<simpson>
I have a Nix expression which builds a link farm from a tree of source files; it recursively explores the tree for source files which it knows how to compile, and individually builds each file.
<simpson>
I want to build a second farm, using the first farm as part of the executable code for the builder. (This is a bootstrapping step; some source files should be compiled after the initial compile.) But I also want the two farms to be merged. Is there a Nix idiom for this, or should I build the two farms independently and then merge them?
<lovesegfault>
Mesa 20 is still not on unstable small :O
* lovesegfault
cries in staging
<blumenkranz>
Hello. I'm attempting to deploy to a libvirtd environment with NixOps. Everything is fine until it gets to the point where postgresql attempts to start, but I get a "timeout" error, and I really don't get what am I doing wrong.
<blumenkranz>
Apparently there used to be an issue with 19.09 (the version I am using for the environment) where this happened, but it was fixed in master. Does anybody know if the current 19.09 tag is bugged?
<bqv[m]>
lovesegfault: you made me check if in advanced again :/
<bqv[m]>
It*
<lovesegfault>
It hasn't
<lovesegfault>
and there are 60k builds on the staging-next queue
<blumenkranz>
Do old NixOS versions still get security updates? I'm considering downgrading.
<b42>
blumenkranz: i've got 19.09 with postgresql running in libvirtd guest and haven't run into this - anything interesting in the logs?
<lovesegfault>
142k builds queued in hydra total
<blumenkranz>
b42: That's the thing, there is no /var/log/postgresql
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/JvFow
<blumenkranz>
I am baffled. Will try with systemctl now, but the console hasn't spewed anything interesting
<infinisil>
blumenkranz: Nope, old versions only get updates for ~1 month after a new version is released
<b42>
blumenkranz: journalctl -b -u postgresql
<b42>
?
mexisme has joined #nixos
bebarker has joined #nixos
<b42>
or maybe even "journalctl -b" to see all messages not just from postgres
<blumenkranz>
b42: Wow. The DB does startup correctly, but then systemd just decides it has timeout'd and it proceeds to kill it
<b42>
o_O
<blumenkranz>
b42: I get "database system is ready to accept connections", then immediately after, systemd kills the service
<b42>
blumenkranz: got any initialScript/ensureUsers/ensureDatabases? looks like that's where service times out
<blumenkranz>
b42: Not really, I just have enabled the postgres service. But I also get the following line
<blumenkranz>
n2ig6d12rk5p00bgim3lzn2cz36csd1k-unit-script-postgresql-post-start[3728]: /nix/store/n2ig6d12rk5p00bgim3lzn2cz36csd1k-unit-script-postgresql-post-start: line 5: kill: (3719) - No such process
<blumenkranz>
It's after the DB has already shut down. I guess it has nothing to do
<blumenkranz>
Just for reference, are you using commit d5291756487d70bc336e33512a9baf9fa1788faf, b42?
<{^_^}>
[nixpkgs] @sikmir opened pull request #84145 → qmapshack: add gdal/routino tools to PATH → https://git.io/JvFoF
jlv52 has quit [Remote host closed the connection]
<b42>
blumenkranz: i'm on d011e4749457af484adf2e90062c83a44ad072a4 - d529175 looks quite old
<b42>
hmm, perhaps even the first 19.09 commit?
<lopsided98>
blumenkranz: you shouldn't use the 19.09 tag; use the release-19.09 branch
<blumenkranz>
b42: Yeah, I know for sure 19.09 used to have this same bug at some point. Might just be using an old version.
<blumenkranz>
lopsided98: Ah, that might be what's wrong. Thanks for the tip!
<b42>
you should be using the nixos-19.09 branch actually
<blumenkranz>
Is there any place where I can read about how does nixpkgs channels versioning work, and how does support work for older versions?
<b42>
it follows release-19.09 but must pass CI and there are guaranteed(tm) packages in binary cache
rogue_koder has quit [Ping timeout: 260 seconds]
<blumenkranz>
Will 19.09 be updated in case of vulnerabilities?
tertl3 has joined #nixos
<blumenkranz>
I mean, nixos-19.09
rogue_koder has joined #nixos
<blumenkranz>
I'm kinda new to channels. Been using nixos-unstable all this time because I need muh bleeding edge, but now I want some stability for this server.
<Kyndig>
Hello, is there a protocol for asking questions?
<simpson>
Yes: Ask questions, get answers.
<Kyndig>
:)
xkapastel has quit [Quit: Connection closed for inactivity]
bvdw has quit [Read error: Connection reset by peer]
<Kyndig>
I'm running nixpkgs, single user with the store in my home dir on an old redhat server; created an overlay to make a custom <prefix>/etc/gitconfig. It worked and I see the config file in the store path, but git is trying to read /etc/gitconfig instead of the prefixed path (evidenced by an strace). Am I wrong to expect it to read the prefixed etc/?
bvdw has joined #nixos
<simpson>
I'd expect it to look at /etc/ first. I'm not sure how one would point it at the store instead.
orivej has quit [Ping timeout: 260 seconds]
orivej_ has joined #nixos
<Kyndig>
I thought that the build tool replaced all the paths with $out dir paths.
<Kyndig>
of course ~/ would still be ~/
<Kyndig>
so i would expect .gitconfig to work
<{^_^}>
[nixpkgs] @eadwu opened pull request #84154 → clight: include location assertion → https://git.io/JvF6D
<{^_^}>
[nixpkgs] @chkno opened pull request #84155 → darcs: Fix build (Use old GHC) → https://git.io/JvF69
<ottidmes>
Kyndig: not paths like /etc, just libs and bins, AFAIK
<Kyndig>
ahh
<Kyndig>
so i should be running nixos, not nixpkgs
<Kyndig>
well i am at home; unfortunately don't have a choice at work
<ottidmes>
Kyndig: at work you could workaround it by making a wrapper setting $HOME to another path that contains your git config
<{^_^}>
[nixpkgs] @eadwu opened pull request #84158 → plymouth: attempt to make flicker-free → https://git.io/JvF6A
<ottidmes>
Kyndig: why could you not create a wrapper named git and give it prio on PATH, should have nothing to do with interactive use
<Kyndig>
ahh; i see
<Kyndig>
a wrapper that's a real store wrapper
<Kyndig>
hmm; that should work
<Kyndig>
dunno why i didn't think of it
<Kyndig>
thanks
<ottidmes>
No problem, good luck!
<Kyndig>
i'm kinda slow
<Kyndig>
new
<Kyndig>
:)
coco has quit [Ping timeout: 258 seconds]
<ottidmes>
Slow but steady works too :) And it takes some time to learn all the tricks needed to work comfortably with Nix(OS)
<blumenkranz>
b42: Hey, I now get "error: cannot figure username" when attempting to deploy nixos-19.09. Apparently this is a known bug, but how did you manage to bypass it?
<morgrimm>
Hey folks - would it be possible to write a derivation for a shell that simply builds as normal, but moves dotfiles into place? (maybe evaluating them with the current variables in scope?)
<morgrimm>
i.e. build fish as normal, but post build, but after build move a set of dotfiles into `$out/etc/fish` ?
<DigitalKiwi>
anyone seen ethernet wired in reverse?
iyzsong has joined #nixos
rardiol has quit [Ping timeout: 260 seconds]
<samueldr>
hm?
<energizer>
morgrimm: sounds like home-manager
<DigitalKiwi>
so say you have a type B from the flat side it's left to right; orange white; orange; green white; blue; blue white; green; brown white; brown; these were wired like that but from right to left
rardiol has joined #nixos
<morgrimm>
energizer: ... Yeah, you're right - home-manager totally works
<morgrimm>
that's sweet
<DigitalKiwi>
so instead of pin 1-1 2-2 3-3 4-4 5-5 6-6 7-7 8-8 it was 1-8 2-7 3-6 4-5 5-4 6-3 7-2 8-1
<DigitalKiwi>
but only on one end the other end was in the right order (but in wall)
jlv has joined #nixos
orivej has quit [Read error: Connection reset by peer]
<{^_^}>
[nixos-homepage] @garbas pushed 2 commits to governance: https://git.io/JvFPX
OrangeTetra has joined #nixos
<OrangeTetra>
I've noticed some packages like KVM and vim-surround have other packages that look like the same thing... How does one tell the difference? (I'm mostly thinking from ad-hoc teriminal perspective)
rogue_koder has quit [Ping timeout: 260 seconds]
rogue_koder has joined #nixos
oranginal has quit [Ping timeout: 265 seconds]
knupfer has joined #nixos
<evils>
OrangeTetra: what other packages?
<davidak[m]>
DigitalKiwi i have never seen that. might be some special cable
<OrangeTetra>
evils: KVM and qemu-KVM are the same, vim-surround and surround seem to be the same...
knupfer has quit [Remote host closed the connection]
captjakk has quit [Remote host closed the connection]
<evils>
> kvm.pname
<{^_^}>
"qemu-host-cpu-only"
bebarker has quit [Ping timeout: 265 seconds]
captjakk has joined #nixos
<{^_^}>
[nixos-homepage] @ryantm opened pull request #379 → fixup Discourse team → https://git.io/JvFXv
<OrangeTetra>
evils: Is there a way to get this information from my terminal?
<evils>
nix repl
OrangeTetra has quit [Remote host closed the connection]
OrangeTetra has joined #nixos
captjakk has quit [Ping timeout: 265 seconds]
<evils>
OrangeTetra: though nix search gives you the name attribute in the parenthesis, which is usually (pname-version)
rogue_koder has joined #nixos
hmpffff has joined #nixos
<OrangeTetra>
evils: I only got the message starting with "though" did you send something before that? I look like you might have.
captjakk has joined #nixos
<evils>
OrangeTetra: my reply was "nix repl", which is exactly what the bot uses for `> kvm.pname` etc
<OrangeTetra>
Ah, okay thank you.
hmpffff_ has quit [Ping timeout: 246 seconds]
<evils>
OrangeTetra: nix search --json will give you an explicit "pkgName" field, which is the pname attribute, and is usually the unique identifier for a package
<OrangeTetra>
It outputs in a difficult to read format, though, so I probably won't use that.
<evils>
OrangeTetra: pipe it to jq, like `nix search --json kvm | jq`
<{^_^}>
[nixos-homepage] @garbas pushed 2 commits to governance: https://git.io/JvFXO
tertl3 has quit [Quit: Connection closed for inactivity]
<OrangeTetra>
evils: ah, nice. Thank you.
<sasycto>
I just noticed in my version of nix it seems like `du` is actually the binary for `sort`
Supersonic has quit [Disconnected by services]
Supersonic112 has joined #nixos
Supersonic112 is now known as Supersonic
<sasycto>
apparently it's getting it from /run/current-system/sw/bin/du
<sasycto>
I need to install whatever derivation has du then, clearly
<clever>
sasycto: what do you get when you run `ls -l` on that path?
<sasycto>
A lot of simlinks, and the du one links to coreutils-8.31 du, so I guess coreutils has a bug, lol
<sasycto>
or the derivation is wrong
knupfer has quit [Remote host closed the connection]
<wrunt>
I think busybox has a thing where all core commands (such as du and sort) are symlinks to a single binary, which behaves differently depending on arg[0]
<wrunt>
maybe coreutils is the same
<wrunt>
that would explain what you're seeing, sasycto
<pjt_014>
clever: is there a platform option that covers just x86-64, i686 and aarch64?
<pjt_014>
If it covers a few that are 'close enough' like gnu64 that's fine too
<clever>
pjt_014: dont think there is one
knupfer has joined #nixos
knupfer has quit [Remote host closed the connection]
oida has quit [Remote host closed the connection]
oida has joined #nixos
captjakk has quit [Remote host closed the connection]
captjakk has joined #nixos
OrangeTetra has quit [Ping timeout: 250 seconds]
kaliumxyz has joined #nixos
captjakk has quit [Ping timeout: 256 seconds]
<samueldr>
wrunt: coreutils has the same
<samueldr>
single call binary
<sasycto>
that's weird.
<samueldr>
multi-call*
oranginal has joined #nixos
<oranginal>
Is there a recommended method of setting a (say) GTK theme in NixOS?
<oranginal>
or just... flipping into dark mode?
lovesegfault has joined #nixos
<{^_^}>
[nixos-artwork] @samueldr merged pull request #49 → 20.03 Markhor art work → https://git.io/Jvdp8
<{^_^}>
[nixos-artwork] @samueldr pushed 3 commits to master: https://git.io/JvFXj
waleee-cl has quit [Quit: Connection closed for inactivity]
<pjt_014>
also evils: If you're around, I've made a few changes to it.
<etu>
pjt_014: I see 6 commits in the log. I only recognize two of them as needed. The rest should be squashed/rebased on master if there was a conflict :)
captjakk has quit [Ping timeout: 265 seconds]
ArchFeh has quit [Remote host closed the connection]
<lovesegfault>
bricewge: Are you around?
fusion809 has joined #nixos
<pjt_014>
It's more the actual package file/build process result and such that I'm worried about. I'd prefer not to wrestle with git for another 2 hours if possible.
<srhb>
lejonet: ceph 14.2.8 is now woring without the horrible patch to cherrypy loading. :-) https://github.com/NixOS/nixpkgs/pull/83749 -- I guess I'll start an Octopus PR soon after that's in.
<kiloreux>
I am trying to compile nethack-ftp on nix. But getting `error: use of undeclared identifier 'ARG_MAX` How can I get the value of ARG_MAX though nix?
<Aleksejs>
Hi, is it possible to see the result sites file that nix generated from .services.nginx.virtualHost entry?
<{^_^}>
[nixpkgs] @filalex77 opened pull request #84172 → kondo: init at 0.3 → https://git.io/JvFQ6
<pjt_014>
mokasin_: Importing it as described would probably be correct. If it modifies the kernel, also something like "boot.kernelPackages = pkgs.linuxPackages_hardended" in the config file unless it already does that.
<symphorien>
Aleksejs: systemctl cat nginx will give you the path to the nginx.conf file
<mokasin_>
Aleksejs, you can get the path to the config looking at systemctl status nginx and have a look there
CMCDragonkai1 has joined #nixos
<mokasin_>
pjt_014, well, I guess only one way to find out :)
<Aleksejs>
thanks :)
<pjt_014>
yup. also if it interferes with your config you may need to use mkOverride if you want your settings to prevail
o1lo01ol1o has joined #nixos
<pjt_014>
foo = pkgs.lib.mkOverride 0 bar I think
Darkmatter66_ has joined #nixos
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
zupo has joined #nixos
Darkmatter66 has quit [Ping timeout: 264 seconds]
<mokasin_>
pjt_014, I am not sure I understand what you are saying. what does mkOverride 0 do?
mexisme has quit [Ping timeout: 246 seconds]
o1lo01ol1o has quit [Ping timeout: 264 seconds]
<{^_^}>
[nixpkgs] @peterhoeg merged pull request #83799 → cantata: make qtmultimedia and vlc optional → https://git.io/Jv7di
<mlatus[m]>
<mokasin_ "pjt_014, I am not sure I underst"> incase the profile and your config both define option foo, the value with `mkOverride 0` will be picked
mexisme has joined #nixos
<mlatus[m]>
mokasin_: i.e. `mkOverride 0` sets higher priority
<mokasin_>
noob question: How do I get access to mkOverride in my configuration.nix? It is not in scope
<ikwildrpepper>
mokasin_: add lib to the function arguments of the module you are in
<ikwildrpepper>
{config, lib, ...}: { ... }
knupfer has quit [Remote host closed the connection]
knupfer has joined #nixos
<ikwildrpepper>
then you can use lib.mkOverride
<mokasin_>
ikwildrpepper, does the order matter?
<ikwildrpepper>
mokasin_: no
<mokasin_>
cool
<mokasin_>
what arguments are passed to a module?
<kenran>
If I have a Haskell `default.nix` and my `shell.nix` returns `(pkgs.haskellPackages.callPackage ./default.nix {}).env`, but I now want more "buildInputs", for example if I want to have `cabal-install` inside my nix-shell. How would I do that?
<tilpner>
mokasin_: Probably the same reason NixOS doesn't use linuxPackages_latest by default, but linuxPackages, which is the current longterm branch
<mokasin_>
tilpner, oh, for some reason I thought it was using a 5.x kernel. my bad. makes sense
<tilpner>
mokasin_: Current longterm is 5.4
<mokasin_>
tilpner, well, then linuxPackages_hardened is a 4.17 kernel
<tilpner>
> linuxPackages_hardened.kernel.name
<{^_^}>
"linux-5.4.30"
<mokasin_>
wtf
<tilpner>
Your nixpkgs is too old
<mokasin_>
r u on unstable?
<tilpner>
It's also on 20.03
<mokasin_>
ah, okay. I'm still using 19.09. :)
hmpffff has joined #nixos
<mokasin_>
Now the confusion is complete ;). And lifted.
<tilpner>
Which makes sense, considering 20.03 has not been released yet
gxt has quit [Remote host closed the connection]
Darkmatter66_ has quit [Quit: ZNC 1.7.5 - https://znc.in]
hmpffff_ has quit [Ping timeout: 240 seconds]
Jackneill has joined #nixos
thibm has joined #nixos
gxt has joined #nixos
Darkmatter66 has joined #nixos
alci has joined #nixos
chloekek has joined #nixos
laduke has quit [Ping timeout: 272 seconds]
dermetfan has joined #nixos
dermetfan has quit [Client Quit]
<alci>
Hi, could anyone tell me where I could get some help regarding a nixos.configuration failed rebuild other than using github issue?
<ratsclub>
Which problem did you have?
<ratsclub>
I would sugest you to try to get help here and perhaps document it on github
linarcx has quit [Read error: Connection reset by peer]
<alci>
the github issue is #83767. Someone closed it saying that it has beens solved while I still can't build it...
<tilpner>
mokasin_: Why do you want to use unstable?
laduke has joined #nixos
<ottidmes>
mokasin_: you could also always opt for a middle ground, using stable as a basis but cherry pick packages to be used from unstable, where its important to use the very latest, I do that for all my systems
<alci>
unstable offers more packages
<benny>
I run unstable for 2 years and it's mostly fine, big updates like with any rolling distribution are an issue, especially things like kde5
<benny>
ottidmes: that seems wise
linarcx has joined #nixos
c0c0 has quit [Ping timeout: 256 seconds]
* tilpner
was going to suggest the exact same as ottidmes, depending on answer
<ratsclub>
As you posted an IPFS link, lemme ask you guys something? How can I upgrade my ipfs version? I'm at 0.4.21 and I feel like I'm not connecting to as many nodes as I should
<chloekek>
Is it possible to stop core.* files from appearing in my home directory when programs crash? systemd.coredump.enable = false; did not do the trick.
<tilpner>
chloekek: Re-read the description of systemd.coredump.enable
<chloekek>
Thanks. Is it possible to have them just never be generated at all?
<chloekek>
So neither appear anywhere nor be processed by systemd.
<tilpner>
What does your ulimit -c say?
<chloekek>
unlimited
<tilpner>
0 would probably prevent it
<tilpner>
But I haven't tested it
<tilpner>
You can set that value globally from security.pam.loginLimits
<ottidmes>
bqv[m]: do you know if there is more information about the outputs of flakes? And how do you deal with creating pkgs beforehand, while your NixOS config might set nixpkgs.config that would be needed to define pkgs?
<{^_^}>
[nixpkgs] @zowoq opened pull request #84177 → maintainers/teams: add podman team → https://git.io/JvFdz
<bqv[m]>
drakonis Oh boy I think its building
linarcx has quit [Quit: WeeChat 2.8]
<mokasin_>
tilpner, sometimes there is a feature in a software, I'd like to have earlier. nextcloud is an example
<mokasin_>
ottidmes, I a bit vary of that, because I am not familiar enough with nix to assess what a backdraft I have to expect
<bqv[m]>
I think flakes can have whatever outputs they like, but don't quote me on that. Packagewise I'm not sure what you mean?
<bqv[m]>
ottidmes: have you seen my repo? Might help
<ottidmes>
bqv[m]: I thought as much with the outputs, seemed arbitrary
<mokasin_>
ottidmes, I've tried it once without issues. But I haven't found out yet, how to use a module form unstable, only how to load a package from another channel
<ottidmes>
bqv[m]: yep, that means you missed my messages yesterday, I commented on how I made a git clone of it and saw a potential issue, not sure if you meant to copy over your entire config to the /nix/store, or does it not contain anything private (not sure how git crypt e.g. works)
<ottidmes>
mokasin_: module from unstable is like this, you just specify the module you want with disabledModules (google it for examples), and then import the unstable module instead: imports = [ "${pkgs.unstable.path}/nixos/modules/..." ];
<bqv[m]>
i resolved on the fact that since the entire config is required to build anyway (and is therefore in the store) then using that is fine
<ottidmes>
bqv[m]: wait, so even if you have a local git repo, it will copy it over to /nix/store when using flakes?
fusion809 has joined #nixos
<bqv[m]>
yes.
<bqv[m]>
at least, it will copy everything tracked by git
<tilpner>
mokasin_: Ahh, if you require the newer modules, then using unstable is reasonable
<ottidmes>
bqv[m]: I have to disagree with that idea though, there is a difference between a git aware copy and a complete snapshot of a directory (the former does not copy gitignore and stuff, while the latter does)
<alci>
<tilpner> thanks a lot. I didn't have any issue for a long time and forgot the necessary update step! Now it's building ok. New issue: tor-browser not building... investigating
<tilpner>
mokasin_: Mixing modules between channels can be done, but good luck if there's some assumption that you're breaking by using module combinations that have never tested before
<tilpner>
*never been tested before
gulplante has joined #nixos
<tilpner>
alci: tor-browser-bundle-bin
<tilpner>
alci: The other package was removed
linarcx has joined #nixos
<bqv[m]>
ottidmes: it's not very secure, agreed, but i also understand why it was done, and i'm happy to use it as an excuse to link my config out
linarcx has quit [Read error: Connection reset by peer]
<mokasin_>
ottidmes, thanks. I have to save that snippet
<ottidmes>
bqv[m]: but I am telling you, there is a complete difference in terms of security between what you are doing and what flakes is doing (I asumme, at least).
<mokasin_>
tilpner, yes, exactly. I fear, I'll break my system and cannot return, because of some database migration, for example
Darkmatter66 has quit [Quit: ZNC 1.7.5 - https://znc.in]
<mokasin_>
tilpner, ottidmes thx for the insights
xkapastel has joined #nixos
roosemberth has quit [Quit: WeeChat 2.8]
<bqv[m]>
ottidmes: what? no, as i understand it nix is already importing the config into the store as part of the flake mechanism. e.g. i'm absolutely certain it does that for flake inputs, and i believe it does it for the flake in question too?
<tilpner>
mokasin_: That can happen, and it will be hard to help you if you have a system that's been patched together from many channels
<tilpner>
mokasin_: FWIW, I've been using unstable matrix-synapse for a few months, and it's been working out okay-ish
<tilpner>
mokasin_: No data loss, but the module began using postgres options that were on unstable only
linarcx has quit [Read error: Connection reset by peer]
<tilpner>
mokasin_: So what do you do? Use the unstable postgres module
<ottidmes>
bqv[m]: yes, but since the git imports need all files to be tracked, I am quite certain it will not just copy over any files in gitignores, like in my case, my secret files that are within the same repo, and doing ${../.} would just copy over the directory recursively, stupidly, including my secret files, while a git clone (which I assume flakes does for git repos), would not end up having them
<tilpner>
mokasin_: My point is: it can work (does for me), but if it breaks you might hear "we told you so"
<tilpner>
TheSirC[m]: How are you using overlays/default.nix?
zaeph1 is now known as zaeph
<TheSirC[m]>
<tilpner "TheSirC: How are you using overl"> ```
<bqv[m]>
huh, i seem to have grown a new gpg key...
<ottidmes>
mokasin_: ^ yep, nixpkgs works together well within the same channel, if not, thats a bug, but outside of that you run the risk of breakage, due to the breaking of assumptions
<tilpner>
TheSirC[m]: There were no words in your last message
<FRidh2>
If you can't use Nix but need a Python interpreter, see https://github.com/FRidh/static-python for statically-linked builds. One major bug is left though
<TheSirC[m]>
tilpner: It was a code snippet, I just import it
<tilpner>
TheSirC[m]: No, there is no "just import it". Your method of importing may be the problem here
<ottidmes>
TheSirC[m]: you probably do not have the format of (self: super: { ...stuff... }), maybe you have just { ...stuff... }, and that's not a valid overlay, even if you don't use the arguments
<TheSirC[m]>
<tilpner "TheSirC: No, there is no "just i"> `imports = [ ../../overlays/default.nix];`
<tilpner>
TheSirC[m]: Okay, you treat it like a module, that's good
<TheSirC[m]>
I know. That's why I included it in the first code snippet
alci has quit [Quit: Leaving]
<tilpner>
I don't recall if value modules work, or if they need to be functions
<FRidh2>
simpson: if we get pypy to work again then that can be the next...i don't have the patience for those builds though :p
<TheSirC[m]>
ottidmes: Hum so you thin that it is one of the overlay that is miswritten
kenran has quit [Ping timeout: 256 seconds]
romildo has joined #nixos
<TheSirC[m]>
* ottidmes: Hum so you think that it is one of the overlay that is miswritten
<TheSirC[m]>
* ottidmes: Hum so you think that it is one of the overlay that is miswritten
<ottidmes>
TheSirC[m]: lol, I see the message three times, but yeah, it could be, since that is what is being reported
<FRidh2>
simpson: the note under the title implies the official builds are portable already nowadays?
_rvl has joined #nixos
<tilpner>
TheSirC[m]: I think overlays/default.nix might end up in your overlays
<tilpner>
TheSirC[m]: But it's a module, so fails the overlay check
<tilpner>
TheSirC[m]: Check from nix repl, or filter it out explicitly
<TheSirC[m]>
tilpner: Wouldn't that trigger infinite recursion
<romildo>
Should backports from master to NixOS 20.03 go into which branch? release-20.03 or nixos-20.03?
<tilpner>
TheSirC[m]: No, I don't think so
<TheSirC[m]>
<tilpner "TheSirC: Check from nix repl, or"> I tried the replk without luck
<tilpner>
romildo: The one with the newer commit
<simpson>
FRidh2: The official builds include portable-pypy-style builds, yes. They do not use Nix or other systematic tools, but have an ad-hoc script.
<tilpner>
romildo: Wrong, sorry. I think it's release-20.03
<nh2>
romildo: release-20.03. nixos-20.03 is what gets auto-advanced by CI when release-20.03 builds successfully
<ottidmes>
TheSirC[m]: no, just the error you got, cause it fails the moment it encounters it again, and its not a recursive module imports either
<TheSirC[m]>
<ottidmes "TheSirC: lol, I see the message "> ottidmes: So I am interested in you view because oddly, the numbe of error message correspond to the number of files directly in the folder 🤔
c0c0 has joined #nixos
<TheSirC[m]>
ottidmes: I will fix the `default.nix` to try to avoid loading itself
kenran has joined #nixos
<ottidmes>
TheSirC[m]: That seems to be the problem indeed, as mentioned by tilpner, if that does not work for you, I would definitely try and debug it by first creating a list of files (not imported yet) and see if you agree with the files being generated
<TheSirC[m]>
Would there be any reason why my argument `(map ..)` would not work in the repl ?
<ottidmes>
TheSirC[m]: could be you are using some lib/builtins that are not available, try this: nix repl '<nixpkgs/lib>'
<bqv[m]>
ottidmes: oh, christ i get you now
<bqv[m]>
i wonder if lib.cleanSource would fix that
<bqv[m]>
ottidmes: hang on, no
<TheSirC[m]>
ottidmes: Could you see my code snippet in my first message ? I didn't realize you guys were on Freenode
<bqv[m]>
even that is demonstrably not true
<bqv[m]>
like i said, nix copies the flake's git'structure first
<bqv[m]>
so e.g. there are files in my directory here that aren't copied into the store
<{^_^}>
TheSirC[m]: The IRC<->Matrix bridge sends multi-line messages as links: `* Someone sent a long message: < https://matrix.org/_matrix/media/v1/download/matrix.org/... >`. This means that highlights don't work, people have to open your link, and it gets less visibility than multiple single-line messages
<ottidmes>
TheSirC[m]: not sure how it works, rather an IRC beginner myself, but I do see matrix pastes being reported
<ottidmes>
bqv[m]: what are you talking about flakes for, ${../.} is flake unrelated, right? or is it made aware of what files flakes have access to and only copies over those?
<TheSirC[m]>
ottidmes: IRC n00b too but I am on the Matrix side, you can find the snippet in the link above if you could not see it the first time
<bqv[m]>
ottidmes: so, in the context of flake evaluation, the current working directory is NOT the directory you evaluate nix from
<bqv[m]>
it's a git skeleton copied into the store
<ottidmes>
TheSirC[m]: I could see it just fine, like tilpner mentioned, I saw that "someone sent a long message" message
<ratsclub>
is there a matrix channel for nixos?
<bqv[m]>
hence, it doesn't matter if you re-copy it into the store
<{^_^}>
[nix] @edolstra pushed to flakes « getFlake: In pure mode, check that the argument is an immutable flakeref »: https://git.io/JvFbe
<TheSirC[m]>
ottidmes: Oh so it renders it as a "spoiler" of some sort ?
<ottidmes>
bqv[m]: Ah, like so! Well, at least it is good to be sure about it :) it would be rather bad if you go through all the trouble to properly setup secret files only to find them world readable due to a nicety
<bqv[m]>
lol
<TheSirC[m]>
ratsclub: Yes ! I think I am in it and you're too but through a bridge
<tilpner>
There is another unbridged channel
<tilpner>
#nix:matrix.org
<romildo>
Trying to checkout release-20.03 gives the error: error: pathspec 'release-20.03' did not match any file(s) known to git. What else should I do checkout this branch in order to cherry pick a commit from master?
mawis has quit [Ping timeout: 256 seconds]
<bqv[m]>
tilpner: it's absolutely dead in comparison, though
<bqv[m]>
because all nix nerds seem to prefer irc
<tilpner>
romildo: git fetch origin release-20.03
<tilpner>
(Or just blanket git fetch)
vykook has quit [Ping timeout: 265 seconds]
<mokasin_>
tilpner, using unstable on one package is like a virus spreading. In the end, using unstable altogether sounds more reasonable? At least, it was tested together to some extent.
<tilpner>
mokasin_: I don't agree with that first sentence
<romildo>
tilpner, that gives: fatal: couldn't find remote ref release-20.03
<tilpner>
romildo: Oh, okay. Maybe just git fetch
<ottidmes>
bqv[m]: what do I gain by going to the matrix? I heard people complain about it being slower (but maybe that is due to the forwarding here)
<tilpner>
ottidmes: There's little point in using Matrix as an IRC client right now, if you already have a bouncer set up via other means
<{^_^}>
[nixpkgs] @gnxlxnxx opened pull request #84178 → swaylock-effects: init at v1.6-0 → https://git.io/JvFbt
<mokasin_>
tilpner, well, if upgrading matrix-synapse requires an updated postgresql, which in turn breaks another package... but maybe this is not so relevant in praxis for a small installation, i dont know
<bqv[m]>
ottidmes: what tilpner said, but you can use both
<tilpner>
mokasin_: That's modules, and in that context I agree. You mentioned packages though, and in that context I disagree
<ottidmes>
tilpner: its on my TODO somewhere :P I just try to keep my weechat open as long as possible :P
<romildo>
tilpner, that does not work either: fatal: 'release-20.03' does not appear to be a git repository
<bqv[m]>
and matrix as a protocol is modern and nice, and not just in the newfangled "slack has emoji!!" sense
<mokasin_>
ottidmes, matrix is actually quite fast if not used on matrix.org, which is just overloaded. But it is indeed a WIP
<tilpner>
romildo: What was the full command you ran?
<romildo>
tilpner, git fetch release-20.03
<bqv[m]>
ottidmes: fyi i use weechat-matrix
<bqv[m]>
i'm actually waiting for nixos-unstable-small to advance past my weechat-matrix update rn
<tilpner>
romildo: Just git fetch
<bqv[m]>
it's not advanced for 23 hours...
<mokasin_>
tilpner, gotcha. I'm still need to get used to nix terminology and mix it often up.
<simpson>
bqv[m]: Just build on your own, if you're impatient. There's no need to wait for the official channel, if you know that your desired packages will build.
vykook has joined #nixos
hmpffff has quit [Remote host closed the connection]
noudle has joined #nixos
hmpffff has joined #nixos
<bqv[m]>
simpson: i can, but perks of a declarative config, i'd quite like to not have to temporarily hack my configuration just to use it
<bqv[m]>
it's not quite as simple as "just build it locally"
<romildo>
tilpner, still does not work.
<bqv[m]>
i have to change my config, build that, run that, and then rebase it every time i make any other change so i can easily revert it later
<TheSirC[m]>
<ottidmes "qy: what do I gain by going to t"> ottidmes: Not slower in my experience I think that you win by having a single point that can talk to multiple protocol
<TheSirC[m]>
At least that what hooked me 🤷♂️
<bqv[m]>
TheSirC: the irc bridge is slow, in my experience, but that's just because freenode is ridiculous. and either way you can still use matrix AND irc
<romildo>
tilpner, git branch -a | grep release-20.03 does not list anything.
ratsclub has quit [Remote host closed the connection]
<TheSirC[m]>
<bqv[m] "TheSirC: the irc bridge is slow,"> I must admit that I never did experiments to check
<bqv[m]>
to be honest, i might actually just cut my losses and point my flake at nixpkgs/master
<mokasin_>
Is nixos.org/channels down? I get a 404
<bqv[m]>
no more waiting, ever
<tilpner>
romildo: What does git remote get-url origin say?
<{^_^}>
[nixpkgs] @kolloch opened pull request #84179 → buildRustCrate: Allow version specific dependency renaming + parameter doc → https://git.io/JvFbz
<tilpner>
romildo: You may have to specify the proper remote, git fetch origin or your equivalent
<bqv[m]>
TheSirC: it will become patently obvious if you ever try and have a rapid conversation, or even just try and use an irc bot from here
<tilpner>
romildo: That's the wrong one then. How did you call nixos/nixpkgs?
<mokasin_>
never mind, typo
<simpson>
bqv[m]: Lucky 10000: Something like $(NIX_PATH=nixpkgs=... nixos-rebuild) will work to temporarily use a local checkout. You can use this to cheaply switch between configured channels and a local nixpkgs checkout.
<romildo>
tilpner, I did no understand your question 'How did you call nixos/nixpkgs?'
<simpson>
bqv[m]: To the extent that it won't, that is WAI, I gather. But sure, overlays and flakes complicate the process somewhat. I'm just saying that one does not need to be beholden to the channels.
<tilpner>
romildo: You must have added upstream nixpkgs (github.com/nixos/nixpkgs) as a remote at some point
<tilpner>
That should also explain why all the previous commands didn't work, your origin was your own fork
<bqv[m]>
simpson: as far as the flake mechanism allows, i actually *do* have to be beholden to channels unless i want to modify my config to not use them, i.e. switch to nixpkgs/master
<bqv[m]>
but to be honest i think i will just do that
<TheSirC[m]>
OK, let's say I have a list with the path of nix files `theList = [ /A/a.nix /B/b.nix ./c.nix ]` why wouldn't `(map (import) theList)` work ? for reference it sends back `cannot coerce a list to a string`
<TheSirC[m]>
It is humiliating but at this point it defies my understanding
<TheSirC[m]>
So let's just improve my understanding
<tilpner>
That should work
boegel has quit [Ping timeout: 240 seconds]
<tilpner>
Are you just executing that, or is there more code you're not showing?
melling has joined #nixos
<tilpner>
FYI All your parentheses are unnecessary
<TheSirC[m]>
<tilpner "FYI All your parentheses are unn"> Thatnks I did not know that
<TheSirC[m]>
<tilpner "Are you just executing that, or "> I am executing that in the repl
<TheSirC[m]>
I can snippet the whole code
<tilpner>
TheSirC[m]: I would prefer if you created a paste of your repl history, so I can see what you tried and the errors
<ottidmes>
TheSirC[m]: mapping import over a list of files should work just fine, it might be due to the file contents though
<TheSirC[m]>
tilpner: Your wishes are my commands
<tilpner>
TheSirC[m]: Use map builtins.pathExists theList to check if all paths are valid and exist
<romildo>
tilpner, thanks
<TheSirC[m]>
<tilpner "TheSirC: Use map builtins.pathEx"> All true
<ottidmes>
bqv[m]: BTW what I meant with pkgs created before config in your flake, is that what if you have say nixpkgs.config.oraclejdk.accept_license = true; be dependent on NixOS config, then it is not known beforehand what is nixpkgs.config, but I guess you would just have to create pkgs twice, one specific for the config, and one specific for the package outputs
kleisli has quit [Ping timeout: 256 seconds]
<TheSirC[m]>
tilpner: They all import well and good with the repl
linarcx has quit [Read error: Connection reset by peer]
<tilpner>
TheSirC[m]: Right now, you're creating a list that contains the map function
linarcx has joined #nixos
<bqv[m]>
ottidmes: ah, yes, i'm not sure on that, that's a situation i haven't considered before. i haven't currently got any packages complex enough to need to worry about that
<TheSirC[m]>
<tilpner "TheSirC: Right now, you're creat"> Isn't taht what `nixpkgs.overlays` expects ?
<tilpner>
TheSirC[m]: When I said your parentheses are unnecessary, that didn't apply to putting the result in a list. Then you need some of them
<simpson>
Yes. Parens don't matter for function application, but they *do* matter for list elements.
<tilpner>
TheSirC[m]: Then it would be: fnA = [ (map import (validFiles ./.)) ];
<TheSirC[m]>
I do not know why but to me `map` consumes the list when it actullay just apply the function to the lists elements
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Unode>
Hi all, does anyone have an example of how to add custom lines to the system's xorg.conf ?
<simpson>
TheSirC[m]: No worries. Everybody is a neophyte at some point; everybody has to learn and to ask questions. Also, that list/paren precedence is *very* quirky, and it trips up even experienced folks.
<simpson>
Unode: services.xserver.extraConfig I think.
<TheSirC[m]>
simpson: You're right but not everyone takes its time to help neophytes
<bqv[m]>
the precedence rules are inconvenient as it stands, coming from haskell
<Unode>
simpson: thanks
<TheSirC[m]>
So I prefer to thank the ones who do
pbogdan has quit [Remote host closed the connection]
zupo has joined #nixos
lsix1 has quit [Ping timeout: 272 seconds]
<ottidmes>
bqv[m]: question about your private files, if paths are relative to the copied repo to /nix/store, won't my current way of directly using say fileContents ./some.pass no longer work? How do you deal with that
zupo has quit [Client Quit]
<bqv[m]>
ottidmes: see for reference ::/users/root.nix which imports from my secrets directory
<bqv[m]>
you can import a file, but it has to be tracked by git, and if it is, you can just navigate relative to the current nix file
fling has quit [Ping timeout: 264 seconds]
<bqv[m]>
unless i'm misunderstanding how fileContents work, is it just the same as builtins.readFile?
<{^_^}>
[nixpkgs] @cpcloud opened pull request #84185 → scmpuff: init at 0.3.0 → https://git.io/JvFxa
<bqv[m]>
ottidmes: ah, then you're fine. it counts as a store path in scope if it's tracked by git, so as long as you use relative paths it will work as before
<bqv[m]>
if you want to avoid it being in git unencrypted then nrxdp's method of scoping it all to a secrets directory and using git-crypt is probably easiest
bbarker has joined #nixos
linarcx has quit [Quit: WeeChat 2.8]
bebarker has quit [Ping timeout: 264 seconds]
NieDzejkob has joined #nixos
<NieDzejkob>
Huh, I just noticed that the homepage URL for patchelf (https://nixos.org/patchelf.html) 404's. Where do I report this?
<ottidmes>
bqv[m]: but of course its not tracked by git, why would I put pass files in my git repo, so my current approach will break, guess I will have to look at workarounds (I am looking into flakes more than actually planning to convert already, but at least now I already know and documented some things to look out for)
<bqv[m]>
if i had realised flakes put the source configuration in the store, i might have developed a better solution (assuming one exists) before switching fully
<bqv[m]>
but it's too late now
<exarkun>
is there a nice pattern for constructing a hash key for CI for the dependencies of a derivation? I want a key that changes when any of the buildInput-type inputs to a derivation change but not when, eg, the src of the derivation changes
<exarkun>
all I can think of is writing a function to walk all the build input attributes and extract their output path and then hash all of that
fendor has quit [Read error: Connection reset by peer]
ambroisie has joined #nixos
domogled has joined #nixos
<{^_^}>
[nixpkgs] @edolstra pushed to master « Don't pin 'nixpkgs' in the system registry by default »: https://git.io/JvFhJ
<{^_^}>
[nixpkgs] @edolstra pushed to release-20.03 « Don't pin 'nixpkgs' in the system registry by default »: https://git.io/JvFhU
linarcx has quit [Read error: Connection reset by peer]
linarcx has joined #nixos
plp_ has quit [Quit: No Ping reply in 180 seconds.]
<exarkun>
That will miss out on some dependencies though, those just referenced in build phases and not included in the explicit input lists
plp_ has joined #nixos
<exarkun>
How do I get those?
<ottidmes>
exarkun: you probably want to use nix-store to gather the info
captjakk has joined #nixos
<gchristensen>
NieDzejkob: can you open an issue?
<exarkun>
is there a nix-store command that will work on the expression and not the store path? I have to compute the cache key before the build for it to be useful.
<simpson>
If only the intensional store RFC were ready.
<bqv[m]>
> stdenv.lib.licenses.gpl3Plus
<{^_^}>
{ fullName = "GNU General Public License v3.0 or later"; shortName = "gpl3Plus"; spdxId = "GPL-3.0-or-later"; url = <CODE>; }
mokasin_ has quit [Quit: Leaving]
ZaraChimera has quit [Quit: Going offline, see ya! (www.adiirc.com)]
<bqv[m]>
cole-h: ty for review, do you think i should use fetchgit over fetchurl? i don't see the benefit of using fetchgit, but happy to switch
<bqv[m]>
no other instances of notabug in nixpkgs apparently, on a quick ripgrep
captjakk has quit [Ping timeout: 260 seconds]
xkapastel has quit [Quit: Connection closed for inactivity]
linarcx has quit [Quit: WeeChat 2.8]
<kenran>
How can I find out which nix package contains the command line tool `which`? For some reason it's not in my pure nix-shell and I want it back!
<peelz>
bqv[m]: for automatic re-indexing or what?
<peelz>
I just have it added to my packages
<bqv[m]>
peelz: yeah
<typetetris>
Can someone give me a few pointers how to package a qt app? I tried following the manual, but had no success. The package in question x2godesktopsharing doesn't have a makefile and one needs to generate one with `qmake` but the result contains paths to `lrelease` which are simply invalid on my machine. (They point to bin in qmake store path, but lrelease resides in qttools store path).
<bqv[m]>
similar to services.locat
<peelz>
bqv[m]: hmm, let me know if you find/make one
<mokasin>
I've upgrade to unstable from 19.09 and I am running into permission issues with ACME. I get errors like [acme-start] Error while loading the certificate for domain example.com
<mokasin>
open certificates/example.com.crt: permission
<mokasin>
Interestingly /var/lib/acme/.lego and subfolders have permission for specific users
<mokasin>
So basically two users, user1 and user2 use acme, and .lego belongs to user1 and is quite restrictive. It looks like a bug or an upgrade issue from simp-le
<ottidmes>
anyone experience with docker on NixOS, any idea why my docker service starts fine, but when its stopped, its stuck, I have to kill it or in the case of poweroff, have it be terminated after a timeout (the reason I mostly want it fixed, dont want to wait 2 minutes for a reboot)
<ottidmes>
when I run the docker command to stop it myself, it works fine
gxt has quit [Ping timeout: 240 seconds]
<bqv[m]>
isn't there a docker2nix thing somewhere
<ottidmes>
there is an option to generate services, which is what I am using
<bqv[m]>
oh right
<bqv[m]>
maintained by LnL, actually
<bqv[m]>
interesting
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
ambroisie has quit [Ping timeout: 264 seconds]
bvdw has quit [Read error: Connection reset by peer]
bvdw has joined #nixos
mallox has quit [Quit: WeeChat 2.7.1]
mallox has joined #nixos
Ariakenom has quit [Read error: Connection reset by peer]
<ottidmes>
bqv[m]: not sure about fetchgit, I believe it has the same problems as builtins.fetchGit in that it includes the history, which does not have to be much of a problem, but 100M vs 1.8G it is for nixpkgs, so I prefer fetchurl/fetchFromGitHub for most things
plutes has quit [Read error: Connection reset by peer]
cybrian has quit [Ping timeout: 240 seconds]
plutes has joined #nixos
linuus has quit [Ping timeout: 256 seconds]
wrunt has quit [Ping timeout: 256 seconds]
wrunt has joined #nixos
silver_hook has quit [Ping timeout: 256 seconds]
linuus has joined #nixos
emilsp has joined #nixos
silver_hook has joined #nixos
jakobrs has joined #nixos
<jonge[m]>
hey there. i would like to create a nixos module that, if my option `foo.enable = true` is set, overrides what `system.build.installBootloader` does (because i need to use a patched install-grub.pl file). so i am not overriding a package but a module attribute. is that possible somehow?
erasmas has joined #nixos
<{^_^}>
[nixpkgs] @lheckemann pushed to structured-attrs « stdenv/linux: fix iconv support in libunistring »: https://git.io/JvbUa
justanotheruser has joined #nixos
<jakobrs>
jonge[m]: Have you tried just setting it via `config` in your module?
<mlatus[m]>
jonge: should be something like `system.build = lib.mkAfter { installBootLoader = <value>; }
gulplante has joined #nixos
<jonge[m]>
mlatus: awesome, i will try that in a few minutes. one question: why not `system.build.installBootLoader = lib. mkAfter ...;` ?
Ariakenom0 has quit [Quit: WeeChat 2.7]
pluplog has joined #nixos
domogled has quit [Ping timeout: 256 seconds]
<mlatus[m]>
jonge: because of the typedef of the `system.build` is `types.attrs` rather than some submodule, so the merging is a little odd (I am no nixos module system expert, this is what I remember from my personal experiment, may be not correct, you could use nixos-option to test if your config is applied)
<jonge[m]>
mlatus: this works. thank you for your help!
<mlatus[m]>
jonge: cheers!
<Freneticks>
Is there a way to see update pending (like apt --upgradable in debian), and check package that have security issue like audit cve ?
<tilpner>
Freneticks: See vulnix
Maxdamantus has quit [Ping timeout: 240 seconds]
rogue_koder has quit [Read error: No route to host]
Darkmatter66_ has quit [Ping timeout: 256 seconds]
rogue_koder_ has joined #nixos
Maxdamantus has joined #nixos
CMCDragonkai1 has quit [Ping timeout: 256 seconds]
<Freneticks>
tilpner: it doesn't seem to have a : Show only packages that have already been fixed
<Freneticks>
i'm not a nix maintainer
Darkmatter66 has joined #nixos
<aanderse>
so i know nothing about mac. i have been given access to a mac and tried to install nix as described by manual
<aanderse>
turns out / is mounted read only
<aanderse>
anyone know anything about mac and can help me out?
<mlatus[m]>
jonge: can confirm, value of `types.attrs` are just merged using //, if you are still interested
<tilpner>
Freneticks: Ahh, you want notifications about vulnerabilites you could fix by updating. I don't know a tool for that
<Freneticks>
yep
<tilpner>
#nixos-security might know one (or it might ignore you for a day)
linarcx has joined #nixos
<jonge[m]>
mlatus: thank you!
<Freneticks>
this kind of tool is kind of like the base of a os
<Freneticks>
Also is there a way to know pending package before nixos switch ?
<tilpner>
You can see what would be installed
<Freneticks>
how ?
<tilpner>
But what would be upgraded? That's not well defined
ixxie has joined #nixos
<tilpner>
Let's say you have three versions of A installed: 1, 2, and 3. And now you update to versions 2, 4, and downgrade the other to 2. How do you display that?
<tilpner>
Because packages can exist in multiple instantiations (and with independent closures), it's not as simple as "check what version A had before, compare to what it has now"
rogue_koder_ has quit [Ping timeout: 258 seconds]
rogue_koder_ has joined #nixos
<Freneticks>
if i do nix-channel --update and then nixos switch (It install new things but it's very very clear what
alina has quit [Quit: leaving]
glitters` has joined #nixos
<tilpner>
Freneticks: No, it doesn't say what changes. It tells you what it needs to fetch/build/realise to change, but not what packages will be replaced with which versions
<Freneticks>
this is crazy
mallox has quit [Quit: WeeChat 2.7.1]
<tilpner>
Freneticks: E.g. if you rollback, and then update again, it doesn't need to do anything, because that system is already built. So it doesn't print anything
markus1189 has quit [Ping timeout: 256 seconds]
aveltras has joined #nixos
glittershark has quit [Ping timeout: 256 seconds]
magnetophon has quit [Read error: Connection reset by peer]
<tilpner>
Freneticks: If you want, you can compare what's in the build-time closures of each system. You can then look up CVEs for each version, and compare those
<Freneticks>
yeah no i'm just playing maybe nixos is not for me
<Freneticks>
I want total control over package
<tilpner>
You have that
<Freneticks>
I mean in a security way
<tilpner>
You can easily script that, and perhaps it's already been done
<tilpner>
(Easily only when you have an API to map (package name, version) -> list of CVEs)
<Freneticks>
I declare my state in *.nix and then nixos switch, but then if i update the channel, new version of package are definied so i do nixos switch and many package is replaced but I can't know what and why, how do you call that control ?
<{^_^}>
[nixpkgs] @primeos pushed 5 commits to release-19.09: https://git.io/Jvbkq
ekleog_ has quit [Quit: WeeChat 2.7.1]
ekleog has joined #nixos
rogue_koder_ has quit [Ping timeout: 256 seconds]
<tilpner>
Freneticks: I can edit, pin, and rollback those channels however I want. I control when things update, and can filter unfree packages by license, and insecure packages by name
<ekleog>
ma27[m]: By chance, are you using a self-hosted nextcloud nixos module? (trying the maintainers one by one, but you're the last one ^^') When I'm trying to use the nextcloud-client nixos package or the Talk android app with my instance, I get an endlessly-looping “please wait” on the “connect” button -- do you also see that, or is it just me?
<tilpner>
Freneticks: What you mean by lack of control is probably more a lack of insight, and then I might agree
knupfer has quit [Ping timeout: 240 seconds]
<Freneticks>
probably that's why I ask
bennofs is now known as ALLES_bennofs
<ma27[m]>
ekleog: yes I do
<ma27[m]>
and no, nextcloud-client works for me
<ekleog>
uhhh :( did you need to configure something special for it, aside from what's in the docs?
<ma27[m]>
do you have $BROWSER set?
lovesegfault has joined #nixos
<Freneticks>
tilpner: when i say update channel and I don't mean major upgrade like 18 > 19 but only refresh the channel
<ma27[m]>
well it opens the login dialog in a browser
<ekleog>
I don't, but it does open a firefox page, which also goes to an endlessly-looping “Grant Access” after I click it
rogue_koder_ has joined #nixos
jakobrs has quit [Ping timeout: 240 seconds]
chloekek has quit [Ping timeout: 256 seconds]
<ma27[m]>
ohh I see....
<ma27[m]>
but no, never experienced this
<tilpner>
Freneticks: Yes, I assumed so
<ma27[m]>
I'm currently on 2.6.4
<ekleog>
oooh just saw Content Security Policy: The page’s settings blocked the loading of a resource at http://cloud.gaspard.io/login/v2/grant (“form-action”).
<ekleog>
(I'm also on 2.6.4git for nextcloud-client; the version on 20.03 basically)
<tilpner>
Freneticks: I realise it must seem strange to someone familiar with e.g. apt, but I've never cared for the full list of version changes
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<ekleog>
oooh found it, there's no `s` to the `http`
<tilpner>
Freneticks: I often check what version a package has on different versions of nixpkgs, but the full list of every change?
knupfer has joined #nixos
<ekleog>
now, why is it not having an `s`…
<tilpner>
Freneticks: That's nothing I ever thought to query. Perhaps there is an easy way that I just never noticed
<yorick>
timokau[m]: so I have tensorflow without mkldnn, why is it spending 70% of its time in mkldnn::impl::cpu::(anonymous namespace)::kernel_mxn<float, false, false>
<Freneticks>
tilpner: okay will search, I just want to know what my system do.
<Freneticks>
If he upgrade a package, he need to print me the version, and maybe the cve fixed
<ekleog>
ooh, looks like services.nextcloud.config.overwriteProtocol is probably it
<ekleog>
weird that it didn't do it before
<tilpner>
Freneticks: Well, that's the conceptual difference from apt to nix. It doesn't upgrade a package in place, it leaves the old package just as it was.
<ekleog>
but then, it was already there on 19.09, so it's probably in the release notes of 19.03 or 19.09 that it should have been, so too late to fix I guess
<tilpner>
Freneticks: And instead it builds your system entirely fresh (reusing what was already there) every time, it doesn't ever look at the old package
rogue_koder_ has quit [Ping timeout: 256 seconds]
rogue_koder has joined #nixos
<Freneticks>
tilpner: yeah i know that
Blessjah has quit [Remote host closed the connection]
<ornxka>
i changed libselinux to unstable and now its rebuilding absolutely freaking everything
mexisme has quit [Ping timeout: 256 seconds]
<ornxka>
do a lot of packages use that or something?
<melling>
cap_sensitive: it passes through your reference from the outer scope, i think it is to allow you to specify a specific nixpkgs for nur to use
<cap_sensitive>
melling: Ha! I see! So that inherit pkgs is passed to the nur! Thanks!
<melling>
cap_sensitive: yes, it's like shorthand for doing { pkgs = pkgs }. you might want to pass through a different nixpkgs from the default, for instance with your own overrides etc
<jco>
ornxka: thx for the advice, gonna rollback :)
jco has quit [Quit: WeeChat 2.7]
drakonis has joined #nixos
emacsomancer has quit [Read error: Connection reset by peer]
emacsomancer has joined #nixos
markus1189 has quit [Ping timeout: 250 seconds]
markus1189 has joined #nixos
<bsima>
when I do `builtins.match ".*(a.).*" "abacadae"` I only get back `["ae"]` but I want to return all of the matches, does anyone know how I can do this?
melling has quit [Quit: WeeChat 2.7.1]
shibboleth has quit [Quit: shibboleth]
<bsima>
hm, i think i'll just use lib.strings to split or duplicate the string i'm working on. not ideal for performance but i can't find a way to do multiple matches with nix regex
<cole-h>
If you want performance, you're using the wrong language :P
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<bqv[m]>
Apologies jgarte and ottomides for not noticing pms
cybrian has joined #nixos
janneke has joined #nixos
mallox has joined #nixos
jbrock has quit [Quit: jbrock]
jbrock has joined #nixos
jbrock has quit [Client Quit]
jbrock has joined #nixos
zupo has joined #nixos
<gchristensen>
hey wait libvirtd is really nice
<gchristensen>
using nixops with the libvirtd backend for the first time feels like being transported to the future
cybrian has quit [Ping timeout: 265 seconds]
<mallox>
hey, does anyone know how to roll out my dotfiles via git/stow in my configuration.nix? (I know about home-manager, but I want mutable dotfiles)
<jakobrs>
Wait you can use nixops with libvirtd? I'd always assumed that it only supported virtualbox for "virtual" deployments
jbox has joined #nixos
ixxie has quit [Ping timeout: 264 seconds]
<gchristensen>
okay the magic is gone, "OSError: [Errno 28] No space left on device" and I don't know why lol
<cole-h>
Quickest about-face in the west
ixxie has joined #nixos
<scoates>
ok, I think my problem is related to systemd; I have files for `wlp3s0` but not `ens9`. Is there a way to generate these post-install?
<{^_^}>
golang/go#37459 (by carlca, 5 weeks ago, closed): SIGILL: illegal instruction on any go tool under macOS
<flokli>
scoates: you might want to try with networkd, which does all that properly also when devices appear later
<flokli>
__monty__: ah, sweet - another issue
pluplog has joined #nixos
<flokli>
well, if you're fancy, you can just checkout nixpkgs to a commit after that merge, and point -I nixpkgs=/path/to/checkout to switch to it
<flokli>
if you want to migitate this for now
<scoates>
uh-oh indeed
<__monty__>
Would that involve lots of building though?
<scoates>
rebooting for shame (-;
<scoates>
remember that part about first-time nixos caller? (-;
<scoates>
(I didn't `nixos-rebuild`… derp) I have a different problem now, but it does look like it created those systemd units
<scoates>
)
<{^_^}>
[nixpkgs] @mmilata opened pull request #84202 → [20.03] nixos/sympa: fix outgoing emails, update package version → https://git.io/JvbqN
lsix1 has joined #nixos
jb55 has quit [Quit: jb55]
markus1189 has quit [Ping timeout: 260 seconds]
thetet has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<flokli>
__monty__: there's only one way to find out ;-)
o1lo01ol1o has joined #nixos
jb55 has joined #nixos
<monokrome>
hmmm... Is there not a way to create an offline mirror of nixos packages?
<monokrome>
Like, since they all use git and such
o1lo01ol1o has quit [Remote host closed the connection]
zupo has quit [Ping timeout: 260 seconds]
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/JvbmT
<bqv[m]>
Goddamn it. Literally MINUTES before I push to my pr changing fetchurl to fetchgit, jon has let us know that we should prefer fetchurl. I have the worst timing
<__monty__>
flokli: I'm not very brave and I have the patience of a saint ; )
<flokli>
__monty__: I don't know how much you rebuild, as I don't know how many packages you pull in ;-)
<__monty__>
monokrome: What do you mean? If you clone the nixpkgs repo you can build everything.
o1lo01ol1o has joined #nixos
jb55 has quit [Client Quit]
<monokrome>
__monty__: like, every package?
<tilpner>
Not really, nixpkgs alone isn't that useful
<tilpner>
You need access to all the sources, or a cache that had
<__monty__>
monokrome: Copying all the build products from the cache wouldn't be hard per se either.
hammernix has quit [Remote host closed the connection]
<__monty__>
tilpner: Ah, true.
<monokrome>
but is there a way to build a completely offline mirror of nixpkgs? or no?
<monokrome>
like if I want to point the nixos installer at a local mirror
<tilpner>
monokrome: Depends. What do you want to with it?
<bqv[m]>
monokrome: how much space do you have :p
<tilpner>
You can easily copy the cache for a single revision
<monokrome>
40TB
<__monty__>
You could clone the repo and evaluate the src attribute for everything I guess?
<monokrome>
hmm
zupo has joined #nixos
<simpson>
monokrome: To anticipate several possible use-cases: How do you want to receive security updates?
<monokrome>
I was considering that but hoping there was a better way
<tilpner>
There is, but answer the question
<scoates>
now I'm seeing `failed to mount /rpool`… pretty sure I'm doing something *different* dumb.
<monokrome>
simpson: by updating my offline mirror...
<scoates>
gchristensen: ^ network problem seems to be solved at least.
<gchristensen>
nice
thibm has quit [Quit: WeeChat 2.6]
<gchristensen>
scoates: let's see your config :)
<exarkun>
I have a `nix-build` command that fails because "The pkg-conf packages does not contain the package: libchallenge_bypass_ristretto (pkg-conf dependency).". How can I get at pkg-conf to poke around and see what's going on? The build command is this, fwiw:
<simpson>
monokrome: Okay. Due to how Nix works, are you okay with rebuilding your entire offline cache every time there's a security update? I'm not sure how disconnected "offline" is; presumably it's still hooked up to some sort of LAN.
<monokrome>
wouldn't it only rebuild things that changed?
<monokrome>
when I `nixos-rebuild`, my whole system doesn't rebuil
<tilpner>
Yes, but with nixpkgs a lot of things can change for little reason
<monokrome>
yeah that's fine
<monokrome>
expected, even
<lordcirth>
monokrome, yes, but "changed" includes "anything in it's closure changed in the slightest"
<tilpner>
Update a core package? All dependent packages have changed
<simpson>
And you currently benefit from the *online* caches, which are pretty sizeable and have a really good hit rate.
<monokrome>
yes I know
<monokrome>
we can assume there's a security issue in musl
<scoates>
heh hard to share without networking (-; need a min
<gchristensen>
I don't mean to be dismissive, but identify the store paths you want, and mirror them. without knowing more about your use case and what exactly you want to mirror, I don't have more info. for example, do you want to mirror every prebuilt thing for a given evaluation?
<simpson>
The main surprise is that you're asking for *everything*, including stuff that is in an expensive long tail. So, like, who's gonna build that, are you really gonna hold all of it, etc. That's all. Not trying to pry *too* much at what is almost certainly a plan for a walled corporate LAN of some sort.
<timokau[m]>
yorick: I don't know. I should really remove myself from the tensorflow maintainers, I'm not really maintaining it :/
<cransom>
'slapping squid' isn't really a thing anymore with https servers. you have to hijack connections now.
<scoates>
it specifically says to not edit. can I edit? (-:
<gchristensen>
yeah :)
* scoates
living on the edge
<gchristensen>
it is just that if you run nixos-generate-config again, it will fearlessly replace that file
jb55 has joined #nixos
<tilpner>
cransom: You don't *have* to use https for this
<lordcirth>
cransom, well, I think you can set your mirror on the client to be your squid server, and have squid rewrite the upstream URL back?
<lordcirth>
Haven't done it that way myself, though
<snicket_>
is it normal that when booting raspberry pi 3 from 'nixos-sd-image-19.09.2358.0ee9ceff38f-aarch64-linux.img' there is no default /etc/nixos/configuration.nix file anymore?
<jared-w>
I'm conflicted. I want linux 5.6.2 so that iwlwifi will finally work with 5ghz wifi again. We upgraded to gigabit at my house and I'm stuck on a slow 2.4ghz connection :(
<jared-w>
On the other hand, I should responsibly wait until the nvidia_x11 PR gets merged...
<jared-w>
RIP, forgot to turn f@h on overnight. Noooo
koleesch__ has quit [Ping timeout: 264 seconds]
<ornxka>
i just have a big git repo with master/release in it and make changes/merge PRs in it as necessary
<ornxka>
waiting for things takes too long
thetet has quit [Quit: Leaving.]
thetet has joined #nixos
<jared-w>
It's more a self control thing so I don't nerdsnipe myself for 6 hours by accident instead of working
<atemu12[m]>
<jared-w "I'm conflicted. I want linux 5.6"> Couldn't you just use 5.4?
thetet has quit [Client Quit]
<jared-w>
atemu12[m]: my wifi card is too new. Support needs 5.5
<ornxka>
<jared-w> It's more a self control thing so I don't nerdsnipe myself for 6 hours by accident instead of working
<jared-w>
Miraculously, 2.4 wifi works pretty well; more or less flawlessly. And most 5ghz wifi networks seem to work great too. It's just 5ghz above a certain speed that makes the card fall over and die
<ornxka>
ah....
<ornxka>
you are smarter than me
<jared-w>
ornxka: only because I've learned through years of stupidity ;)
<{^_^}>
[nixpkgs] @peti pushed 552 commits to haskell-updates: https://git.io/JvbYm
o1lo01ol1o has joined #nixos
markus1189 has quit [Ping timeout: 256 seconds]
<ornxka>
i am literally doing nixos stuff when i should be working at this very moment
<ornxka>
i have much to learn i guess
domogled has quit [Remote host closed the connection]
<atemu12[m]>
jared-w: So what about linuxPackages_5_5 then?
domogled has joined #nixos
<jared-w>
heh, same. I tend to tinker around until I get my system into a stable state and then I freeze it and force myself to not do anything to it for like a month or two. Then I tinker around and lose a day or two of max productivity and repeat the cycle. Seems to work pretty well
fuzzy-id has joined #nixos
<ornxka>
that sounds like me
<jared-w>
atemu12[m]: I'm on 5.4 right now, my bad; my wifi card is the AX200 wifi 6 card from intel. I could probably do 5.5 but it's more broken for wifi than 5.4 is; 5.4 won't crash loop the wifi card fast enough to freeze my computer if I connect to my 5ghz wifi network. 5.5 will
<ornxka>
i try not to update when i can avoid it but "chrome security update" sounds pretty important
<jared-w>
ornxka: I just use a firefox overlay that's non-deterministic, so that marches forward regardless of nixpkgs. Pretty nice
<fuzzy-id>
i use "fetchFromGitHub" to hook a package of mine into the system. as sha256 attribute i use the one given by `nix-prefetch-url https://github.com/OWNER/REPO/archive/REV.tar.gz'. but that results in the error "hash mismatch in fixed-output derivation …".
<fuzzy-id>
so, what's the correct url to pass to nix-prefetch-url?
<tilpner>
fuzzy-id: Try with --unpack
<fuzzy-id>
yeah, that's it. thanks!
<armin>
just recognized mandelbulber2 is not packaged for nix :(
<armin>
considering how much i LOVE fractals and that i have quite some complex setup with remote rendering and such, that's probable something i like to change.
<armin>
tilpner: i'm new to nixOS though - maintaining a couple of packages in various distributions, but fearing nixOS packages a bit still...
<armin>
tilpner: a co-worker (he's also hiding in the depth of this channel) got me into this rabbit-hole and now there doesn't seem to be a way back out
<armin>
damn you, eric.
<cole-h>
I would argue NixOS packaging is easier than any other distribution. Make it work (the hardest part), submit a PR upstream, wait for it to get merged, done.
<armin>
cole-h: come on, every packager of every distribution would say that about their packaging mechanism ;)
<lordcirth>
It's easy, unless the official build process is "run this easy custom script that downloads all the deps"
<lordcirth>
(looking at you, CKAN)
<armin>
:D
<tilpner>
armin: That line was meant to tell you it's already packaged
<cole-h>
Except for Debian, RedHat, etc, where, for official inclusion, you need to get a sponsor or something
<FRidh>
cole-h: packaging != maintaining
<armin>
tilpner: oh!
<armin>
tilpner: so why didn't i find it even with nix search -u?
<tilpner>
armin: Old cache, or old nixpkgs
Guest62117 is now known as atlas_
atlas_ has quit [Quit: WeeChat 2.8]
<tilpner>
Or wrong search term
<{^_^}>
[nixpkgs] @peti pushed 2 commits to haskell-updates: https://git.io/JvbYj
<tilpner>
Perhaps you included "2" in the search term
<armin>
tilpner: ok thanks a lot for your kind help. :)
<armin>
i tried both!
<tilpner>
It shows up in the search here
<tilpner>
Does nix search hello work for you?
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/JvbOv
<armin>
yes that works :D
<tilpner>
Odd
<armin>
it's very likely layer8 error, i'm new to all of this and still try to find my way around
<tilpner>
Ahh, old nixpkgs
<tilpner>
It seems to be more recent
<armin>
oh okay :D
<tilpner>
Not your fault then
<armin>
again, thx for telling me so!
<armin>
i'm super addicted to fractals btw, if you like some 4k wallpapers, maybe this is for you: fhttps://unix.porn/fractals/
<gchristensen>
FYI: looks like there is some phishing going for github users, saying some private file was modified withe curlx. don't click it :P
<morgrimm>
Does anyone have any really good examples of config repos with home-manager? I'm trying to wrap my head around how it needs to be configured (if it needs everything in nix configs, or if it can symlink existing dotfiles, or what)
<worldofpeace>
cole-h: so from looking at those logs, as a qa person, I'd verify "Version output displays `git` revision information."
<worldofpeace>
it seems also you made the commit, so I trust it much more 😃
<cole-h>
:P
<worldofpeace>
will merge when ci passes
<cole-h>
There's no easy way to get the actual commit info stuff to display without keeping .git
<worldofpeace>
you could keep it by fetching it different. but I don't really think people should do that
<cole-h>
Could go quick-and-dirty and just use the commit hash (after `leaveDotGit = true;`), but it would still differ from `git describe --tags`'s output.
<cole-h>
Right, that's why I didn't :P
<cole-h>
My local overlay displays the hash, but that's because I actually have the source on-disk (and `lib.commitIdFromGitRepo` it)
iceypoi has quit [Quit: WeeChat 2.3]
iceypoi has joined #nixos
mexisme has quit [Ping timeout: 260 seconds]
domogled has joined #nixos
<cole-h>
worldofpeace++ Thanks for keeping me honest -- been trying to get better about having commit messages that explain the change (even if it's just a link to what's changed)
<{^_^}>
worldofpeace's karma got increased to 87
<gchristensen>
<3 cole-h worldofpeace++
<{^_^}>
cole-h's karma got increased to 12, worldofpeace's karma got increased to 88
iceypoi has quit [Client Quit]
<gchristensen>
mog: are you planning on doing some 20.03 artwork? :)
<worldofpeace>
cole-h: as a merger, we can integrate so much faster if we have that information, and be more sure of any side-effect or if it's backportable.
<worldofpeace>
gchristensen: yep, asked for that in advance
<gchristensen>
isn't that a beaut'
<cole-h>
👍 That's why I want to get better at it -- make your (collective) jobs easier
<samueldr>
worldofpeace: confirmed, all 100px wide
<samueldr>
monokrome: let me add that this is not an exhaustive list, but hosts the community had success with :)
<ottidmes>
monokrome: Although I moved to some cheap black friday deal VPS in germany for mine, I have had a good experience at TransIP before, although you do have to bootstrap NixOS it worked perfectly
<samueldr>
so there may be other hosts where it's trivial
<samueldr>
monokrome: I personally use vultr to run {`-`} (the logging bot) on
<samueldr>
the nixos.wiki coincidentally also uses a vultr vps
<worldofpeace>
samueldr: thanks so much. it's hard to figure all these little pieces.
cosimone has joined #nixos
<cole-h>
worldofpeace++ With you being so exhaustive (meant in a good way), hopefully this will be clearer for whoever you pass the RM torch to :)
<{^_^}>
worldofpeace's karma got increased to 89
nobrak` has quit [Remote host closed the connection]
fuzzy-id has quit [Remote host closed the connection]
fuzzy-id has joined #nixos
nobrak has joined #nixos
<ottidmes>
I am now extra careful with any host allowing custom ISOs, my current one claimed to support it, but it turned out you could request an ISO to be added to the global list, and I do not want my custom build NixOS installerd to be added to the global list... so I still had to bootstrap
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/Jvbs6
<worldofpeace>
cole-h: I have lots of notes whenever I have to do something. So, if my brain permits me to write docbook, or I can find someone, it could be added to the manual.
<cole-h>
You could also wait for that one RFC to allow us to switch to something not-XML-based ;)
<adisbladis>
I've had great success using kexec on many hosting providers
cransom has quit [Quit: WeeChat 2.6]
fusion809 has quit [Remote host closed the connection]
pbogdan has quit [Remote host closed the connection]
fuzzy-id has quit [Ping timeout: 265 seconds]
cransom has joined #nixos
<ottidmes>
yep, kexec worked for me on all hosts I had, so I don't worry about running NixOS anymore on any VPS (the KVM ones at least)
<bqv[m]>
i think i used something i bastardised from nixos-infect
<bqv[m]>
to be honest, if you're sufficiently flexible with your cli skills, you can always boot into a ramfs system and install whatever the McHeck you want. doubly easy if it has a small disk image around that you can just dd
<bqv[m]>
cole-h: so, shall i revert it to fetchurl?
kqb has quit [Quit: Leaving]
knupfer1 is now known as knupfer
<{^_^}>
[nixpkgs] @rnhmjoj pushed 2 commits to release-20.03: https://git.io/JvbGc
<colemickens>
Hi. I have `nix`, same version on two machines. But it seems that `--experimental-features` is only recognized on one? Am I doing something obviously wrong?
<bqv[m]>
check your nix.conf
gxt has quit [Remote host closed the connection]
gxt has joined #nixos
lovesegfault has quit [Quit: WeeChat 2.8]
<colemickens>
I don't see anything in user or system nix.conf that looks at all related
<colemickens>
or any real differences at all
<ornxka>
how long does chromium typically take to compile
<bqv[m]>
traditionally, longer than you're willing to wait
<ornxka>
i have 8 cores/16 threads
hmpffff has quit [Read error: Connection reset by peer]
<ornxka>
i see
<gchristensen>
any where from 20 minutes to 48 hours, depending on cores, memory, cpu speed, and storage type
<ornxka>
i would like to see the machine that can compile chromium in 20 minutes
<bqv[m]>
me too
<colemickens>
ah, system-features = nixos-test, maybe thats it
<peelz>
Is there a difference between `{ a, ... }@args` and `args@{ a, ... }` ?
<gchristensen>
the first one is preferred and the second one makes me go "wat"
<peelz>
lol
<colemickens>
bqv: it turns out I was looking at my two machines that don't require it. on the machine that does, its running nix unstable.
<peelz>
gchristensen: is this syntax inspired by another language? not too familiar with functional programming
<ornxka>
the last compiler warning said it was on file eighteen thousand out of about thirty-seven thousand, and that was about 45 minutes ago, so hopefully it will finish within the hour
<gchristensen>
peelz: I dunno, actually!
<bqv[m]>
wait what, really?
<bqv[m]>
gchristensen: the second is the order in haskell
<bqv[m]>
why does it confuse you
<bqv[m]>
Cole Mickens: heh
<gchristensen>
bqv[m]: well, I don't know haskell
<bqv[m]>
there's your problem :p
<peelz>
personally I prefer `{}@args`
<gchristensen>
{}@args is the prevailing style in nixpkgs
<peelz>
ah gotcha
<bqv[m]>
as a haskeller, {}@args seems extremely inpleasant, but i guess i'll switch to using that then
<ornxka>
i think of it more like "the computer is how C works" rather than the other way around
<bqv[m]>
someone tell me the good news, do fetchurl and fetchTarball have the same hashes for the same url?
<ornxka>
in C there is undefined behavior etc but there is not really any such thing in real life, the holes gets defined by the computer itself
smatting has joined #nixos
<peelz>
bqv[m]: no, fetchTarball unpacks it first
<ornxka>
so in the end everything is defined, just not necessarily in an intentional (or known at all) way
<bqv[m]>
peelz: suffering
<peelz>
bqv[m]: you can get the new hash with `nix-prefetch-url --unpack`
<bqv[m]>
i'm sure i can, it's just a pain to when editing derivations
<peelz>
true
<cole-h>
Just `sha256 = stdenv.lib.fakeSha256;` and try to build :)
<bqv[m]>
stdenv.lib.fakeSha256? christ i've been typing random numbers of zeroes all this time
<cole-h>
I still do ;)
<gchristensen>
I just ,tofu in here and then copy/paste each time
domogled has joined #nixos
<cole-h>
,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. See: tofu-vim
<cole-h>
You seriously copy-paste those 52 zeroes? haha
<gchristensen>
yep.
gxt has quit [Remote host closed the connection]
<peelz>
tofu-vim?
<bqv[m]>
oh, fetchTarball is a builtin. cole-h: why fetchTarball over fetchurl?
<gchristensen>
,tofu-vim
<{^_^}>
<esc>52i0<esc>
<peelz>
lmao
<bqv[m]>
isn't fetchTarball the one that doesn't require a hash
<bqv[m]>
ahahah
<cole-h>
¯\_(ツ)_/¯
gxt has joined #nixos
<cole-h>
Just what I'd do. It's your package, so do what you feel right
<bqv[m]>
i just find a hash already in the file and ver0
<peelz>
bqv[m]: I would use fetchgit too. It just looks look cleaner than an interpolated URL. Personally, I default to git unless the repo is massive.
<bqv[m]>
don't do this to me guys
<bqv[m]>
i literally just pushed changing it back
<peelz>
LOL I'm sorry, just giving my opinion. I'm sure it's fine either way
<ottidmes>
What is going on, if I run any application through rofi desktop launcher I get the error: No GSettings schemas are installed on the system
<ottidmes>
When I run the application directly from the command line, it works without issue
<ottidmes>
Seems like an environment variable issue, but not sure what is the cause
<ottidmes>
I am running latest nixos-20.03 with a few latest nixos-unstable packages (not used when testing this)
chloekek has quit [Ping timeout: 260 seconds]
<ottidmes>
I am using i3 as my window manager, and thought maybe something changed causing the breakage, but even when I enable xfce with no window manager and desktop, it still fails with the same error, so it does not seem like I am missing any fundamental desktop manager related
<worldofpeace>
ottidmes: Can you try gnome-calculator for me? I know this is wrapped
<ottidmes>
worldofpeace: will do, all the apps that crash are wrapped, but guess thats to be expected, your point is about it being fixed, right?
mexisme has joined #nixos
<worldofpeace>
ottidmes: ohh. I can try it by myself by doing rofi -show drun right?
<ottidmes>
worldofpeace: yes,
<ottidmes>
worldofpeace: I tried just now, gnome-calculator works, although it ran with run/rather than drun, but then again trying run with vlc through rofi fails too
lovesegfault has joined #nixos
<worldofpeace>
ottidmes: It all works for me. Are you using gnomeplatform for qt gtk styling?
<worldofpeace>
note, I've ran rofi like `XDG_DATA_DIRS=/run/current-system/sw/share ./result/bin/rofi -show drun`. That is because my environment will be polluted by being launched with a program that is wrapped
<ottidmes>
worldofpeace: only gnome-calculator works for me. I was running no desktop manager before, just i3 with some fixes (which I called desktop lite) that has worked for me for ages, I enabled xfce to be sure, so should those styling things work for xfce too?
<ottidmes>
worldofpeace: XDG_DATA_DIRS=/run/current-system/sw/share rofi -show drun fails for me too for vlc
<ottidmes>
worldofpeace: Ah, maybe that's it, I have switched to gnomeplatform!
relaxmax has quit [Ping timeout: 265 seconds]
<worldofpeace>
ottidmes: I was about to tell you. vlc is a pure qt app (I think)
<ottidmes>
worldofpeace: yep, must be that, I am reverting: { qt5 = { enable = true; platformTheme = "gnome"; style = cfg.qt5Theme; }; } to my old config, hopefully that fixes things
<worldofpeace>
ottidmes: vlc is just an example. it's any app that uses qt and then uses the qgnomeplatform theme. It depends on gsettings-desktop-schemas, but you can't wrap .so
<worldofpeace>
*platformtheme
ixxie has quit [Ping timeout: 265 seconds]
<worldofpeace>
should be in nixos-unstable in a few days
jumper149 has quit [Quit: WeeChat 2.7.1]
smatting has joined #nixos
ixxie has joined #nixos
<ottidmes>
worldofpeace: cool, thanks a lot for the help, I have had too many changes recently so it was not easy to pinpoint the cause
<ottidmes>
worldofpeace++
<{^_^}>
worldofpeace's karma got increased to 90
<ottidmes>
Its working again :D
Naptra has quit [Remote host closed the connection]
<worldofpeace>
ottidmes: you're welcome. You were lucky in that there's already a committed fix too
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/Jvbc2
<cole-h>
gchristensen: Is it OK to be listing the entire paths from the builders? Looking at a log, idk if "sensitive" information could ever be leaked as a result of my "add paths to error" change
<gchristensen>
yep thats fine!
ixxie has quit [Ping timeout: 265 seconds]
<cole-h>
OK, great./
nek0 has joined #nixos
__monty__ has quit [Quit: leaving]
nek0 has quit [Remote host closed the connection]
aranea has quit [Quit: aranea]
<bqv[m]>
> (lib.fix lib.mergeAttrsWithFunc) { "a" = {}; } { b = "c"; }
<bqv[m]>
why does this fail
<{^_^}>
{ a = <CODE>; b = <CODE>; }
<cole-h>
> :p (lib.fix lib.mergeAttrsWithFunc) { "a" = {}; } { b = "c"; }
<{^_^}>
value is a string while a set was expected, at /var/lib/nixbot/nixpkgs/master/repo/lib/deprecated.nix:183:28
<bqv[m]>
oh, what
<bqv[m]>
it doesn't
<bqv[m]>
but it does in repl
<bqv[m]>
ah
<bqv[m]>
yeah, why does that fail, is there something about mergeAttrs i'm not seeing
nek0 has joined #nixos
thetet has left #nixos [#nixos]
<bqv[m]>
it's almost like it's trying to form a type for the toplevel attrsets. that or it's actually comparing {} to "c" for whatever reason
erasmas has quit [Quit: leaving]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<gchristensen>
cole-h: github just can't keeep their stuff together.
<gchristensen>
Error("missing field `created_at`", line:3, column: 1),
<{^_^}>
[nixpkgs] @bhipple pushed 3 commits to release-20.03: https://git.io/JvbCi
ZaraChimera has joined #nixos
<{^_^}>
[nixpkgs] @worldofpeace pushed 3 commits to staging-next: https://git.io/JvbCQ
kiloreux has quit [Remote host closed the connection]
kiloreux_ has quit [Remote host closed the connection]
<immae>
Hello, I’m a bit surprised by how much the apache configuration (nixos/modules/services/web-servers/apache-httpd/default.nix) has become invasive in latest channel, like forcing in every virtual host the DocumentRoot directory to be "Require all granted" for instance (which seems like a non-wanted enforced option when we want non-public vhosts), did I miss some option?