<krey>
Myrl-saki: I thought that my channels are just local copies of nixpkgs
<Myrl-saki>
krey: They are. They're local copies of nixpkgs, which are saved to the /nix/store.
<{^_^}>
[nixpkgs] @oxij opened pull request #41088 → Merge master into staging → https://git.io/vhqfq
das_j has joined #nixos
newhoggy has joined #nixos
<das_j>
Hey everyone, I have a package build that is currently running make -j1. How can I increase the number of jobs? It's taking really long this way
<das_j>
nix.BuildCores doesn't solve it
<Myrl-saki>
das_j: --cores
<Myrl-saki>
Also, nix.buildCores, IIRC.
<clever>
das_j: you also have to set enableParallelBuilding = true; on the derivation
<Myrl-saki>
Let me check my configuration.nix.
<Myrl-saki>
# nix.buildCores = lib.mkDefault 4;
<Myrl-saki>
Don't know why I had that disabled.
* Myrl-saki
shrug
<infinisil>
> option "nix.buildCores"
<{^_^}>
"This option defines the maximum number of concurrent tasks during\none build. It affects, e.g., -j option for make. The default is 1.\nThe special value 0 means that the builder should use all\navail...
<angerman>
eacameron, clever: building haskell on x86_64 Linux for deployment, can be done with docker and nix.
<Myrl-saki>
Is that an actual Nix function??
<Myrl-saki>
infinisil:
<krey>
Myrl-saki: but the nixpkgs repo doesn't contain a nixpkgs folder
<das_j>
Myrl-saki: --cores actually solved it
<{^_^}>
[nix] @grahamc closed pull request #1455 → Create a profile suitable for multi-user installs (for master) → https://git.io/vQi6H
<infinisil>
Myrl-saki: Nah
<Myrl-saki>
infinisil: Oh. :(
<krey>
Myrl-saki: it's added after the fact somehow
<das_j>
infinisil: I tried setting it to 0 and 8, but neither helped
<ldlework>
I have my access key in ~/.aws/credentials under [nixops]
<noefk>
in nixops with the none backend I have trouble getting the nodes.*.config.networking.publicIPv4 attributes - works great in AWS, is it supposed to work?
<ldlework>
noefk: maybe you can spot why I can't get going too
justanotheruser has quit [Ping timeout: 245 seconds]
<clever>
ldlework: what is the error?
nur0n0 has joined #nixos
<ldlework>
there's no error
<ldlework>
I showed the output in the gist
<clever>
everything looks normal to me
<ldlework>
all it did was create a keypair thing
<noefk>
idlework: looks fine to me
<ldlework>
it didn't create an actual host
<ldlework>
with an ip and stuff
<clever>
ah
<clever>
let me look over it closer
<ldlework>
noefk: heh you just reminded me of Westworld "Doesn't look like anything to me."
<noefk>
:)
<ldlework>
These reproducable delights will have reproducable ends.
<clever>
ldlework: change the name of your key
<clever>
ldlework: its colliding with the ec2 host
<noefk>
I have my own frustrations with nixops
<ldlework>
oh I see
bbarker has quit [Quit: Leaving]
iyzsong has joined #nixos
blankhart has quit [Ping timeout: 245 seconds]
<ldlework>
Cannot delete this deployment because it still has resources.
<ldlework>
ah destroy
<{^_^}>
[nixpkgs] @peterhoeg closed pull request #41017 → [WIP] nixos: make several modules not run as root → https://git.io/vhkX7
<ldlework>
yay it is working
<clever>
nice
<ldlework>
I wonder what the NixOps vs Cloudformation coverage differentials are.
<ldlework>
Probably substantial
justanotheruser has joined #nixos
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<ldlework>
Waiting for SSH for days.
<ldlework>
Wonder if something went wrong.
stepcut has joined #nixos
<clever>
ldlework: that can happen if you didnt allow port 22 in the firewall
<ldlework>
uh oh
fragamus has joined #nixos
<ldlework>
Shutting down for days. XD
hamishmack has joined #nixos
stepcut has quit [Remote host closed the connection]
<ldlework>
clever: do people use nixops to build docker containers, create nixos hosts on cloud services, and deploy nixos docker containers to nixos ec2 instances?
stepcut has joined #nixos
<ldlework>
Maybe that's what Disnix is for
<clever>
ldlework: i would just skip docker entirely
<reachingfourpeas>
that's the output of the nix-shell command in the channel header
sanscoeur has quit [Remote host closed the connection]
<reachingfourpeas>
Here's the output of nix-shell:
<clever>
ldlework: and if you try wireshark on that port?
<reachingfourpeas>
error: while evaluating the attribute 'shell' at /home/username/clash-playground/release.nix:94:14:
<reachingfourpeas>
while evaluating anonymous function at /home/username/clash-playground/nix/nixpkgs.nix:1:1, called from /home/username/clash-playground/release.nix:13:13:
<reachingfourpeas>
while evaluating anonymous function at /home/username/clash-playground/nix/fetchnix.nix:7:1, called from /home/username/clash-playground/nix/nixpkgs.nix:8:9:
<reachingfourpeas>
while evaluating 'ifThenElse' at /home/username/clash-playground/nix/fetchnix.nix:14:16, called from /home/username/clash-playground/nix/fetchnix.nix:18:1:
sbdchd has quit [Remote host closed the connection]
<reachingfourpeas>
Some background: I'm new to Haskell, starting to learn so I can explore Clash for FPGA synthesis. I was suggested to use the clash-playground via Nix, and I ran into this error
iqubic has joined #nixos
<clever>
reachingfourpeas: run `nix-prefetch-url --unpack` on that URL, then use that as the unpacked hash
<clever>
ldlework: how much ram on this machine?, any swap?
<clever>
ldlework: what does dmesg say?
Neo-- has quit [Remote host closed the connection]
Neo-- has joined #nixos
sir_guy_carleton has quit [Quit: WeeChat 2.0]
Ross has quit [Ping timeout: 268 seconds]
MP2E has joined #nixos
Ross has joined #nixos
<ldlework>
clever: dmesg doesn't say anything when the segfault happens
<ldlework>
oh maybe this
<ldlework>
[ 773.971372] znc[2039]: segfault at 10 ip 00007faed56273b8 sp 00007ffd15fd2dd0 error 4 in libc-2.26.so[7faed55a9000+1a9000]
<ldlework>
799M free memory, zero swap
sir_guy_carleton has joined #nixos
<clever>
ldlework: how much ram available?, free -m
<ldlework>
clever: ~800M
stepcut has joined #nixos
<{^_^}>
[nixpkgs] @lopsided98 opened pull request #41093 → spl, zfs: 0.7.8 -> 0.7.9, unstable to latest → https://git.io/vhqI6
<clever>
ldlework: not sure then
johnw has joined #nixos
drakonis is now known as Drakonis_
<ldlework>
clever: using nixops can I just use a local pem file and write it to the remote host in the right spot?
<clever>
ldlework: nixops scp --to
sanscoeur has joined #nixos
Drakonis_ is now known as drakonis
johnw has quit [Ping timeout: 256 seconds]
sanscoeur has quit [Ping timeout: 268 seconds]
<ldlework>
this is madness
<ldlework>
disabling ssl from my laptop to the znc, and znc to freenode, still can't connect
newhoggy has joined #nixos
<clever>
ldlework: anything in the znc logs or journal?
<ldlework>
how do I access znc logs?
<ldlework>
nothing in journal when the connection fails
<clever>
lsof -p <PID>
<clever>
see if it has any log files open
<reachingfourpeas>
clever: Your advice worked. It also turned out that a bunch of dependencies were broken (not sure of the Nix terminology) so I ran `nix-shell --repair` and it ended up rebuilding everything from GHC up to the package I ultimately wanted to use
<sir_guy_carleton>
where do you pass compile flags in nixos? in nixpkgs or in systemPackages?
<ldlework>
I just don't get it
<ldlework>
I have it running in the foreground
<ldlework>
It doesn't say anything when I attempt to connect
<ldlework>
clever: how do I tell if znc is listening on the port?
<ldlework>
Is there a way I can like check the port from the outside?
stepcut has quit [Remote host closed the connection]
<ldlework>
I can't even connect to znc locally from the host using weechat
Rusty1_ has quit [Quit: Konversation terminated!]
iyzsong has quit [Read error: Connection reset by peer]
<ldlework>
clever: when I netstat it, all I get is
<iqubic>
Not sure, I haven't tried nix 2.0 in a while.
<iqubic>
Alright, I just removed the "nix.pkg = pkgs.nix1;" from my configuration.nix, but the rebuild used the prior version of nix. Now I have nothing I need to update. How shall I go about testing nix 2.0?
<Myrl-saki>
I have a feeling that this is a result of ELF pruning.
semilattice has quit [Remote host closed the connection]
<jasongrossman>
ldlework: There must be hardware interactions, because headphones work fine on a vanilla NixOS on my Macbook. (My built-in speakers, OTOH, don't work realiably.)
<jasongrossman>
ldlework: I assume when you say "in a headphone jack" you mean you're plugging the headphones in using a jack. Do NOT try to get your headphones actually INSIDE the jack.
<ldlework>
jasongrossman: I just didn't have them plugged in all the way lol
<ldlework>
new laptop, the port is fresh and sturdy
<Myrl-saki>
Well
<Myrl-saki>
This was a new bug
<Myrl-saki>
I did not declare Reflex.Bulma.Internal in my other modules.
newhoggy has joined #nixos
<jasongrossman>
ldlework: :-)
simukis has joined #nixos
simukis has quit [Read error: Connection reset by peer]
simukis has joined #nixos
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<{^_^}>
→ 996849ab by @cstrahan: kubernetes: update to 1.10
<{^_^}>
→ 3e3917bb by @johanot: kubernetes: 1.10.0 -> 1.10.3
<{^_^}>
→ 55fa98dd by @azazel75: Kuberbetes DNS addon 1.14.4 -> 1.14.10
NightTrain has quit [Quit: Lost terminal]
<juhe>
Hello, is it possible to create symlink to a file via configuration somehow? Specifically, I need to symlink diodmount to mount.diod so I can use diod type in fileSystems. I can hack it with security.wrappers, though it's not nice. Does anybody know better solution?
<sphalerite>
juhe: I'd probably create a simple wrapper package containing just the symlink. Something like (pkgs.runCommandNoCC "mount.diod" {} "mkdir -p $out/bin ; ln -s ${pkgs.diod}/bin/diodmount $out/bin/mount.diod") and stick that in systemPackages
<juhe>
sphalerite: hm, thanks, I'll try that!
<juhe>
sphalerite: works great, thank you!
<sphalerite>
\o/
<ldlework>
In NixOS is there any way to write a file outside the store at all? With NixOps?
<ldlework>
There's no way to use Nixlang to describe how to provision remote data?
<ldlework>
A program uses /var/lib/foo as a datastore, there's no way to put something there?
<ldlework>
I have to use bash to script `nixops scp`?
<sphalerite>
no, you can't do that in the nix language. That doesn't mean it's impossible to do.
<ldlework>
like with an override or something?
thc202 has joined #nixos
<sphalerite>
you could make a systemd service to create the directory, for instance: systemd.services.create-var-lib-foo = {serviceConfig.type = "oneshot"; wantedBy = ["multi-user.target"]; script = "${coreutils}/bin/mkdir -p /var/lib/foo";};
<sphalerite>
no, you use the hooks that nixos gives you :)
<ldlework>
hooks?
<ldlework>
preStart?
<ldlework>
how do I add some preStart to the znc service already defined?
<ldlework>
How do I figure out where the systemd unit definitions are?
ldlework has quit [Client Quit]
<Myrl-saki>
Uhh
<Myrl-saki>
How do I have an overlay when I already have a nixpkgs?
newhoggy has quit [Remote host closed the connection]
<Myrl-saki>
`(import <nixpkgs> {})` How would I add an overlay *without* putting it in the {}
newhoggy has joined #nixos
<sphalerite>
by putting it in ~/.config/nixpkgs/overlays
<Myrl-saki>
sphalerite: It's a custom overlay for a package.
<Myrl-saki>
Apparently, one way is to reimport it using `import pkgs.path`
<sphalerite>
yes, although that won't affect pkgs.
<{^_^}>
[nixpkgs] @mkaito opened pull request #41098 → oauth2_proxy: Handle attributes being derivations → https://git.io/vhqZw
<Myrl-saki>
sphalerite: Oh, I don't mean the global pkgs.
<sphalerite>
righ
<sphalerite>
Hm, ssh has support for multiple "streams" within a single session right? I want to write a script that does multiple things on a remote host via SSH but only needs to authenticate once
<sphalerite>
(and these multiple things need to be connected to different local commands, so I can't just do ssh foo 'a; b; c'
<Myrl-saki>
a &; b &; c
<Myrl-saki>
:D
newhoggy has quit [Ping timeout: 264 seconds]
<Myrl-saki>
Not sure if that works, tbh
<sphalerite>
that doesn't help pipe in different local commands to a and b.
<sphalerite>
It would work as you'd expect though, except that putting both & and ; is a syntax error (you'd want a & b & c )
<Myrl-saki>
Yay
<Dezgeg>
I believe you specify the ControlPath or something
<sphalerite>
ooh that looks like exactly what I want. Thanks!
Acou_Bass has quit [Ping timeout: 252 seconds]
<Myrl-saki>
Well
<Myrl-saki>
Nginx is better than I expected.
* Myrl-saki
shrugs
ldlework has joined #nixos
<jasongrossman>
In what way?
<Myrl-saki>
Usability.
<Myrl-saki>
I've always been cautious of using... behemoth softwares.
<jasongrossman>
OK cool.
<sphalerite>
well considering that apache is what it replaces… :p
<Myrl-saki>
sphalerite: I always just went for $HASKELL_LIBRARY for anything.
<Myrl-saki>
sphalerite: Then after getting exposed to nginx for a non-Haskell thing.
<Myrl-saki>
twas okay
<Myrl-saki>
sphalerite: And now I'm using nginx for the front end of my Haksell thing.
<sphalerite>
Is there a way to get the registration info for a path's closure imperatively?
<symphorien>
what does mean registration info ?
<Myrl-saki>
Okay, I take that back. Who can help me with nginx? lmao
<Myrl-saki>
Basically
<sphalerite>
I want to do something like make-system-tarball, but instead just copy the stuff over to a target host via SSH without the overhead of a big tarball each time
<Myrl-saki>
I want to make it /nix/store/something/foo
<sphalerite>
symphorien: the thing that goes in /nix-path-registration in system tarballs
<mkaito>
what tools exist to debug nix expressions aside from builtins.trace? there's no debugger, not even sure one can write unit tests, and trace is the closest thing to printf debugging. how do people dig into errors? --show-trace I've found less than useful too.
<Myrl-saki>
I used the root directive, which clearly wasn't the right thing.
<symphorien>
hum I don't even know what this is
<sphalerite>
mkaito: the REPL. That's about it AFAIK.
<sphalerite>
it's an unfortunate situation.
thblt has joined #nixos
<Myrl-saki>
Oh, apparently I want to use alias, not root.
<mkaito>
Nix is awesome, except for the part where there's almost no documentation and any debugging is black magic. hah!
<jasongrossman>
sphalerite: Right, apache is not good. I've been using lighttpd.
<Myrl-saki>
Yay, works.
<sphalerite>
mkaito: oh there's also the various kinds of debug output that you get by spamming -v
<mkaito>
if those are anything like --show-trace, I kind of doubt that'd be useful :P
Turion has joined #nixos
sanscoeur has joined #nixos
<sphalerite>
symphorien: it's a serialisation of the nix database info for the paths in the tarball, which can be loaded into a fresh database using nix-store --load-db
jasongrossman has quit [Remote host closed the connection]
<Myrl-saki>
Anyone else geting
<Myrl-saki>
builder for '/nix/store/mr9516zcczsc9p42hmd48hjp25v2x7lk-shell.drv' failed to produce output path '/nix/store/s6ygzcvzjc5q947mzc7pb1amd3zcdhsw-shell'
<Myrl-saki>
I just use nix-build $(nix-instantiate ...)
<Myrl-saki>
and it works
<jD91mZM2>
sphalerite: Ooooo nice, thanks! (although ew no ad hoc installs :P)
seafood has quit [Quit: seafood]
averell has quit [Quit: .]
<sphalerite>
Myrl-saki: typically shell derivations aren't meant to be built
obadz- has joined #nixos
<Myrl-saki>
sphalerite: That's my point. How am I getting shell derivatiosn?
<sphalerite>
what are you instantiating?
<Myrl-saki>
cgit.
obadz has quit [Ping timeout: 248 seconds]
obadz- is now known as obadz
<Myrl-saki>
FWIW
<Myrl-saki>
`nix-build nixpkgs.cgit` seems to work.
<Myrl-saki>
So does
<Myrl-saki>
myrl@myrl:~$ nix-build '<nixpkgs>' -A cgit
<Myrl-saki>
It's just the `nix-build -p cgit` that doesn't.
averell has joined #nixos
Turion has quit [Quit: Leaving.]
<Myrl-saki>
Also, I mean `nix build nixpkgs.cgit`
coot has quit [Quit: coot]
<sphalerite>
weird. It should print the .drv file that it tried to build on failure, right? Maybe compare that with nix-instantiate '<nixpkgs>' -A cgit?
<Myrl-saki>
sphalerite: It tried to build '/nix/store/mr9516zcczsc9p42hmd48hjp25v2x7lk-shell.drv'
<Myrl-saki>
I think that's actually the last thing it tries to build.
<sphalerite>
huh.
<Myrl-saki>
Yep.
<Myrl-saki>
I don't get how I'm getting shell.drv from `nix-build -p cgit`
<phdoerfler>
hi! What video chat programs are available for nixos?
tzemanovic has quit [Read error: No route to host]
tzemanov_ has joined #nixos
<sphalerite>
phdoerfler: personally I use jitsi meet in firefox
abcrawf has joined #nixos
smallville7123 has quit [Remote host closed the connection]
<sphalerite>
Dezgeg: SSH's ControlPath is amazing, thanks so much!
<phdoerfler>
sphalerite: ok I'll give this a shot
<phdoerfler>
we tried hangouts in the browser but for some reason the other person (who is on nixos) can't hear me
Neo-- has quit [Ping timeout: 264 seconds]
<phdoerfler>
my mic is working and I tried 3 different devices and the speakers in hangouts of the nixos person are also working (they play the test sound in hangouts just fine) and I am at the end of my wisdom here ^^
<mpickering>
I don't know how Setup searches for C libraries really
<mpickering>
What is failing to build exactly?
<cinimod_>
`hmatrix-sundials` fails to build
<mpickering>
does it work before you patched it?
<mpickering>
The postgresql server doesn't work for me. After I have enabled it, when switching, the error is - FATAL: could not create lock file "/tmp/.s.PGSQL.5432.lock": Permission denied
<cinimod_>
> does it work before you patched it?
<{^_^}>
error: syntax error, unexpected ')', expecting ID or OR_KW or DOLLAR_CURLY or '"', at (string):42:1
<cinimod_>
No
acarrico has quit [Ping timeout: 248 seconds]
<cinimod_>
But if I build hmatrix-sundials using `stack --nix` then it does build
<{^_^}>
[nixpkgs] @xeji pushed commit from @romildo to master « tint2: 16.2 -> 16.4 (#41097) »: https://git.io/vhqCj
seafood has joined #nixos
muffins_Madcat_ has quit [Ping timeout: 248 seconds]
muffins__Madcat_ has quit [Remote host closed the connection]
muffins__Madcat_ has joined #nixos
greymalkin has quit [Ping timeout: 240 seconds]
sanscoeur has joined #nixos
<nh2[m]>
if two modules assign an option of type list, the lists are usually concatenated. Can I use one of the mkForce* style modifiers to ensure the concatenation doesn't happen?
stepcut has joined #nixos
<mpickering>
anyone have an example of deploying a web server with a postgresql db to the same server using nixops?
karlguy has joined #nixos
<nh2[m]>
in my case I want to do that to totally override `networking.nameservers`, no matter what other modules set in there.
<{^_^}>
[nixpkgs] @xeji pushed commit from @r-ryantm to master « pan: 0.144 -> 0.145 (#40963) »: https://git.io/vhqld
<sphalerite>
not sure there is one.
acarrico has quit [Ping timeout: 240 seconds]
stepcut has joined #nixos
jbgi has joined #nixos
johanot has joined #nixos
<MarcWeber>
I tried setting seutid on a shell script chowning as root -> but running does not change uid. Does anybody know why? Is my understanding of setuid wrong?
<MarcWeber>
Or do I need wrapper executables?
fendor has quit [Ping timeout: 245 seconds]
<symphorien>
setuid does not work on scripts with shebang, I think
asuryawanshi has quit [Remote host closed the connection]
asuryawanshi has joined #nixos
<jasongrossman>
Yeah, there are general limits on setuid - enforced by the kernel, I guess? A web search should find you much more about this.
<sphalerite>
yep
<jbgi>
what is status of linux_latest in stable channels? Is it allowed to be updated to a newer version of the kernel after the nixos stable release?
<sphalerite>
jbgi: yes. Kernels that are neither the default version nor _latest are even removed
<jasongrossman>
jbgi: A newer version, yes. Or do you mean a newer major version or something?
<sphalerite>
when their upstream support ends
<jbgi>
jasongrossman: yes, updated to a newer major version. The issue is that currently linux_lastest in 18.03 point to an unmaintained kernel (5.15).
<jbgi>
*4.15
<jasongrossman>
I doubt that there's a policy about it. You'd have to ask who's updating them.
sanscoeur has joined #nixos
jasongrossman is now known as JasonGrossman
asuryawanshi has quit [Remote host closed the connection]
asuryawanshi has joined #nixos
newhoggy has quit [Remote host closed the connection]
newhoggy has joined #nixos
<MarcWeber>
ls
sanscoeur has quit [Ping timeout: 265 seconds]
newhoggy has quit [Remote host closed the connection]
thblt has quit [Remote host closed the connection]
jtojnar has quit [Read error: Connection reset by peer]
jtojnar_ has joined #nixos
jtojnar_ is now known as jtojnar
johanot has quit [Quit: Leaving.]
ixxie has joined #nixos
andrewmiller1 has quit [Ping timeout: 260 seconds]
<gchristensen>
interestin, jbgi
asuryawanshi has quit [Ping timeout: 252 seconds]
stepcut has quit [Remote host closed the connection]
srl295 has joined #nixos
lostman has joined #nixos
<Myrl-saki>
[2/41/290 copied (69.8/1008.9 MiB)] copying path '/nix/store/ilha1h5gr3hc9nhbjig05n9iz1iv7wgd-network-2.6.3.2' to 'ssh://me@adrianpar.vin'
<Myrl-saki>
1000 MiB
<Myrl-saki>
On my 1 Mbps connection.
<ixxie>
Are there any terminal-based IRC clients that have a NixOS module?
<sphalerite>
Myrl-saki: hf
<Myrl-saki>
Prelude> 1008.9*8/60/60
<Myrl-saki>
2.242
<Myrl-saki>
2 hours. Very nice.
<infinisil>
> 1008.9 * 8 / 60 / 60
<{^_^}>
value is a float while an integer was expected, at (string):41:1
<Myrl-saki>
lmao
<Myrl-saki>
> 1009 * 8 / 60 / 60
<{^_^}>
2
<Myrl-saki>
Uhhh... I'm surprised Nix has division support.
<lostman>
I'm trying to build a tiny docker image with nix that includes some derivations. However, for some reason gcc is included in the image. I tracked down the derivation that causes that but I'm not sure why nix pulls it in. gcc is only required to build the derivation. Any ideas how can I investigate this?
<clacke[m]>
I have generated the private key on the server, fed it to nix-serve.
<clacke[m]>
and I'm relying on the server locally, and I've fed the public key into nix.conf
<Myrl-saki>
infinisil: I'm guessing so. I'm just looking at the patterns, not at the actual code or anything formal.
<clacke[m]>
But I get `warning: substituter 'https://hydra.fractalide.com' does not have a valid signature for path '/nix/store/wzgr69i39bv4cdcb7jdydnijzhr8nvmj-typed-racket-more'
<clacke[m]>
`.
jtojnar has quit [Quit: jtojnar]
<clacke[m]>
So it's at least talking to the server, so it knows it has `typed-racket-more`.
jtojnar has joined #nixos
<{^_^}>
[nixpkgs] @primeos pushed to release-18.03 « signal-desktop-beta: Mark as broken (really outdated) »: https://git.io/vhqBD
jbgi has quit [Quit: WeeChat 2.1]
<clacke[m]>
Where can i sniff to find out if the package on the server carries any signature, and what signature?
<sphalerite>
clacke[m]: nix path-info --sigs
camsbury has joined #nixos
<sphalerite>
I'm trying to get a script working that will replace ubuntu on a scaleway server iwth nixos. It's working mostly, but I want to preserve the original SSH host keys, and ubuntu's sshd seems to default to sending ecdsa-sha2-nistp256 keys while nixos prefers ed25519, so after the OS replacement the client thinks something bad is going on. I've tried adding `HostKeyAlgorithms ecdsa-sha2-nistp256` to the
<sphalerite>
sshd_config, but that hasn't helped, now it doesn't offer *any* host key algorithms to the client. Any ideas?
freeman42x]NixOS has joined #nixos
<infinisil>
I recently added services.openssh.kexAlgorithms to add additional supported algs to the server
<lostman>
sphalerite: trying to use why-depends... seems a bit tricky. tried why-depends mypkg pkgs.gcc7 but that says mypkg doesn't depend on gcc7. how can I just list all dependencies?
<sphalerite>
infinisil: not key exchange, host key
<infinisil>
Oh
sanscoeur has joined #nixos
<sphalerite>
lostman: try with pkgs.gcc7.cc, or just the store path to the gcc that's appearing there unexpectedly
<jD91mZM2>
Being sphalerite and having questions must be almost like being a stackoverflow developer searching for answers on stackoverflow.
<sphalerite>
lol
<lostman>
sphalerite: why-depends shows me this: lib/libclang.so.6.0: …inux-gnu/gcc/......../nix/store/bm7pb1s7rx1ad80706b5xqrznq7fgpgx-gcc-7.3.0./opt/rh/devtoolset-6/…
Xal has joined #nixos
<Dezgeg>
is that just the libstdc++?
<Dezgeg>
or the gcc proper?
<lostman>
and when I ldd libclang.so.6.0 I get this: libstdc++.so.6 => /nix/store/4zd34747fz0ggzzasy4icgn3lmy89pra-gcc-7.3.0-lib. so /nix/store/...-gcc-7.3.0-lib makes sense but when I copy the closure I also get /nix/store/...gcc-7.3.0 which doesn't seem necessary
leat has joined #nixos
stepcut has joined #nixos
<clacke>
Also does anyone know why #freenode_#nixos:matrix.org is silent? I can see my clacke[m] user here, but on Matrix side there hasn't been any traffic for over a month.
<srk>
Dezgeg: hey, thanks for the arm mirror! :)
<sphalerit>
clacke: works for me
<srk>
trying to get xorg & etnaviv running after few days of compiling
<sphalerite>
(sphalerit is my matrix one)
<Dezgeg>
I tried etnaviv once, iirc it needs packaging the xf86 driver to work
<{^_^}>
→ 8f3a541a by @samueldr: nixos/grub: Updates `splashImage` description to match reality.
iyzsong has quit [Ping timeout: 256 seconds]
<srk>
right
<{^_^}>
→ 0ccfe14e by @samueldr: nixos/grub: Implements use of file format for splashImage.
<{^_^}>
→ 1edc7ae6 by @worldofpeace: xbmcPlugins.osmc-skin: init at 17.0.4
<jD91mZM2>
Oh, speaking of mirrors, does NixOS have an easy way to setup a mirror? And can those mirrors be verified for integrity?
<srk>
yes!
<sphalerite>
jD91mZM2: nix copy!
<rotaerk>
is there a quicker way to build a haskell project besides running nix-build every time? for instance if I'm inside a nix-shell
<sphalerite>
although actually no, it's not really mirroring
<srk>
with cabal
<rotaerk>
ah cabal build
<srk>
yes, or cabal new-build
<srk>
or stack build
<Dezgeg>
but vc4 gpu acceleration with mesa works totally out-of-the-box
<srk>
anything that can do partial rebuilds
<rotaerk>
k thanks
<srk>
Dezgeg: wow
<jD91mZM2>
sphalerite: Yeah nix copy is just for copying cache to another machine right?
<srk>
Dezgeg: gonna try packaging xf86-video-armada I guess
<Dezgeg>
and nouveau on the jetson tk1 with some patches to wayland (dunno if they're needed these days tho)
<srk>
then xf86-video-armada-novena and cnfigs..
<Dezgeg>
yeah, I suppose
<sphalerite>
jD91mZM2: no, you can copy to various things, including flat-file binary caches (that can hten be served by a dumb HTTP server), S3, or chroot stores
fragamus has joined #nixos
<sphalerite>
copying to the nix store on another host is just one of the many things you can do with nix copy :D
<jD91mZM2>
sphalerite: Can you verify it for integrity later?
<sphalerite>
jD91mZM2: yes, signing is supported
<Dezgeg>
I would use whatever has the latest commits
<Dezgeg>
probably
<jD91mZM2>
sphalerite: Wait, who would sign what?
<srk>
wohoo
<sphalerite>
jD91mZM2: see nix-store --generate-binary-cache-key, nix sign-paths, nix path-info --sigs, and the `trusted-public-keys` setting (copying will be refused unless you're a trusted-user or the paths being copied are signed by a trusted key)
<srk>
gonna be fun
<sphalerite>
jD91mZM2: the secret-key-files setting in nix.conf makes a machine sign all the things built on it with that key, or you can use nix sign-paths to sign stuff explicitly
<srk>
like you can see that pretty fullhd ips display while it's in u-boot and then it goes dark
<srk>
and you have compile for days to see it again :D
<rotaerk>
hmm I should probably actually learn cabal
<rotaerk>
it's just been this thing that I occasionally deal with superficially, thus far
<srk>
cabal's getting lot better
<Dezgeg>
hmm, I think at least wandboard had non-accelerated graphics support?
<rotaerk>
is new-build supposed to be good enough to make nix redundant?
<srk>
I might even ditch stack
<Dezgeg>
on the hdmi port
<srk>
new-build can't integrate with nix iirc
<srk>
(yet)
<Dezgeg>
but maybe that's different on whatever board you have
<rotaerk>
I don't use stack, because it seems redundant with nix
<jD91mZM2>
sphalerite: Are we talking about signed mirror -> client transaction or verifying that the mirror actually is producing the correct binaries?
<srk>
but you only need simple shell.nix with cabal/ghc
<srk>
rest is handled by cabal
<bgamari>
srk, do you have any ideas regarding what new-build/nix integration would look like?
<rotaerk>
interesting
<srk>
many ways to build haskell tho
<srk>
bgamari: like it says it uses nix-style isolation. nix integration could just provide ghc I guess
<srk>
I'm not an expert tho, only started playing with that recently as I wanted to try backpack
<sphalerite>
jD91mZM2: we're talking about signing "this output path was produced by this derivation"
<sphalerite>
jD91mZM2: so you can build and sign it on some machine that you trust, copy it to an untrusted mirror, and be able to tell if it's been tampered with
smallville7123 has quit [Quit: rosa]
<bgamari>
srk, I was thinking perhaps you wanted to have new-build use packages built by nix
<srk>
possibly, I don't know
<bgamari>
srk, or perhaps even have new-build produce derivations
<srk>
for development I don't really care, for deployment I want everything handled by nix
<srk>
that would be cool for sure
<srk>
nice thing is that you can go like dhall2cabal -> cabal2nix ..
<bgamari>
right
<jD91mZM2>
sphalerite: Oh okay, thanks for clarifying
<srk>
bgamari: what I really like about nixos that I only have to publish stuff on hackage and it propagates to nixos
<{^_^}>
→ 10c973ef by R. RyanTM: nanomsg: 1.1.2 -> 1.1.3
<{^_^}>
→ 059a71a6 by @grahamc: lib/options: teach showOptions about funny option names
<{^_^}>
→ 996849ab by @cstrahan: kubernetes: update to 1.10
<clacke>
sphalerit: You can use #freenode_#nixos:matrix.org? That's odd. So maybe some issue with my home server then.
<clacke>
Hm, ok, I didn't see the sphalerit message on matrix either.
<clacke>
So weird, other matrix.org channels work, including other freenode ones
<clacke>
> the secret-key-files setting in nix.conf makes a machine sign all the things built on it with that key, or you can use nix sign-paths to sign stuff explicitly
<{^_^}>
error: syntax error, unexpected IN, expecting ')', at (string):41:30
<clacke>
thank you sphalerite that's what I needed to know
<LnL>
then users can override it like the darwin stdenv
<jcrben>
LnL: and that happens automatically under the hood right? looks like the latest version (https://github.com/paulej/AESCrypt) has diverged significantly from the release and is now using CMake anyhow
<jcrben>
latest version isn't an actual release tho, just a development version. guess I'd have to email a patch
<sphalerite>
oh jeez, "nix daemon out of memory" on my 16GB laptop…
Ariakenom has quit [Ping timeout: 252 seconds]
<symphorien>
is the nar format nix-specific ?
<Dezgeg>
it is
Ariakenom has joined #nixos
<{^_^}>
[nix] @grahamc opened pull request #2183 → Omnibus backport: #2178: doc fix for --check, #2181: MITM HTTPS proxy fix at install time → https://git.io/vhquG
<clacke>
Is there a way to make nix regenerate all its packages after I have given it secret keys?
<LnL>
jcrben: yeah, the build environment contains CC=clang environment variables and make/cmake will use those unless explicitly overriden like this example
<sphalerite>
clacke: no, but you can sign all the paths, if you don't have anything you don't trust in your nix store, using nix sign-paths --all
* sphalerite
would quite like if nix generated a key and signed everything by default
<clacke>
sphalerite: thx!
<clacke>
Oh yes, I think it's a big surprise that it doesn't have a key generated on install.
<gchristensen>
interesting, clacke
freusque has joined #nixos
<freusque>
hello! For you haskell users out here, what's your standard response when a package doesn't build?
<freusque>
usually, due to bounds not being met
<infinisil>
freusque: Use stack :P
<freusque>
ok, one vote for stack, anyone else ? :-)
<gchristensen>
I recommend going for a walk outside
<freusque>
one vote for stack and one vote for anger management techniques
novum has quit [Quit: Lost terminal]
<clacke>
nix sign-paths --all went ridiculously fast
<clacke>
I wonder if it did anything
<clacke>
but I guess just signing checksums can go pretty fast
<gchristensen>
I think so
joehh has quit [Ping timeout: 264 seconds]
<sphalerite>
clacke: yeah, and it doesn't compute the checksums either, just grabs them from the database
acarrico has quit [Ping timeout: 240 seconds]
novum has joined #nixos
<elvishjerricco>
freusque: Usually the package isn't on Stackage, or else the bounds in nixpkgs would have been right. So stack doesn't help much. You probably either need to bump it or just use doJailbreak
novum has quit [Client Quit]
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
thblt has quit [Remote host closed the connection]
<freusque>
I guess someone was an ac/dc fan
<freusque>
when they named that. I'll look it up, thank you
clacke has quit [Ping timeout: 260 seconds]
seanparsons has quit [Quit: ZNC 1.6.5 - http://znc.in]
seanparsons has joined #nixos
smallville7123 has joined #nixos
Ariakenom has joined #nixos
<jD91mZM2>
Does nix have anything like `guix import`?
<drakonis>
well, sort of
xy2_ has quit [Ping timeout: 248 seconds]
<jD91mZM2>
I mean I guess that's what all the _2nix things are for?
<drakonis>
yes
<drakonis>
it is all over the place though
<drakonis>
nix 2.0 has plugin support, which would effectively allow it to have a guix import equivalent
<drakonis>
built into nix itself
<jD91mZM2>
That's awesome!
<jD91mZM2>
Speaking of nix 2.0... What's `nix edit` for? At least without specifying -f it feels pretty useless
andymandias has quit [Ping timeout: 256 seconds]
<gchristensen>
what is guix import?
<LnL>
various generators
<gchristensen>
oh cool
<freusque>
jD91mZM2: it's really cool when you have your own package set in scope
<freusque>
nix edit -f '<mypkgs>' -A mypackage
<freusque>
ah sorry, I didn't read through you whole message.
<freusque>
>:)
<jD91mZM2>
freusque: It would be useful for me if my EDITOR variable wasn't overriden by /etc/zshenv constantly :P
<tilpner>
It's also useful in combination with EDITOR=echo and EDITOR=cat
<freusque>
yeah I do the echo and cat thing too hehe
<tilpner>
Though I dislike that it doesn't fallback to unsafeGetAttrPos :/
johnw has joined #nixos
blankhart has joined #nixos
<Guest64915>
hi all. On latest 18.03 I'm having a strange X startup issue after boot.
sanscoeur has joined #nixos
<Guest64915>
the system boots but X doesn't. Only after login on console X starts (immediately after login)
<Guest64915>
any clue why this would be happening?
Guest64915 is now known as Unode
<jD91mZM2>
/var/log/X.0.log
<jD91mZM2>
give it to me
<Unode>
:D
<Unode>
one moment
<Unode>
erm... there is none
<Unode>
odd...
<jD91mZM2>
Oh, that's odd
<jD91mZM2>
Did you enable a display manager and stuff in the nix config?
<Unode>
yes, this has worked before. I think the most recent change (after which things started behaving oddly) was to increase the number of processes in ulimit
reinzelmann has joined #nixos
<jD91mZM2>
`journalctl -xe display-manager`?
<Unode>
Failed to add match 'display-manager': Invalid argument
<elvishjerricco>
Don't you use -u to specify a unit with journalctl?
<Unode>
ok -u seems to help
<Unode>
checking...
sanscoeur has quit [Ping timeout: 260 seconds]
<jD91mZM2>
Do you get to your display manager at all or does it start directly into X?
nkaretnikov has joined #nixos
<Unode>
At the moment: I get linux login prompt, no X. I then login on the prompt and X starts.
<Unode>
What's even weirder is that I don't actually need to login successfully
<jD91mZM2>
oh god
<nkaretnikov>
could someone direct me to a sample ghcjs project that uses nix? things i've seen so far are either deprecated or huge.
<jD91mZM2>
Unode: So no display manager like LightDM or whatever?
<jD91mZM2>
It just starts directly?
<Unode>
for instance now I saw a message saying "nixos-help" for NixOS manual. and I simply wrote that as login. No password. And X started right after
<Unode>
jD91mZM2: after this I get the usualy login screen in X
<tilpner>
krey - It's interesting that you re-use /nix/var/nix/profiles/per-user/root/channels, but that allows nixos-rebuild and nix-channel to update your allegedly pinned sources :/
<tilpner>
Which you may consider a feature if you want
<tilpner>
Also needs support for non-GitHub git repos
<krey>
tilpner: fair point about updating a single channel at a time
ggp0647 has joined #nixos
<krey>
tilpner: and yeah, I just gave up on nix-channel
<tilpner>
Ironic how your nix-pinning script explicitly forces non-pinned python3 on the user
<krey>
tilpner: well, I need something to handle the json
yurb has joined #nixos
<krey>
tilpner: but yeah, maybe it should be pinned, haha
<ryantm>
Is leaveDotGit in the fetchGit okay to use? Or, are we trying to get rid of it?
<gchristensen>
ideally you don't want it
<gchristensen>
your hash will become out of date with leaveDotGit
<krey>
tilpner: so is this kinda similar to what you have?
<tilpner>
krey - Does "import json" not work on older python3 versions?
novum has joined #nixos
<krey>
tilpner: no it works. I jus meant that it would've been nicer not to have a python dependency
<krey>
tilpner: I tried to do it in bash/nix, but it wasn't going very well
<ryantm>
gchristensen: How does it get out of date? Does it get out of date the minute someone makes a new commit to the git repo?
<gchristensen>
hmm I'm not sure exactly
<tilpner>
krey - It is similar, but my source descriptions are scripts, and I have a separate json file for every pin
<gchristensen>
but if the repository repacks, or creates new branches / tags / etc. it may be out of date
<sphalerite>
hm, I started diffoscope this morning at 09:16 UTC. It's now 17:54 UTC and it's still eating 100% of one core. I feel like something might not be going right here..?
<tilpner>
krey - tx0.co/tarball and tx0.co/github , then e.g. tx0.co/Z
<nkaretnikov>
i'm attempting to update cabal2nix to the latest version on darwin. the nix manual says that i need to do `nix-channel --update nixpkgs && nix-env -u '*'`, but it does nothing visible on my system. `nix-channel --list` is also empty. do i need to subscribe to unstable first?
<LnL>
jcrben: yeah, libiconv
<gchristensen>
sphalerite: when diagnosing diff issues I'll generally `split -n<bignumber>` the file, then sha1sum each, find the different ones, and diffoscope on those
<tilpner>
krey - Could be prettier by setting PATH from the parent script
<boxofrox>
srhb: oh, just saw your pastebin about feeding the config.php.mysqlnd into the nixpkgs import. your way also works. I hadn't considered it because setting config.php.mysqlnd in my /etc/nixos/configuration.nix was generating errors for a containerized php+mysqlnd and the override was the only syntax I could figure out that worked. thanks for the knowledge. will keep it in mind for later.
<LnL>
jcrben: oh
<sphalerite>
gchristensen: Dezgeg good to know :D
clacke[m] has left #nixos ["User left"]
<nkaretnikov>
in which repo do i need to open an issue wrt the nix manual? should it be done in `nix`?
<sphalerite>
oddly enough, diffoscope on the nix-store --export of the paths I'm comparing took much less time
tzemanovic has quit [Ping timeout: 256 seconds]
<sphalerite>
nkaretnikov: yes, if it's the nix manual and not the nixpkgs or the nixos manual
<clacke>
Wow. When I try to join #freenode_#nixos:matrix.org I am now noticing that I'm actually redirected to #nixos:pierrebeaucamp.com . I wonder how that happened.
<yurb>
When I run 'nix-env -iA nixos.supercollider_scel' I'm getting the following error: 'error: attribute 'nixos' in selection path 'nixos.supercollider_scel' not found'
<gchristensen>
clacke: man, I've tried to fix the NixOS on Matrix situation a few times but the tools just don't exist.
<gchristensen>
yurb: you'll need to do nixpkgs.... instead of nixos...
<yurb>
gchristensen: oh, so there are two repositories?
<clacke>
Hey sphalerit could you tell me the real hash-like name of the real #nixos room that you're in?
<sphalerite>
clacke: the room has many addresses, #nixos:pierrebeaucamp.com is one of them
<jcrben>
LnL: worked :) - I'll make a PR
<clacke>
ok
<gchristensen>
yurb: the "nixos." prefix and "nixpkgs." prefix refers to the "channel" by which you have the Nixpkgs package set setup locally
<jcrben>
LnL: also send an email to paul about fixing it upstream
<clacke>
I'll poke my homeserver admin
<sphalerit>
!AinLFXQRxTuqNpXyXk:matrix.org
<sphalerite>
clacke: ^
<jcrben>
(I will send that email)
<gchristensen>
yurb: or NixOS, it defaults to "nixos." off NixOS, it defaults to "nixpkgs"
<gchristensen>
yurb: I feel the difference is a bit vestigial at this point
Guest84464 has quit []
MatrixTraveler[m has joined #nixos
<yurb>
okay, thanks for explaining:) Now, if there is a 'supercollider_scel' package listed in nixos, it can be missing in nixpgks?
<clacke>
sphalerite: thx!
kaychaks has joined #nixos
<gchristensen>
yurb: it should be there for nixpkgs too
<clacke>
Ok, that takes me to my same echo chamber. So this needs to be kicked on my matrix instance.
<gchristensen>
but on macos, `sudo` doesn't update the HOME variable to be ~root (`sudo env | grep HOME` will still point to your user's home)
<gchristensen>
thusly, `sudo nix-channel --update` doesn't do The Right Thing, and is confusing for macos users
<gchristensen>
A question becomes, should we even have a "default" profile on macOS, owned by root? I don't know
<nkaretnikov>
gchristensen: okay, then i'll wait for me input on the issue
<nkaretnikov>
not that it's a dealbreaker. just might be confusing for newbies.
<gchristensen>
it is definitely confusing for newbies ... and oldies ...
<nkaretnikov>
lol
<gchristensen>
it is weird, because the default channel is important for questions like "who installs the nix-daemon?"
vaninwagen has joined #nixos
<yurb>
I have another question... The supercollider_scel package should contain some emacs lisp files, how can I make emacs find them?
<sphalerite>
if you install it in your user profile, just add ~/.nix-profile/share/emacs/whateveritis to your emacs search path (sorry I can't remember the exact path and name of the variable)
<yurb>
sphalerite: thanks a lot! same applies to mu I guess.
<clacke>
I have done `nix-store --delete /nix/store/ajsdölfkjslajff-foo` on my hydra
<sphalerite>
:D
<clacke>
But my local machine still says 'https://hydra.fractalide.com' does not have a valid signature for path
<clacke>
where does my machine remember that it fetched something from hydra?
<sphalerite>
there's a local cache of what's available in the binary cache
<sphalerite>
~/.cache/nix/binary-cache-v5.sqlite
<nkaretnikov>
so do i need to do anything special when running cabal2nix if my project uses ghcjs? i've tried the compiler option with the ghcjs argument, but it doesn't seem to make a difference. or do i just need to pass --argstr compiler ghcjs when running nix-shell? even if it works, how would folks install via nix-env if the default.nix file has nothing ghcjs-specific.
drakonis has quit [Remote host closed the connection]
<gchristensen>
25 points gets you a sticker, 100 points gets you a t-shirt, 1000 verified points gets you a free trip to nixcon *restrictions apply, must be verifiable points, given by grateful people, in channels I'm in
<clacke>
I nuked my ~/.cache/nix
<clacke>
What's the system cache?
<sphalerite>
/root/.cache/nix
<sphalerite>
oh and /nix/var/nix/binary-cache-v3.sqlite ? I think? Not sure if that is used or not…
<{^_^}>
[nixpkgs] @ryantm opened pull request #41108 → prefix haskell packages with "haskell-" → https://git.io/vhqaN
coot has joined #nixos
<clacke>
Ok, I only looked in /nix/var and saw no cache, and then /var/cache which had nothing related to nix
<clacke>
oh right
<clacke>
directly in /nix/var/nix
<clacke>
thanks again
<gchristensen>
sphalerite: ^ seriously
<sphalerite>
mine hasn't been touched since february, so I think this might be only for nix 1.x which you're probably not using :)
<sphalerite>
nkaretnikov: does this occur consistently?
<nkaretnikov>
sphalerite: yes, i tried a couple of times just now
<sphalerite>
:/
sanscoeur has joined #nixos
<nkaretnikov>
i'd appreciate if someone could reproduce at least. i created a hello world haskell project with cabal2nix, then ran `nix-shell --argstr compiler ghcjs`as mentioned in the puffnfresh's twit.
xy2_ has quit [Ping timeout: 240 seconds]
{^_^} has quit [Remote host closed the connection]
<sphalerite>
oh boy. Running out of memory makes you really appreciate how slow block storage actually is.
{^_^} has joined #nixos
{^_^} has quit [Changing host]
{^_^} has joined #nixos
<gchristensen>
how ya doin', {^_^}
<sphalerite>
♥ haskell
<sphalerite>
gchristensen: you should totally give it a vanity IP address :D
<gchristensen>
haha why yes I should
<gchristensen>
samueldr: I just merged a PR and the github to irc thing didn't pick it up
sanscoeur has quit [Ping timeout: 256 seconds]
<samueldr>
any log outout in the github to irc thing?
<tazjin>
I'm trying to use `build.googleComputeImage` atm - it seems to launch qemu during its build to create the VM image, but that very quickly fails with `no space left on device` and the build exits
<gchristensen>
samueldr: literally none
<tazjin>
I'm not sure what device is running out of space here, it's not the host, so I checked if maybe it uses tmpfs, but that's not the case either
<sphalerite>
tazjin: it'll be within the VM probably
<sphalerite>
raaaaaaaagh so many OOMs trying to build pandoc on 2GB
<tazjin>
sphalerite: it would seem like it, not quite sure what to do about it yet so I guess I'll dig into that code
<gchristensen>
sphalerite: man, haskell requires 1tb right out of the gate /simplified
<sphalerite>
gchristensen: not on ARMv7, there's not enough address space there ;)
<nkaretnikov>
oh, it could be just me using a proxy. perhaps there are some rules that look for ip-addresses and user-agents, i dunno. now it fails somewhere else :(
<samueldr>
gchristensen: have you previously stopped the service or was it running since day one?
saltsa has quit [Quit: leaving]
<gchristensen>
samueldr: interesting question
<sphalerite>
clacke: any luck with invalidating the cache?
<gchristensen>
samueldr: last time I restarted it was April 22
<sphalerite>
You may also need to restart the nix daemon after removing the database file
<samueldr>
in theory, nothing I changed should cause it to... do nothing, but I may be wrong!
<gchristensen>
:D
<samueldr>
(I'm starting up the rabbitmq test environment to verify on my end)
<nkaretnikov>
i don't understand why it fails. the .cabal file has a dependency on semigroups >= 0.11 && < 0.19 and `nix-env -f '<nixpkgs>' -qaP -A haskellPackages | grep semigroups` lists 0.18.4
<nkaretnikov>
trying with -j1 now... :/
stepcut has quit [Remote host closed the connection]
johnw has joined #nixos
<gchristensen>
samueldr: I dunno :/
tertle||eltret has joined #nixos
<samueldr>
I haven't left enough information about my dev setup hah, I'm looking up the bits I forgot to have it work locally
<samueldr>
ah, good, gchristensen when you say no output, do you mean there isn't even "connecting..." "connected" "Waiting for events..." at start?
<nkaretnikov>
also, how come i get warnings like this? the ghc is supposed to be much newer
not_a_bot has joined #nixos
dottedmag has quit [Quit: QUIT]
not_a_bot is now known as Guest48325
alex`` has quit [Quit: WeeChat 2.1]
palo_ has joined #nixos
palo_ has quit [Changing host]
palo_ has joined #nixos
lord| has joined #nixos
alex`` has joined #nixos
<Guest48325>
In my config.nix I have a few different buildEnvs I use to manager packages, using `pkgs.name` to add a package to the list.
<Guest48325>
How do I add a local custom derivation to that list of pkgs
palo has quit [Ping timeout: 260 seconds]
<nkaretnikov>
collected some garbage just in case, trying again
ixxie has quit [Ping timeout: 245 seconds]
chimay has quit [Ping timeout: 240 seconds]
fendor has quit [Remote host closed the connection]
MP2E has joined #nixos
muffins_Madcat has quit [Read error: Connection reset by peer]
muffins_Madcat has joined #nixos
<yurb>
supercollider coredumps:( is there a way to ask nix to build it from source instead of using a prebuilt binary?
<infinisil>
yurb: You'd get the exact same result, but you can do that with --check
<nkaretnikov>
argh! it seems to be what gchristensen told me about (thanks!): did nix-channel --update as root and now nix-shell picks different versions :(
ixxie has joined #nixos
<yurb>
thanks infinisil
johnw has quit [Ping timeout: 240 seconds]
muffins_Madcat_ has joined #nixos
qqlq has quit [Remote host closed the connection]
simukis has joined #nixos
simukis has quit [Client Quit]
muffins_Madcat has quit [Ping timeout: 240 seconds]
simukis has joined #nixos
reinzelmann has quit [Quit: Leaving]
sanscoeur has joined #nixos
johnw has joined #nixos
worldofpeace has quit [Remote host closed the connection]
sanscoeur has quit [Ping timeout: 260 seconds]
johnw has quit [Ping timeout: 240 seconds]
lielazive has joined #nixos
<sphalerite>
Is there a simple way to undo a nix-channel --rollback? Or do I need to use nix-env -p?
blankhart has quit [Read error: Connection reset by peer]
acarrico has joined #nixos
blankhart has joined #nixos
<infinisil>
nix reflog
<infinisil>
Oh wait
<infinisil>
Not git
goaway has joined #nixos
<lielazive>
Is there a NixOS package for Gala WM? I like the idea of NixOS, but I'd like it to be very pretty like ElementaryOS :D
muffins_Madcat_ has quit [Ping timeout: 260 seconds]
initiumdoeslinux has joined #nixos
stepcut has quit [Remote host closed the connection]
<initiumdoeslinux>
Hi, I'm trying to run nix-build on a stack2nix generated package definition, but I get this error: https://lpaste.net/7602893292824428544 What does this mean?
tzemanovic has quit [Ping timeout: 256 seconds]
xy2_ has quit [Read error: Connection reset by peer]
<clever>
initiumdoeslinux: did you tell nix-build which package to build?
stepcut has joined #nixos
freusque has quit [Quit: WeeChat 2.0]
logzet has quit [Ping timeout: 245 seconds]
muffins_Madcat_ has joined #nixos
<initiumdoeslinux>
@clever Will it not build the ones in ./default.nix? Or do I have to call it like so: nix-build lens
logzet has joined #nixos
<clever>
initiumdoeslinux: if you dont give a specific attribute, it will try to build every single one
<clever>
initiumdoeslinux: and stack2nix generates one for every single package in stackage
<clever>
initiumdoeslinux: so you need to do `nix-build -A yourproject`
johnw has quit [Ping timeout: 240 seconds]
xy2_ has joined #nixos
muffins_Madcat has quit [Ping timeout: 240 seconds]
blankhart has quit [Ping timeout: 276 seconds]
ilyaigpetrov has quit [Quit: Connection closed for inactivity]
blankhart has joined #nixos
<initiumdoeslinux>
clever: Oh! I see. Why does stack2nix define an attribute for every single package in stackage?
simukis has quit [Quit: simukis]
<clever>
initiumdoeslinux: that was a recent change, prior, to that it only defined attributes for your entire dependency tree
camsbury has joined #nixos
simukis has joined #nixos
<initiumdoeslinux>
clever: are you aware of the reason? 40k lines for a project dependency definition is a lot
<clever>
initiumdoeslinux: i think it was just a side-effect of some recent refactoring, and it wasnt that big of an issue at the time
muffins_Madcat_ has quit [Ping timeout: 252 seconds]
<initiumdoeslinux>
clever: alright, at least I dont have to melt my laptop compiling packages :) Thanks!
simukis has quit [Client Quit]
<jcrben>
nkaretnikov gchristensen that discussion around https://github.com/NixOS/nix/issues/2184 could really use a writeup in the wiki or something - still not really clear on what's going on
<initiumdoeslinux>
lol, nvm. that command is also melting my laptop :)
simukis has joined #nixos
srdqty has joined #nixos
<clever>
initiumdoeslinux: which process is consuming cpu?
<initiumdoeslinux>
ghc
<spear2>
initiumdoeslinux: how old laptop?
<clever>
initiumdoeslinux: if you check `ls -ltrhd /tmp/nix-build*` what is it building?
<initiumdoeslinux>
I was under the impression that the packages would ship as binaries
<spear2>
initiumdoeslinux: i'm not sure how to pass flags to ghc in this case, but --jobs=1 usually makes ghc go single threaded, if you are overheating
<clever>
initiumdoeslinux: stack2nix overrides the versions of everything, so it wont get any help from cache.nixos.org
<initiumdoeslinux>
`drwx------ 12 nixbld13 nixbld 384B May 26 22:19 /tmp/nix-build-Cabal-2.0.1.1.drv-0`
robstr has quit [Quit: WeeChat 1.9.1]
<initiumdoeslinux>
oh, what a bummer. that was the reason I wanted to use nix
<clever>
initiumdoeslinux: if you want help from the cache, you need to apply minimal overrides to haskellPackages
<clever>
initiumdoeslinux: https://github.com/cleverca22/machotool/blob/master/default.nix is another simple example, this one overrides macho and creates machotool, so i can now `nix-build -A machotool` and freely modify the source of the macho library (it had many bugs)
<initiumdoeslinux>
clever: how will this play out with stack? will `stack-repl` use these or try to build everything itself?
<clever>
initiumdoeslinux: stack will never use anything made by either stack2nix or the 2 examples above
<clever>
initiumdoeslinux: stack will always try to rebuild everything on its own
<initiumdoeslinux>
so I can strictly use nix for the building phase?
newhoggy has joined #nixos
<clever>
initiumdoeslinux: let me paste an example of how to repl and build with the machotool example above
dottedmag has joined #nixos
dottedmag has quit [Changing host]
dottedmag has joined #nixos
<initiumdoeslinux>
sure, I would really appreciate that
<initiumdoeslinux>
I am really interested in seeing how I can reduce the amount of compiling I currently have to do
<clever>
initiumdoeslinux: on line 1 i ask for an env suitable for building machotool, which will build everything it depends on using nix, then drop me into a shell
<clever>
initiumdoeslinux: on lines 3 and 6, i then run the configure and repl commands, which gets me into ghci
<clever>
initiumdoeslinux: those commands can be used with both stack2nix, and the simpler machotool examples
<clever>
but stack2nix will involve more rebuilding the first time its ran
<initiumdoeslinux>
Wow! That is neat :)
<clever>
and nix-build will just build the project
<clever>
another handy thing, you can do `ghcid -c "runhaskell Setup.hs repl"`
<initiumdoeslinux>
Does it work with the first example as well? Or do you need an attribute to get into the env?
<clever>
and then ghcid will :r the project every time you save any file, and show the errors
<initiumdoeslinux>
Oh, nice :)
<clever>
arcstats bypasses all of the haskell support in nixpkgs, so you get a usable shell without .env
<clever>
arcstats doesnt even have a cabal file, i generally avoid cabal when i can
<clever>
but machotool does have a cabal file, so it can make use of the haskell support in nixpkgs
newhoggy has quit [Ping timeout: 245 seconds]
sanscoeur has joined #nixos
<initiumdoeslinux>
alright, thanks a lot for helping me get started
<yurb>
I have built supercollider manually from the source tarball and it doesn't crash. How can I debug the crash in the nixpkgs version?
<yurb>
(to possibly fix it in nixpkgs?)
<manveru>
grahamc: thanks
cnidario has joined #nixos
<symphorien>
yurb: to build it the way nixpkgs does, run nix-shell --pure instead of nix-build
<symphorien>
in the resulting shell, run $genericBuild
<symphorien>
hum without $
<sphalerite>
On my netboot machines, I want wake-on-lan to be enabled whenever possible. However, services.wakeonlan seems to want specific interface names, which I don't know for the general case. Is there a way around this or do I need to write a script for it myself?
<clever>
sphalerite: it looks like it has to run ethtool on every interface at shutdown
sanscoeur has quit [Ping timeout: 264 seconds]
<sphalerite>
clever: so basically make my own script that will go through /sys/class/net and try it on all of them?
<NYXT>
Is there a way to have NixOS put all header files installed into the GCC search path rather than only when I enter a nix-shell with those libraries specified?
<clever>
sphalerite: or turn off predictable interface names and just hard-code eth0
<sphalerite>
some of the machines may have more than one network interface
<clever>
sphalerite: i think it is ran in a context without `set -e`, so you could just list every interface that exists on every machine, maybe
<sphalerite>
unrelated, earlyoom is really handy.
<sphalerite>
clever: but I don't know all of those :p I think I'll go with the /sys/class/net way
<clever>
sphalerite: or make a proper look and set its script in powerManagement.powerDownCommands
<clever>
s/look/loop/
<sphalerite>
yeah
<yurb>
symphorien: I ran nix-shell --pure -p supercollider, then in the shell I ran genericBuild and nothing happens (sorry I'm very new to nix)
<clever>
yurb: you need to use -A supercollider
<clever>
-p gives you an env with the (broken) supercollider already in $PATH
<clever>
-A gives you an env with the source, that is suitable for building it
<ldlework>
Does anyone have a snippet that will basically recursively visit a folder of files and do something with each of them? I have a folder of files, and I'd like to generate a systemd preStart script that write all those files to disk using the contents from the source files.
<joepie91>
ldlework: fwiw, the concept you're looking for is 'walk'
<clever>
ldlework: the ${./source} will cause nix to copy the entire source into the nix store, and then deploy it to the remote machine
<clever>
ldlework: and if that is in a prestart script, it will copy to /destination on the remote machine, during prestart
<yurb>
clever: and in what directory should I run the command? It seems to be looking for a default.nix, should I the nixpkgs and run it in the supercollider directory?
<ldlework>
clever: I need to do it in the form of "cat >${path_of_current_file} <<EOL ${contents_of_current_file} EOL
<clever>
yurb: you can also `nix-shell '<nixpkgs>' -A supercollider --pure` to tell it to find the default.nix in NIX_PATH
<ldlework>
I realize the contents of all the files will end up being in that preStart script
<symphorien>
yurb: either do this at the root of your checkout, or nix-shell --pure "<nixpkgs>" -A supercollider
<ldlework>
But it is the only way to provision files in nixops
<clever>
ldlework: why do the contents have to be in the prestart script?
<ldlework>
clever: there's no other place they can be
<clever>
ldlework: nix can copy it to the remote machine for you if you just "cp ${./source} destination"
<ldlework>
what
<clever>
ldlework: thats how dependencies work in nix
<ldlework>
so I can do znc.nix with a znc.conf next to it, and inside znc.nix in the preStart for systemd.services.znc.preStart to cp ${./znc.conf} /var/lib/znc/configs
<ldlework>
?
<ldlework>
seems doubtful
<clever>
ldlework: yes, thats exactly how you can do it
<ldlework>
that's nuts
<ldlework>
clever: thanks again
<clever>
ldlework: just beware that nix will rename the file to /nix/store/hash-znc.conf, and then cp will preserve that name, and create /var/lib/znc/configs/hash-znc.conf on you
<ldlework>
clever: you can only do this in script-y places right?
<clever>
ldlework: so you want to tell cp the full path
<ldlework>
like, there's no module option that will just let you specify files and destinations to copy remote files to?
<clever>
ldlework: you can do it in any string in nix, you just need to be aware of when that string is being executed
<ldlework>
like if I just wanna copy a bunch of crap to /var/lib/znc/moddata without going through systemd?
<symphorien>
ldlework: activationCommands
<clever>
ldlework: you need to use either systemd or an activation script to run the string at bootup
<symphorien>
or activationScripts
<ldlework>
I see.
<clever>
and activationScripts are more fragile, one error in your script and the machine wont even boot
<ldlework>
Ouch.
goaway has quit [Read error: Connection reset by peer]
<clever>
i helped one guy that tried to do network in his activation script, and it worked during `nixos-rebuild switch`, and he had since done garbage collection
<sphalerite>
yeah activationScripts are to be avoided.
<clever>
months later, after forgeting he made such a change, he had a power outage and had to reboot
<clever>
due to the error, it aborted in the middle of the activation scripts, and now systemd isnt in $PATH
<clever>
and confusing errors spew forth
<sphalerite>
the option should probably be marked as internal.
drakonis has quit [Read error: Connection reset by peer]
<ldlework>
a well formalized nixpkg options thing, that took in paths/destinations and then produced a sturdy activationscript for copying files to a host would be good.
<sphalerite>
or, you know, you could just use a systemd service.
<sphalerite>
Or maybe systemd-tmpfiles even
<clever>
ldlework: any time you do ${./foo} in a string in nix, nix will copy the path to /nix/store, and substitute in that store path
<ldlework>
It seems very strange to use a systemd service who's only function is copying some files when it starts...
<clever>
ldlework: and it automatically becomes a dependency, so if your string winds up in a script, nix will copy those files before your script, at all times
<sphalerite>
ldlework: what's conceptually wrong with it? It even allows you to declare how other services depend on it.
nur0n0 has joined #nixos
clacke has quit [Quit: Page closed]
joehh has quit [Ping timeout: 252 seconds]
<yurb>
Okay, so I've rebuilt supercollider from nix-shell and the crash is reproducible. So it seems to be related to some versions of libraries, probably the jack library (I see an assertion failing right before the crash). What approach should I take to attempt to fix it?
<clever>
yurb: if you build it from the same copy of nixpkgs, then it will have the same versions of everything
<symphorien>
did you look at the output of ldd ?
krey has quit [Ping timeout: 252 seconds]
dottedmag has quit [Quit: QUIT]
Cypi has joined #nixos
<yurb>
clever: yes, I first installed supercollider with nix-env and the crash was there. When compiling manually (outside nix), it doesn't crash.
<Cypi>
Hi, I encountered this issue https://github.com/NixOS/nixpkgs/issues/33274 while upgrading teamspeak-client. I am not knowledgeable to understand what the suggested fixes were, what should I do?
krey has joined #nixos
<clever>
yurb: what about editing the source, then building it with both nix-shell and nix-build (using an override), and comparing the output, add print statements near the point of failure and compare values
tomberek has joined #nixos
<yurb>
clever: I'm afraid my current knowledge of nix is too limited to understand your suggestion:( Can you clarify what is override? Also, by point of failure you mean when the crash happens in the supercollider source?
<clever>
yurb: yeah
tzemanovic has joined #nixos
<symphorien>
Cypi: is it installed with nix-env ?
<Cypi>
Yes it is
<sphalerite>
meh, the wake-on-lan isn't working, even though ethtool doesn't complain
<symphorien>
qt is still impure with nix, so make sure all qt-related package use the same qt version
<symphorien>
ie: uninstall and reinstall them all
<Cypi>
I don't know how to distinguish a qt-related package from a non-qt-related package
<tomberek>
anyone know if you can instantiate a systemd service file directly on non-NixOS? Even just building it so it can be linked to, but uses all the proper /nix/store paths inside.
Ariakenom has quit [Read error: Connection reset by peer]
<infinisil>
tomberek: Yeah hold on
<Cypi>
Also, does that concern solely what I installed with nix-env, or also everything in my system-wide configuration.nix?
simukis has quit [Quit: simukis]
<symphorien>
your systemwide configuration is built from a single revision of nixpkgs so a single qt version; those one should be in sync
simukis has joined #nixos
<sphalerite>
tomberek: nix build -f '<nixpkgs/nixos>' 'config.system.build.units."nix-daemon.service".unit' should work if you have a /etc/nixos/configuration.nix
<sphalerite>
tomberek: if not, pass `--arg configuration path/to/configuration.nix`
<tomberek>
infinisil: sphalerite (how did you find that answer?) (teach a man to fish.. etc)
<Cypi>
I'm trying to upgrade everything remotely related to any kind of UI in my nix-env, I'll tell you if it's enough
<sphalerite>
tomberek: it doesn't seem to generate the [Install] section though, I think nixos handles dependencies itself by creating .wants directories or something
tzemanovic has quit [Ping timeout: 252 seconds]
<sphalerite>
tomberek: nix repl '<nixpkgs/nixos>', then explored the attrs in config.system.build
<hodapp>
add qt5 to buildInputs, get "error: cannot coerce a set to a string, at /nix/store/wp7dj4zm44gb84ppg0dhx3mbqcj2qns0-nixos-18.03.132083.06c576b0525/nixos/pkgs/stdenv/generic/make-derivation.nix:148:11"
<clever>
hodapp: qt5 is a set
<clever>
hodapp: you have to pick an attribute inside qt5
<clever>
error: undefined variable 'config' at (string):1:2
<symphorien>
Cypi: a possible scenario is: 1/ you install vlc with qt v5.9 2/ you upgrade you channel 3/ you install teamspeak with the version of the new channel: v5.10 4/ crash
<infinisil>
--arg/-A?
<clever>
`nix build` uses some weird algos to decide if your refering to a file or a nix expression
<tilpner>
clever - Didn't notice anything like that
<infinisil>
clever: Maybe check again, I think it's been improved
<clever>
tomberek: with the above, it now expects you to have a configuration.nix file in the current dir
<tilpner>
clever - I agree about nix build though, it's been worse to use than nix-build
drakonis has joined #nixos
<sphalerite>
infinisil: you don't have to write C++, you can also just write up a wonderful alternative of how the UI should work and if it's good I'm sure someone else will be willing to implement it :)
<infinisil>
sphalerite: That would be something yeah
don_quijote[m] has joined #nixos
<tomberek>
(side conversation), nixpkgs is the de-facto standard library for Nix, and has many useful function all over the place, but discoverability is poor.
<sphalerite>
"Note: this program is EXPERIMENTAL and subject to change." :)
<infinisil>
sphalerite: Then it shouldn't be 2.0
blankhart has quit [Ping timeout: 260 seconds]
<nur0n0>
maybe some sort of wiki to compile UI deficiencies would be helpful?
<sphalerite>
infinisil: There is no reason for it not to be.
fragamus has quit [Max SendQ exceeded]
<hodapp>
tomberek: yes, I rather consistently have problems figuring out effective ways to accomplish commonplace things in it
<sphalerite>
infinisil: nix-build and friends are still the stable CLI, they grew a bunch of features with the 2.0 release like alternate stores.
<infinisil>
Hmm I guess
<sphalerite>
Who knows, maybe a completed new UI will warrant a 3.0.
dbe has quit [Ping timeout: 252 seconds]
<sphalerite>
none of wakelan, wol and pythonPackages.wakeonlan seem to allow specifying which interface to send wake-on-lan packets on, which is unfortunate because they choose the wrong one for me. Does anyone know a tool that does allow specifying this, or sends the packet on all interfaces?
<clever>
sphalerite: let me check my old notes...
<clever>
404, notes not found
<tomberek>
sphalerite: infinisil: okay, i got my own services working now with a super minimal dummy-nixos configuration.nix
<sphalerite>
tomberek: \o/
lord| has quit [Ping timeout: 260 seconds]
<WilliamHamilton>
what's the equivalent of cabal2nix for the python world? I found some alternative but I'm confused by their usage: for example, what's the equivalent of the .cabal file to pass to it?
<tomberek>
WilliamHamilton: buildPythonPackage is pretty good
<drakonis>
pypi2nix
<WilliamHamilton>
tomberek: should I write a .nix file or can I invoke this command from the shell?
<WilliamHamilton>
drakonis: installing
<tomberek>
WilliamHamilton: that is for use inside a nix expression, see pkgs/top-level/python-packages.nix for examples
<tomberek>
sphalerite: infinisil: how would you do the same thing as calling that "nix build" but having it all be done in a nix-expression?
<sphalerite>
tomberek: the thing in the parens is the nix expression :)
<tomberek>
copy
<tomberek>
i guess that was a dumb question
<WilliamHamilton>
drakonis: in the --help of pypi2nix it's written that I should use "requirements.txt" generated by pip freeze. However, if I try `pip freeze` it complains about an old version of pip (9.0.1). What should I do?
<ben>
do yall know why i need to compile nvidia-settings locally rather than getting a build from a binary cache?
<drakonis>
WilliamHamilton, you should direct it to others in this channel
<infinisil>
ben: Sounds like that would be unfree software, which can't be provided for download
<drakonis>
you compile the nvidia module yourself
sanscoeur has joined #nixos
johnw has quit [Ping timeout: 245 seconds]
johnw has joined #nixos
sanscoeur has quit [Ping timeout: 248 seconds]
<tomberek>
is is correct to say that while evaluating a nix expression it does the equivalent of nix-instantiate?
<clever>
tomberek: yeah
<clever>
tomberek: previously, nix-build was a perl script that did nix-instantiate and nix-store -r
<clever>
it has since been replaced with a c++ binary that directly links to the nix libs and calls those internally
aarvar has joined #nixos
camsbury has quit [Quit: WeeChat 2.0]
<tomberek>
clever: and then using a nix expression inside a "let ... in.." construct is the same as nix-instantiate --eval?
<clever>
tomberek: depends on what is in scope
<clever>
nix-repl and/or `nix repl` can also be used to test things more easily
<tomberek>
yeah, my primary use is inside a nix expression that will be a declarative hydra build. I just want to have the right mental model as I test things and interactively set everything up.
Guanin has quit [Ping timeout: 260 seconds]
<sphalerite>
clever: managed to get it working by setting the broadcast address to a narrower one, reachable via the interface
<clever>
sphalerite: oh, and something else about wake-on-lan, it will activate if the card sees the magic bytes, in any packet that reaches it
<clever>
sphalerite: so you can either target its mac (hard with it not answering to ARP) or target the mac level broadcast addr
<sphalerite>
yeah
<clever>
i have made it work before by port-forwarding a port to 192.168.2.255 on the router, and then sending the magic packet from outside
<ldlework>
I just tried 'nix-env -i discord' since there is a `discord` package in nixpkgs but it can't be found that way. When I can I 'nix-env -i something' and when can't I? And in the cases I can't, how do I get foo installed? I can't seem to internalize the underlying pattern here.
<clever>
ldlework: nix-build '<nixpkgs>' -A discord
<ldlework>
yes, but what is the underlying pattern here
hiratara has quit [Ping timeout: 256 seconds]
<ldlework>
why can somethings be 'nix-env -i foo' installed?
<ldlework>
and other things not? How do I tell?
<clever>
ldlework: the first problem is that -i wont show unfree packages if you havent turned on unfree
<clever>
ldlework: and the 2nd problem is that `nix-env -iA nixos.discord` wont show the unfree warnings
<ldlework>
Interesting.
<ldlework>
does nix-build actually install it ?
<clever>
it puts it in /nix/store and creates a result symlink pointing to it
<clever>
but it wont be in $PATH by default
<drakonis>
why do we hide nonfree software?
hiratara has joined #nixos
<ldlework>
clever: you mean until I restart the terminal?
<clever>
nix-build will never add it to PATH
<ldlework>
I see. Why did you suggest it then? Not a challenge, I'm curious.
freeman42x[nix] has quit [Ping timeout: 245 seconds]
Acou_Bass has quit [Ping timeout: 245 seconds]
Tethys has joined #nixos
palo_ has quit [Ping timeout: 240 seconds]
palo has joined #nixos
palo has quit [Changing host]
palo has joined #nixos
tzemanovic has joined #nixos
roconnor_ has quit [Ping timeout: 256 seconds]
sanscoeur has joined #nixos
lord| has quit [Ping timeout: 252 seconds]
foldingcookie has joined #nixos
<foldingcookie>
how does one specify to patch a package in /etc/nixos/configuration.nix?
<foldingcookie>
I have as much as `pkgs.foobar.overrideDerivation (oldAttrs: { patches = [ "/path/foobar.patch" ]; });` but I'm not sure how to replace the original pkgs.foobar with it
Acou_Bass has joined #nixos
sanscoeur has quit [Ping timeout: 252 seconds]
sbdchd has joined #nixos
<foldingcookie>
it appears I need to wrap that with parentheses and put it in environment.systemPackages?
smallville7123 has joined #nixos
<elvishjerricco>
foldingcookie: You should use an overlay
<elvishjerricco>
Put something like that in configuration.nix
lord| has joined #nixos
<elvishjerricco>
I made a couple tweaks. Namely I made sure to keep any old patches, and I used `overrideAttrs` instead of `overrideDerivation`
<elvishjerricco>
IIUC, `overrideAttrs` directly overrides the arguments to mkDerivation, but `overrideDerivation` overrides something much lower level so it's not usually suitable
<elvishjerricco>
Though they often work for the same things