{^_^}>
[nixpkgs] @knl opened pull request #73819 → Multithreaded perl on darwin → https://git.io/Je6Gj
<ij>
tilpner, ah, you're right! sorry. from command line either way, but I heard libvirt is hard to configure, so I guess qemu first :)
<jared-w>
If you've never used qemu before it's probably a good idea to get it working through a GUI. If you already have it all working and just want to run the "same thing" from the terminal, that's different
<ij>
I've used virt-manager. I'm curious how it works from the command line, because ... I like command-line stuff I guess
<jared-w>
Because then instead of having to figure out if your command and/or setup is wrong, then you can fix one at a time.
cosimone has joined #nixos
Izorkin has joined #nixos
hyper_ch2 has quit [Remote host closed the connection]
Ariakenom has quit [Ping timeout: 240 seconds]
Izorkin has quit [Client Quit]
<jared-w>
But if you're familiar with virt-manager it shouldn't be terribly difficult to go through the qemu documentation and go from there. There's a lot of settings you tend to have to pass that the GUI does for you
<{^_^}>
#73720 (by ggPeti, 1 day ago, open): pgsidekick: init at unstable-2014-11-23
<ggpeti>
if you have a comment, please leave it on the PR
evils has quit [Ping timeout: 245 seconds]
<aranea>
How do I "mix" channels? I set up my system to track nixos-19.09, but I also want to install a package I just wrote (in an overlay) which depends on a few libraries in nixos-unstable.
<aranea>
nix fails at the skawarePackages assignment with the error "infinite recursion encountered".
<aranea>
Oh duh, now that I asked I think I understand what's wrong. :D
<aranea>
So I probably just need to import nixos-unstable with overlays = [].
domogled has joined #nixos
<aranea>
Though on second thought, I suppose it'd be saner not to do that, remove the skawarePackages assignment, and then install unstable.s6-linux-init.
selfsymmetric-mu has quit [Disconnected by services]
selfsymmetric-mu has joined #nixos
selfsymmetric-mu has quit [Remote host closed the connection]
selfsymmetric-mu has joined #nixos
zupo has joined #nixos
<selfsymmetric-mu>
I need to procure Python 3.7.2. Nix only provides 3.7.5. Specific, I know, but I'm in a specific enterprise environment whose tests will fail if I work with a different version.
<{^_^}>
[nixpkgs] @jonringer pushed commit from @kwohlfahrt to master « freebayes: 2017-08-23 -> 1.3.1 (2019-06-03) »: https://git.io/Je6Cn
<selfsymmetric-mu>
As far as I can tell nix isn't trying to build my special Python at all. I specifically put in a dummy sha256 so that it would fail.
<selfsymmetric-mu>
Wow, thank you FRidh, that looks very promising. I'm just trying to merge it with my `shell.nix`'s `mkDerivation` so that I can test it out.
orivej has quit [Ping timeout: 265 seconds]
<{^_^}>
[nixpkgs] @kmcopper opened pull request #73823 → nixos/ksm: Enhancements and new ksmtuned package. → https://git.io/Je6CP
zeta_0 has joined #nixos
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
knupfer has quit [Quit: knupfer]
<selfsymmetric-mu>
I have a `mkDerivation` that references `python372`. How do I combine that with this overlay?
ambro718 has joined #nixos
<selfsymmetric-mu>
I'd like it to be local to the `shell.nix` if possible, to keep it isolated.
<FRidh>
if you have a function that calls stdenv.mkDerivation, just add python372 as parameter to that function and it gets filled in by callPackage
<selfsymmetric-mu>
Without overlays, just localized overrides.
orivej has quit [Ping timeout: 240 seconds]
<selfsymmetric-mu>
I have something distinctly Python3.7.2-looking compiling right now.
<selfsymmetric-mu>
`Python 3.7.2 (default, Dec 24 2018, 03:41:55)`
<selfsymmetric-mu>
yessssss
<selfsymmetric-mu>
FRidh++
<{^_^}>
FRidh's karma got increased to 13
<selfsymmetric-mu>
Thanks for the help!
<selfsymmetric-mu>
FRidh: Question though. What's with the `self = py` trick? That seems to be the key to the whole thing.
<zeta_0>
selfsymmetric-mu: ok but how though is the installation similar to installing it on a laptop with a flash drive (usb) ?
<Church->
question, what would be the way to run `ip rule add fwmark 1 table mail.out` in nixOS declaratively? Using the iproute2 service to add a table called mail.out to my routing table
<Church->
But I need to set a rule as well to match against it
<Church->
As well as add the route itself via ip route
knupfer has joined #nixos
<selfsymmetric-mu>
zeta_0: You're asking how NixOS on a Raspberry Pi is installed in general? In that case, see installation notes here: https://nixos.wiki/wiki/NixOS_on_ARM/Raspberry_Pi. I believe your Raspberry Pi has a USB port.
<selfsymmetric-mu>
FRidh: Does `self = py` make this an "override with fixed point", similar to `rec`?
<Church->
Ah I see how I think for adding the routes.
<Church->
Not so much the rules, let me check docs again then
<Church->
Hmm, not sure I can set a routing table when via networking.interfaces.NAME.ipv4.routes...
<Church->
So I can set my mail packets outbound to be routed via my static vps
Ariakenom has joined #nixos
nexgen has quit [Quit: Leaving]
<FRidh>
selfsymmetric-mu: in the python expression it needs to refer to itself because we need some information. Now that we need to override it, we need to modify the "outer" python and the "inner" one.
<selfsymmetric-mu>
Wild, thank you.
selfsymmetric-mu has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @socksy opened pull request #73825 → session-manager-plugin: init at 1.1.50.0 → https://git.io/Je6Wy
<{^_^}>
[nixpkgs] @LnL7 opened pull request #73826 → git: disable failing test on darwin → https://git.io/Je6W9
<zeta_0>
can i use the same nixos installation image i used for my laptop also for my raspberry pi ?
<zeta_0>
or is there one specific to raspberry pi ?
<lordcirth>
zeta_0, you will need one built for ARM
<samueldr>
though it doesn't exactly respect the rules set by the generic aarch64 image
knupfer has quit [Quit: knupfer]
bvdw has quit [Read error: Connection reset by peer]
bvdw has joined #nixos
<das_j>
AmandaC: Just found you in the channel logs. Did you fix your "cannot allocate memory" from cptofs eventually?
<{^_^}>
[nixpkgs] @tfc opened pull request #73827 → Mosquitto test python → https://git.io/Je6ls
<AmandaC>
das_j: no, but I've only tried the once since getting my laptop back, confirming it happens on my laptop with 16G of RAM as well. My best guess is qemu-system is limiting the memory, and for whatever reason the cptofs is running under it. (Maybe hardware differences in some FS drivers? Not sure what other justificaation there could be. )
<das_j>
AmandaC: Yes, the same applies to qemu-user. The system has ~60GB of free RAM :/
<das_j>
Also building on disk, not tmpfs
<AmandaC>
er, yeah, qemu-user, not system. I alwasy get them mixed u
<das_j>
well, maybe qemu-system could be a solution
<AmandaC>
I don't think it effects builds on native arm64 hosts, at least I saw no indication of the prblem in the hydra logs for NixOS/Nixpkgs itself.
<das_j>
btw, if you let the build complete, you get a non-empty image so it shouldn't be too bad
<das_j>
I still couldn't get it to run on my pi 4
<das_j>
(bootloader foo)
<AmandaC>
The image wouldn't boot, for me on my pi3+
<AmandaC>
Got into a boot loop after stage 1, IIRC
mexisme has joined #nixos
<das_j>
at least you got to stage 1 :/
<das_j>
I'm stuck at the colors screen
<AmandaC>
pi4 isn't supported by mainline linux yet, is it? or mainline uboot or something mainline, IIRC it's not supported by anything but raspbian yet
Jackneill has quit [Remote host closed the connection]
<das_j>
yes, mainline uboot. but should work with extlinux, shouldn't it?
<das_j>
or maybe I should look around for an aarch64 server
<das_j>
for building my pis there
<zeta_0>
samueldr: so i can also build my own image as well ?
orivej has joined #nixos
<samueldr>
yes
<samueldr>
das_j: mainline linux won't work with the pi4 yet
<das_j>
samueldr: Also not with boot.loader.raspberryPi.version = 4?
<samueldr>
but the goal is to delete the raspberry pi 4 specific bits ASAP
mexisme has quit [Ping timeout: 252 seconds]
jgt1 has quit [Ping timeout: 245 seconds]
<das_j>
but that's good enough for my hydra. I'll come back once the load of the server is < 80
<samueldr>
it's already a bad precedent enough to have a specific platform in there :)
hmpffff has joined #nixos
<zeta_0>
`SBCs` ?
<samueldr>
Single Board Computers
<zeta_0>
samueldr: thanks
<zeta_0>
samueldr: i'm still very confused, i am going to spend more time reading the documentation before i attempt to install
<das_j>
zeta_0: But that's only half the fun ;)
<zeta_0>
samueldr: i thought this would be trivial
<samueldr>
I suggest also reading about other raspberry pi distributions, trying them, dissecting them even
<samueldr>
NixOS does things one way, but there are multiple ways
zupo has joined #nixos
reallymemorable has joined #nixos
<zeta_0>
samueldr: i have had so many issues with other distributions(debian, ubuntu), i want to stick with nix(nixos) (guix sounds good but i haven't tried it yet)
domogled has quit [Quit: domogled]
<samueldr>
yeah, I wasn't telling you to switch, but to observe the raspberry pi specific things in them :)
<samueldr>
if I hadn't had the knowledge from previously using other distros on the pi, I think it would have been harder to grok the raspberry pi specifics in nixos
wildtrees has joined #nixos
Chiliparrot has joined #nixos
<Aleksejs>
maybe someone know the quick answer. After upgrading to 19.09, as usually things are broken randomly. I'm trying to print a document and some specific characters are broken - I see crosses. However, the same characters are not broken. Also, all broken charactrs are in paragraphs that are not italic in pdf but are italic when printed out. Also characters are not broken when bolded
<lordcirth>
Aleksejs, what program is this document open in, and what font is showing this problem?
<{^_^}>
[nixos-org-configurations] @grahamc pushed to master « alerts: hopefully make them less flappy »: https://git.io/Je68I
<Aleksejs>
lordcirth: trying to print directly from libre writer and from imported pdf opened with ocular. Font is source sans pro
<lordcirth>
Aleksejs, ok. If you make a copy of this document and change to, eg, arial, does it have the same problem?
drakonis_ has joined #nixos
iclanzan has joined #nixos
iclanzan has quit [Client Quit]
fendor has joined #nixos
<zeta_0>
samueldr: so i would like to try building my own image then place it in an sd card, then installing it on my raspberry pi, do i need to do anything else other than that ?
<samueldr>
you can directly use the image I linked previously for the raspberry pi 4
jgeerds has joined #nixos
<samueldr>
"burn" it to an sd card using dd
<samueldr>
then it should boot
<Aleksejs>
lordcirth: with other font it works perfectly. But I used the very same document with that font before 19.09 on the very same printer and it worked
<samueldr>
then, you install, but with a twist: you install "in-place"
<jco>
hey! how do you find the crypto hash sha256 for a newly created package?
cosimone has quit [Quit: Terminated!]
<jco>
when I fetch the sources from github I often get the required hash but this time not
<gchristensen>
,tofu jco
<{^_^}>
jco: To get a sha256 hash of a new source, you can use the Trust On First Use model: use probably-wrong hash (for example: 0000000000000000000000000000000000000000000000000000) then replace it with the correct hash Nix expected. See: tofu-vim
gxt has quit [Quit: WeeChat 2.6]
Church- has quit [Quit: WeeChat 1.6]
<jco>
ok so I try a random one and correct it based on the compilation output?
oida has joined #nixos
noudle has joined #nixos
<jco>
thanks!
<gchristensen>
yep
<jco>
i used the hash code from another package and it compiles though
werner291 has joined #nixos
<jco>
an other package which isn't installed on my computer
<gchristensen>
you'll need to change the hash
<gchristensen>
so change one of the characters in the middle a little bit
shibboleth has quit [Quit: shibboleth]
<jco>
okay!
<gchristensen>
Nix uses the hash as the identity of the build's result. this means it first tries to find a build which matches that hash before trying the build a second time
zupo has joined #nixos
<gchristensen>
so if you use a hash which is a known-good hash for another package, it risks just getting that package's source :)
gxt has joined #nixos
<jco>
oh it's exactly what it did :)
<jco>
going to fix that
hmpffff has quit [Quit: nchrrrr…]
doublex is now known as doublex_
<zeta_0>
are there monads in nix like any other functional programming languages
<zeta_0>
?
<gchristensen>
zeta_0: as somebody who is not an expert in functional progamming, or functionally familiar with monads, I would say no ... but perhaps there are technically monads somewhere and I just don't recognize them as such.
aleph- has joined #nixos
<zeta_0>
gchristensen: oh, that makes sense, so you don't know haskell ?
brano543 has joined #nixos
<gchristensen>
I can read some Haskell, but I don't know enough to be productive
<darthdeus>
any idea how I can fix `warning: name collision in input Nix expressions, skipping '/home/darth/.nix-defexpr/channels_root/nixos'`?
<brano543>
Hello, guys. I am having a problem creating expression in which Python would be able to found pkg-config path. The strange thing is it always works inside nix-shell, but never inside nix-build. Is this scenario unsupported or am I doing something wrong?
<brano543>
Hello guys, I am on non-nixos system (Oracle Linux 8 ) and I am trying to build Nix expression for ScyllaDB. I am having trouble with the build as the library needs to execute Python configuration file. During nix-build I am getting error that Nix can't find pkg-config (while trying to execute this piece of code
<brano543>
When I try to replicate the steps inside nix-shell for the same expression, it can find pkg-config without problem.
kisonecat has joined #nixos
kisoneca1 has joined #nixos
kisoneca2 has joined #nixos
shibboleth has joined #nixos
<brano543>
Isn't there really anybody who would be so kind and tell me how do I import the same Nix environment variables which are inside nix-shell where it works also to nix-build?
kisoneca3 has joined #nixos
irl25519 has quit [Quit: My computer has gone to sleep. ZZZzzz…]
kisonecat has quit [Remote host closed the connection]
kisoneca1 has quit [Remote host closed the connection]
kisoneca2 has quit [Remote host closed the connection]
Blessjah has quit [Remote host closed the connection]
kisonecat has joined #nixos
<lassulus>
brano543: maybe exporting PATH in the configurephase? something like `export PATH=${pkgconfig}/bin:$PATH`
Blessjah has joined #nixos
cosimone has joined #nixos
xkapastel has quit [Quit: Connection closed for inactivity]
irl25519 has joined #nixos
<brano543>
lassulus: This is strange, I printed the whole PATH and I can see the pkg-config in there (/nix/store/xqy71s4d455bj58xyhx0vih83vfglqvy-pkg-config-0.29.2/bin), yet still the python executed from inside nix-build can't see it. The strange thing is from inside nix-shell if I manually call Python, import subprocess and then try to call the same command
<brano543>
which fails in nix-build,it can find pkg-config just fine.
<brano543>
lassulus: Works exactly the same. No issues at all. I think the Python within nix-build is not getting the PATH variable for some reason.
<jtojnar>
brano54339 try strictDeps = false; in the derivation
<jtojnar>
never mind it is not a python package
smatting has joined #nixos
<brano543>
jtojnar, lassulus: What is even more unbelievable is that Cmake which is executed before the configure.py script can find the pkg-config. -- Found PkgConfig: /nix/store/xqy71s4d455bj58xyhx0vih83vfglqvy-pkg-config-0.29.2/bin/pkg-config (found version "0.29.2")
<lassulus>
hmm, so maybe its something in the configure script which interferes?
<jtojnar>
brano54339 can you share the build log, I would not expect cmake to run when configurePhase exists
<brano543>
jtojnar: My first suspicion was that shell=True needs to be sut, but maybe explicit env removes everything and just sets PKG_CONFIG_PATH only.
<lassulus>
how does this work on other distributions? or does subprocess fallback to something like /bin ?
michiel_ has joined #nixos
noudle has quit [Remote host closed the connection]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<michiel_>
Question: I have a new machine that I've installed NixOS on. All is well, until I want to use a graphical shell. The weird thing is: the "non-configured" live USB installer gives me a fine xorg with Plasma - but after I have installed and reboot, I cannot in any way get things to work. I have an AMD card, and the amdgpu module is even loaded (first thought that was the issue). Does anyone have a hint on
<{^_^}>
[nixpkgs] @Lassulus pushed commit from @filalex77 to master « hyphen: init at 2.8.8 »: https://git.io/Je6Rg
Lears has quit [Remote host closed the connection]
magneticduck has quit [Remote host closed the connection]
<brano543>
lassulus, jtojnar: Okay, to fix this I will need to use absolute paths instead of relative paths and it will work. Thank you for your help, guys. Much appreciated. You know I wasn't sure where to look for that error. Thank you for opening my eyes and fixing the problem between monitor and chair. Have a great evening !
<lassulus>
no problem, you too have a good evening!
LysergicDreams has quit [Ping timeout: 276 seconds]
<{^_^}>
[nixpkgs] @Lassulus merged pull request #73566 → nixos/adb: don't make android tools the system-wide mke2fs, lowPrio → https://git.io/Jeo5G
<{^_^}>
[nixpkgs] @Lassulus pushed commit from @dtzWill to master « nixos/adb: don't make android tools the system-wide mke2fs, lowPrio »: https://git.io/Je6RV
<{^_^}>
[nixpkgs] @zimbatm opened pull request #73836 → travis: add pry to the dependencies → https://git.io/Je6RM
gxt has quit [Ping timeout: 260 seconds]
shibboleth has quit [Ping timeout: 260 seconds]
oida has quit [Ping timeout: 260 seconds]
<{^_^}>
[nixpkgs] @nh2 opened pull request #73837 → polkit: Don't build elogind on non-Linux. → https://git.io/Je6Ry
phreedom has quit [Ping timeout: 260 seconds]
oida has joined #nixos
shibboleth has joined #nixos
gxt has joined #nixos
phreedom has joined #nixos
<VaNilLa[m]>
Hi all! Recently, I decided to try out Haskell's miso library, which is powered by GHCJS. The recommended way of installing ghcjs is supposedly through Nix, so I decided to try set up a NixOS VM and play around. What would be the best way of integrating the nix-based dependency on GHCJS into an existing Haskell / Stack project? I'm very inexperienced with Nix, so sorry if this is stupid
markus1189 has quit [Ping timeout: 265 seconds]
drakonis_ has joined #nixos
knupfer has quit [Quit: knupfer]
brano543 has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @fpletz pushed commit from @rardiol to release-19.09 « hedgewars: 1.0.0-beta2 -> 1.0.0 »: https://git.io/Je6EI
<{^_^}>
[nixpkgs] @jonringer opened pull request #73842 → python3Packages.tensorboard: add wheel as runtime dep → https://git.io/Je6EZ
<Nyanloutre[m]>
hello, how do you connect an iscsi target with open-iscsi ? When using iscsiadm it's complaining about a missing config file in `/etc/iscsi`