<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 3 commits to master: https://git.io/JL7mi
supersandro2000 has quit [Disconnected by services]
supersandro2000 has joined #nixos
mallox has joined #nixos
new2nix has quit [Remote host closed the connection]
Dr8128 has quit [Ping timeout: 260 seconds]
<pushqrdx>
jasom abathur yeah i just changed my mkShell into mkDerivation blindly and tried nix-build and it seems to be building without any additional config, where can i learn more about other stuff can nix handle automatically
<pushqrdx>
like where are these defaults specified, i just wanna look at them so they don't seem like pure voodoo to me :D
dduduruh has quit [Remote host closed the connection]
<abathur>
pushqrdx: as far as documentation goes, I think a few subsections in the stdenv section of the nixpkgs manual are helpful; https://nixos.org/manual/nixpkgs/stable/#sec-stdenv-phases is helpful as well as the setup-hook section jasom mentioned
<pinpox>
anyone here using awesomeWM? since my last system-update firefox is starting in floating mode and refusing to tile, anyone else experiencing this or is it just my config?
<donofrio>
how would I get partclone on my osx imac?
OJ[m] has joined #nixos
<abathur>
is it known to work on macOS?
<abathur>
I went to the site and just see "Operating system: GNU/Linux"
scott has quit [Quit: Ping timeout (120 seconds)]
scott has joined #nixos
<donofrio>
abathur, osx is bsd...
<Yaniel>
and BSD is not GNU/Linux
o1lo01ol1o has joined #nixos
v0|d has joined #nixos
o1lo01ol1o has quit [Ping timeout: 240 seconds]
<pushqrdx>
abathur thanks for the links they are very helpful
<{^_^}>
[nixpkgs] @SuperSandro2000 merged pull request #107465 → audacity: switch to cmake to fix missing plug-ins → https://git.io/JL7Ot
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JL7Oq
<{^_^}>
[nixpkgs] @github-actions[bot] pushed 58 commits to staging-next: https://git.io/JL7Os
<{^_^}>
[nixpkgs] @github-actions[bot] pushed 59 commits to staging: https://git.io/JL7OG
<Rusty1>
onixos versions
<abathur>
donofrio: these tend to go 1 of 3 ways: you set `nixpkgs.config.allowUnsupportedSystem = true;` and 1.) it just works (and PR to update the package platform); 2) it doesn't work and you have to find/PR fixes for macOS; 3) same as 2, but eventually fix your way to something that is intrinsically incompatible
mbrgm_ has joined #nixos
<abathur>
donofrio: so, figuring out if people who aren't using Nix have it running on macOS is a good signal that you won't just waste your time trying to get it working only to find, hours or days later, that it or one of its dependencies just isn't at all compatible
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JL7Ow
arghzero[m] has joined #nixos
<KarlJoad>
Is there a way to have a pure nix-shell when using the :s or :u command in `nix repl`?
<KarlJoad>
Wait, nevermind. I got an alternative working.
pushqrdx has quit [Remote host closed the connection]
<KarlJoad>
Although, I am running into another issue. I can't get python3 into a default.nix. I am passing it as a buildInput, but python is needed at runtime. Should it go in another inputs list?
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 3 commits to master: https://git.io/JL73R
<abathur>
what language? how is it invoking python?
<KarlJoad>
The GNU Octave package "symbolic" uses sympy and python as a back-end to perform symbolic calculations.
<{^_^}>
[rfc39-record] @grahamc pushed commit from rfc39 to main « Automated team sync results. »: https://git.io/JL73D
spudly1 has quit [Ping timeout: 246 seconds]
<abathur>
I think putting it in propagatedBuildInputs may be what you need if it literally needs to be a runtime PATH lookup, but I get the impression that propagating is a bit of a last-resort
<{^_^}>
[nix] @acx0 opened pull request #4412 → Fix `configure` error in introduction doc → https://git.io/JL73p
<KarlJoad>
Yeah, a runtime PATH lookup is probably what it does.
<KarlJoad>
But, I just changed python and some of its libraries to propagatedBuildInputs, and they still don't show up in a sub-shell.
<abathur>
I think the PATH lookup isn't preferred because PATHs are still a bit of a crude instrument and there's no guarantee it won't clash or misbehave, and that wrappers or patching would be preferred--but I'll be honest that I'm still feeling out the when/where/why/how of this myself
supersandro2000 has quit [Quit: Ping timeout (120 seconds)]
<KarlJoad>
abathur: What do you mean? Do you want to see what I'm doing?
<abathur>
are you using withPackages?
supersandro2000 has joined #nixos
zangi has quit [Ping timeout: 246 seconds]
<abathur>
just commenting further on what I meant by "I get the impression that propagating is a bit of a last-resort"
<{^_^}>
[nixpkgs] @AndersonTorres pushed 2 commits to master: https://git.io/JL7nf
boschma has joined #nixos
ransom has joined #nixos
<{^_^}>
[nixpkgs] @mjlbach opened pull request #108093 → sumneko-lua-language-server: init at 1.9.0 → https://git.io/JL7ns
donofrio has quit [Ping timeout: 246 seconds]
boschma has left #nixos [#nixos]
boschma has joined #nixos
boschma has quit [Client Quit]
boschma has joined #nixos
ransom has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
anderslundstedt has joined #nixos
donofrio has joined #nixos
<colemickens>
I have a shell script that relies on a couple things being available in PATH
<colemickens>
normally its used in a nix-shell so I don't mind
anderslu1dstedt has quit [Ping timeout: 256 seconds]
<colemickens>
but I also want to expose it for use with `nix run ...` so I want to expose it as a derivation with propagatedbuildinputs. I'm thinking a standard derivation with no build phase and propagatedbuildinputs might do what I want?
<abathur>
oh, huh, I guess this is rawtaz from restic? are you always hanging around here and I just haven't noticed? :)
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 3 commits to master: https://git.io/JL7cD
<{^_^}>
[nixpkgs] @siraben opened pull request #108097 → ctodo: enable build on unix → https://git.io/JL7cS
<{^_^}>
[nixpkgs] @SuperSandro2000 opened pull request #108098 → pythonPackages.pillow-simd: Init at 7.0.0.post3 → https://git.io/JL7cH
IRCsum has quit [Remote host closed the connection]
gustavderdrache has quit [Quit: Leaving.]
o1lo01ol1o has joined #nixos
inkbottle has quit [Remote host closed the connection]
inkbottle has joined #nixos
<{^_^}>
[rfc39-record] @grahamc pushed commit from rfc39 to main « Automated team sync results. »: https://git.io/JL7cp
okt has quit [Quit: WeeChat 3.0]
o1lo01ol1o has quit [Ping timeout: 264 seconds]
orivej has joined #nixos
skeptor has joined #nixos
<skeptor>
hi
<skeptor>
i am on nix on droid
<skeptor>
how can i make nix as essential and controll over my android
<humancalico[m]>
Hey can anyone help me with setting up OCaml and utop? I have this https://paste.rs/hMv in my `shell.nix` but on typing `open Base;;` in `utop` I am getting `Error: Unbound module Base`
<energizer>
supersandro2000: that's only two of the four things, and `setup.py test` is deprecated anyway
<energizer>
unless propagatedNativeBuildInputs and buildInputs aren't used, in which case it would be helpful if it said that
endformationage has quit [Quit: WeeChat 2.9]
orivej has quit [Ping timeout: 240 seconds]
kalbasit has quit [Ping timeout: 240 seconds]
orivej has joined #nixos
kalbasit has joined #nixos
<quinn>
supersandro2000: hey bro, not trying to be rude but is there any reason you haven't merged 107997? it looks like you reviewed it as ok, but no merge, is there anything else i need to do?
<{^_^}>
[nixpkgs] @FRidh pushed commit from @r-burns to staging-next « gnutls: fix build on darwin »: https://git.io/JL7EJ
<{^_^}>
[nixpkgs] @FRidh closed pull request #108087 → [staging-next] gnutls: fix build on darwin → https://git.io/JL7Yl
cole-h has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @gurjeet opened pull request #108109 → Vim: Added vim_darwin package for macOS → https://git.io/JL7EI
davidv7 has joined #nixos
davidv7_ has quit [Ping timeout: 240 seconds]
__monty__ has joined #nixos
cole-h has joined #nixos
<{^_^}>
[nixpkgs] @fabaff opened pull request #108110 → badchars: init at 0.4.0 → https://git.io/JL7Et
quinn has joined #nixos
mananamenos has joined #nixos
arjen-jonathan has joined #nixos
orivej has quit [Ping timeout: 240 seconds]
o1lo01ol1o has joined #nixos
cole-h has quit [Ping timeout: 256 seconds]
nly has joined #nixos
Dr8128 has joined #nixos
Mateon2 has joined #nixos
nly has quit [Ping timeout: 256 seconds]
novotny has joined #nixos
arjen-jonathan has quit [Ping timeout: 260 seconds]
Mateon1 has quit [Ping timeout: 256 seconds]
Mateon2 is now known as Mateon1
sangoma has joined #nixos
hnOsmium0001 has quit [Quit: Connection closed for inactivity]
mananamenos has quit [Ping timeout: 240 seconds]
<immae>
It seems like I cannot use a "variable" name in imports (a variable coming from _module.args), I keep getting a "infinite recursion error", is there a way to workaround it?
mananamenos has joined #nixos
<immae>
(the variable name is the server hostname, I want to import a file whose name depends on that host name)
<akavel>
immae: I do have import based on contents of /etc/hostname in my own home-manager config. But I don't understand what you mean by _module.args?
<Ke>
does eatmydata refuse to run for other people when it thinks user wants to run eatmydata as command and can't find itself
<akavel>
immae: could the infinite recursion be result of something else? can you import the file if you hardcode the path?
<immae>
akavel: morph adds the server name to the _module.args, and I want to use it. But I could as well use config.deployment.targetHost
<immae>
Yes it works if I hardcode the path
<Ke>
it seems that eg. for the wrapper script somehow $1 is eatmydata for cmdline eatmydata true
<immae>
akavel: note that I’m not building locally so I cannot use the content of /etc/hostname
Kozova1 has quit [Remote host closed the connection]
<akavel>
immae: uh; no idea what is morph :/ sorry I'm kinda early-intermediate nix user...
<immae>
No problem :)
fortuneteller2k has joined #nixos
<immae>
morph plays little role here apart from the fact that it’s building a configuration for another system
<immae>
it’s just : trying to use a file with a variable name in "imports" that leads to infinite recursion
<immae>
(by variable I mean anything that comes from the "{ config, ... }" args)
<akavel>
hm ok I checked: in my config I'm not exactly using import, but rather `imports = [ "${home}/.config/nixpkgs/modules/host-${hostname}.nix" ];`
<immae>
and what is ${hostname} in your case?
<akavel>
`readFile /etc/hostname`
<immae>
ok
<immae>
so it’s not coming from the module args ;)
<immae>
I guess I’ll have to find another way to do my imports
fortuneteller2k has quit [Quit: WeeChat 3.0]
<akavel>
hm; I guess infinite recursion makes sense then IIUC what _module.args is: imports can give you new modules
<akavel>
so generating new inputs based on results of modules... can lead to a loop...
<steveeJ>
akavel: a look at $NIX_PATH might be an easy place to discover that
<steveeJ>
akavel: ^ is referring to where the unpacked channel is stored
<akavel>
steeveeJ: thanks, I'll check it!
<akavel>
immae: one thing that came to my mind is maybe you could always import all the files, but only enable some of their parts conditionally based on the hostname?
<akavel>
immae: I think this might be how the whole modules system is constructed?
<immae>
yes it’s indeed what I did so far (enabling parts conditionnally) but I have a huge infrastructure and I wanted to make things a bit more "separate"
<immae>
I thought variable imports would be the way but it seems not :)
<akavel>
steveeJ: works, thanks!!! <3 <3
<akavel>
immae: I mean, I think effectively the same result would be if you did `imports = [ "hostname-foo" "hostname-bar" ...]`, then
<akavel>
each "hostname-X" would define its own module
<immae>
yes just thought of that 1 minute ago :D
<akavel>
and in main config you'd have something like: `hostname-foo.enable = (hostname == "foo")`
<akavel>
or maybe *this* could actually be internalized into those modules already based on _module.args
<akavel>
I mean, maybe there's some way to auto-import hostname into the modules logic; but I'm not sure
<immae>
there is
<immae>
The hostname is available everywhere
<immae>
So instead of conditionnal import I can import everything and then in the import itself decide if I do something of it or not
gentauro has quit [Read error: Connection reset by peer]
gentauro has joined #nixos
MidAutumnHotaru has quit [Quit: Quit 啾]
MidAutumnHotaru has joined #nixos
orivej has joined #nixos
ym555 has joined #nixos
<{^_^}>
[nixpkgs] @rycee merged pull request #103646 → lombok: add bin output with lombok wrapper script → https://git.io/JkmqD
<{^_^}>
[nixpkgs] @rycee pushed to master « lombok: add bin output with lombok wrapper script »: https://git.io/JL7gD
<aasg>
Just to check if I'm missing something: there's no way to pass `specialArgs` to container configs other than calling eval-config.nix or nixosSystem yourself, right? I'm trying to pass a flake's inputs down to containers so they can pick which modules to import.
<danderson>
aasg: I struggled with this and figured a thing, one sec
<aasg>
@danderson yes, I was doing something similar but it ends up in an infinite recursion if I then use `flakes` in `imports`, hence why (as I understand it) I need `specialArgs`
<danderson>
ah yes, I remember that. I had to stop doing that in my imports
<danderson>
unfortunately I don't think that can get fixed without fixing how containers get built in the nixpkgs source
<leibniz[m]>
If two packages need the same lib, (for example gtk-lib), does nix install 2 libs or uses 1 and shares??
<immae>
It depends if the two packages use the same nixpkgs or not
kim0 has joined #nixos
ym555_ has joined #nixos
<{^_^}>
[nixpkgs] @danieldk opened pull request #108121 → Use cargoHash in buildRustPackage crates with SRI hashes → https://git.io/JL7az
<immae>
aasg: I had the same issue as you just yesterday (see discussion above from this morning)
<leibniz[m]>
immae: The dependencies are listed in the nixpkgs manifest, right?
<immae>
aasg: However I’m able to import a flake using the flake-compat repo, the issue is not with being a flake, it’s with using pkgs or lib in doing the import
<leibniz[m]>
So its up to the nix manifest maintainer to not bloat the apps by trying to use as much shared libs as possible
<immae>
leibniz[m]: yes
<immae>
aasg: if you want an example: https://git.immae.eu/?p=perso/Immae/Config/Nix.git;a=commitdiff;h=74a10960d8659eaffa8224cee31fcf217c69609f
<leibniz[m]>
Why Snap or flatpak didn't do this?
ym555_ has quit [Client Quit]
<immae>
leibniz[m]: does Snap handles "dependencies"?
ym555 has quit [Ping timeout: 264 seconds]
<immae>
I thought they were all standaloe packages
<leibniz[m]>
immae: as far as I know, it just puts everything an app needs beside it
<leibniz[m]>
complete isolation...but there must be a reason they did this
<leibniz[m]>
I wonder what's the price for it in nix
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
<hexa->
would it be possible to get these links posted without the git.io shortener ^ I'm filtering them with my keyboards, and all of the recent links are annoyingly alike. Meanwhile, if I could just write the issue number and press enter, that would be awesome.
gustavderdrache has joined #nixos
waleee-cl has joined #nixos
<hexa->
gchristensen: ^
<gchristensen>
samueldr: ^ :)
<exarkun>
isn't the number there too? eg 108126, 108027
<Yaniel>
not in the url
euandreh has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<pinpox>
I don't fully understand how to activate that config on plasma5, the list of packages at least doesn't seem to get installed on setting plasma5.enable = true;
<roym>
Hi. anyone have a pointer to how to add a systemd service for a package? I install the package (in /etc/nixos/configuration.nix using `environment.systemPackages = with pkgs....` and have a `systemd.services."navidrome"....` section defining the systemd service config. My problem seems to be that I don't know the correct path to the executable.
<roym>
journalctl shows `Jan 01 09:29:41 nixos systemd[21692]: navidrome.service: Failed at step USER spawning /run/current-system/sw/bin/navidrome: No such process` and that works on the command line
<{^_^}>
[nixpkgs] @fabaff opened pull request #108135 → ldapdomaindump: init at 0.9.3 → https://git.io/JL7oa
<sterni>
Mic92: is there any difference between how nixpkgs-review detects if a package is broken and nix? I can build certain packages locally with allowBroken = false without any problems, but in nixpkgs-review they are detected as marked as broken.
<sterni>
tried to look into the code, but couldn't figure it out
<sterni>
packages are ocamlPackages.arp ocamlPackages.mirage-nat ocamlPackages.mirage-vnetif ocamlPackages.tcpip
respawn_ has quit [Quit: Leaving]
attila_lendvai has joined #nixos
attila_lendvai has joined #nixos
attila_lendvai has quit [Changing host]
<Mic92>
sterni: nixpkgs-review does not build broken packages
<Mic92>
sterni: mhm. Is ofborg reporting the package as broken?
<{^_^}>
[nixpkgs] @vcunat pushed 16 commits to release-20.09: https://git.io/JL7KT
<mikroskeem>
i asked that before, but asking again in case it was missed - any plans for upgrading docker to 20.10 in any near future?
<mikroskeem>
if not, then i could try doing it myself
<supersandro2000>
I don't see any PR and don't know anyone doing it right now
<supersandro2000>
FYI main repo changed from docker/docker-ce to moby/moby IIRC
<mikroskeem>
yes
<mikroskeem>
tbh that's where i got stuck also before while updating my overlay...
<afreak>
i got an error similar to this https://github.com/NixOS/nixpkgs/issues/97433 when I ran nixos-rebuild boot, and now I cant boot the latest image in bootloader, but loading 2nd latest was fine, which i am running atm, but should I now update channels and rebuild again while on 2nd latest bootentry?
<brano543>
Hello everyone and happy new year 2021 !
<brano543>
I am trying to cross-compile openssl_1_1 with MingW64 toolchain, but I am getting "configure: error: could not determine how to read list of mounted file systems". OpenSSL 1.0.2 works fine. Does anyone have a clue what could be causing this in configure script so that we can patch it?
<mikroskeem>
i wonder why openssl even tries to do that
<mikroskeem>
my one bet is that it's trying to read /etc/mtab instead of /proc/mounts
<mikroskeem>
oh wait, it's mingw lol
<mikroskeem>
no clue; maybe you'll be able to find something by grepping those paths
endformationage has joined #nixos
<brano543>
mikroskeem: They must be using some call inside configure which wasn't there before, but I am unable to find which one as the file has over 3k lines
<brano543>
just before it fails it says
<brano543>
checking for BEOS mounted file system support functions... noconfigure: error: could not determine how to read list of mounted file systems
<brano543>
mikroskeem: when I check the configure script for 1.0.2 which works with the exactly same mingw toolchain, I don't see it would try to invoke the perl.
<mikroskeem>
openssl's Configure script is written in perl, no?
<brano543>
mikroskeem: Sorry, there is just different way how it invokes in old script by using eval
<{^_^}>
[nixpkgs] @AndersonTorres opened pull request #108141 → wio: init at unstable-2020-11-02 → https://git.io/JL7M2
<{^_^}>
[nixpkgs] @Ma27 opened pull request #108142 → grafana-image-renderer: init at 2020-12-01 → https://git.io/JL7Ma
hiro99 is now known as hiro98
<brano543>
mikroskeem: The problem seems to be in how the configure script is generated. I have tried to compile the same by using Nix + Conan openssl 1.1.1 recipe and there it works. It definitely needs to be in way how it generates the makefile
<brano543>
mikroskeem: Hmm this is strange. I have tried to manually run the perl configure through nix-shell and it succeeded
<brano543>
mikroskeem: What is strange that there were no "Checking" commands even. perl ./Configure mingw64Configuring OpenSSL version 3.0.0-alpha10-dev for target mingw64Using os-specific seed configurationCreating configdata.pmRunning configdata.pmCreating Makefile
<mikroskeem>
it must be something else
<mikroskeem>
in that case
<brano543>
it literally finished in 4 rows
<brano543>
I will try to add the config options one by one as nix recipe does
respawn_ has joined #nixos
nek0 has quit [Remote host closed the connection]
<brano543>
but wait a second, I just noticed I just tried the latest master. Maybe the bug is only in 1.1.1
<mikroskeem>
but that doesn't make any sense
<mikroskeem>
BEOS thing ain't referenced anywhere in openssl
<{^_^}>
[nixpkgs] @fabaff opened pull request #108144 → python3Packages.python-mystrom: init at 2.0.0 → https://git.io/JL7D9
sss2 has joined #nixos
<brano543>
mikroskeem: I believe my problem is lack of understanding of overlays. Can I share my nix files with you? I thought that just defining overlay for given cross package is enough. But somehow it tries to incorrectly use build steps for linux and combines it with correct compiler....
<mikroskeem>
you can share but i wouldn't want to deal with figuring out the issue today, sorry
<mikroskeem>
bit busy with other things
zebrag has quit [Quit: Konversation terminated!]
zebrag has joined #nixos
v0|d has quit [Read error: Connection reset by peer]
ManiacOfMadness has quit [Ping timeout: 264 seconds]
<brano543>
mikroskeem: I think you can help me in one minute. I have identified the trouble. I don't know how can I tell the derivation that it should use the cross system I defined. https://pastebin.com/Cn3YLMFb
<brano543>
mikroskeem: I think that even when I tell him mingw.openssl he doesn't know I need steps for mingw, so he tries all routines for linux anyways
ilios has joined #nixos
<mikroskeem>
cross compiling has always been hell of a mess with nix imho
<mikroskeem>
source: i know a dude who cross-compiled nixos for armv7
<mikroskeem>
i don't know how to help with that :/
<brano543>
mikroskeem: I try to run the build as "nix-build --argstr system 'windows'". Maybe I am missing something, I have literally read every documentation I could find on this topic
<brano543>
mikroskeem: I think the problem is somehow he doesn't try to invoke the ./Configure with perl. I am going to try one thing to try appending perl within the configureScript
lassulus has joined #nixos
fuzzypixelz has quit [Ping timeout: 240 seconds]
mikroskeem has quit [Quit: i'll set up a bouncer again one day ~ back tomorrow]
<{^_^}>
[nixpkgs] @ryantm opened pull request #108147 → doc: add anchors for Markdown sections; other fixes → https://git.io/JL7y5
<{^_^}>
[nixpkgs] @veprbl pushed commit from @bcdarwin to master « itk: 5.1.1 -> 5.1.2 (#107502) »: https://git.io/JL7SV
fuzzypixelz has joined #nixos
mallox has joined #nixos
<{^_^}>
[nixpkgs] @nagy opened pull request #108151 → unbound: optionally support DNS-over-HTTPS → https://git.io/JL7So
mlen has quit [Quit: WeeChat 2.9]
<donofrio>
abathur, I have no AVX extention on my 2008 imac
mlen has joined #nixos
lassulus has quit [Quit: WeeChat 2.9]
lassulus has joined #nixos
<brano543>
mikroskeem: I have found the issue. the postPatch part somehow invokes configure
<brano543>
mikroskeem: If I remove postPatch part and comment out configureScript,it will say no configure script exists. However when there is only postPatch section and configureScript commented, it invokes the wrong configure
<abathur>
`nix-info -m` if you have it installed, or `nix-shell -p nix-info --run "nix-info -m"` if that doesn't work
<higherorder>
Hello! I'm trying to run Deno with Nix on MacOS via `nix-shell -p deno`, but getting the following error: https://pastebin.com/8yFYWyJM. Could someone hint me as to what might be going wrong? I'm on Catalina (not M1).
zebrag has quit [Remote host closed the connection]
<higherorder>
Also: nix-shell (Nix) 2.3.7
cr4y1 has quit [Ping timeout: 256 seconds]
jmeredith has joined #nixos
Lord_of_Life_ has joined #nixos
pushqrdx has joined #nixos
<pushqrdx>
i remember that there was a way to extend the packages inside `with import <nixpkgs>` expressions, i don't mean overlay but actuall add more packages with callPackage
<pushqrdx>
can't find or remember the syntax for that though
<donofrio>
this is why I was compiling....at least it was working, I must have messed something up ;(
mkaito has joined #nixos
mkaito has joined #nixos
neiluj has quit [Ping timeout: 260 seconds]
<abathur>
I'm not sure; definitely something weird here
<abathur>
your nixpkgs is fairly recent, so that rules out the possibility you were just missing some fixes that I saw in the issue tracker over the fall
<abathur>
do you see anything if you run `nix show-derivation /nix/store/zyy834ibw1f1k25cgx05g2w5gmpgb343-source`?
<pushqrdx>
why isn't it easy to do something like this http://ix.io/2Kx8 i want the import <nixpkgs> to propagate down to these derivations
hnOsmium0001 has joined #nixos
<abathur>
higherorder: deno may just have been broken in nixpkgs at that point, you may just need to update your channel/checkout unless you're overriding it; can see on https://hydra.nixos.org/job/nixpkgs/trunk/deno.x86_64-darwin/all that all but 1 of the hydra builds at 1.3.3 had problems
<{^_^}>
[nixpkgs] @nh2 opened pull request #108157 → [20.09] thunderbird: Add gpg/gpgme dependencies to fix smartcard usage → https://git.io/JL7FO
zebrag has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @peti pushed 5 commits to haskell-updates: https://git.io/JL7FW
Chiliparrot has joined #nixos
<govanify>
hey, i've been trying to patch system.build.installBootLoader for a bit now; it's a bit of a pain as grub's is defined as a perl script in a let. I ended up making a correct perl script override but am getting stuck on the actual overlay overriding part, either it doesn't interpret it or leads to infinite recursion with a traceVal, any idea what might be wrong? relevant snippet:
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « haskell: drop more obsolete overrides »: https://git.io/JL7bf
ctlee has joined #nixos
<govanify>
this currently doesn't get evaluated at all, and adding a lib.debug.traceVal before the lambda leads to infinite recursion
<{^_^}>
[nixpkgs] @r-burns opened pull request #108159 → [staging-next] pulseaudio: fix build on darwin → https://git.io/JL7bL
sangoma has joined #nixos
<{^_^}>
[nixpkgs] @minijackson opened pull request #108160 → nixos/shiori: fix SystemCallFilter after libseccomp update → https://git.io/JL7bY
ctlee has quit [Remote host closed the connection]
<steveeJ>
repeating my message from this morning (here): (for the first time :-D) I set up nginx to use ACME and I'm surprised that firefox doesn't trust the certificate. is that expected?
<q3k[m]>
lassulus: so how it's going? i got the pusher to build (grpc in nodejs is cursed), not sure if i should try doing the rest, or wait for you
<lassulus>
q3k[m]: supersandro2000 finished most of the components yesterday, so they build in some way, but we didn't verify it further then just starting the components: https://github.com/SuperSandro2000/workadventure-nix
<q3k[m]>
yeah i just saw that via another channel :)
<lassulus>
I will try to build some modules when I got some other stuff finished, earliest tomorrow evening
<q3k[m]>
could you drop in a LICENSE/COPYING file there?
<q3k[m]>
or well, could you, supersandro2000, do it?
<jdnixx[m]>
does nixos change any of the completely "stock" KDE/plasma default settings?
<jdnixx[m]>
like is there anything in ~/.config that would be different at all from say a totally fresh Kubuntu installation?
<jdnixx[m]>
because I know in the nixos live ISO's at least, I've noticed a couple things like 2-finger touchpad scrolling enabled which I'm pretty sure isn't present on other KDE distros
<jdnixx[m]>
and then of course the convenience aliases like `ll` and `l` for ls -alF
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « ghc-9.0.1: set utf-8 locale during the build so that Haddock succeeds »: https://git.io/JL7Nc
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JL7Nl
<jdnixx[m]>
btw, I reinstalled the other day from the unstable ISO (was on 20.09) and loving the crap out of plasma 5.20
<jdnixx[m]>
that was so needed lol
ManiacOfMadness has quit [Ping timeout: 256 seconds]
gustavderdrache has quit [Quit: Leaving.]
<supersandro2000>
q3k[m]: gimme a secon
<supersandro2000>
d
<supersandro2000>
MIT is fine?
<q3k[m]>
yep, that works
<cole-h>
samueldr: Maybe we could trial run for a day or two and collect feedback?
fuzzypixelz has left #nixos [#nixos]
<samueldr>
yeah, I'm asking first and foremost to see if there's an obvious reason not to do that
<supersandro2000>
abathur: don't know. Downloading something wrong. try to run nix-collect-garbarge
<abathur>
donofrio:^
<supersandro2000>
q3k[m]: MIT license added. Feel free to do PRs to it or adapt it to work with nixpkgs. Drop me a message when you need anything.
ManiacOfMadness has joined #nixos
<q3k[m]>
supersandro2000: thank you!
o1lo01ol1o has joined #nixos
<supersandro2000>
no problem. I just quickly dumped it somewhere.
aleph- has quit [Read error: Connection reset by peer]
Church- has joined #nixos
fooker has quit [Ping timeout: 272 seconds]
<{^_^}>
[nixpkgs] @peti pushed 2 commits to haskell-updates: https://git.io/JL7Am
<q3k[m]>
supersandro2000: i would like to depend on this from a repository with a pinned nixpkgs - would you mind if I sent a PR that changes the .nixes to be callPackage compatible functions instead of self-standing derivations? or would you like to go about this differently?
<{^_^}>
[nixpkgs] @mweinelt opened pull request #108164 → solanum: init at unstable-2020-12-14 → https://git.io/JL7xT
<{^_^}>
[nixpkgs] @rmcgibbo opened pull request #108165 → elfx86exts: init at 0.4.3 → https://git.io/JL7x3
<lukegb>
DigitalKiwi: I'm using libvirt for that, but nondeclaratively, so it's all state. That's mostly because VMs are reasonably ephemeral though.
peelz has quit [Remote host closed the connection]
zebrag has joined #nixos
<{^_^}>
[nixpkgs] @fabaff opened pull request #108169 → python3Packages.connect-box: init at 0.2.8 → https://git.io/JL5f6
sangoma has quit [Ping timeout: 264 seconds]
<donofrio>
supersandro2000, that will reset everything thought - won't it (took 12+ hours to compile everything (needed because I do not have AVX instruction set on my core duo imac circa 2008)
<DigitalKiwi>
lukegb: is there a way to use libvirt declaratively that you know of?
ddellacosta has joined #nixos
<DigitalKiwi>
anyone want to build nixops-plesk
<naom>
samueldr: thanks! btw, another question: if i recall it correctly you decided against using nixos-mobile for the pinebook. Are you planning on abstracting the similarities via some kind of lib? or do you want to keep them strictly seperate (with duplicate code)?
<samueldr>
good question, it's not anything required for the pinebook pro
<samueldr>
the mobile nixos tooling is more for touchscreen-first devices, where there is no classic bootloaders
<samueldr>
the pinebook pro is gaining display at the firmware/bootloader level, so there's no need for that indirection
<samueldr>
though, that stage-0/stage-1 code is abstracted away to be usable outside of strictly mobile nixos, I just added support for standard UEFI x86_64, and will soon dogfood it on my laptop
<samueldr>
and since it already somewhat supports rk3399 and u-boot, it's already trivial to use on the pinebook pro if you really wanted to, though not really that useful
<samueldr>
am I missing something in the answer?
<naom>
samueldr: a counter-example would for example be: ./modules/luks.nix
<naom>
samueldr: but i get your point
<samueldr>
oh, I think I see, you're more interested in the graphical boot aspect?
<naom>
samueldr: yes, i was playing arround with feature/gfx-u-boot - which caused me to end up there
<samueldr>
hmm, let me rephrase to ensure we aren't getting mixed-up: you can do luks without a GUI boot splash using the gfx-u-boot branch, where like on most other computers, the console asks you for the passphrase
<naom>
samueldr: furthermore i assumed that once the pinebookpro adds the GSM module (or if) this might be another overlap between the devices
<naom>
samueldr: interesting! in this case i broke something else, thanks :P
<samueldr>
n/p
<samueldr>
with gfx-boot there shouldn't really be a difference compared to the usual x86_64 kind of hardware for the "graphical" boot flow; not like before where some of it is happening hidden
<samueldr>
the display is initialized by the firmware (u-boot), which means the bootloader can use it (u-boot still, here), then the kernel may use it, though here it has to reset it, but it *should* be reset before stage-1 init is started IIRC
<samueldr>
so at that point we have the VT as usual, where the nixos stage-1 can ask you for the passphrase
<supersandro2000>
q3k[m]: we need to do that anyway when upstreaming to nixpkgs
<{^_^}>
[nixpkgs] @fabaff opened pull request #108171 → python3Packages.opensensemap-api: init at 0.1.5 → https://git.io/JL5JO
<q3k[m]>
supersandro2000: yep, you have a PR for that now ^^
<pinpox>
Do flake output *have* to be a package? I would like to manage my windowmanager's configuration with flakes and created this flake.nix, but it's not strictly a package, just a bunch of config files I would like to symlink here (and replace that let .. in definition)
<naom>
samueldr: on my device the boot-menu / nixos-version-selection works as expected, but instead of a passwordprompt i get a glitchy blackscreen, and after that an unrelated error is printed. But this was probably my bad due to messing with the kernel parameters/patches.
<naom>
samueldr: anyways, great work! thanks for your help :)
<samueldr>
naom: it might be that I'm assuming stage-1 starts after the display is ready, while maybe it isn't
<naom>
samueldr: but the display was ready during the uboot dialogs etc?
<naom>
pinpox: no, you can use flakes as modules-only
<pinpox>
naom: Woud that go under nixosModules then?
<samueldr>
naom: yeah, the kernel has assumptions there that the display has to be initialized again
<naom>
pinpox: exaclty, just add your modules there and include the flake in your "main" one
<pinpox>
I can only find more complex examples, not how to just "provide some files"
<naom>
pinpox: which kind of files? is this just about your window manager?
astrofog has quit [Ping timeout: 256 seconds]
<pinpox>
naom: For now yes, but I'd need a genereal example of how to do this. I'd like to just be able to drop a flake.nix in some of my github repos (like my awesomeWM dotfiles) and import them in my main flake.nix file
<pinpox>
DigitalKiwi: Do you know how flakes work? naom seems to have gone away
<DigitalKiwi>
i've used them a bit but not with home manager
<pinpox>
I'm getting an "infinite recursion error" and cant figure out why
<pinpox>
DigitalKiwi: can I build a flake.nix even if it only specifies modules not packages?
cosimone has quit [Quit: cosimone]
<colemickens>
what does it mean to "build" a module?
<DigitalKiwi>
nixops has an output that builds docs
<supersandro2000>
dongcarl: which options did you overwrite?
<pinpox>
colemickens: I have a flake that is supposed to specify a module. Can I "run" or "test" or "build" it somehow to verify it works correctly without rebuilding my whole system?
fendor has quit [Remote host closed the connection]
<pinpox>
I have to always: do changes, push changes, update systemconfig's flake input, rebuild system. Everytime I want to test it
<colemickens>
idk not really, it's a module, it needs to be evaluated fundamentally
<pinpox>
colemickens: Can you help me out with the flake inputs? I'm not finding why it fails
<DigitalKiwi>
at that point i start to make .bashrc functions
<colemickens>
pinpox: if you're doing local development, you can do `nix flake update --update-input nixpkgs /home/cole/code/nixpkgs` to change the nixpkgs to a local path
<colemickens>
pinpox: then you can just execute that command instead of push/pulling for each change.
<colemickens>
pinpox: slightly better.
<colemickens>
pinpox: I can't tell if "flake inputs... it fails" is a separate issue, but what's wrong?
<pinpox>
colemickens: hm. Ok. I posted the relevant lines above, do you see why it would result in infinite recursion?
<pinpox>
colemickens: I'm trying to have a repository be "importable" as flake input if that makes any sense.
ddellacosta has quit [Ping timeout: 256 seconds]
<colemickens>
sure, I have nixpkgs-wayland that I import from my nixcfg.
<colemickens>
or I have nixos-azure that just exports a module that I use from nixcfg
<colemickens>
so that is a reasonable thing to want
<colemickens>
I don't know whats not working though
<pinpox>
colemickens: That's pretty much what I want to do: Keep the dotfiles separate so I can use it with other OS's but have my home-manager config pull it in as a flake input (which in turn get's deployed with the system config)
<cizra>
Hello! I'm trying to compile a program that requires mariadb-connector-c and boost. I created a shell.nix that adds both to buildInputs. Both boost and mariadb-connector-c are added to NIX_CFLAGS_COMPILE, inside the shell, but mariadb-connector-c is not added to NIX_LDFLAGS (boost is added there). As a result, gcc fails to link against -libmysql. What might be wrong?
cole-h has quit [Quit: Goodbye]
riksutin has joined #nixos
<pinpox>
colemickens: It's just specifying two imports: "./home.nix" and "{ nixpkgs.overlays = [ nur.overlay ]; }" which are both modules
cole-h has joined #nixos
<pinpox>
colemickens: I updated the flake to be a module, but it still complains about infinite recursion
cole-h has quit [Client Quit]
<colemickens>
inf recursion is a huge pita
<colemickens>
maybe I'll just try to build it with a newer nix and see if I get a better error
<{^_^}>
[nixpkgs] @dasJ closed pull request #108177 → samba4: Properly use Python 3 and fix PYTHONPATH → https://git.io/JL5Tb
<colemickens>
and/or the package is a few bugfixes behind stable, apparently, maybe try bumping it too
<colemickens>
pinpox: can you remove that second "module" that you just described
<colemickens>
it also doesn't look like a module
<colemickens>
but maybe it was working previously, idk?
<pinpox>
colemickens: yes it was working
<{^_^}>
[nixpkgs] @jakeisnt opened pull request #108178 → nodePackages.esy: init at 0.6.7 → https://git.io/JL5k4
<pinpox>
with the dotfiles just being a let .. in statement that fetched them with fetchgit in awesome.nix
<pinpox>
I just don't want that let.. in fetchgit statement in there because I have to sepecify a hash and I much rather would have that be a flake input that is managed via the lockfile
<colemickens>
how are you testing this?
<colemickens>
(and sorry but I don't know what fetchgit you're talking about yet)
<pinpox>
colemickens: whoot.. sorry I think I just fixed it by accident?
<pinpox>
let me verify
<pinpox>
I'm just rebuilding my system's config that pulls in my home-manager config for my user as a flake input, that in turn pulls the flake.nix from my dotfiles
<cizra>
colemickens: hm, can't see anything obviously wrong. Anyway, thanks for the tips.
<colemickens>
pinpox: I guess one other tip would be to be wary of nix caching git rev lookups
<pinpox>
colemickens: Oh, maybe that was an error. I felt like I was changintg things and getting the same errors
<pinpox>
It seems to work mostly now
<pinpox>
I'll do some more testing and write it down
<colemickens>
pinpox: you'd need to have updated your nixos-home config, regenerated the lock, after fixing dotfiles-awesome
<colemickens>
pinpox: if you didn't then you'd just keep getting the bad "module".
<colemickens>
There is? should be? a flag to disable that caching. I wish. I would pretty much just always disable it. Its mostly just annoying for my usage.
lordcirth has quit [Remote host closed the connection]
riksutin has quit [Ping timeout: 260 seconds]
cz3 has joined #nixos
<pushqrdx>
so i keep hitting this `error: attempt to call something which is not a function but a set` every now and then and i still don't understand it
riksteri has joined #nixos
ransom has joined #nixos
<pushqrdx>
i just want to specify a let foo = { package = callPackage ../somedep {}; } in stdenv.mkDerivation ... and use foo.package inside my build inputs
<pushqrdx>
why does this end up throwing that error