<{^_^}>
[nixpkgs] @bhipple pushed 2 commits to release-20.03: https://git.io/JfzRK
Darkmatter66 has joined #nixos
m0rphism has quit [Ping timeout: 265 seconds]
Darkmatter66_ has quit [Ping timeout: 258 seconds]
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<anton-latukha[m]>
Good time to everyone. Someone knows how to get the `nixpkgs` commit (of `nixos` channel) from which the generation is built? It is quite useful to recommend someone the the commit right away.
<anton-latukha[m]>
* Good time to everyone. Someone knows how to get the `nixpkgs` commit (of `nixos` channel) from which the generation is built? It is quite useful to recommend someone the commit right away.
orivej_ has quit [Ping timeout: 240 seconds]
Max63 has joined #nixos
<{^_^}>
[nixpkgs] @Ericson2314 opened pull request #88222 → WIP: gobject-introspection: Support cross → https://git.io/Jfz0f
xd1le has quit [Ping timeout: 258 seconds]
Max63 has quit [Remote host closed the connection]
<bbigras>
you're right that the documentation could always be better. I have trouble at first with gpt and mbr, and not using grub. and having partitions not aligned.
lunik16 has joined #nixos
<freeman42x[m]1>
bbigras: if documentation was either .md or explained easily which software to use to edit it
rardiol has quit [Quit: No Ping reply in 180 seconds.]
rardiol has joined #nixos
<freeman42x[m]1>
bbigras: yeah, being docbook was the problem for me. The editor I used changed too much and my improvements never got in. And no-one else did them either... waste of our time
<freeman42x[m]1>
now that it already installed. can I do a nixos-rebuild switch or do I have to completely reinstall just to change where to boot from?
<freeman42x[m]1>
and which setting tells it where to boot UEFI from?
<freeman42x[m]1>
and even if I did... what can I do? can I nixos-rebuild switch?
<Gaelan>
if you boot the installer CD and mount things properly (root -> /mnt, boot partition -> /mnt/boot), you can run nixos-install again to change the configuration
<freeman42x[m]1>
do I need to nixos-install again?
<Gaelan>
afiak that won't destroy anything, just the equivalent of a rebuild
srid2 has quit [Quit: Connection closed for inactivity]
camsbury has quit [Ping timeout: 265 seconds]
<freeman42x[m]1>
if I re-do things from section 2.3 Installing, that should allow me to try new things by just slightly tweaking existing configuration.nix ?
<freeman42x[m]1>
Gaelan: in which folder should I run tree ?
<cole-h>
Do I just have to do `pkgs.writeShellScript "asdf" ''${some_external_package}/bin/some_external_command''` ?
<freeman42x[m]1>
I got tree installed
<bbigras>
cole-h: I'm going to sound like an idiot because you very likely know a lot more than me but do you mean like a bash script with wrapProgram?
<Gaelan>
freeman42x[m]1 /mnt/boot
<colemickens>
nix-shell?
<cole-h>
colemickens: I've tried. Unfortunately, the service that tries to run the script gets really mad when I use a nix-shell shebang
<Gaelan>
cole-h could you generate the script from a derivation?
<Gaelan>
nothing obviously wrong there freeman42x[m]1
<bbigras>
like this `wrapProgram script.sh --prefix PATH : "${pkgs.git}/bin"` ?
<cole-h>
Yes, but then I have to get `wrapProgram` somehow.
<Gaelan>
freeman42x[m]1 does vmware give you any indication of why it's not booting? can you see the disk in any sort of bios menu?
<Gaelan>
sorry if we already asked this, but you are sure you've done whatever you need to to turn EFI on in vmware, right?
<freeman42x[m]1>
Gaelan: I never needed to do any special setting to get NixOS installed in a VM
<freeman42x[m]1>
but I think I always installed NixOS using the BIOS method
<Gaelan>
freeman42x[m]1 ok. at least on virtualbox there's an "enable EFI" checkbox but i've never used vmware
<freeman42x[m]1>
so I should reset everything and do BIOS method I think
rardiol has quit [Ping timeout: 264 seconds]
<freeman42x[m]1>
Gaelan: VMWare is like a better VirtualBox
rardiol has joined #nixos
<freeman42x[m]1>
much less buggy
<freeman42x[m]1>
Gaelan: if I want to reset all and install using BIOS... is it enough to just delete all partitions and start from step 1 ?
<Gaelan>
freeman42x[m]1 yep, would do just fine. if you wanted to be safe you could make a new VM, i don't think it adds that much trouble
<clever>
freeman42x[m]1: if you want BIOS booting on gpt, its pretty simple
<clever>
freeman42x[m]1: just make a bios boot partition (its a special type), 1mb, not formatted, not mounted, then set boot.loader.grub.device = "/dev/sdX";
felixfoertsch23 has joined #nixos
<clever>
freeman42x[m]1: if grub supports whatever is on / then you can just do without the /boot partition, and you can just keep the /
<Gaelan>
freeman42x[m]1 google suggests that you might need to enable EFI in vmware by adding firmware = "efi" to a .vmx file. Might be worth trying before you start all over
felixfoertsch has quit [Ping timeout: 256 seconds]
felixfoertsch23 is now known as felixfoertsch
<clever>
freeman42x[m]1: you could also just turn the 1mb at the end into a bios boot partition
<clever>
then it will work for both at once
<freeman42x[m]1>
<clever "freeman42x: you could also just "> 8 GB at the end, see picture above
<clever>
freeman42x[m]1: just after that 8gig swap, is 1mb of unallocated
<clever>
freeman42x[m]1: just allocate it, to bios boot partition
<freeman42x[m]1>
clever: hold on, there are too many ways you are telling me to do things
<freeman42x[m]1>
for now I will try: enable UEFI in VMWare player
<freeman42x[m]1>
and 2nd: set boot.loader.grub.device = "/dev/sda";
<freeman42x[m]1>
not sure if both needed
<clever>
the grub.device only works once you make a bios boot partition
<clever>
and you have 1mb of unallocated space, that is perfect for that
proofofkeags has quit [Remote host closed the connection]
<typetetris>
etu: Just wrap a program from another package and provide some arguments. Can I use makeWrapper for this, or does it need the program to be in $out/bin ?
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<shyim>
Have anyone else problems with node on unstable? node: /nix/store/784rh7jrfhagbkydjfrv68h9x3g4gqmk-gcc-8.3.0-lib/lib/libstdc++.so.6: version `GLIBCXX_3.4.26' not found (required by node)
cr4y1 has quit [Remote host closed the connection]
cr4y1 has joined #nixos
maxter has quit [Ping timeout: 256 seconds]
<{^_^}>
[nixpkgs] @peterhoeg merged pull request #88127 → handbrake: allow building from checkout → https://git.io/JfuRn
<{^_^}>
[nixpkgs] @FRidh pushed commit from @eonpatapon to master « pre-commit: use buildPythonPackage »: https://git.io/Jfz5q
maxter has joined #nixos
justanotheruser has quit [Ping timeout: 260 seconds]
noudle has joined #nixos
<EatThem>
how can i let config.nix detect the system itself and load the appropriate config file
<EatThem>
with like uname or sth
teto has joined #nixos
luz666 has joined #nixos
luz666 has left #nixos [#nixos]
alp has joined #nixos
luz666 has joined #nixos
<pistache_>
what kind of things should be tested in NixOS tests ?
<pistache_>
I'm adding "hitless reload" support in the haproxy service, but I'm not sure if I should only be testing the reloading operation itself, or if I should make sure that there are no lost HTTP requests during the reload
agam has joined #nixos
dm9 has quit [Quit: WeeChat 2.8]
<MichaelRaskin>
I would say, as much as you can tolerate debugging given the overhead of VM tests…
<azazel>
anyone using containers with packages/config at an older nixos version? i.e having the main host at 20.03 and some containers using 19.xx config/packages? Is it feasible?
<adisbladis>
azazel: It's feasible but the option to add a custom nixpkgs for a container was only added recently on unstable
<goibhniu>
azazel75: you can also do that with imperative containers
<{^_^}>
[nixpkgs] @adisbladis opened pull request #88402 → nixos-containers: Add support for custom nixpkgs argument (20.03 backport) → https://git.io/JfzFO
matthewcroughan has joined #nixos
<adisbladis>
azazel: ^
<typetetris>
Should I use `systemd.service.<name>.serviceConfig.ExecStartPre` or `systemd.service.<name>.preStart` ?
<clever>
typetetris: .preStart will generate a script with #! for you, and accepts multiple lines
<hadrian[m]>
<adisbladis "azazel75: Sure, we could backpor"> @azazel75:matrix.org: Isn't that a bit risky given the potential for abuse of the hotfix feature?
<clever>
typetetris: ExecStartPre accepts only one line, which must start with an absolute path to a program
<azazel>
adisbladis: Thanks a lot!
<typetetris>
clever: good to know, thanks
<clever>
typetetris: which you use, depends on how many lines youll need to run
07EAAZCID has quit [Remote host closed the connection]
<azazel>
hadrian: sorry, I don't follow you... could you explain better?
alp has quit [Ping timeout: 252 seconds]
orivej has quit [Ping timeout: 258 seconds]
gxt has quit [Ping timeout: 240 seconds]
orivej has joined #nixos
<typetetris>
clever: Is there an alternative to `systemd.service.<name>.serviceConfig.User` ?
<clever>
typetetris: DynamicUser ?
agsdheidjd has quit [Ping timeout: 240 seconds]
coco has quit [Quit: WeeChat 2.7.1]
<hadrian[m]>
<azazel "hadrian: sorry, I don't follow y"> I'm sure Mike / EatThem can elaborate better than I, but it's my understanding that if this change is implemented that a pull request from a non-forked repo can trigger a hotfix on any machine onto which malicious code has been placed into the ./dist directory? e ode
<typetetris>
clever: I was somehow expecting, that there would also be a key outside of serviceConfig for this, too. To avoid the lowercase/uppercase mixing .. but of course, its completely irrelevant :)
teto has joined #nixos
jkarni_ has quit [Ping timeout: 246 seconds]
jkarni_ has joined #nixos
<hadrian[m]>
And runtime detection of an "infected" system or group thereof by identifier(s) could lead to sponsored corporate espionage, state-led initatives, Etc.
orivej has quit [Read error: Connection reset by peer]
orivej has joined #nixos
<Woutifier>
how is /nix/var/nix/profiles/per-user/[my-user]/profile supposed to be created?
<Woutifier>
(it does not exist on my fresh NixOS system)
<Woutifier>
there is a .nix-profile in my home dir pointing to the non-existant path
reallymemorable has joined #nixos
simba1 has quit [Ping timeout: 260 seconds]
drakonis has joined #nixos
<sshow>
I'm having issues with `fetchgit` when running nix-build, as I'm targeting a private git server, I'm getting "Please make sure [....] access rights. Host key verification failed." What ssh config does it use?
<bqv>
Woutifier: automatically on first use/need
<Woutifier>
yes running nix-env -i hello did the trick. The home-manager service was failing to start because it could not find the home-manager.lock file in my profile-dir
<Woutifier>
thanks
rauno has quit [Remote host closed the connection]
orivej has quit [Ping timeout: 258 seconds]
wnklmnn has quit [Read error: Connection reset by peer]
orivej has joined #nixos
wnklmnn has joined #nixos
vidbina has quit [Ping timeout: 260 seconds]
Darkmatter66_ has joined #nixos
linarcx has quit [Ping timeout: 246 seconds]
Darkmatter66 has quit [Ping timeout: 260 seconds]
is_null has joined #nixos
vidbina has joined #nixos
kaliumxyz has quit [Remote host closed the connection]
<kaliumxyz>
I'll build against the command line tools for now like everyone else, maybe I can build against the nix repl?
matthiaskrgr has joined #nixos
<alj[m]>
I rebooted my device after yesterdays rebuild (went over night, my changes worked out, very proud of that) and now i want to make a change and this stupid thing wants to rebuild gcc, perl and similar stuff AGAIN. looking at the store paths there is no gcc in there, even tho i just built that shit yesterday night. What the heck is going on? Does anyone know?
<simpson>
kaliumxyz: How do you want to handle transactions? It's hard to beat the nix-instantiate/nix-build/nix-shell API, just emitting a text .nix file to a temporary directory or providing it inline.
<alj[m]>
I enabled automatic garbage collection, but gcc and perl are needed no? they should not get cleaned up right? I just dont know anymore
<simpson>
alj[m]: Build tools are only needed during the build, and GC is smart enough to remove them if there's no builds happening. You can either do GC less often, or you could explicitly depend on certain build tools, effectively "pinning" them. "pin" in the GC sense, not the package-manager versioning sense!
<alj[m]>
<simpson "alj: Build tools are only needed"> so you're saying I've been deleting gcc and other buildtools every night, without knowing? but this stuff is *needed* to run nixos-rebuild! What does nixos expect me to do??? My other Nixos system doesnt shred itself every night when GC is ran! wtf. Also how do i "pin" that?
<simpson>
alj[m]: I'm looking to see if there's an elegant technique; the inelegant technique is to add whatever desired tool to environment.systemPackages.
<alj[m]>
I'm sorry if i sound rude, but this is quite irritating. I've already lost a whole day to this exact crap
<simpson>
Generally, one doesn't run nixos-rebuild nightly.
<simpson>
It's understandable. You might want to turn off automatic GC if bandwidth is a problem.
<evanjs>
simpson: yeah, every night?? who does that? Mine is set up to auto-run every 6h :P
<alj[m]>
my problem is not bandwidth, its an armvl7 system. its literally rebuilding the whole build toolchain this way.
<simpson>
alj[m]: For user environments, and not the system NixOS config, there's direnv, leading up to lorri. But that won't solve system issues.
brandonhamilton has joined #nixos
<simpson>
Oh! Hm, I would pick direnv. But that's probably not a foolproof-enough solution for the fact that ARM rebuilds can take days.
<simpson>
BTW there is a #nixos-aarch64 channel dedicated to helping with the fact that ARM is a tough build target.
<kaliumxyz>
simpson Good to know, I'll go down that route then.
etu1 has joined #nixos
<simpson>
kaliumxyz: Well, how would you like to handle transactions?
ctp__ has joined #nixos
<kaliumxyz>
Changes in state?
<alj[m]>
simpson: The issue is with nixos config rebuilds, so direnv is not possible for me. I'll just turn off GC for now and head over to the other channel. thanks for the quick answers!
<simpson>
alj[m]: Also if you just need a GCC that doesn't change, it's possible to make new GC roots, and it can be as easy as running nix-build.
etu has quit [Ping timeout: 246 seconds]
ctp has quit [Ping timeout: 246 seconds]
laval[m] has quit [Ping timeout: 246 seconds]
ctp__ is now known as ctp
Keith[m] has quit [Ping timeout: 246 seconds]
<kaliumxyz>
I'm currently trying to port a terraform plugin for nix to pulumi but would also like to extend the capabalities of the plugin and eventually add some support for langauge bindings to nix so that you can get code completion, type checking, etc.
<typetetris>
I don't get it, somehow I cant use stuff from `EnvironmentFile` in `ExecStartPre` strange.
Keith[m] has joined #nixos
<kaliumxyz>
pulumi is basically nixops but with typescript, C#, javascript, etc, as languages you write the scripts in
morgrimm has quit [Ping timeout: 258 seconds]
<alj[m]>
simpson: I'll get back to you on that, if i cant figure out another way
<simpson>
kaliumxyz: Ah. I think that code completion and type-checking can be done with code editor plugins, more or less; there's a language server project at https://github.com/domenkozar/hnix-lsp
<kaliumxyz>
Yes but this is for other languages.
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<simpson>
kaliumxyz: I was wondering how you wanted to handle evaluation which causes changes to the Nix store.
<kaliumxyz>
so would need to transpile nix to type definitions for those.
<kaliumxyz>
ah, I want nix to handle transactions.
<kaliumxyz>
It should match the state pulumi holds and provide feedback on if it was able to do so or not.
<kaliumxyz>
if a transaction failed pulumi should know.
klntsky has quit [Remote host closed the connection]
<simpson>
Hm. I can't see this being a good match. Nix is more like a protocol for a directory, like git, rather than a language. The Nix expression language is just the language for instructing the reference toolchain.
klntsky has joined #nixos
fusion809 has quit [Remote host closed the connection]
<simpson>
But OTOH I don't want to discourage experimentation.
<kaliumxyz>
This is a solved problem though, just embed nix into your code.
<{^_^}>
[nixpkgs] @adisbladis merged pull request #88402 → nixos-containers: Add support for custom nixpkgs argument (20.03 backport) → https://git.io/JfzFO
<{^_^}>
[nixpkgs] @adisbladis pushed 2 commits to release-20.03: https://git.io/JfzhB
<kaliumxyz>
There already is a terraform plugin that does that.
<kaliumxyz>
I want to extend it to have basic functionality so that the user does not need to write in nix but rather can write in their language of choice.
chrispickard has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
<kaliumxyz>
Pulumi is just a set of bindings to terraform in popular programming languages
<kaliumxyz>
Terraform and Pulumi are both basically nixops but they are more widely adapted and handle state better.
jkarni_ has quit [Remote host closed the connection]
jkarni_ has joined #nixos
jb55 has quit [Ping timeout: 240 seconds]
rauno has joined #nixos
o1lo01ol1o has quit [Ping timeout: 272 seconds]
<kaliumxyz>
The idea would be to have typedefinitions for nixpkgs in the language you want to work in, then describe your desired nix using the langauge you want to work in. Pulumi will provision and manage the server and the pulumi plugin for nix will compile the nix files, ssh into the server and run the nix bfiles.
<kaliumxyz>
Just use whatever goes as a struct in the langauge you want to work in.
<evanjs>
lassulus: nixos-generate -> VBox seems to work fine, cool. So my config is fine. Wonder why HyperV (azure) is acting all weird and refusing to boot 🤔
<evanjs>
Assuming the two are compatible, at least in the context of nixos-generate outputs
o1lo01ol1o has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @zowoq opened pull request #88431 → .editorconfig: various → https://git.io/Jfzj4
<kaliumxyz>
evanjs: if you are part of the project: is rnix along enough to be used to generate type definitions from nixpkgs for typescript or some other target language?
linarcx has joined #nixos
<kaliumxyz>
like can it go trough everything in nixpkgs in reasonable time and help me somehow loop trough all it has read to spit out the type definitions for nixpkgs in another lang?
<{^_^}>
[nixpkgs] @Kloenk opened pull request #88432 → nixos/minecraft-server: add fifo controll and stop warning → https://git.io/Jfzjz
<{^_^}>
nix-community/poetry2nix#107 (by dhess, 9 seconds ago, open): poetry2nix breaks on these two applications
etu1 is now known as etu
<dhess>
adisbladis: ok cool, adding the homepage attribute to my local pyproject.toml fixed that issue, anyway.
justanotheruser has joined #nixos
<adisbladis>
dhess: Alright :) Good
<adisbladis>
I don't really like the prospect of adding a dummy homepage
<adisbladis>
But I guess we'll do it anyway
<drozdziak1[m]>
kdenlive appears to be qt-borked on my 20.03 system (qt can't load "xcd" upon project render). If I understand the guides correctly I need to add `wrapQtAppsHook` to `nativeBuildInputs` on the derivation. How do I bring it into scope? I'm trying to experiment with this on the repl
<NickHu>
Is there a document anywhere which tracks the progress of the nix 2 cli?
<NickHu>
I've heard vague things like nix-build still covers some usecases that nix doesn't, but it would be nice if all this sort of stuff were collated on a wiki page or something
chloekek has joined #nixos
AmandaC has joined #nixos
<EatThem>
<adisbladis> do i have to put the posted script inside the imports = [ ] brackets or where? and where should i enter my directorypath with the nix files?
<EatThem>
<adisbladis> sorry for beeing dumb :D
AmandaC__ has joined #nixos
<shyim>
How can i find which dependency is using p7zip?
<{^_^}>
[nixpkgs] @kalbasit pushed commit from @marsam to master « prototool: 1.9.0 -> 1.10.0 (#88369) »: https://git.io/Jfgf1
<emily>
it probably has nothing to do with nixos and everythign to do with driver/hw issues though
kim0 has joined #nixos
<tarelerulz>
would it help looking at say my Ubuntu to see what drivers it uses ?
agam has quit [Ping timeout: 256 seconds]
mallox has quit [Quit: WeeChat 2.8]
mallox has joined #nixos
chloekek has joined #nixos
red[evilred] has joined #nixos
<red[evilred]>
Is there any reason that /etc/nixos can't have permissions 700 ?
vyorkin has joined #nixos
<vyorkin>
hi, I'm trying upgrade my NixOS config to use flakes, but got the following error: attribute 'currentSystem' missing, at /nix/store/l3i4dxnghgxh3dcs33vbwd4ljgpyfdk3-source/pkgs/top-level/impure.nix:87:34
<vyorkin>
looks I'm missing smth, but I have no idea what it could be
felixfoertsch has quit [Quit: ZNC 1.7.4 - https://znc.in]
ninjin has quit [Remote host closed the connection]
felixfoertsch has joined #nixos
ninjin has joined #nixos
felixfoertsch has quit [Client Quit]
<red[evilred]>
thanks - appreciate that!
wfranzini has joined #nixos
never_released_ has joined #nixos
never_released has quit [Ping timeout: 265 seconds]
proofofkeags has joined #nixos
<unclechu>
hi. in nixos i can't scroll using touchscreen, instead it selects text when i try to do so. i couldn't find any related option in `xinput list-props`
rodrigo_ has joined #nixos
<unclechu>
does anyone faced the same issue? and maybe you found a way to change this behavior?
<Cadey>
never mind, i apparently don't have a working microsd card lol
cr4y1 has quit [Quit: Leaving]
noudle has joined #nixos
justanotheruser has joined #nixos
noudle has quit [Client Quit]
brandonhamilton has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
srid2 has joined #nixos
wfranzini has quit [Quit: wfranzini]
wfranzini has joined #nixos
<{^_^}>
[hydra] @knl opened pull request #764 → Render the jobset page correctly when there are fetch errors → https://git.io/JfgTP
<red[evilred]>
I'm trying to get NixOS to manage my vim configuration, and I see examples that talk about using .nixpkgs/config.nix to configure your local user's vim stuff.
<red[evilred]>
How does one "activate" / "execute" that file?
<red[evilred]>
since my /etc/nixos/configuration.nix makes vim globally available on nixos-rebuild
smatting has quit [Ping timeout: 260 seconds]
<chloekek>
Is it possible to see the content on discourse.nixos.org without connecting to discourse cdn?
<chloekek>
Ah! Seems disabling JS makes the page visible, because it puts the thread in <noscript>. Neat.
brandonhamilton has joined #nixos
<boolman>
im adding ip-up.d scripts with environments.etc.<filenam>.text = '' ''; how do I make this executable?
<red[evilred]>
drakonis (IRC): the only issue I have is that his package definition creates vim as an executable instead of just vi. I need an alias or a symlink :-)
<infinisil>
emily: Why do I need that just to clone a repo?
<emily>
oh, huh, apparently git actually lets you clone without the .git
<emily>
infinisil: you can make git redirect https urls to ssh for pushing and stuff, that's my only hypothesis
<infinisil>
Yeah saw that too
<infinisil>
But I don't even want ssh here
<emily>
strace time?
<{^_^}>
[nixos-homepage] @garbas pushed 12 commits to landing-page: https://git.io/JfgLt
<emily>
infinisil: ok, now show me /root/.config/git/config
<emily>
(and maybe be scared that it exists :p)
<infinisil>
Ah!
<infinisil>
Well darn you're genious emily, I have the ssh redirect in there!
<infinisil>
emily++
<{^_^}>
emily's karma got increased to 28
<emily>
^_^
<emily>
maybe to access private repos from deploy scripts or something?
<infinisil>
I just assigned that in my config, and it gets applied to all users I use
<emily>
you can set it to be for push only
<emily>
ah
<infinisil>
emily: How to do it for push only?
<infinisil>
Currently it's [url "git@github.com:"] \n insteadOf=https://github.com/
<infinisil>
Ah pushInsteadOf
<infinisil>
Thanks `man git-config`
<emily>
pushInsteadOf
<infinisil>
I'm now wondering why I have that in there anyways, I can only remember that something didn't work with it
<jasom>
What's a good way to have nix-build ignore ~/.config/nixpkgs?
<infinisil>
Ah, so I wouldn't have to change remotes that hub configures
<infinisil>
I think
<infinisil>
emily: It scares me a bit how easily this can break Nix's purity
<infinisil>
Maybe Nix should use its own git config for fetching
<emily>
infinisil: fetchGit/fetchTarball/... are kind of gross hacks, yes. I'd suggest avoiding them whenever possible
<infinisil>
I mean with sha256 = ... they're okay imo
<{^_^}>
[nixpkgs] @oxalica opened pull request #88441 → systembus-notify: init at 1.0 → https://git.io/JfgLu
<emily>
infinisil: have you considered flakes?
<emily>
I think flakes supports fetching private git repos in a more principled way
<infinisil>
This fetchGit is deep inside some dependency that I use
<emily>
right
<emily>
that's why they're not okay
noudle has quit []
<emily>
fixing the hashes helps ensure reproducibility, but it still means you have impurities and external dependencies embedded implicitly N layers deep :(
<emily>
if we have distinct instantiate -> build stages, we should have distinct fetch -> instantiate -> build stages rather than mingling the first two together, flakes moves a bit closer to that model
<infinisil>
emily: Since fetchGit now supports recursive modules, I have been dreaming of this being the way to manage dependencies. Everybody uses git submodules and the end-user just uses `fetchGit { fetchSubmodules = true }` to get all dependencies in one single evaluation step
ddellacosta has joined #nixos
<{^_^}>
[nixos-homepage] @garbas pushed to landing-page « Resurrect nix and nixos about pages as one features page »: https://git.io/JfgLo
<boolman>
from what package does 'ip' comes from?
rodrigo_ has left #nixos ["Leaving"]
<adisbladis>
iproute2
<emily>
,locate bin/ip
<{^_^}>
Found in packages: busybox, iproute, iproute_mptcp
orivej has quit [Quit: No Ping reply in 180 seconds.]
<infinisil>
Or actually, the end user *also* uses a git submodule in their system repo
<infinisil>
And then uses `fetchGit { url = ./.; sha256 = "..." }`
<emily>
sigh, pname = "iproute2" but the attribute is iproute
<infinisil>
Oh and with submodules of course again
<emily>
I really wish there was consistency for that
<infinisil>
,locate bin ip
<{^_^}>
Found in packages: busybox, iproute, iproute_mptcp
NeoCron has quit [Ping timeout: 260 seconds]
wnklmnn has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @jasom opened pull request #88442 → gnucash: Add libdbiDrivers env var to wrapper → https://git.io/JfgtL
<adisbladis>
Ugh... The upstream name is iproute2 so that should really be the attr name
<{^_^}>
[nixos-homepage] @garbas pushed to landing-page « remove hideTitle and menu option from layout.tt »: https://git.io/Jfgtq
wnklmnn has joined #nixos
<{^_^}>
[nixpkgs] @oxalica opened pull request #88443 → earlyoom: patch absolute dbus path and make nixos module up to date → https://git.io/JfgtY
<boolman>
thanks, I thought I tried both iproute and iproute2 but i guess i didnt
<cole-h>
infinisil: Looking through your config for transmission inspiration. How does `config.private.passwords` work? Do you have those passwords as plaintext and git-crypt'd or something?
<{^_^}>
[nixpkgs] @veprbl pushed commit from @gnidorah to master « tes3mp: fix working (#88061) »: https://git.io/Jfgmn
meh` has quit [Remote host closed the connection]
zaeph has quit [Ping timeout: 256 seconds]
sigmundv has quit [Remote host closed the connection]
<T0pH4t>
infinisil: I've been looking into my issue more and I think i have tracked down the problem. Not sure if this is a nix bug, but its seems the sha256 is generated post configure of a package. Meaning I have a cmake file that places some header files during cmake configure and that seems to be screwing over the sha256
<pie_>
quinn: maybe with import?
<pie_>
quinn: you still wont have stuff in scope though
<T0pH4t>
i would have expected nix to generate sha256 right after git pull
<betawaffle>
i figured it out. i just wasn't sure if it would be persistent, since i'm in erase-your-darlings mode
noudle has joined #nixos
MmeQuignon has quit [Quit: WeeChat 2.8]
alp has quit [Ping timeout: 246 seconds]
<{^_^}>
[nixos-homepage] @github-actions[bot] pushed commit from @edolstra to master « Update flake.lock and blogs.xml [ci skip] »: https://git.io/JfgY4
skomorokh has joined #nixos
orivej has joined #nixos
meh` has joined #nixos
proofofkeags has quit [Remote host closed the connection]
proofofkeags has joined #nixos
<sphalerite>
Is there a way to set createHome to false for all users without assuming which users exist?
jb55 has quit [Remote host closed the connection]
<infinisil>
sphalerite: Something like `options.users.users = mkOption { type = loaOf (submodule { createHome = true; }); }
user_0x58 has quit [Quit: Leaving]
proofofkeags has quit [Read error: Connection reset by peer]
proofofkeags has joined #nixos
<infinisil>
This includes system users too though, but you can further customize this with `submodule ({ config, ... }: { createHome = mkIf config.isNormalUser true; })` or so
jkarni_ has joined #nixos
<sphalerite>
whoa what
mtn has joined #nixos
brandonhamilton has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<emily>
infinisil: you can do that???
<mtn>
I have a really basic question: I included neovim in my `environment.systemPackages`, but I get an outdated version even after running `nixos-rebuild switch --upgrade`
<infinisil>
emily: Yup yup!
timmw has quit [Quit: Ping timeout (120 seconds)]
<mtn>
How can I the updated version? It looks like it's in nixpkgs
jb55 has joined #nixos
timmw has joined #nixos
<emily>
mtn: either you're on the wrong nixos channel or the channel hasn't updated yet
<infinisil>
This works because multiple option definitions get merged, and for options with submodules, this means applying all defined submodules
<sphalerite>
mtn: are you on 20.03?
<emily>
if you're on a stable release then the canonical source is the release- branches, not master
<mtn>
Nope, probably not, haven't touched the system in awhile
<mtn>
I'll check
<mtn>
And I'll also see what on the release branch, in case that's it
<mtn>
Okay, the updated versions on `release-` for 20.03, so I must be on an old version
jkarni_ has quit [Ping timeout: 272 seconds]
davidv7 has joined #nixos
<emily>
well, it also depends where the channel is (nixos-20.03 branch)
<emily>
but if it was updated on the 20.03 branch a while ago then indeed your nixpkgs is probably somehow out of date
<drakonis>
hmm, how troublesome, i need to get javafx for java 8 but its only available with oracle jdk
jb55 has quit [Ping timeout: 240 seconds]
RRedcroft has joined #nixos
hoijui has quit [Quit: Leaving]
<mtn>
Am I on the stable channel for my nixos version if `nix-channel --list` doesn't return anything about nixpkgs?
<bbigras>
use sudo
<bbigras>
you can have channels on the user and system levels.
o1lo01ol1o has joined #nixos
<sphalerite>
infinisil: well, it works, thanks for that. I don't like it though x)
magnetophon has quit [Read error: Connection reset by peer]
<mtn>
Thanks bbigras!
alp has joined #nixos
<bbigras>
You're welcome 🙂
magnetophon has joined #nixos
<mtn>
Okay so very broad question and then ima go -- I configured my system awhile ago, broke stuff now, and wanna understand better what's going on. Is there a doc or reading path that's recommended?
<mtn>
I read the nix pills and a subset of the manual previously
brandonhamilton has joined #nixos
dongcarl has quit [Quit: Ping timeout (120 seconds)]
<mtn>
I've mainly just written a few shell.nix and used them to manage environments
dongcarl has joined #nixos
<T0pH4t>
so do nix-build and nix-store verify-path use different things to validate checksums or something. nix-build works fine then nix-store post build fails to verify path
<T0pH4t>
there is literally no diff on the directories
<T0pH4t>
i can only suspect a bug in nix somewhere
<sphalerite>
T0pH4t: what exactly are you doing with which expressions?
hmpffff has quit [Quit: nchrrrr…]
<T0pH4t>
sphalerite: i'm trying to publish a package that is using c++, its just a simple package that fetches code from git and then builds it with cmake. The probablem is the sha256 is failing to validate after cmake is ran
<T0pH4t>
if i disable the configure/build of hte package the sha256 checks out
jb55 has joined #nixos
<sphalerite>
T0pH4t: you probably don't want a sha256 in the package itself, only the src attribute
<T0pH4t>
but if I enable the configure of cmake, then sha356 fails to validate
<T0pH4t>
its in the src
<{^_^}>
[nixpkgs] @romildo opened pull request #88446 → fbmenugen: init at 2020-05-20 → https://git.io/JfgOR
<skomorokh>
Hi! I'm intimidated but intrigued, I hope this is as good a place as any to ask very naive questions that arise while skimming the docs as I work up the courage to start trying to replace Debian with this on a server...
<T0pH4t>
its a bit wonky right now since i have been debuging htis issue since yesterday
<T0pH4t>
normaly string and build would be anbled
magnetophon has quit [Read error: Connection reset by peer]
<T0pH4t>
enabled*
<drakonis>
its the place, yes.
<skomorokh>
First one would be does rebuild only restart the minimum set of impacted services or does everything restart when you rebuild?
mtn has quit [Remote host closed the connection]
<sphalerite>
T0pH4t: what happens if you remove sourceRoot?
<drakonis>
restarts all changed services i think?
<T0pH4t>
well package won't build, but i also suspect maybe something is going on there
<gchristensen>
skomorokh: yep, as drakonis said, it only stops/restarts/starts services which are not needed/changed/new
<T0pH4t>
like the sha checks out if i add dontConfigure = true;
<sphalerite>
T0pH4t: sharing the expression uncensored would probably be another bit more helpful
magnetophon has joined #nixos
<T0pH4t>
but fwiw, i have other packages that use a sourceRoot that is a sub directory
brandonhamilton has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<T0pH4t>
its a private package for my company :/
<sphalerite>
but you generally don't want to refer to src in sourceRoot
<sphalerite>
try just sourceRoot = "foo" maybe
<T0pH4t>
well then it doesn't find folder
<morgrimm>
Has anyone played with Hydra and GH actions? Or triggering Hydra builds on GH webhook events?
<T0pH4t>
i can try again tho, maybe an ordering thing
<sphalerite>
what…
<T0pH4t>
nix-store doesn't seem to like any other args when repairing
<T0pH4t>
not sure if thats a bug
sn3rd has quit [Client Quit]
<T0pH4t>
yeah no go, get error: no flags expected
<skomorokh>
Can you opt something out of Nix-ness? I have a commercial PHP script my brother uses for invoicing, it has an auto updater thing that I think I'd rather just use than bothering to build a package for it that I can't even distribute anyone and have to fiddle with it on every version... is there a cookbook type example of that sort of practical compromise? (though the cookbook link seems to be down)
o1lo01ol1o has joined #nixos
<sphalerite>
T0pH4t: I'm getting more and more confused. You used the command exactly as I wrote, as in with --option sandbox relaxed before --repair-path?
<T0pH4t>
yup
<T0pH4t>
ran ur command exactly but supplied my path
<T0pH4t>
and nix-store errors out
ebopp has joined #nixos
<T0pH4t>
it doesn't seem tot support any other flags when --repair-path is used
<T0pH4t>
tho fyi, ur preconfigure suggestion worked. Not sure why that package is so special tho
rauno has quit [Remote host closed the connection]
<sphalerite>
lol well
o1lo01ol1o has quit [Ping timeout: 246 seconds]
<T0pH4t>
but still curious as to why sourceRoot isn't working :/
<sphalerite>
anyway, verifying a path should only fail when it's been modified under nix's nose
<sphalerite>
if you do it again now, it should hopefully work
<sphalerite>
(though "modified under nix's nose" could also be attributable to filesystem corruption or nasty things like that)
<sphalerite>
(I've had lots of "fun" with broken SD cards)
<T0pH4t>
well those were my thoughts
<T0pH4t>
but its only this package
<sphalerite>
I need to go now, ttyl
<T0pH4t>
cya thx
<{^_^}>
[nixpkgs] @matthewbauer merged pull request #87528 → llvm{6..10}/compiler-rt: fix compilation on Darwin → https://git.io/JfWMn
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to staging: https://git.io/Jfg3m
knupfer has quit [Quit: knupfer]
knupfer has joined #nixos
knupfer has quit [Client Quit]
knupfer has joined #nixos
<srhb>
skomorokh: It Depends.
<skomorokh>
srhb: :)
<srhb>
skomorokh: If you're willing to just let it live in some sandboxed directory and have Nix never touch it, it might be doable.
<srhb>
skomorokh: If it has a lot of FHS assumptions, you're in for a ride.
<srhb>
skomorokh: As in: It'll probably be harder than actually doing the proper Nix packaging.
brandonhamilton has joined #nixos
<srhb>
skomorokh: But you can always try and see how far you can get with just giving it a shell with the "right dependencies"
<skomorokh>
srhb: That's kind of what I was thinking. Was hoping it won't assume much at all being PHP so is expecting to just be dumped somewhere and the updater hopes to run as a user with sufficient permissions to overwrite itself, that kinda thing.
o1lo01ol1o has joined #nixos
<srhb>
skomorokh: Yes, that seems more like not opting in rather than opting out, if you get the distinction. Sounds doable.
<srhb>
skomorokh: I do want to add that I have loads of nix expressions that I never intend to distribute, and they give me huge value, personally.
<srhb>
It's all peace of mind and sanity.
o1lo01ol1o has quit [Remote host closed the connection]
<srhb>
If it's not too bothersome, you may find that you feel the same way.
<skomorokh>
Was hoping it was usual enough usecase (autoupdating php webapps) that someone would have recollection of a guide to this sorta thing eg. maybe an expression that kinda sets up everything around it modulo the software itself and any specific version.
<srhb>
None that I know of, but I don't use PHP myself.
chloekek has quit [Read error: Connection reset by peer]
<vyorkin>
nix flakes question: if I have an input of type "git" then what other fields should be in the input?
<vyorkin>
"url"?
<skomorokh>
srhb: I imagine I will feel that way about most things, that's why I'm interested in Nix after all--why bother otherwise right :) I just didn't want to deal with this one since I mostly just leave it alone and occasionally back it up.
<vyorkin>
I wonder if I can use the ssh url... it seems like I can't, because it says "error: 'git@github.com:vyorkin/dotsecrets.git' is not a valid URL"
<skomorokh>
I'll see if I can sorta treat it like data...
<srhb>
vyorkin: Is that with git+ssh:// ?
<vyorkin>
srhb: yes
hmpffff has joined #nixos
<srhb>
vyorkin: git+ssh://github.com/vyrokis/dotsecrets.git should work
<srhb>
If I could spell your name :)
<vyorkin>
srhb: nevermind, the name is not that important :)
<vyorkin>
srhb: thx!
<srhb>
vyorkin: For other fields, I think it has changed a bit, but there's `inputs.foo.inputs.[...]` especially for follows, to propagate dependencies downwards.
<srhb>
(But this is not mandatory of course)
<vyorkin>
srhb: yep, I think "url" will be enough in my case
<srhb>
RRedcroft: "installed" in the sense of being "runnable" via the houdini commands means that some derivation should be in the system profile and have a /bin/houdini path
<srhb>
RRedcroft: The package you installed, "houdini" does not have such a path.
rogue_koder has joined #nixos
<{^_^}>
[hydra] @cransom opened pull request #765 → Add hydra-notify to devshell → https://git.io/Jfgsa
<RRedcroft>
houdini itself is normally installed to '/opt/hfs/' and inside there is normally a bin directory, but no nothing gets installed to /bin, /usr/bin, or /usr/local/bin
<srhb>
I meant relative to the store path
<srhb>
so /nix/store/houdini-.../bin/houdini
<RRedcroft>
oh, there is and should be
brandonhamilton has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<srhb>
I don't think there is.
<srhb>
But check :)
orivej has quit [Ping timeout: 246 seconds]
<RRedcroft>
trying to run via the path 'runs' (version 18 requires libsmime3 and some additional libs which i need to add in the overlay)
<RRedcroft>
/nix/store/lbcwz98a3z90y3rhvnzlsyrhvrddfvgz-houdini-runtime-18.0.460/bin/hkey-bin: error while loading shared libraries: libsmime3.so: cannot open shared object file: No such file or directory
<srhb>
RRedcroft: Wrong package -- that's houdini-runtime, not houdini.
<srhb>
RRedcroft: You installed houdini, not houdini-runtime, right?
orivej has joined #nixos
<RRedcroft>
yeah
<srhb>
Only the paths in houdini matter for what gets in your system profile's bin folder.
<srhb>
(Because that's the one in your systemPackages)
<RRedcroft>
oh, ok
<RRedcroft>
then no it doesnt look like it, how do i get them into houdini?
<RRedcroft>
i do have a single binary /nix/store/i6vw5q8mglx1xy9d0hw8sh9qbzxg7jls-houdini-18.0.460/bin/houdini-18.0.460
<srhb>
RRedcroft: I'm not sure with FHS stuff. :) You may have to look at some of the other FHS launchers in nixpkgs to figure it out. I was about to suggest steam-run as one that I know does it, but that looks pretty complex.
<RRedcroft>
srhb: no worries, thank you for the help :)
<srhb>
RRedcroft: Oh actually, it totally looks like its runScript does the trick
<RRedcroft>
srhb: so adding to the runscript should work?
<srhb>
RRedcroft: I think it already works, but the name of the binary is the entire string you're passing to name in buildFHSUserEnv
simba1 has joined #nixos
<srhb>
If I'm reading this right.
johnnyfive has quit [Quit: I quit now.]
<srhb>
RRedcroft: So try making that just houdini
<srhb>
Er, not binary, script* in the bin directory.
johnnyfive has joined #nixos
<RRedcroft>
srhb: is it possible to expose multiple binaries?
<unclechu>
hey, can someone explain to me what magic makes dependency appear in the derivation when i just interpolate something like `${pkgs.app}/bin/app` in a string of `writeScript`
<srhb>
RRedcroft: For FHS, you'd probably need a buildFHSUserEnv for each of them, but you can automate that. It seems like the function creates one executable equal to the name.
<unclechu>
how does nix know that he need `pkgs.app` as a dependency for a derivation?
<srhb>
unclechu: At the lowest level, because there's a nix store path in there.
agam has joined #nixos
<RRedcroft>
srhb: ahh that makes sense, i'll give that a go, ty
<cole-h>
infinisil: Basically copying your tranmission setup. However, setting `home = "/var/lib.torrent"` and then starting the unit fails to create that directory (permission denied). Did you get this as well?
<unclechu>
srhb: how nix find derivation paths when evaluating a derivation like this? does it hooks all the nix interpolations in strings?
<unclechu>
or any access to the values and when a value has `type = "derivation"` it adds this value to the dependencies list when builds a derivation?
<unclechu>
srhb: yes, i know, that's what makes me wonder, how nix is able to extract dependency from just a string
shafox has quit [Remote host closed the connection]
<srhb>
unclechu: It is a string that uniquely points to a specific version of a specific package
<srhb>
unclechu: As with all /nix/store/foo-hash paths
<srhb>
unclechu: So basically, grep would suffice!
<unclechu>
i'm just using string but in `/nix/store/*.drv` file i see also the dependency, like `/nix/store/*-hello-1.2.3.drv`
<srhb>
Just pick up all strings that are store paths, and you have all your dependencies :)
dongcarl has quit [Quit: Ping timeout (120 seconds)]
<unclechu>
srhb: so, nix just does the grepping for /nix/store/*? 🤔
orivej has quit [Quit: No Ping reply in 180 seconds.]
<unclechu>
for `/nix/store*` paths in the string of my script?
<srhb>
Well, as you said, when you instantiate the derivation all the build time deps are also recorded
<RRedcroft>
probably doing something stupid, but how do i modify ld_library_path in an overlay trying this hasnt working https://pastebin.com/raw/EVvKnitP
orivej has joined #nixos
<unclechu>
srhb: how nix collect those "build time deps"? i'm just using strings
<srhb>
unclechu: For the specifics of that you'll probably want to read Eelcos thesis, specifically section 5.4 I think
<Bryan[m]>
unclechu: I remember seeing string context be a sticking point for hnix at one point - I guess nix really does just keep track of which store path is referred to and uses it for dependency tracking.
<gchristensen>
unclechu: Nix remembers that you interpolated a package in to the string, and attaches that information to the string as String Context
<Bryan[m]>
Reminds me that I also want to actually read the thesis sooner rather than later
<srhb>
unclechu: But the short story is that each store path in all the attributes of the input is (at least) a build time dependency
<srhb>
unclechu: So yes, basically "grep" once again.
<unclechu>
Bryan: so said "nix does the grepping for /nix/store/* pattern"? otherwise i could imagine that nix track all the operations with values and record resolved values which has `type = "derivation"`.
<unclechu>
by "resolved" i mean evaluated lazy value
o1lo01ol1o has joined #nixos
<gchristensen>
after a build, nix searches the build results for /nix/store/* patterns to look for run-time dependencies. for build-time dependencies, it remembers string interpolations
<cole-h>
Is there a way to force a uid change, even if another user has the same uid?
dongcarl has joined #nixos
<unclechu>
now i assume that when nix evaluate strings it tries to extract some so called "context" from each interpolation
<srhb>
cole-h: There is not protection against two users having the same uid, if that's what you mean.
<unclechu>
and when grep for `/nix/store/*` is successful it adds it to the context set
<srhb>
cole-h: But you'll have a fun time.
<cole-h>
srhb: Well, they're not interactive users, but ones for running a service
<srhb>
unclechu: It was pointed out to me that maybe my "lies to children" might be taken too literal re. the grepping.
<unclechu>
but plain strings stay plain strings and don't affect the context
<srhb>
unclechu: That's right, context-ful strings are special.
<unclechu>
if i put plain path to the string it wouldn't be added to the context
<srhb>
unclechu: Correct.
<cole-h>
srhb: The main reason I ask is because when setting user, group, and home of `services.transmission`, the prestart script fails to create /var/lib/torrent, but if I don't manually set the user and group, it gets created just fine...
<unclechu>
it seems it doesn't tell the whole truth
yutyo[m] has joined #nixos
<infinisil>
unclechu: Yeah Nix context can't be hacked together, since it's a language feature so to say
<infinisil>
The derivation builtin specifically makes the result have context
turlando_ has quit [Quit: Leaving]
turlando has joined #nixos
orivej has quit [Quit: No Ping reply in 180 seconds.]
orivej has joined #nixos
<{^_^}>
[nixpkgs] @kalekseev opened pull request #88456 → pre-commit: hardcode path to virtualenv's binary → https://git.io/Jfgnd
<unclechu>
so, the assumption that derivation is just an attrset with proper `type` attribute is wrong. that page is a bit misleading, it tells nothing about underlying implicit context a real derivation has
<unclechu>
> builtins.getContext "${pkgs.hello // { type = "lol"; }}"
<unclechu>
as i see even laziness doesn't play any important role here. you touch a value that has context at any level you inherit that context.
orivej has quit [Ping timeout: 260 seconds]
orivej has joined #nixos
shibboleth has joined #nixos
<unclechu>
this is actually matching my first guess. that nix tracks all the access of values under the hood. well, a bit rephrased but idea was correct.
<srhb>
unclechu: That about sums it up, yeah.
<chin-tastic>
Hello friends. Is anyone aware of any examples of building derivations out of webcomponents/react components? (I'm not a javascript expert, so this has been particularly difficult.)
xd1le has quit [Read error: Connection reset by peer]
meh` has quit [Ping timeout: 265 seconds]
xd1le has joined #nixos
dansho has quit [Ping timeout: 256 seconds]
fusion809 has joined #nixos
brandonhamilton has joined #nixos
morgrimm has quit [Ping timeout: 240 seconds]
knupfer has quit [Remote host closed the connection]
knupfer has joined #nixos
dansho has joined #nixos
fendor has quit [Remote host closed the connection]
magnetophon has quit [Read error: Connection reset by peer]
<hexagoxel>
are there any best practices around overlays and directory-specific behaviour? I have a package that can be built in two ways: a) can be built on a range of recent official nixpkgs revisions b) requires a custom overlay. Locally I want to use b) by default, but I am hesitant to hardcode that into default.nix/shell.nix
rawr has quit [Quit: I can't explain, you would not understand, this is not how I am]
pjt_014 has joined #nixos
orivej has quit [Ping timeout: 272 seconds]
grumble has joined #nixos
orivej has joined #nixos
proofofkeags has quit [Remote host closed the connection]
proofofkeags has joined #nixos
<RRedcroft>
b42: im getting "anonymous function at /home/redsith/tmpdev/nixpkgs/pkgs/applications/misc/houdini/runtime.nix:1:1 called without required argument 'wrapQtAppsHook'"
quinn has joined #nixos
<cole-h>
infinisil: Figure out my Transmission problem :D Needed to set the `downloads` user's home (and let it be created) lol
morgrimm has quit [Ping timeout: 264 seconds]
ebopp has quit [Remote host closed the connection]
<cole-h>
s/Figure/Figured
<infinisil>
Ah I missed those messages
<infinisil>
Glad you figured it out :)
<infinisil>
(well I can't remember having to do this myself though)
<cole-h>
Hehe, because you didn't change the user/group
<RRedcroft>
b42: nvm fixed it, doing libs.forqt5.callpackage, then just using mkDerivation
<cole-h>
I set up a new "downloads" user+group for transmission + sonarr + etc
xcmw has joined #nixos
<cole-h>
infinisil: I'm wondering if you know how I would set up nginx (or anything) to redirect some URL like `transmission.lan` to `localhost:9091/transmission/web` and `sonarr.lan` to `localhost:8989`?
orivej has quit [Quit: No Ping reply in 180 seconds.]
fuzen has joined #nixos
orivej has joined #nixos
<etu>
cole-h: i do that for my sonarr
<cole-h>
Oh yeah, you do. But I noticed your stuff is in a container -- mine isn't; will that make a difference?
<etu>
cole-h: on the host you were looking at before, check the services directory
<pistache_>
cole-h: multiple "server { ... }" blocks, each with server_name and proxy_pass directives
<etu>
cole-h: That rewrite is in nginx outside of the container :)
<cole-h>
Oh, I see. Btw, is there a reason you run sonarr and friends in a container? etu
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
zebrag has quit [Ping timeout: 246 seconds]
<{^_^}>
[nixpkgs] @sternenseemann opened pull request #88463 → ocamlPackages.curly: init at 2019-11-14 → https://git.io/JfgWm
lsix has joined #nixos
reallymemorable has joined #nixos
drewc has quit [Ping timeout: 260 seconds]
morgrimm has quit [Ping timeout: 256 seconds]
<{^_^}>
[nixpkgs] @adisbladis pushed to master « nixos/display-managers: Also set DBUS_SESSION_BUS_ADDRESS in the wrapper script context »: https://git.io/JfgWc
vyorkin has joined #nixos
<vyorkin>
looks like my /nix/store/.links dir is larger than it should be
lsix has quit [Quit: WeeChat 2.8]
orivej_ has joined #nixos
RRedcroft has quit [Ping timeout: 240 seconds]
knupfer has quit [Ping timeout: 260 seconds]
<vyorkin>
`nix-collect-garbage -d` and `nix optimise-store` don't help
orivej has quit [Ping timeout: 246 seconds]
__monty__ has quit [Quit: leaving]
<infinisil>
vyorkin: That's where everything is!
<vyorkin>
I believe it should be 2 or 3 times smaller
<infinisil>
.links contains all files of the whole /nix/store
<infinisil>
And the /nix/store/*-* paths just hardlink with the .links dir
<vyorkin>
yeah, understand...
chloekek has quit [Ping timeout: 272 seconds]
<vyorkin>
but I still think that it larger than it should really be
<{^_^}>
#88449 (by satshabad-cr, 3 hours ago, open): Can't build numpy cross package for armv7 due to failure compiling gfortran needed in lapack and blas dependencies
orivej has quit [Read error: Connection reset by peer]
orivej has joined #nixos
erasmas has quit [Quit: leaving]
cosimone has quit [Ping timeout: 260 seconds]
xd1le has quit [Read error: Connection reset by peer]
xd1le has joined #nixos
<evanjs>
lassulus: wowwww I did not realize how straightforward nixos-generators really is. And that hyperv (or rather, HyperV gen 2) image support was nust added to nixpkgs...
justanotheruser has quit [Ping timeout: 264 seconds]
<cole-h>
RIP. The web is a mysterious place.
kim0 has quit [Quit: Connection closed for inactivity]
<hpfr[m]>
using the systemd module to write some services for suspend and post-resume. I've specified wantedBy but when I view the generated service files in /etc/systemd/system, they don't have a wantedBy line? does this matter?
<{^_^}>
[nixpkgs] @sternenseemann opened pull request #88467 → ocamlPackages.mirage-crypto 0.6.2 → 0.7.0 and updated/new dependencies → https://git.io/Jfglh
<hpfr[m]>
`before` and `after` have generated fine, but not `wantedBy`
<clever>
hpfr[m]: i think it uses overrides.conf to add it to the thing your wantedby
<clever>
instead adding Wants= to the other thing
<hpfr[m]>
oh yeah, so it's not stateful somehow. thanks!
<hpfr[m]>
clever++
<{^_^}>
clever's karma got increased to 430
<hpfr[m]>
thanks for that big-parallel tip, I can actually build the patched kernel for my surface now without it going nuclear
MmeQuignon has quit [Ping timeout: 252 seconds]
<T0pH4t>
can a nixos module create a new user or would that have to be done at a container/os configuration.nix level?
civodul has quit [Quit: ERC (IRC client for Emacs 26.3)]
cole-h has quit [Quit: Goodbye]
<clever>
T0pH4t: configuration.nix is a nixos module!
<T0pH4t>
i guess that its true
<clever>
so both modules and configuration.nix can create users
<clever>
T0pH4t: and did you add your module to module-list.nix ?
xd1le has quit [Read error: Connection reset by peer]
<T0pH4t>
no, i addded to the imports of a container
justanotheruser has joined #nixos
xd1le has joined #nixos
<clever>
T0pH4t: ah, containers dont restart when you nixos-rebuild
<clever>
T0pH4t: you need to restart the container manually
<T0pH4t>
but its declarative
<T0pH4t>
shoudl that not restart it?
<clever>
nope, it doesnt for some weird reason
orivej has quit [Quit: No Ping reply in 180 seconds.]
<clever>
systemctl restart container@name
SaraCurvy has quit [Client Quit]
<T0pH4t>
oh really! wow, ok.. thx for the info lol
cole-h has joined #nixos
orivej has joined #nixos
<T0pH4t>
actually it doesn't get restart if the config for the module changes
<T0pH4t>
since i see it error out i think
<T0pH4t>
tho i guess something odd is going on since i don't see any of the prestart invoking or was the user i wanted created so :/ will keep looking lol
jgeerds has quit [Ping timeout: 265 seconds]
<{^_^}>
[nixpkgs] @lilyball opened pull request #88468 → ffsend: don't require openssl on darwin → https://git.io/Jfg8L
vyorkin has quit [Quit: WeeChat 2.8]
jgeerds has joined #nixos
ssk has quit [Remote host closed the connection]
jgeerds_ has joined #nixos
RRedcroft has joined #nixos
jgeerds has quit [Ping timeout: 260 seconds]
RRedcroft has quit [Read error: Connection reset by peer]
orivej has quit [Ping timeout: 240 seconds]
<T0pH4t>
so my user is there. just fyi.
<T0pH4t>
preStart doesn't seem to be running tho :/
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<T0pH4t>
nvm foudn it buried in journalctl
maxdevjs has joined #nixos
mbrgm_ has joined #nixos
kvda has joined #nixos
ixxie has quit [Quit: Lost terminal]
mbrgm has quit [Ping timeout: 240 seconds]
mbrgm_ is now known as mbrgm
orivej has quit [Ping timeout: 265 seconds]
orivej_ has joined #nixos
meh` has joined #nixos
Mateon1 has quit [Ping timeout: 264 seconds]
Mateon3 has joined #nixos
Mateon3 is now known as Mateon1
<agam>
Hi, question about "local dotfiles" with Nix, specifically my `~/.zshrc`. I've been using this for a while now: https://github.com/agam/dotfiles/blob/master/.zshrc, and obviously can't refer to an _absolute_ path for the `antigen.zsh` helper script, what's the right way to do this in Nix?