<instantepiphany_>
I think a system level service is failing due to a missing build dependency (netdata needs iproute2) - whats the quickest way to get the system to build netdata with iproute as an additional build input? I'm sure ive seen this in the docs somewhere but can't find the right page.
gurjeet has quit [Ping timeout: 256 seconds]
<instantepiphany_>
If I am right about this dependency, I already know how to add that dependency to nixpkgs and open a PR, just want to test it first (and have it work in the meantime).
rcshm has quit [Remote host closed the connection]
rcshm has joined #nixos
eadwu has joined #nixos
eadwu has quit [Client Quit]
vk3wtf has quit [Ping timeout: 264 seconds]
jluttine has joined #nixos
<jonreeve[m]>
yorick: Apparently that built it correctly, but now I can't do `import pandas`. In my shell.nix file, I have the line: ` pandas.overrideAttrs(oldAttrs: { doCheck = false; })`. Am I doing that right?
eadwu has joined #nixos
yl has joined #nixos
drakonis has quit [Quit: WeeChat 2.3]
isgy[m] has joined #nixos
sigmundv has joined #nixos
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fh7Cb
aanderse has joined #nixos
rcshm has quit [Remote host closed the connection]
Rusty1 has quit [Quit: Konversation terminated!]
rcshm has joined #nixos
<{^_^}>
[nixos-weekly] @worldofpeace opened pull request #78 → March: Pantheon DE → https://git.io/fh7Wf
yl has quit [Quit: WeeChat 2.2]
yl has joined #nixos
acarrico has quit [Ping timeout: 245 seconds]
yl has quit [Read error: Connection reset by peer]
<oborot>
I got as far as generating the database index of all files in my channel with nix-index, but getting an error
<oborot>
The error says an attribute is missing in one of my packages.
<oborot>
It's a package I'm pretty sure I don't need... is it just a matter of removing that package?
orivej has joined #nixos
shibboleth has joined #nixos
<oborot>
It's in the default-v10.nix package.
<oborot>
Looks like it might just be a missing dependency
<oborot>
Nope, that wasn't it.
<oborot>
Let's just edit this nix package manually and see what happens...
<oborot>
Yay, that worked
<oborot>
I'm guessing nix-index is gonna take a while
<ottidmes>
its worth the wait :)
endformationage has joined #nixos
DrLambda has joined #nixos
shibboleth has quit [Remote host closed the connection]
shibboleth has joined #nixos
shibboleth has quit [Remote host closed the connection]
<justanotheruser>
for some reason having xserver.displayManager.sessionCommands = "${pkgs.feh}/bin/feh --bg-scale wallpaperpath.jpg" results in my wallpaper showing up for a second right when I log in but not persisting. If I do that command after logging in from my terminal emulator, the background sticks. Where should I put that command so it always sticks?
bhipple has quit [Remote host closed the connection]
ottidmes has quit [Ping timeout: 246 seconds]
sigmundv has quit [Ping timeout: 246 seconds]
rcshm has quit [Ping timeout: 244 seconds]
joehh has joined #nixos
<joehh>
hello, is there any way I can "chroot" into a newer version of nixos and leave my existing installation available and the default?
<joehh>
I'm running 17.09 and for a bunch of reasons, want to execute stuff with 18.09, but keep my main/default environment at 17.09 for a little longer
<{^_^}>
[nixpkgs] @samueldr opened pull request #55744 → nixos/tests/switch-test: Ensures the test fails on failure → https://git.io/fh7l7
praetorg_ has quit [Read error: Connection reset by peer]
aw has quit [Remote host closed the connection]
spacefrogg has quit [Remote host closed the connection]
aw has joined #nixos
spacefrogg has joined #nixos
eadwu has quit [Quit: WeeChat 2.3]
phizzz has quit [Ping timeout: 258 seconds]
<delroth>
if my package needs a binary only during "make check", is that buildInputs or nativeBuildInputs?
<delroth>
I'd tend to say the former since technically we should be running tests on target not host...
<delroth>
but I have no idea how nix cross compilation actually works
<spacekitteh[m]>
LnL: are you the LnL of LnLGames?
hiroshi[m] has joined #nixos
<oborot>
How do I get out of nix-repl?
<oborot>
What's the interupt seq?
hiroshi[m] has left #nixos ["User left"]
hiroshi[m] has joined #nixos
yl has joined #nixos
joehh has quit [Ping timeout: 250 seconds]
<oborot>
Found it, it's Ctrl + Z
<jackdk>
that will suspend it (`fg` will bring it back). Ctrl+D sends EOF and terminates is
<clever>
that file is from ancient times, 2011, back before nixos and nixpkgs got merged
Ariakenom has joined #nixos
reinhardt has joined #nixos
alex`` has joined #nixos
MinceR has quit [Ping timeout: 258 seconds]
MinceR has joined #nixos
alex`` has quit [Ping timeout: 246 seconds]
schjetne has joined #nixos
zupo_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
ym555 has quit [Ping timeout: 250 seconds]
ilya-fedin has joined #nixos
<ilya-fedin>
Hello, I have problem: Java unable to find OpenGL: `OpenGL: ~~ERROR~~ NoClassDefFoundError: Could not initialize class org.lwjgl.glfw.GLFW`. hardware.opengl.enable = true, java (oraclejdk8) installed via nix-env
lawlesseel has quit [Remote host closed the connection]
mbrock has joined #nixos
lawlesseel has joined #nixos
m0rphism has joined #nixos
arjen-jonathan has joined #nixos
<Yaniel>
you need lwjgl and GLFW in your environment
fractal has joined #nixos
Jetien_ has joined #nixos
__monty__ has joined #nixos
<ebzzry>
In my buildPhase, I’m getting this:
<ebzzry>
/nix/store/5vyv136pqs75pj0b8vcpdyc03dmn9p0n-binutils-2.30/bin/ld: warning: libfddrivers.so.4, needed by lib/libfdscheme.so, not found (try using -rpath or -rpath-link)
<LnL>
but yeah, a file->package entry for each file in every package :)
<ebzzry>
10M is no longer a lot, these days.
<ebzzry>
the good thing is that nix-index works on all platforms.
<srhb>
I honestly don't think it's a lot. Try calculating the lower bound using an estimation of average path length, number of packages, number of files per package.
<srhb>
I think you'll find that the overhead is not terrible.
<__monty__>
All platforms? Even those that are ram starved? I didn't have much luck.
<LnL>
it makes a _lot_ of http requests when building the index
<{^_^}>
[nixpkgs] @orivej-nixos pushed commit from @orivej to master « slirp4netns: fix call »: https://git.io/fh704
<LnL>
__monty__: try: nix-index --requests 10 or some other lower number
<__monty__>
I've given up on keeping an index up to date. Just use ,locate now.
zupo_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<{^_^}>
[nixpkgs] @dtzWill opened pull request #55753 → awesome: drop asciidoc, no longer needed when moved to asciidoctor → https://git.io/fh70u
Ariakenom has quit [Remote host closed the connection]
Ariakenom has joined #nixos
<sb0>
I'm trying to use the src attribute of pkgs.releaseTools.channel, but whatever I am putting there, hydra keeps using the location of the nix expression of the jobset
<sb0>
is that a known problem?
<sb0>
the result is, the contents of the channel is the directory containing the nix expression of the jobset, and not what the src attribute points to
xkapastel has joined #nixos
<sb0>
is that what is supposed to happen? how can I make the content of the channels become what the src attribute points to?
knupfer has quit [Ping timeout: 264 seconds]
<sb0>
the tarball nixexprs.tar.xz has the correct contents
<sb0>
but when doing nix-channel --add and nix-channel --update, the wrong files are downloaded
<sb0>
seems there's some bug with the channel management. removing the channel + nix-collect-garbage + adding the channel back (all on the subscriber) fixed that problem
<Guest11311>
I would expect this config to block incoming requests on ports 80 and 443 since they are not explicitly listed, but I can access them without issue... Am I missing something?
<{^_^}>
[nixpkgs] @dotlambda pushed 3 commits to python-unstable: https://git.io/fh7Er
Guest11311 is now known as ramses_
<boogiewoogie>
hey, I'm looking to patchelf a binary that throws the following error: https://hastebin.com/obuhocavuw.md. from what I understand, the rpath needs to be fixed, is this correct? will I have to collect and add the paths to all of those individually or is there another way?
<ramses_>
It seems to me like these connections should end up in nixos-fw, then get send to nixos-fw-log-refuse and get dropped
Tucky has quit [Quit: WeeChat 2.2]
<sphalerite>
ramses_: sounds right
bpa has quit [Remote host closed the connection]
<sphalerite>
boogiewoogie: not sure, this might be a bit different since that's not the normal message the dynamic linker throws when a library is missing
<ramses_>
sphalerite: but still I can access... So the connections do not actually get dropped
<sphalerite>
boogiewoogie: check if patchelfing one of the libraries in makes the error line go away
<sphalerite>
ramses_: what about the nat table? iptables -S -t nat
<ramses_>
sphalerite: these are production servers. I was assuming that docker was automatically opening the ports but yesterday I took a look at the FW and noticed that actually there are no such rules. So I don't understand at all what
<ramses_>
... what's happening now
yokyolu has joined #nixos
Tucky has joined #nixos
<sphalerite>
ramses_: I'd suspect that the nat table changes the destination address, which results in the incoming packets never hitting the INPUT chain but the FORWARD chain instead
<rauno>
How can i manage multuple routing tables with nixos configuration?
<sphalerite>
rauno: my first guess would be that you can't, at least not elegantly
<rauno>
with networking.iproute2.rttablesExtraConfig i can define additional tables but the rules and routes for specific table ?
<sphalerite>
maybe networkd can do stuff
<ramses_>
sphalerite: I added the output of that command to the same gist. Hmm, ok this is beyond my simple knowledge of iptables, I should read up on that, I guess. I don't see any mention of the public IP of the server though (172.16.0.0/16)
<sphalerite>
but really I don't know
<rauno>
okay, so basically with netowrking.localCOmmands ?
<sphalerite>
rauno: maybe :p
<teto>
rauno: it's not declarative but I use networkmanager hooks to create dynamically routing tables
<sphalerite>
ramses_: yep looks like that's the case
<ramses_>
sphalerite: so you say that before the request hits the INPUT chain, the source address gets rewritten by NAT and the request ends up in FORWARD instead? Which line in the gist is doing this?
<ramses_>
I will read up on this anyway, thanks for the pointer!
<sphalerite>
ramses_: the last ones
<sphalerite>
ramses_: so what you'll want to do is create rules in the FORWARD chain to deny the access
<sphalerite>
that or set docker up to use the userland proxy or not to do the forwarding itself at all
jtojnar has joined #nixos
<ramses_>
sphalerite: but those are specific to this br-* docker interface, right? Why do they match traffic on the enp* interface on which the packets are arriving?
<boogiewoogie>
sphalerite: alright, I'll try! how can I look up the path of a library in nixos? :p
<sphalerite>
boogiewoogie: nix-locate is usually the easiest way
<sphalerite>
,locate libX11.so
<{^_^}>
Found in packages: xlibs.libX11
<sphalerite>
nix-locate is in the nix-index package :)
<boogiewoogie>
sphalerite: thanks!
<sphalerite>
ramses_: there's an ! in there ;)
<Taneb>
sphalerite: ooh, I didn't know it was, like, something I can run
<sphalerite>
:)
<ramses_>
sphalerite: *facepalm* Ok :) Thanks a lot, I think I kind of get it. I'll need to study this NAT business in iptables, clearly
<{^_^}>
[nixpkgs] @FRidh pushed 306 commits to python-unstable: https://git.io/fh7uF
shabius has quit [Ping timeout: 252 seconds]
<{^_^}>
[nixpkgs] @dotlambda pushed 2 commits to python-unstable: https://git.io/fh7ux
<callahad>
How should I go about packaging a plugin for another derivation? I'd like to add WebP support to Imlib2, which means somehow getting a webp.so into lib/imlib2/loaders/. Does this mean I need to override the imlib2 derivation itself?
<{^_^}>
[nixpkgs] @dotlambda pushed commit from @timokau to python-unstable « python.pkgs.entrypoints: simplify tests »: https://git.io/fh7zf
<callahad>
My first attempt (https://gist.github.com/callahad/453f95279f17df36ce9557fe5cec2321) compiles fine, but install predictably fails: install: cannot change permissions of '/nix/store/g5x17k4zy6j2hzhll680kjgyi6dz4jyn-imlib2-1.5.1/lib/imlib2/loaders/': Operation not permitted
<ottidmes>
callahad: not that I ever packaged such a thing myself, but I'd imagine you make it install to $out/lib/imlib2/loaders/ rather than ${imlib2}/lib/imlib2/loaders/. Do you do that installation yourself, or is it doing that itself with the gist you showed? If you do it yourself, try just changing to $out (they should merge folders when installed), otherwise you will have to patch something to configure it to
<ottidmes>
callahad: yep, so makeFlags = "LOADERDIR=$(out)/lib/imlib2/loaders/";
agander_ is now known as agander
polman has quit [Ping timeout: 246 seconds]
<callahad>
Okay, so that builds the derivation correctly and the resulting path in the store looks reasonable (...-imlib2-webp-2015.06.01/lib/imlib2/loaders/webp.so vs ...-imlib2-1.5.1/lib/imlib2/loaders/jpeg.so), now to figure out how to get applications that use imlib2 to see it.
<sb0>
when I add a "git checkout" input to hydra, is there a way the build can know what git revision was checked out? other than by doing a deep clone?
<callahad>
(which is probably more of a general nix question than anything specific to this; I'm happy to poke around a bit)
Soo_Slow has joined #nixos
polman has joined #nixos
<callahad>
(thinking I may have to override buildInputs on those?)
<WilliamHamilton[>
has a `gomp` package recently been removed from nixpkgs?
<ottidmes>
callahad: I am afraid that you might need to change imlib2 after all, if it was just to get the .so file in the lib folder, the $out/lib works, but those applications will look in the hardcoded locations, so you would then indeed need to change their buildInputs, but then I think it is better to just change imlib2 instead
<ottidmes>
WilliamHamilton[: cannot find gomp in my stable, nor with a Google search, only references to it when I search my Nixpkgs locally are disabling libgomp and a steam .deb lib for it
<WilliamHamilton[>
ottidmes: if you search "z3" in pkgs/development/haskell-modules/hackage-packages.nix you will see at the last line
<MasseR>
Hey guys. We just spent something like 3 hours to have a coworkers darwin-machine ready for developing our haskell-based executable, until we realized that our local test environment is on a linux machine on virtualbox
<MasseR>
His darwin based binaries can't be copy-closured to the virtualbox
<MasseR>
We have a linux based build-machine downstairs
<MasseR>
Can we hook that up to his buildchain?
rprije has quit [Ping timeout: 245 seconds]
<MasseR>
can we somehow specify the architecture for the builds?
<callahad>
Is there an easy way to pretty print a .drv?
<srhb>
callahad: nix show-derivation
<callahad>
<3 thanks
<callahad>
Oh, looks like that's covered in Nix Pills. I need to re-read that.
xkapastel has quit [Quit: Connection closed for inactivity]
joehh has quit [Ping timeout: 246 seconds]
juliendehos has joined #nixos
<boogiewoogie>
okay, I did 'patchelf --set-rpath /nix/store/cvs6xv383qadls36r3nszwqmk9a81v0v-alsa-lib-1.1.6/lib/libasound.so.2 program', still gives me the same error as before: https://hastebin.com/obuhocavuw.md :(
<ottidmes>
boogiewoogie: if you could paste the whole expression, then others might be better able to help
<boogiewoogie>
it's a steam game, so I'm coming from here https://nixos.wiki/wiki/Steam. steam-run throws the same error, so I'm looking at the "patchelf" and "using LD_LIBRARY_PATH" steps
<boogiewoogie>
ottidmes: oh, it's no expression unfortunately :(
joehh has joined #nixos
<joehh>
thanks sphalerite
<joehh>
I ended up deleting the record from the deployments.nixops file
<{^_^}>
[nix] @edolstra merged pull request #2579 → nix-store: make --dump-db take a list of paths to dump → https://git.io/fpdVd
<callahad>
ottidmes: Okay, running up against a conceptual wall. Can you give me a hint as to how you'd think about writing the patched imlib2 derivation? I somehow need to combine the output of the original imlib2 derivation and my own imlib2-webp derivation into a single thing. But imlib2-webp needs imlib2's headers, which seems to make everything circular.
rprije has joined #nixos
reinhardt has joined #nixos
<callahad>
I was thinking I could maybe do something in a postInstall hook for imlib2, but that doesn't seem right. I think what I want is a union of the output of imlib2 and imlib2-webp
<ottidmes>
callahad: An union is possible with buildEnv, but I think in this case you just want to merge imlib2-webp into imlib2, maybe by default, or if there is any reason to make it optional, behind an enableWebp or something
<dramforever>
callahad: Maybe you want something like buildEnv { name = "my-imlib2-0.2.whatever"; paths = [ imlib2-webp imlib2 ]; } ?
<ottidmes>
callahad: nice! for me it was the barrier to entry, only once I made my own package was I confident enough in it to fully go NixOS
<ottidmes>
callahad: the buildEnv solution probably works if you have an overlay, you could just overwrite imlib2 with your combined buildEnv package, but then you have to learn about overlays too (good to know about anyways)
<{^_^}>
[nixpkgs] @dotlambda pushed 3 commits to python-unstable: https://git.io/fh7g7
ng0 has quit [Ping timeout: 256 seconds]
acarrico has joined #nixos
<ottidmes>
callahad: ah, I just checked, I would go with https://github.com/NixOS/nixpkgs/blob/master/pkgs/applications/audio/audacious/default.nix the buildEnv is not really appropriate for libs, its OK for applications, because it would not require massive rebuilds (i.e. nothing would depend on the app itself, unlike with a lib). An example of an application doing something plugin based is:
<callahad>
I think I understand overlays (I'm at Mozilla, so I've been doing a lot of reading about our overlay for Firefox, Rust, etc.)
<callahad>
And I had to override the sedutil derivation with a patch so my laptop would resume from sleep properly... so I'm getting there :)
<callahad>
ottidmes: Thank you so much for those examples
<ottidmes>
I have a few sedutil things I still need to make PRs for
rauno has quit [Ping timeout: 246 seconds]
<ebzzry>
callahad: because of that, I may have to add an introduction section about overlays in my introductory article.
asymmetric has quit [Ping timeout: 250 seconds]
init_6 has joined #nixos
<ebzzry>
callahad: including handing patches, and things like that
<ebzzry>
*handling
<callahad>
ebzzry: Hey! I ran across your website and the Nix introduction post last night. I have it bookmarked for further reading. I really liked the typographic design of the site, too. Great work :)
jtojnar has quit [Read error: Connection reset by peer]
jtojnar has joined #nixos
iyzsong has joined #nixos
<joepie91>
samueldr: was there any resolution on the steam-run + pango fonts issue a few days ago? I'm running into the same problem with the Terasology launcher
<joepie91>
I don't really care about the icon issues, but it's difficult to pick a folder if you can't read text because it's all blocks :D
<samueldr>
yeah, not my issue, wouldn't know, maybe look into figuring out what it tries to load using a good dose of strace and add it to the FHS chroot?
<joepie91>
right, so that issue was never resolved 'properly' that you know of?
<joepie91>
also, a quick search around suggests that it might not be finding a pango.modules file
<joepie91>
which seems like more than just a missing dep
<samueldr>
oh, the rendering thing itself can't be loaded, I guess no amount of fonts will help
<joepie91>
samueldr: had a look at the ardour package, which I know renders fonts correctly; does no pango-specific things as far as I can tell, just includes pango as a dependency, nothing special about the call signature in all-packages.nix either
<joepie91>
so it really does seem specific to chroots..
<joepie91>
do we have any packages outside of steam-run and appimage-run that use chroots?
<joepie91>
(and/or how do I find those)
<joepie91>
oh, a search for buildFHSUserEnv yields results
rauno has joined #nixos
rprije has quit [Ping timeout: 258 seconds]
rauno has quit [Read error: Connection reset by peer]
<samueldr>
joepie91: gut feeling; could it be glib, through libgmodule that allows loading of modules, and it would resort to pango.modules if that isn't available? (I don't know much GTK)
<joepie91>
I know very little GTK as well, so no clue, tbh
<samueldr>
it *looks* like, stracing ardour, it would load libgmodule using LD_LIBRARY_PATH; which is set in the wrapper script
<sphalerite>
Can anyone recommend a javascript pretty-printer in nixpkgs?
<joepie91>
sphalerite: I usually just use jsbeautifier in-browser :P
joehh has joined #nixos
<joepie91>
samueldr: the command I'm running for Terasology is this, btw: steam-run `which java` -jar ./TerasologyLauncher.jar
<sphalerite>
joepie91: that works, thanks!
<joepie91>
samueldr: executed from within lib/ in the ZIP that you get from the site
<{^_^}>
[nixpkgs] @dotlambda pushed 2 commits to python-unstable: https://git.io/fh7aw
johanot has quit [Quit: WeeChat 2.2]
<joepie91>
samueldr: oh, to clarify, I have java installed in systemPackages
asymmetric has joined #nixos
<sicklorkin>
If I use 'localhost' in my config; I am able to connect to my database on one machines but on the other machine it won't connect; however 127.0.0.1 does work. I'm running inside nix-shell and am wondering what underlying thing (resolvconf?) is inconsistent on these machines?
<sicklorkin>
gethostbyname maybe?
<dramforever>
firewall?
<sicklorkin>
no fw
<sicklorkin>
ping localhost returns 127.0.0.1
<sicklorkin>
it's when I use localhost as the host in a connection string
<symphorien>
Does /etc/hosts mention localhost ?
xkapastel has joined #nixos
<sicklorkin>
yes, both machines have an entry
<tilpner>
sicklorkin: localhost can be ipv6, 127.0.0.1 is reliably ipv4
<tilpner>
See getent ahosts localhost
joehh has quit [Ping timeout: 272 seconds]
<sicklorkin>
tilpner: both produce the same output
<sicklorkin>
`::1 STREAM localhost`
<tilpner>
Both what? What commands are you comparing the output of?
<sicklorkin>
` getent ahosts localhost | grep local
<sicklorkin>
::1 STREAM localhost
<dramforever>
sicklorkin: Just checking: by default nixos has a firewall that blocks incoming connections. You got rid of that?
<dramforever>
blocks by default
fendor_ has joined #nixos
<sicklorkin>
You're gonna slap me when I tell you this.. I'm not running nixOS, i'm running tests inside nix-shell. trying to get my CI environment ready
<dramforever>
Eh
<dramforever>
I need to slap myself
<sicklorkin>
i've check and the database it's listing on tcp6/4
<sicklorkin>
the odd thing is, before I was always using localhost in my config, but it wasn't until I started using nix-shell to build run tests that I ran into this issue..
reinhardt has quit [Quit: Leaving]
<sphalerite>
sicklorkin: mysql has a really stupid behaviour where it will try to connect via the unix socket if you pass localhost — but not if you pass a corresponding IP address
<sicklorkin>
i'm fine to use ipv4 (127.0.0.1) but I just wondered if this could possibly be a bug in any nix related code.. if not,
<sphalerite>
sicklorkin: I'm guessing it's a case of that.
<sicklorkin>
sphalerite: not using mysql but mssql may suffer from this as well..
<sicklorkin>
in any case I can connect to localhost if I'm not using nix-shell
<sicklorkin>
weird.
<sicklorkin>
something seems borked.. anyway.. thanks for helping
<sphalerite>
you could try diffing the output of `env` inside and outside the nix-shell and copying values one by one until you get the behaviour outside the nix-shell
<sphalerite>
if you really want to get to the bottom of this ;)
<sicklorkin>
5:53am.. hummm
asymmetric has quit [Remote host closed the connection]
asymmetric has joined #nixos
Profpatsch has quit [Quit: WeeChat 0.4.3]
<{^_^}>
[nixpkgs] @edolstra pushed 2 commits to release-18.09: https://git.io/fh7Vm
<{^_^}>
[nixpkgs] @oxij opened pull request #55764 → nasm: build from source → https://git.io/fh7Vs
pinkypie has quit [Remote host closed the connection]
obfusk has joined #nixos
<siers>
manveru, hi! have you seen "could not find gem 'ruby (~> 2.3.3.0)' in the local ruby installation."? ruby -v shows up the right one and that's the one that's in the Gemfile
<attente[m]>
hi, does anyone know why the sage package is running 8 hours worth of tests when installing?
<manveru>
siers: ruby is not a gem
mmlb has quit [Ping timeout: 246 seconds]
<siers>
manveru, I wouldn't say otherwise. I'm supposed to use -l or -m?
<tilpner>
attente[m]: You wouldn't want to install a possibly-broken sage, right? The tests help make sure that you don't
<siers>
that's within a zsh loaded with shell.nix's environment that has the 2.3.3 ruby, bundix, but no bundlerEnv
WhittlesJr has joined #nixos
<manveru>
what's the Gemfile?
<siers>
I ran bundix -m and now bundix -l worked… :( and I have no idea why
<manveru>
:P
<manveru>
that's why it's called --magic
<siers>
it's a heisenburger
<attente[m]>
tilpner: how do you disable the tests?
elibrokeit has joined #nixos
<tilpner>
attente[m]: I think you want to either set doCheck or doInstallCheck to false
<tilpner>
I'm not sure how withDoc interacts with the tests here
<tilpner>
You can set these with .overrideAttrs
mmlb has joined #nixos
<attente[m]>
isn't withDoc for documentation?
<tilpner>
Perhaps, but in the sage expression it says "doInstallCheck = withDoc;"
<tilpner>
Which runs "$out/bin/sage" -c 'browse_sage_doc._open("reference", testing=True)'
<tilpner>
And I don't know if that testing=True takes a long time
waleee has joined #nixos
<{^_^}>
[nixpkgs] @Lucus16 opened pull request #55766 → nixos/quassel: Add support for certificate file → https://git.io/fh7rJ
<attente[m]>
sage-tests seems to pass in a `--long` parameter to the tests by default
dermetfan has quit [Ping timeout: 272 seconds]
kisonecat has quit [Ping timeout: 245 seconds]
init_6 has quit [Ping timeout: 245 seconds]
drakonis has joined #nixos
xok has left #nixos [#nixos]
dramforever has quit [Ping timeout: 250 seconds]
endformationage has joined #nixos
<{^_^}>
[nixpkgs] @dotlambda pushed to python-unstable « python.pkgs.virtualenv: rebase virtualenv-change-prefix.patch »: https://git.io/fh7r8
<srk>
,overlay mikky
<{^_^}>
mikky: Overlays look like `self: super: { foo = ...; }`. Use the self argument to get dependencies, super for overriding things and library functions (including callPackage). More info: https://nixos.org/nixpkgs/manual/#sec-overlays-install
kisonecat has joined #nixos
<mikky>
srk: thx
DrLambda has quit [Ping timeout: 268 seconds]
<freusque>
Is there a way to build a package outside the store? I would like to specify both the build and output directories ideally.
asymmetric has quit [Quit: Leaving]
<Taneb>
Where does nix.conf end up on a single-user install?
<freusque>
~/.config/nix/ Taneb
<gchristensen>
/etc/nix/nix.conf or maybe $XDG_CONFIG_DIR/nix/nix.conf (I thinK)
<tilpner>
nix.conf(5) has more details
<tilpner>
With Hydra I still need to apply a random PR for git submodules, right?
<tilpner>
I tried cp ${nix-prefetch-git}/bin/* $out/bin/ in postInstall
<tilpner>
But that fails with errors that tell me nothing
knupfer has quit [Quit: knupfer]
<tilpner>
"path '"url":' is not in the Nix store at /nix/store/8mw4awjykvwz9y2gfia5gambvxp529lm-hydra-2019-02-01/libexec/hydra/lib/Hydra/Plugin/GitInput.pm line 219."
<{^_^}>
#21119 (by fubarbaz, 2 years ago, closed): error: cannot download flash_player_ppapi_linux.x86_64.tar.gz from any mirror
<tilpner>
attente[m]: Oooh, I think you're overwriting all of latest
prooftechnique has joined #nixos
<tilpner>
attente[m]: That expands to latest = { rustChannels = { nightly = { ..., which removes all of the other elements in latest added by the mozilla overlay
<tilpner>
It works in modules due to the merging magic, but not here
<attente[m]>
ahhhhh
<jakov>
{^_^}: I guess it needs to be reopened.
<attente[m]>
tilpner: ok, thank you! i never would've figured it out without your help!!
<attente[m]>
i'll try it, thanks!
hedning has quit [Remote host closed the connection]
hedning has joined #nixos
<jakov>
How could i go along to fix chromium 'enablePepperFlash' ?
cswl has joined #nixos
Makaveli7 has joined #nixos
<tilpner>
You can't fix it locally, AFAICT
<tilpner>
I'm building it from master, to see if it's still broken
xkapastel has quit [Quit: Connection closed for inactivity]
<tilpner>
Looks fine
dermetfan has joined #nixos
bgamari has quit [Ping timeout: 258 seconds]
<srhb>
jakov: I think I already merged someone's fix for that
<tilpner>
Ahh, yes
<srhb>
jakov: But next time it happens (and it does, often) you need to update the url and the sha :)
<tilpner>
32.0.0.114 -> 32.0.0.142
<srhb>
Er, maybe it wasn't me. Could have sworn..
<tilpner>
jakov: It appears to be in nixos-unstable-small, if you want to try it
bgamari has joined #nixos
<tilpner>
Nevermind, nixos-unstable-small does not appear to have chromium caches
kisonecat has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @dotlambda pushed 2 commits to python-unstable: https://git.io/fh76U
<{^_^}>
[nixpkgs] @swflint opened pull request #55769 → z3: Patch file to get rid of python error → https://git.io/fh76m
<WhittlesJr>
Hey, just wanted to say that NixOS is amazing and you guys are all heroes. /love
<gchristensen>
<3 <3 <3
<tilpner>
WhittlesJr++
<{^_^}>
WhittlesJr's karma got increased to 1
yokyolu has joined #nixos
<jakov>
tilpner: Thank you for your quick answer. I am currenlty still building the chromium package from the nixpkgs github repository, master branch, using 'nix-build -A chromium'. I guess that this won't fix the issue, since i did't specify it to build with pepperFlash enabled. How would i do that? And Where did you get the version numbers from?
<yokyolu>
can i make nix-build only fetch bin caches but not build ?
jrddunbr has joined #nixos
<tilpner>
jakov: I got them from pkgs/applications/networking/browsers/chromium/plugins.nix
<wedens>
can openvpn be used regularly (without NM openvpn module) when network is configured via NM?
<{^_^}>
[nixpkgs] @dtzWill closed pull request #55734 → quick patch to bump BINPRM_BUF_SIZE from 128 to 1024 → https://git.io/fh7G9
rabbitz has joined #nixos
rabbitz has quit [Client Quit]
rcshm has joined #nixos
<jrddunbr>
Hi. I'm trying to install NixOS on a cluster connected by Infiniband. What is a good place to start for finding what packages I will need to write into my configuration? I was able to do the base installation just now, but I am not sure what would be the top level package I need.
<jrddunbr>
For example, I can find libibverbs, but I know that's not all I will need.
grw has quit [Ping timeout: 252 seconds]
<jrddunbr>
Can someone point me in the right direction?
<srhb>
wedens: iirc all you need for that is the relevant nm openvpn package in systemPackages.
<clever>
,locate libibverbs
<srhb>
wedens: It's a runtime thing.
<{^_^}>
Found in packages: rdma-core
<clever>
jrddunbr: looks like that may be in rdma-core
<clever>
jrddunbr: but a local search claims its in a package called libibverbs
<jrddunbr>
Hmm. Would that be all I need then? I may need some mlx4 driver stoo?
<srhb>
wedens: Actually, there's a networking.networkmanager.packages nowadays..
<jrddunbr>
clever: ^^
<wedens>
srhb: isn't it the same as configuring openvpn connection via NM?
Tucky has quit [Quit: WeeChat 2.2]
xkapastel has joined #nixos
<srhb>
wedens: Is that not what you asked for? Maybe I misunderstood :)
<srhb>
Oh, you didn't mean module as in "NixOS module"
<wedens>
yeah
<srhb>
wedens: Yes, you can sudo openvpn your.ovpn
<wedens>
I want to have regular openvpn config, but my interface is configured via NM
<srhb>
Yeah, that's fine.
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<srhb>
Well, as fine as NM and OpenVPN can be.. :-)
<jakov>
tilpner: Eh, i mean, if they change again, where to find the new version number ?
<wedens>
srhb: I'm not a big fan of NM. I've used netctl in Arch, but I can't find something similar in nixpkgs.
<tilpner>
jakov: Probably on some Adobe website
<srhb>
wedens: Me neither. I used that intel thing for a while, I forget its name.
<wedens>
iwd
<srhb>
Yeah, thanks.
<srhb>
wedens: fwiw using network manager through nmtui helps me block out the fact that nm is horrid most of the time.
<wedens>
srhb: yeah, I use nmcli-rofi
winem_ has quit [Ping timeout: 244 seconds]
yokyolu has quit [Ping timeout: 246 seconds]
<Baughn>
So the NixOS znapzend module has a startup/reload script that's O(n) in the number of datasets being backed up, and also slow; O(2 seconds) per dataset.
<Baughn>
I finally reached the point where it times out on startup.
<Baughn>
Well, this sounds like something I can fix.
<Baughn>
clever: By the way, remember my NixOS test issues? It turned out to be a Linux kernel bug. Yeeeah.
<pbb>
How can I customize some of the files in xkeyboard-config without recompiling half nixpkgs on every change?
<Baughn>
Without digging into the details of how that package works, and possible alternatives...
<srhb>
pbb: If that is sufficient for your use case.
<bbigras>
It seems rustPlatform.buildRustPackage rebuilds every dependencies every time I run nix-build. Any way around that? samrose ottidmes maybe?
<srhb>
pbb: (You need to push the change away from the root of the dependency tree)
<pbb>
looks like I will create my own keyboard layout
<Baughn>
pbb: There is a way to override the contents of a package without recompiling anything that depends on it, by rewriting the references. It's not considered to be a good idea, and I can't seem to find the documentation, but if you absolutely need it it's possible.
<symphorien>
bbigras: use carnix instead
<Baughn>
The manual I read about it in suggested doing so in case of critical security bugs.
<pbb>
that actually sounds like what I want to do, on other distributions I would just go and edit the packaged file
<bbigras>
symphorien: I'll try. Thanks!
<pbb>
but since it's so undocumented I'll try creating the custom xkbmap first
<srhb>
Baughn: Sounds like replace-dependency.
Ariakenom has quit [Ping timeout: 268 seconds]
<infinisil>
bbigras: (Hello, I'm the one who wrote that znapzend thing)
<Baughn>
infinisil: Did you want to ping me?
<infinisil>
Baughn: ^
_kwstas has joined #nixos
<infinisil>
Yes :2
_kwstas has quit [Remote host closed the connection]
realrokka has quit [Read error: Connection reset by peer]
<infinisil>
Nice
simukis has quit [Quit: simukis]
neminis has joined #nixos
<Baughn>
infinisil: `znapzendzetup import` is slow because each zfs set operation requires a txg sync -- that's behaviour we depend on elsewhere to force a sync before shutdown.
<Baughn>
But there's no limit on how many set operations can be included in one sync.
grw has joined #nixos
<infinisil>
Ah I see
ddellacosta has joined #nixos
realrokka has joined #nixos
<srhb>
tilpner: If you limit your requirements a bit further, you can query hydra for the latest succesful build of a certain job in a certain jobset in a certain project.
<srhb>
tilpner: From there, you get an output path to play with.
<tilpner>
srhb: But I always have to manually retrieve the output and plug that into nix-store?
<srhb>
tilpner: Yes.
<tilpner>
Hmm. Does Hydra create symlinks roots for the latest eval somewhere? Maybe I can point nginx at them
<Baughn>
infinisil: Do you want to PR this, or should I?
<Baughn>
...I don't have my workstation set up yet. Just moved. So it might be a while.
<infinisil>
Baughn: Feel free to, but if you want me to do it I can
<Baughn>
Well, see above.
<Baughn>
Hmm. I wonder if I can get the github website to accept a patch?
<Baughn>
..nope.
nschoe has quit [Quit: Program. Terminated.]
<infinisil>
Baughn: Alright I'll do it
<Baughn>
👍
<neminis>
LnL: i figured what happened with google test, somehow `ar` create an archive that clang++ cannot link against later on, so i could not use the official pkg, i am not sure why though
drakonis has quit [Quit: WeeChat 2.3]
<LnL>
neminis: sorry? don't remember anything recent about google test
<neminis>
LnL: yesterday i was struggling with linking a project to google test, all was set properly except clang++ cannot link against this libgtest.a file for some reason
<neminis>
LnL: don't bother i was just saying in case you wanted to know
<LnL>
hmm, .a?
<LnL>
that's a static lib
<{^_^}>
[nixpkgs] @Infinisil opened pull request #55771 → nixos/znapzend: Run znapzendzetup import in parallel → https://git.io/fh7i9
<infinisil>
Baughn: ^
Jetien_ has quit [Ping timeout: 250 seconds]
<neminis>
yes
alex`` has quit [Quit: WeeChat 2.3]
<LnL>
what's the error, I can't recall
ambro718 has joined #nixos
justanotheruser has quit [Ping timeout: 250 seconds]
<neminis>
i followed the tutorials on GoogleTest github docs, and basically you have to link at compile-time with libgtest.a, but it does not work if `ar` is used to make an archive of the .o file built when compiling googletest
<yl[m]>
is it possible to write a derivation that uses `cp` and `patch` entirely based on builtins (without nixpkgs)?
<infinisil>
yl[m]: Probably not, why?
<tilpner>
yl[m]: You can use "${(import <nix/config.nix>).coreutils}/bin/cp"
<gchristensen>
(config.nix isn't guaranteed to be a stable API, but I do that wackiness too)
<yl[m]>
infinisil: I have an issue that getting me somehow into an infinite recursion when I pass pkgs down to a derivation that patches nixpkgs or home-manager. Let me push what I have to get you a link
jakov has left #nixos ["ERC (IRC client for Emacs 26.1)"]
<infinisil>
yl[m]: You can usually just import nixpkgs once again to avoid infinite rec: `(import <nixpkgs> {}).foobar` instead of `pkgs.foobar`
bigvalen has quit [Ping timeout: 272 seconds]
<tilpner>
But then your derivation can't go into nixpkgs
<yl[m]>
infinisil: that's what I'm currently doing, but isn't that wasteful?
infinisil has quit [Quit: Configuring ZNC, sorry for the joins/quits!]
infinisil has joined #nixos
nikivi has joined #nixos
<Twey>
Now that we have `python.withPackages`, should we stop using propagatedBuildInputs in our Python packages?
q6AA4FD has joined #nixos
gagbo has joined #nixos
<{^_^}>
[nixpkgs] @offlinehacker opened pull request #55773 → rippled-validatior-keys-tool: init at d7774bcc → https://git.io/fh7Pt
<infinisil>
yl[m]: Yeah I don't know of a way to avoid it there
<yl[m]>
infinisil: I'll just import it again for now, thanks!
<infinisil>
You'd need like a fixed pkgs argument in the module that doesn't change with setting nixpkgs.overlays
<Twey>
It's quite awkward that a bunch of packages wrap and set PYTHONPATH — means you can't invoke Python from inside those things
<Twey>
But I'm not sure how to combine buildPythonPackage with python.withPackages
<clever>
Twey: i recently ran into the problem that nix-shell with nixops in the inputs, sets PYTHONPATH to look at nixops, by force
<clever>
Twey: if you then run a totally different nixops via ./result/bin/nixops, it uses the source of that 1st one!
<clever>
so the 2nd nixops your trying to debug, ignores every single change
rotty has joined #nixos
<Twey>
clever: Yeah, I've run into that too :(
<Twey>
clever: Currently having similar problems with Ansible
Thra11 has joined #nixos
<tilpner>
How would a systemd service use the latest successful hydra build? It runs on the Hydra machine, everything's already in the store. Does it need to call the API, or does Hydra maintain a profile/symlink directory somewhere?
<tilpner>
(I could have another service building that symlink directory by using the API, but that seems wrong)
timon37 has joined #nixos
<Twey>
clever: I think that issue actually killed NixOps at my company :(
<Twey>
I wasn't the one playing with it but we were hoping to use it for a bit, but it didn't pan out
kisonecat has joined #nixos
sigmundv has quit [Ping timeout: 244 seconds]
kisonecat has quit [Ping timeout: 258 seconds]
Ariakenom has joined #nixos
<{^_^}>
[nix] @shlevy pushed to master « nix.sh: Be set -u compliant. »: https://git.io/fh7PK
Olgierd has quit [Quit: WeeChat 1.9.1]
Olgierd has joined #nixos
hedning has left #nixos [#nixos]
<eyjhb>
Does anybody have experience with nixos and charging thresholds ?
hedning has joined #nixos
<tilpner>
What's the problem you're asking for?
Olgierd has quit [Quit: WeeChat 1.9.1]
<eyjhb>
tilpner: basically if it works... :p On my old Ubuntu system, I used tpacpi-bat, but wondering if tlp on nixos supports it
<eyjhb>
Or if tpacpi-bat works at least
Olgierd has joined #nixos
<tilpner>
Like START_CHARGE_THRESH_BAT0 and STOP_CHARGE_THRESH_BAT0?
ddellacosta has quit [Ping timeout: 272 seconds]
<tilpner>
Those work, according to someone who unintentionally activated them
fusion809 has quit [Remote host closed the connection]
fusion809 has joined #nixos
slack1256 has joined #nixos
<{^_^}>
[nixpkgs] @LnL7 opened pull request #55777 → haskellPackages.conduit-extra: fix darwin sandbox build → https://git.io/fh71e
mounty has joined #nixos
<jomik>
hat lockscreens/screensavers are people using? I am pondering what I can set up for XMonad.
<samrose>
bbigras: I was going to be looking into carnix next
<jomik>
s/hat/What
<tilpner>
jomik: i3lock or slock
<averell>
i3lock, works fine with xmonad for me.
<{^_^}>
[nixpkgs] @FRidh pushed 2 commits to python-unstable: https://git.io/fh71v
<rotty>
I'm having trouble getting nix set up: `nix-channel --update` --> while setting up the build environment: executing '/bin/bash': No such file or directory
<bbigras>
samrose: nice.
<samrose>
bbigras: (I definitely do not want to be rebuilding my deps each time either)
<{^_^}>
[nixpkgs] @dylex opened pull request #55778 → sympow: fix patch url to working revision → https://git.io/fh71J
<jomik>
Oooh, I did ponder using i3lock - there is a wrapper that lets us use a blurred screenshot of the desktop before the lock, right?
<jomik>
I'll try that then, thanks.
<samrose>
bbigras: were you every able to get rust overlay working with nix-build?
<rotty>
(This is using nix 2.2.1, built from source, running on a Debian buster amd64 installation)
<ikitat>
What's a good strategy for developing private packages against nixpkgs? Something like Nixpkgs Quick Start to Adding a Package, but for something you don't intend to push upstream.
<bbigras>
samrose: yeah otherwise nix-shell would be better for me. I might have been able to get the overlay with nightly working but for some reason it didn't seem to work with a git dependency (in Cargo.toml).
<tilpner>
,overlays
<{^_^}>
tilpner: Did you mean overlay?
<{^_^}>
Overlays look like `self: super: { foo = ...; }`. Use the self argument to get dependencies, super for overriding things and library functions (including callPackage). More info: https://nixos.org/nixpkgs/manual/#sec-overlays-install
<tilpner>
ikitat: Overlays are great for private packages
<samrose>
bbigras: I came close but got an infinite recursion issue.
freeman42x has joined #nixos
slack1256 has quit [Remote host closed the connection]
simukis has joined #nixos
<{^_^}>
[nixpkgs] @LnL7 opened pull request #55779 → WIP stdenv: only set __darwinAllowLocalNetworking on darwin → https://git.io/fh71O
<ikitat>
My goal is to eventually use the package in nixops
<tilpner>
ikitat: That's possible. Personally, I'm too lazy to register every package in an overlay
<tilpner>
I have an overlay that loads packages based on their filename, so I just create foo.nix, and suddenly the package foo is available everywhere
<samrose>
ikitat: I have been defining my packages in in overlay where I am building an over-lain customization on top of nixos 18.09/nixpkgs
<{^_^}>
[nixpkgs] @FRidh pushed 5 commits to python-unstable: https://git.io/fh71B
matijja has joined #nixos
<ikitat>
tilpner: mind sharing that? Do you then reference foo.env in a shell.nix for a shell?
alex`` has joined #nixos
<ikitat>
I feel like I have the shell story for this project working well, but my mental model falls apart when I try to deploy it via nixops
<{^_^}>
[nixpkgs] @LnL7 opened pull request #55781 → pythonPackages.httpretty: fix darwin sandbox build → https://git.io/fh71R
<jomik>
tilpner: Ah, thanks! Why are you saving to /run/user/$UID/screen.png ? Why not mktemp
<tilpner>
jomik: There's no scenario where I want multiple screen.pngs
<tilpner>
jomik: And I don't want other users to be able to access screen.png
<tilpner>
samrose: I think super.callPackage is the same as super.lib.callPackageWith super
<samrose>
thanks tilpner
superherointj has joined #nixos
<ikitat>
tilpner: thank you for the advice
<WhittlesJr>
When submitting a new package with a bunch of new dependency packages, should I have a different PR for each package, or one big PR? If one PR, should each package have its own commit?
fusion809 has quit [Remote host closed the connection]
shibboleth has joined #nixos
fusion809 has joined #nixos
<Yaniel>
one package per commit is a lot cleaner to look at in the commit log
<tilpner>
WhittlesJr: Depends on what packages they are. Are they related? Can they be accepted independently? Are some more controversial than others?
<Yaniel>
and much easier to find when things changed
<slabity>
Aren't commits flattened on merging though?
<Yaniel>
not by default I think+
<Yaniel>
unless nixpkgs specifically has that policy
<jomik>
tilpner: Ah, yeah okay. I guess I figured it'd work the same if you put it in temp? But, I guess this way you just overwrite the old.
<slabity>
pkgdesc="Docker powered mini-Heroku in around 100 lines of Bash."
<slabity>
Kill it with fire
fusion809 has quit [Remote host closed the connection]
<WhittlesJr>
I've identified `gliderlabs-sigil`, `herokuish`, `plugn`, `procfile-util`, and `sshcommand` as new deps. Maybe one or two of those are already packaged under other names?
<tilpner>
jomik: It's also removed after unlocking
cryptomonad is now known as cryptomonad_
<samueldr>
Yaniel: it's expected to see one commit per "change", where change doesn't mean your fixups for oopsies (if any), but e.g. one commit "xyz: init at 1.2.3" or "abc: 1.2.3 -> 1.2.4"
FRidh has quit [Quit: Konversation terminated!]
<tilpner>
WhittlesJr: Quick search says they're all new
<samueldr>
PRs are generally not squashed, and only squashed if there was one commit or one "logical" commit anyway
<WhittlesJr>
So I gather: one PR, with one commit per new package?
<tilpner>
Probably a good idea. Inter-PR dependencies aren't fun
jabranham has joined #nixos
<jomik>
tilpner: why `export PATH`?
<samueldr>
oops, mis-identified the one asking; sorry WhittlesJr, yeah, one PR should do for "main subject and dependents"
<tilpner>
Is your question "why PATH=" or "why export"? The PATH needs to be set to find all the executables I'm calling, like convert and i3lock
<jomik>
Hmm, okay, that makes sense. I guess alternatively you could use `${pkgs.scrot}`, etc?
<tilpner>
That means you don't need to install imagemagick yourself for this script to work
<{^_^}>
[nixpkgs] @xeji pushed commit from @r-ryantm to master « conan: 1.11.2 -> 1.12.0 (#55335) »: https://git.io/fh7MH
ym555 has joined #nixos
<laas>
how do I solve `error: Nix daemon out of memory`?
<laas>
it happens when I nix-shell
aanderse has joined #nixos
<laas>
the nix expression is just 64-bit wine
<clever>
laas: are you doing src = ./.; ?, what args did you give nix-shell?
<clever>
laas: how much ram on the system?
<laas>
I have 8 GB
<laas>
and yeah I am doing src = ./.
<clever>
laas: how big is . ? (du -h)
<laas>
oh does it read the entire source tree or something?
<laas>
oof it's big
<tilpner>
Wasn't this fixed with newer Nix versions?
<clever>
yeah
<laas>
2.3G
<clever>
laas: nix-store --version
<laas>
is there a way I can prevent it from doing that?
Glider_IRC_ has quit [Ping timeout: 240 seconds]
<laas>
2.2
<clever>
laas: maybe use an override in shell.nix to just src = null;
<laas>
yeah I'll do that..
Ariakenom has joined #nixos
alex`` has joined #nixos
<laas>
is there a technical reason nix-shell can't ignore `src`?
<laas>
thanks btw
<clever>
laas: mostly that src = ./.; happens at eval time, and nix-shell cant easily mutate that region
<clever>
laas: 64bit wine is also already packaged in nixpkgs
requisitebits has quit [Quit: leaving]
requisitebits has joined #nixos
bgamari has joined #nixos
matijja has joined #nixos
<ddellacosta>
folks, I'm going through the nix pills and I'm on #9, and I'm a bit confused by the relevance of NAR archives here--for example, in 9.3 it talks about dumping the derivation as a NAR, then examining that for runtime dependencies. However when I follow the steps below involving nix-instantiate and nix-store, I don't see anything NAR-related. Can someone help me understand better why it's relevant to discuss NAR
<ddellacosta>
is that simply describing what nix-store is doing behind the scenes when you call it with -q and --references, perhaps?
<clever>
tilpner: yeah, nix does search the nar form of a path forthehashes, at the end of a build
<tilpner>
I always just assumed it scanned the store path itself, file by file, but if you had a streaming and fast NAR exporter... maybe this works too
zupo has joined #nixos
<clever>
behind the scenes, nix will basically do `nix-store --dump $out | egrep "foo|bar|baz" -o` to check for deps
<clever>
and it will search for the paths that are in the build-time closure (the inputs)
<clever>
it then saves that to db.sqlite
<clever>
and --references is just a query against db.sqlite
<ddellacosta>
got it, okay, that's basically what I was assuming and trying to confirm, thanks tilpner and clever
vk3wtf has quit [Ping timeout: 268 seconds]
<ddellacosta>
I think it's extra confusing in that particular doc because you don't seem to actually produce any NAR file yourself if you follow the steps, so it's helpful to understand that is what nix-store is doing (generating the NAR file and grepping it)
<ddellacosta>
also I have no idea if it would be considered good or not but I find this C++ pretty readable, which is something considering I haven't touched C++ for more than a decade
<clever>
tilpner: this lets me bake the init=/nix/store/foo/init into the file command-line
_kwstas has joined #nixos
<clever>
tilpner: but, because i discarded all context, i dont depend on /nix/store/foo at build-time
<clever>
and runtime deps are a subset of that, so i dont depend on it at runtime either
<clever>
(and a copy is already inside the squashfs on line 118)
<tilpner>
I needed exactly that for my squashfs generation too
<tilpner>
Will remember this for next time I work on that
<clever>
tilpner: one thing to keep in mind, is that if you use unsafeDiscardOutputDependency at the wrong time, a lack of nix sandboxing can hide problems, because the path still exists
<clever>
so having the nix sandbox on, forces the path to "not exist" and expose such mistakes
<tilpner>
I always have sandboxing on
<clever>
as for what it does, every single string in nix has a context list on it
<clever>
when you later pass that string to builtins.derivation, it will collect the context for all of its strings, and that new derivation depends on those things
o1lo01ol1o has quit [Remote host closed the connection]
<clever>
elvishjerricco: by attrpath, not really possible, but by name, you can, it could also help to check by attrs on its derivation, but it depends heavily on the context, got a path as an example?
rpg has joined #nixos
boogiewoogie has quit [Quit: Leaving]
dermetfan has joined #nixos
<gchristensen>
clever: ok your arm binfmt trick was cute
<gchristensen>
clever: but I have your next challenge when you're ready
superherointj has quit [Quit: Leaving]
<clever>
gchristensen: ?
<gchristensen>
have it actually execute the code on an arm device
<clever>
ah
<clever>
i have mixed qemu-user and real arm build slaves before
<clever>
main problem was building v6 binaries, because v7 opcodes would leak in and poison the hydra
<elvishjerricco>
clever: I'm trying to use `system.replaceRuntimeDependencies` to replace utillinux with utillinuxMinimal, but for some reason the activation script is using some completely different utillinux.bin (`/nix/store/18dixsx2kd16vmxv9dwxygljx324f8ai-util-linux-2.33.1-bin`) that I can't find at any attrpath
<elvishjerricco>
Looking at the activation-script.nix file, it just uses `pkgs.utillinux`, so I can't figure out why it's getting a different utillinux
<clever>
elvishjerricco: use `nix-store --query --deriver` to get the .drv for that, and others, then `nix show-derivation` them
<clever>
and maybe `nix-diff` the 2 drv's
<elvishjerricco>
TIL about nix-diff. That sounds useful
dermetfan has quit [Ping timeout: 258 seconds]
Ariakenom has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @xeji pushed commit from @romildo to master « shades-of-gray-theme: 1.1.4 -> 1.1.5 (#55767) »: https://git.io/fh7Da
<tilpner>
clever: Ah, thanks, didn't know that in detail either :)
<{^_^}>
[nixpkgs] @cleverca22 opened pull request #55786 → improve perl shebang lines by switching to `use lib ...;` → https://git.io/fh7Dw
fusion809 has quit [Remote host closed the connection]
knupfer has joined #nixos
hjst has joined #nixos
<clever>
tilpner: the context is also like a virus, if you use string manip to extract just the "/" from "/nix/store/foo", that "/" still has all of the context on it
<tilpner>
D:
<clever>
tilpner: the only way to get rid of context is via unsafeDiscardStringContext
<tilpner>
That makes sense in this context, losing context is worse and keeping it
<clever>
wait, i linked the wrong builtin above, and you named an entirely diff one, lol
<tilpner>
*than
fusion809 has joined #nixos
<clever>
what did unsafeDiscardOutputDependency do... lol
<clever>
tilpner: wait, i was going to say, you have no way to view context ... lol
<clever>
tilpner: that commit also explains why half the context functions are missing
<clever>
they got moved, just yesterday!
<elvishjerricco>
clever: Actually, I think this might be some kind of issue with replaceRuntimeDependencies... The reason I can't find the attr path for this utillinux version is that it is the one I was trying to replace, but with the replacement applied to itself.
<tilpner>
That was January
<elvishjerricco>
`"buildCommand": "$nixStore --dump /nix/store/mbkp2r3ifzrqfy83a1yqln87md0mzg6q-util-linux-2.33.1-bin | sed 's|mbkp2r3ifzrqfy83a1yqln87md0mzg6q-util-linux-2.33.1-bin|'$(basename $out)'|g' | sed -e 's|kq99jm2qrcbd62rm49bvj36n7y7w1av2-util-linux-2.33.1|78l2rzvqk88za6aijydbysvrsdkd56qb-util-linux-2.33.1|g' | $nixStore --restore $out\n"`
o1lo01ol1o has joined #nixos
<tilpner>
But neat, that looks useful too, when it reaches the channels
<phizzz>
if i want to take the default.nix from a not installed nix package to change the version and nix-build it (fetchFromGitHub), how do i get the sha256 from the new version to put in the default.nix?
<emily>
,tofu
<{^_^}>
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.
<haslersn>
Follow up question: Is there a package for Nix (the package manager itself) which I can overlay using a NixOS overlay?
<clever>
elvishjerricco: of note, i think replaceRuntimeDependencies is recursive, and will create new versions of everything, with the paths swapped around
<elvishjerricco>
Maybe utillinux.bin is getting patched with utillinuxMinimal.bin, and then that dependency is being preserved?
<clever>
elvishjerricco: maybe, ive never really used replaceRuntimeDependencies, and i thought minimal was already the default
<phizzz>
heh tofu
<elvishjerricco>
minimal is not the default unfortunately
<phizzz>
thanks emily
<clever>
elvishjerricco: ive run into problems due to minimal being the default, `cal` cant highlight the current date
<elvishjerricco>
though maybe I could just change it so it is :P
<clever>
elvishjerricco: could be that somebody swapped it around, and your going to swap it once more? :P
<elvishjerricco>
clever: Looks that way lol
slack1256 has quit [Remote host closed the connection]
matijja has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @xeji merged pull request #55744 → nixos/tests/switch-test: Ensures the test fails on failure → https://git.io/fh7l7
<{^_^}>
[nixpkgs] @xeji pushed commit from @samueldr to master « nixos/tests/switch-test: Ensures the test fails on failure (#55744) »: https://git.io/fh7yk
ikitat has quit [Remote host closed the connection]
ikitat has joined #nixos
ilya-fedin has quit [Quit: Connection closed for inactivity]
<{^_^}>
[nixpkgs] @cleverca22 pushed 0 commits to fix-perl-shebang: https://git.io/fh7y6
jackdk has joined #nixos
jasongrossman has joined #nixos
mrus has joined #nixos
<mrus>
hey there! I'm trying to build a system with mdadm RAID1, with Luks on top, with LVM on top. Unfortunately, every time I try to boot, it waits for the /dev/mapper/root device and then reports it wasn't found. Is there any trick to get md0 started, luks opened and lvm activated?
<gchristensen>
wow!
_kwstas has quit [Quit: _kwstas]
ym555 has joined #nixos
Makaveli7 has quit [Quit: Leaving]
kisonecat has quit [Ping timeout: 250 seconds]
<clever>
mrus: why both mdadm and lvm?
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<pie__>
does anyone have a good way of dealing with per project tooling integrations with emacs?
<mrus>
clever: I use mdadm for the RAID1 across two SSDs and LVM within the Luks that's on top of md0 to have multiple partitions (e.g. root, swap, ...).
<clever>
mrus: lvm can already do mirroring, so you could just lvm mirror 2 luks devices
UltraPhil has joined #nixos
<ikitat>
If I've defined overlays in a default.nix, is there a way for me to also utilize the overlay from a shell.nix? I was hoping to do something like (import ./default.nix).env but that doesn't work
magnetophon has quit [Remote host closed the connection]
<jackdk>
pie__ what project specifically? I have .dir-locals.el files for most of my haskell projects, so dante knows to use nix-shell and which cabal target
<pie__>
jackdk, yeah im going to start trying to learn emacs for haskell
<pie__>
but i meant the question generally because i figured this might be a provlem?
<mrus>
clever: but then I'd have lvm -> luks -> lvm?
<clever>
mrus: you can just do 2*luks -> lvm -> filesystems
<mrus>
clever: long story short, would like to keep the mdadm -> luks -> lvm setup, as I had it running on other distros before and it usually works very good :)
<clever>
mrus: you would do all of that at the lvm level
<rotty>
is the config for https://hydra.nixos.org/ available somewhere? it would be quite helpful for people who want to play around with their own installation...
<inx0>
Hi everyone! I'm trying to add a file to an existing nixpkgs package. The file needs to be in /nix/store/... I found writeText, but I can't figure out how to use it when overriding an existing package.
<inx0>
I've also read section 6.6.2 of the Nixpkgs manual, but that seems to place files in ~/.nix-profile if I'm not mistaken?
<clever>
inx0: all paths in the store are immutable, writeTextFile always creaes a new /nix/store/foo/
<tilpner>
inx0: You can create a new package and copy the old packages files in there, then add your new file. That costs more storage, but is faster
<clever>
inx0: so you need to apply an override to a package, to modify how its build
<tilpner>
inx0: Or you can modify the package to create your new file at the end, but then you need to rebuild the package every time you change the file, which is slower but doesn't need as much storage (best case)
<{^_^}>
#36883 (by veprbl, 48 weeks ago, open): A derivation to merge multiple outputs
<tilpner>
And I forgot about that, huh
<rotty>
tilpner: thanks
shibboleth has quit [Quit: shibboleth]
<inx0>
wow, y'all are amazing. Thanks a lot! This is a temporary change, so I tried to create the file manually, but of course ran into the read-only file system error.
<inx0>
I'll give symlinkjoin a try. Thanks everyone! \o/
<ikitat>
... then it occurs to me that nix-shell will use ./default.nix if ./shell.nix doesn't exist. Not sure why I was thinking I needed both
<pie__>
hmm kind of makes me thinkits a shameyou cant just copy a store item to a random mutable localdirectory
<pie__>
clever, ^
<pie__>
tilpner, its just a totally random thing a friend told me about while saying "i discovered this the other day and people want it all the time but noone knows about it"
<emily>
aya shameyou
<pie__>
just replace me with a bot that says "maybe you want symlinkmerge"
<tilpner>
pie__: I usually remember it, I used it a lot
<pie__>
it happens :>
<pie__>
still better than when i wasted two days trying to figure out why a python program wasdying with no errors
<pie__>
*was dying
<pie__>
im still salty about that. the file with a shebang i was calling (artifact of a reorganization) wasnt the actual starter, it just had main(). nothing was calling the main() ....
<pie__>
</rant> /me goes back to shaving emacs
jasongro` has joined #nixos
gchristensen is now known as SvenSlootweg
SvenSlootweg is now known as gchristensen
simukis has quit [Quit: simukis]
jasongrossman has quit [Ping timeout: 250 seconds]
jasongro` has quit [Read error: Connection reset by peer]
mek42_laptop has joined #nixos
<NemesisD>
so i've got a derivation and i'm adding it to a shell and to a docker image under contents. where's an appropriate place to attach *runtime* dependencies to a derivation?
<clever>
NemesisD: nix computes runtime deps automatically, based on what storepaths your $out refers to
<NemesisD>
right now i'm just adding an arbitrary runtime-dependencies attr, then in the docker contents i'm doing [mypkg] ++ mypkg.runtime-dependencies;
<NemesisD>
clever: hmm, so this is a haskell package that i'm using callCabal2nix on. so i'm not sure how those would be factored in
<clever>
NemesisD: you could then make a bash script that wraps it, and sets $PATH first
kisonecat has joined #nixos
<NemesisD>
clever: that wraps what? are you talking about providing builder somehow to callCabal2nix?
dramforever has joined #nixos
jasongro` has joined #nixos
cryptomonad_ is now known as cryptomonoid
jasongro` has quit [Client Quit]
<mek42_laptop>
if two packages share deps, as long as those deps are the same version, etc., each package will have a link to the same dep in the store, so only one dep will exist on disk? Am I understanding this right?
dramforever has quit [Remote host closed the connection]
<clever>
mek42_laptop: yep
<mek42_laptop>
ok, so NixOS, in and of itself, doesn't change why one would want (or not) zfs deduplication?
fusion809 has quit [Ping timeout: 250 seconds]
<gchristensen>
don't use zfs dedup
ajs124 has quit [Quit: Gateway shutdown]
<cransom>
mek42_laptop: not really. there's also an optional mechanism you can run against hte store that will look for identical files and hard link them.
<cransom>
and yes, zfs dedup is a performance killer, even when you have the resources for it i've found.
<gchristensen>
and makes for very unhappy users when you hit the limit
<{^_^}>
[nixpkgs] @dotlambda pushed 6 commits to python-unstable: https://git.io/fh7Si
<mek42_laptop>
ok, tyvm - i think I'll give NixOS (without zfs dedup) a spin when my new laptop arrives
<{^_^}>
[nixpkgs] @eadwu opened pull request #55793 → linux: allow for interpreter to be truncated → https://git.io/fh7Sx
romildo has quit [Quit: Leaving]
kreisys has joined #nixos
eadwu has quit [Quit: WeeChat 2.3]
<infinisil>
elvishjerricco: I think I'll merge it now
<infinisil>
elvishjerricco: You tested that it doesn't cause the problem in the issue anymore right?
<elvishjerricco>
infinisil: Yea
<infinisil>
elvishjerricco: So if I understood this right, we nullSrc every selected input, then we get all deps from all of those, nullSrc the deps, and remove all that equal our inputs (which are also nullSrc'd so this works)
<infinisil>
Correct?
justanotheruser has joined #nixos
<elvishjerricco>
infinisil: Yea. So we only check for use of the selected packages as immediate dependencies, not transitive, but that should be sufficient for proper use of `shellFor`
<{^_^}>
[nixpkgs] @Infinisil pushed 2 commits to release-18.09: https://git.io/fh79Z
WhittlesJr has quit [Quit: Page closed]
<elvishjerricco>
matthewbauer[m]: Say I wanted to try to make a super minimal NixOS config that's based on musl (ripping out everything that fails to build). Is `nixpkgs.localSystem = lib.systems.examples.musl64;` the right way to do that?
alex`` has quit [Ping timeout: 245 seconds]
npmccallum has joined #nixos
<elvishjerricco>
Ericson2314: ^
<matthewbauer[m]>
I think that should work! I am not sure if it will actually build a full system though
<elvishjerricco>
matthewbauer[m]: Yea I'm finding a lot of things that break :P
<elvishjerricco>
But that's ok; this system is supposed to be as minimal as possible anyway
<{^_^}>
[nixpkgs] @worldofpeace opened pull request #55794 → atom updates → https://git.io/fh794
<Ericson2314>
@elvishjerricco I'm on my phone but soon going back to computer. But looks like Matt beat me to it :)