<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkSGm
Ariakenom__ has quit [Read error: Connection reset by peer]
Ariakenom__ has joined #nixos
m4tsa_ has quit [Quit: bye]
<rogerr>
i tried to run nix-shell -p openssl but it error that unable to download ... SSL peer certificate or SSH remote key was not OK (60). what's that mean?
<unclechu>
hi. what would you recommend to change default applications? when i try to “open containing folder” in of a downloaded file in firefox it opens smplayer
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkScw
<bqv>
clever: so, ping has a missing libcap-ng library and doesn't work, but my browser still works fine, and i can curl/dig things fine. /etc/resolv.conf looks accurate, and i restarted systemd-resolved to no avail
<bqv>
i think this is a feature of this being a longstanding repl
ris has quit [Ping timeout: 264 seconds]
<bqv>
or not... :r doesn't help
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/JkSc1
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkSWF
<rogerr>
i install pkgs.rust-analyzer in home.packages then add coc-rust-analyzer to programs.neovim.plugins but when i start nvim with a rust file for the first time it says it can't find rust analyzer and asks if i wanna download it. what i'm doing wrong pls?
<KarlJoad>
If I were to start on some infrastructure for supporting GNU Octave packages in nixpkgs, what would be the best way to go about it? I know https://github.com/NixOS/nixpkgs/issues/65398 exists.
<{^_^}>
#65398 (by matthuszagh, 1 year ago, open): Octave packages support
heyitsrama has quit [Read error: Connection reset by peer]
gthm has quit [Remote host closed the connection]
<YeongSheng>
Hi
<YeongSheng>
I need help removing/uninstalling a broken nix install from macOS Catalina
<YeongSheng>
Is there any writeup I can follow?
<YeongSheng>
I had a working nix install while in macOS Catalina, and since upgrading to Big Sur, my nix is no longer working.
<YeongSheng>
I tried to reinstall nix with multi-user mode, but was told installer detected a previous /nix install and gave me a bunch of instructions to remove the nix install
slack1256 has quit [Remote host closed the connection]
<YeongSheng>
However, when I executed those scripts, the uninstall was reporting errors in Big Sur
iH8c0ff33 has quit [Ping timeout: 260 seconds]
iH8c0ff33 has joined #nixos
Supersonic112 has joined #nixos
Supersonic has quit [Ping timeout: 264 seconds]
Supersonic112 is now known as Supersonic
iH8c0ff33 has quit [Ping timeout: 272 seconds]
<rogerr>
i install pkgs.rust-analyzer in home.packages then add coc-rust-analyzer to programs.neovim.plugins but when i start nvim with a rust file for the first time it says it can't find rust analyzer and asks if i wanna download it. what i'm doing wrong pls?
<raghavsood>
If your config.json is managed by a nix derivation (such as home-manager, or a module), you could put in `${pkgs.rust-analyzer}` and it will output the path in the nix store - can do `${pkgs.rust-analyzer}/bin` if you need a specific sub directory
<rogerr>
raghavsood it's not managed by home manager. any way to have it dynamically looked up so i can hardcode it?
<raghavsood>
`ll `which rust-analyzer` should work
<raghavsood>
Remove the first backtick, IRC doesn't seem to like nested ones
endformationage has quit [Quit: WeeChat 2.9]
<{^_^}>
[nixpkgs] @zakame opened pull request #104974 → perlPackages.CommandRunner: init at 0.103 → https://git.io/JkSiC
Morfio has quit [Quit: This computer has gone to sleep]
<{^_^}>
[nixpkgs] @zakame opened pull request #104975 → perlPackages.ParallelPipes: init at 0.005 → https://git.io/JkSP5
Morfio has joined #nixos
berberman_ has quit [Ping timeout: 240 seconds]
berberman has joined #nixos
marcusr has quit [Ping timeout: 240 seconds]
supercoven has joined #nixos
supercoven has quit [Max SendQ exceeded]
supercoven has joined #nixos
supercoven has quit [Max SendQ exceeded]
supercoven has joined #nixos
supercoven has quit [Max SendQ exceeded]
supercoven has joined #nixos
supercoven has quit [Max SendQ exceeded]
supercoven has joined #nixos
red[evilred] has joined #nixos
<red[evilred]>
oh adisbladis (IRC) - my wallet just winced in pain after seeing you mention the SOPINE clusterboard
<{^_^}>
[nixpkgs] @mudrii opened pull request #104976 → add new user as maintainer into maintainers-list.nix → https://git.io/JkS1i
<DigitalKiwi>
red[evilred]: do you have crypto coins?
waleee-cl has quit [Quit: Connection closed for inactivity]
alp has joined #nixos
<rogerr>
why can't i run `cargo install wasm-bindgen-cli` after `nix-shell -p openssl`? it panics because "could not find directory of openssl installation" what im doing wrong pls?
<mizukota[m]>
rogerr your openssl is isolated from your cargo. What are you trying to accomplish? If you need just rust and cargo to work, probably you can install rust and cargo from nixpkgs
<mizukota[m]>
veleiro: it's very hard thing yet to combine a lot of flake inputs together, there's no documentation about how to, for example, add overlay to nixpkgs and pass result as flake output
<rogerr>
mizukota[m] already said what i wanna do, run `cargo install wasm-bindgen-cli`
<mizukota[m]>
@rogerr that is the source of your problem, nix wont patch binaries downloaded using rustup, you need cargo and rustc from nixpkgs
<rogerr>
mizukota[m] ohhh so when i should do is install rustup, cargo, and rustc with systempackages or similar, then use rustup only for the other components that i can't get directly from pkgs?
malook has joined #nixos
<mizukota[m]>
you cant use rustup at all, it downloads binary blobs. You can use steam-run to launch those binaries manually, but i don't think you really want that. So you should use nix for everything
<mizukota[m]>
for example there's a "wasm-bindgen-cli" package in nixpkgs
<mizukota[m]>
also with cargo installed from nixpkgs you should be able to `cargo install` this thing since it will build from source. But you have to ensure that you have all dependencies installed in your current nix shell
<rogerr>
oh wow. mizukota[m] why is rustup in pkgs then?
<DigitalKiwi>
i don't know why we have that but dpkg comes in handy when all you've got is a .deb and you want to make a nix derivation out of the files
<veleiro>
thanks mizukota[m] ! so most flakes only take a few inputs in like nixpkgs?
<rogerr>
mizukota[m]++
<{^_^}>
mizukota[m]'s karma got increased to 1
<veleiro>
or there's no way to easly provide a nixpkgs derivation with an overlay
<mizukota[m]>
veleiro yes, i've never seen flake configuration with more than 10 inputs
<{^_^}>
[nixpkgs] @peterhoeg opened pull request #104987 → ###### Motivation for this change → https://git.io/JkSxE
<mizukota[m]>
you can do weird trick of import inputs.nixpkgs { overlays = [ ... ]; } and put that as output of your flake, but in next flake you won't be able to repeat that since imported nixpkgs with custom overlays cannot be used as nix store path
<veleiro>
why can't it? because its an overlay i suppose
<mizukota[m]>
because imported nixpkgs is not a derivation. And if it would be a derivation, you'll have separated nixpkgs directory for each variant of nixpkgs with overlays you have used, and it's a big waste of space. It's not very expandable, but from the other side it keeps your flakes simple to use and read
<{^_^}>
[nixos-artwork] @garbas pushed 2 commits to master: https://git.io/JkShn
iH8c0ff33 has joined #nixos
<DigitalKiwi>
does ofborg build haskellPackages.* PRs?
<{^_^}>
[nixos-hardware] @dependabot[bot] pushed to dependabot/github_actions/cachix/install-nix-action-v12 « build(deps): bump cachix/install-nix-action from v10 to v12 »: https://git.io/JkSjI
<{^_^}>
[nixos-hardware] @dependabot[bot] opened pull request #213 → build(deps): bump cachix/install-nix-action from v10 to v12 → https://git.io/JkSjL
<pinpox>
I want to deploy a docker container declaratively, which usually takes a password/key via Environment variables. How can I deal with that so they don't end up in the nix store?
<veleiro>
thanks for the explanations mizukota[m] ! this is very good information
vidbina has joined #nixos
<pinpox>
I did "docker run -e PASSWD=12345 .. " But I'd like to put the config on github and not have the password there or in the nix store
<{^_^}>
[nixos-hardware] @Mic92 pushed to dependabot/github_actions/cachix/install-nix-action-v12 « set nixpkgs to unstable »: https://git.io/JkSjr
<veleiro>
flakes is kind of a departure from the monolitic repo i suppose
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkSj1
sangoma has joined #nixos
<veleiro>
pinpox: there's git crypt
<veleiro>
not sure if that helps
<V>
supersandro2000: thank you, I thought that was never going to get merged
<pinpox>
veleiro: I want to run a bitwardenRS docker container, The env vars are used for admin tokens and secrets. I don't want other users of the system to be able to read those
<eyJhb>
pinpox: source a .env you don't commit? -E PASSWD=$SOMETHING_PASSWD
<pinpox>
eyJhb: Where would I soure it? Can I do that in my configuration.nix?
<mizukota[m]>
for docker-compose things you can define both docker nodes and their nixos insides using arion ( https://docs.hercules-ci.com/arion/ ). But i have no idea how to keep passwords in secret, maybe with environment variables
<eyJhb>
Because you can just use --env-file for starting docker, and then it reads it frem there. Then do not commit it, and ensure that no other user can read the env file
<eyJhb>
pinpox: Well you can just stay using Docker for now. Can't really seem to find a way to use the .config withouth it ending in the store
<DigitalKiwi>
pinpox: no, not yet, been thinking about it tho
<redcedar[m]>
hello, I'm running into an odd issue with an emacs package: I've added org-caldav to my emacs packages expression and successfully rebuilt the system, but the package is not in /nix/store (and not accessible in emacs). I've never had this kind of issue adding a package before. What kind of problem could it be? (Things I've tried: updating the channel; trying to pull the package in via elpa or orgPackages, but this gives me
<redcedar[m]>
an error (I'm currently getting it from melpa stable)).
<pinpox>
eyJhb: I think I'll do that and use the --env-file. Seems like a reasonable option. I'm not sure if the env vars are needed onece the container is started for the first time since it create a config file from them in the volume
<eyJhb>
Prop not, but doesn't hurt having it there
<eyJhb>
The real trouble is just if you plan on other people having access to Docker on the server...
<veleiro>
but think about moving to nixos containers =)
<pinpox>
veleiro: nixos containers?
domogled has quit [Ping timeout: 256 seconds]
<veleiro>
instead of docker
LnL has joined #nixos
<pinpox>
veleiro: didn't even know that existed. Bitwardenrs comes packaged as docker, that's my main reason for going with docker.
<supersandro2000>
V: if you have anything other long waiting send me a link
<supersandro2000>
I see if I can do something
<pinpox>
Using native bitwardenrs would also be nice if I find a good way to do it
<V>
supersandro2000: will bear that in mind <3
<veleiro>
yeah i'm currently exploring nixos containers but long term it looks good
<veleiro>
extra-container builds on them to make them declarative instead of imperative
<veleiro>
and extra-container works on any system with systemd, by systemd-nspawn
cr4y1_ has joined #nixos
<eyJhb>
I use nixos containers for plex and other stuff I don't trust
<tomturbo>
hello, on nixos I can use security.pki.certificates to manage the trusted root certificate authorities. is there an option in nix-shell which allows this to be done on a per-shell basis?
acowley has joined #nixos
civodul has quit [Remote host closed the connection]
<DigitalKiwi>
i have every device possible and bitwarden has clients for all of them and i don't need the features that they charge $10 a year for but i gladly do (i used to have 1password which was a lot more expensive and like bitwarden better) keepassxc was a pain to use on everything tons of ad-hoc dropb box and syncthing with different clients and i just never liked the UI but i did use it for a long time
supersandro2000 has joined #nixos
<DigitalKiwi>
i swear bitwarden and signal should pay me for how much i shill for them :|
domogled has joined #nixos
<veleiro>
i just remember my passwords so far, havent felt the need for a manager
<gthm>
I use 1password because it comes with nice apps and guis I can convince my wife to use
<DigitalKiwi>
i have hundreds of passwords
<DigitalKiwi>
and i don't like typing lol
ilmu1 has quit [Ping timeout: 260 seconds]
<DigitalKiwi>
not 10 minutes ago i was pondering making a joke like "i'm tired of typing my gpg key passphrase" (and i use keypass and a long timeout...so i don't even type it that often ) "should i remove the password or buy a yubikey?"
<{^_^}>
[nixos-hardware] @Mic92 pushed 2 commits to master: https://git.io/Jk9LQ
zupo has quit [Ping timeout: 240 seconds]
<gthm>
DigitalKiwi: no, but I'll have a look. thanks for the tip
<DigitalKiwi>
https://simplelogin.io/ :o i've been thinking about making something like this for a long time >.>
<V>
DigitalKiwi: -chat is a better place for stuff like this
<DigitalKiwi>
:(
<gthm>
I'm trying to run jupyter inside a niv environment. It worked a few minutes ago, then I added pandas, and now launching jupyter fails with "[...] Argument list too long"
<gthm>
the whole error is "/nix/store/pcgz4g8dllm2gmg3b7l33sppq6v98k3f-python3.8-notebook-6.0.3/bin/jupyter-notebook: line 4: /nix/store/pcgz4g8dllm2gmg3b7l33sppq6v98k3f-python3.8-notebook-6.0.3/bin/.jupyter-notebook-wrapped: Argument list too long"
veleiro has quit [Read error: Connection reset by peer]
<octe>
i'm a bit confused about nativeBuildInputs vs buildInputs, even after reading the documentation.. buildInputs is for things that are also required at runtime, otherwise it goes into nativeBuildInputs?
werner291 has quit [Ping timeout: 272 seconds]
werner292 is now known as werner291
<mizukota[m]>
thing about cross-compilation. When you build something for windows, you use pkg-config for linux to discover dependencies for windows. So pkg-config is nativeBuildInput. NativeBuildInputs are needed only for build-time and for native architecture you run nix from. All cross-compilers for example should go to nativeBuildInputs. Cmake shoud also go there
zupo has joined #nixos
Guest40130 has quit [Read error: Connection reset by peer]
<mizukota[m]>
@daende are you sure you need a patch? Do you have 100GB of disk space and at least 16 gigs of RAM to compile the chromium?
m1cr0man has joined #nixos
<daende>
yes i am sure and prepared :)
<mizukota[m]>
I call it "semi-open-source", when it looks like opensource but only 1/1000000000 of people can build it because of those hardware requirements...
<daende>
the patches are in common.nix, so i need to get grip on mkChromiumDerivation, am i right?
Ariakenom_ has joined #nixos
anderslu1dstedt has joined #nixos
<thibm>
daende: just a question, do you want to apply to patch and follow updates or is this a one shot try?
<DigitalKiwi>
mizukota[m]: it's not a real google project if you don't need a fiber connection and google data center to download and store it, now is it?
<daende>
at the moment its a one shot, but if it works, well, than this is going to stay
<thibm>
And at google they don't compile on their work machine and they never compile twice the same files. So they don't really bother about the 100G/16G and compile time
<thibm>
daende: I would say to copy/paste the derivation and just insert your patch at the right place before bothering with overriding. You'll see if it works
<thibm>
(If it does, you won't need to compile again with the more elegant solution anyway)
mkaito has quit [Quit: WeeChat 2.9-dev]
anderslundstedt has quit [Ping timeout: 265 seconds]
<daende>
there is a nix-cache here, a clean solution would be best ;)
<thibm>
Sure, a clean solution is best. That depends on your goal ;)
<thibm>
A intermediate solution is to change the nixpkgs files to be able to insert a patch deep in common.nix
<daende>
thibm: there must be another way :( guess i have to copy&paste until i find it, or someone points it out
<thibm>
They are several ways to do it. Then upstream your change to nixpkgs and you keep updated
<thibm>
daende: honestly I'm not sure. But maybe wait for someone else advice
<daende>
its such a special usecase, lets see if the chromium devs bother to fix my problem if i can verify the bug.
<daende>
until then, a clean fix for me would be great :3
<daende>
thibm: thanks for your input :)
anderslundstedt has joined #nixos
Ariakenom__ has joined #nixos
anderslu1dstedt has quit [Ping timeout: 240 seconds]
<thibm>
You're welcome. The let…in is a really bad pattern. You can take a look at pkgs/applications/science/electronics/kicad/default.nix to see a way to workaround, if you're interested
<pinpox>
Hey, I was asking about bitwarden_rs earlier. Maybe someone can help: Would it be possible to overwrite this line to point to a "fixed" path on my system like "/secrets/bitwarden_envfile" instead of using the generated config?
<pinpox>
Do I need to create a completly new package for that?
<pinpox>
I'd like to just write the env file myself and put it manually somewhere
<supersandro2000>
How can I pass an optional input to nix-build? --argstr does not want to work
<rogerr>
why doesn't home manager env get into my $PATH when i rdp in but does when i ssh in? my rdp startwm.sh https://termbin.com/i1gq and .xprofile https://termbin.com/mcxp what is break pls?
<supersandro2000>
Currently I have nix-build -A postfix --argstr withMySQL true
<adisbladis>
mizukota[m]: Why don't you PR it yourself?
<mizukota[m]>
I do not have github account, and i'm lazy. Does that matter?
thblt has left #nixos ["ERC (IRC client for Emacs 27.1)"]
werner292 has joined #nixos
<adisbladis>
mizukota[m]: You're not just asking someone to make a PR. You're asking them to go through an entire review cycle.
werner291 has quit [Ping timeout: 246 seconds]
werner292 is now known as werner291
<mizukota[m]>
yes. For 45 lines of code, with 5-8 modifications from old version of same package
LilleCarl has quit [Ping timeout: 260 seconds]
<adisbladis>
Like most contributions, they still need to go through review, and someone needs to own that process.
<adisbladis>
I think it's not in good form to ask other people to proxy contributions because you are lazy.
SomeoneSerge has joined #nixos
marcusr has joined #nixos
mvnetbiz_99 has joined #nixos
lankatze has joined #nixos
lankatze has quit [Client Quit]
<mizukota[m]>
I do help people with their code, for free, without calculating my power consumption and resource usage, and I expect the same. The chat isn't just you, and maybe somebody even needs fresh zig compiler with version 0.7.0. I only ask to PR, i'm not forcing it, so please don't tell me to stop beleiving in people, throw away my code and not to ask anything because it's not easy
iH8c0ff33 has joined #nixos
<averell>
i wish there was a work queue for stuff like that to shovel diffs into. but i doubt there's many people enjoying that part.
red[evilred] has joined #nixos
<red[evilred]>
what adisbladis (IRC) said. If people are willing to learn we're more than happy to assist.
SomeoneSerge has quit [Ping timeout: 272 seconds]
lankatze has joined #nixos
<red[evilred]>
oof
<V>
matrix is so weird
<V>
seeing someone join and then reply to something said before they entered is... interesting
<red[evilred]>
I think the problem there is "I expect the same"
<red[evilred]>
it certainly can be
<lukegb>
V: it's super disconcerting, hah
<red[evilred]>
V (IRC): also when you type something and it shows in your channel version - before you even join the channel in irc for real
werner291 has quit [Ping timeout: 272 seconds]
<V>
mizukota[m]: also there's a difference between tossing a random paste into a chat and expecting someone to carry it to the finish line (without even explaining what it does; no diff or anything)
<thibm>
mizukota[m]: well, first of all, diff show that you disabled check phase. Why?
<mizukota[m]>
because I have no idea how it works, how it should work and how to check that check phase works. Instead I've used the power of manual checking that thing works.
shah^ has quit []
Fare has joined #nixos
<red[evilred]>
The problem with manual checking is:
<thibm>
mizukota[m]: so imagine someone created a PR to you
<thibm>
the someone asked him/her this question
<thibm>
how can this someone answer it? You see the problem?
<thibm>
(And your answer is not valid IMHO)
<mizukota[m]>
yeah, sorry i'll never share my code again
<red[evilred]>
1. reproducabilty.
<red[evilred]>
It's not a case of not sharing
<red[evilred]>
it's a case of ...
<red[evilred]>
Okay - got it.
<averell>
the world will mourn the loss. we have a bot to bump numbers in version strings.
<thibm>
… that's not what I said.
<thibm>
I just exposed one of the problem of your approach.
<red[evilred]>
When you're working on a Linux Distro - we literally have tens of thousands of packages
LilleCarl has joined #nixos
<red[evilred]>
If you've done the work to solve a problem - we appreciate that - we really do
<red[evilred]>
but in order to use it, you have to help us get it into the state it can be injected
<red[evilred]>
otherwise we can't keep up with the demand
<red[evilred]>
our PR queue is current 2,400ish
<red[evilred]>
if you can understand our issue
<red[evilred]>
btw - can you email me the diff since I can't see it?
<red[evilred]>
I like ziglang so may go in that direction at some point
Dotz0cat has quit [Ping timeout: 272 seconds]
<red[evilred]>
(I maintain ponyc and corral because I use both those languages)
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkHv4
<mizukota[m]>
i wrote package definition, i even made a patch. To reproduce a package you can use nix-shell. Do I need to apologize for each line i changed or to ask more nicely, to beg for my patch to be added? Maybe I have some psychological problems but I cant do that or do not want that enough.
<V>
Ah.
<V>
that makes sense
Ariakenom_ has joined #nixos
<thibm>
mizukota[m]: again, the problem is not on the work you've done.
<adisbladis>
mizukota[m]: All you need to do is to go through the same contribution pipeline as everyone else.
<thibm>
mizukota[m]: you are basically asking people here to do extra work because you're lazy.
<red[evilred]>
Hell, I spent 45 minutes in a zoom call with someone the other day teaching them how to build their first package definition and submit a PR
<red[evilred]>
We'll welcome anyone who is a positive contributor to our community
<red[evilred]>
(ziglang is the same too)
nikivi has joined #nixos
Ariakenom__ has quit [Ping timeout: 256 seconds]
<thibm>
The first time can be more or less difficult whether your know well nix, git, github. You can ask for help.
<{^_^}>
[nixpkgs] @ymarkus opened pull request #105034 → [20.09] postfix: fix "cant find <mysql.h>" when building with "withMySQL = true" → https://git.io/JkHfW
joesventek has quit [Quit: Quit]
jonatanb_ has joined #nixos
nikivi has quit [Client Quit]
<mizukota[m]>
Even if I'd send PR, i'm still not able to stand infinite PR rejects and attacks of "hey, whats here? Why did you change that? Your indentation is not appropriate! Merge those 3 commit into 1 commit" red[evilred] Did you teach your student how to bypass that?
<thibm>
mizukota[m]: feedback also helps you to improve your knowledge/skills
<red[evilred]>
we have to be that strict in order to try and keep entropy down
<thibm>
(Like how does the check phase work)
<etu>
mizukota[m]: The feedback process is meant to be done in a positive way for both parties to learn and improve. It's both about learning to give good feedback, but also to recieve feedback. Both are skills that is good to work on.
<red[evilred]>
but I really do understand why it would feel that way
<red[evilred]>
I felt that at the start
jonatanb has quit [Ping timeout: 240 seconds]
<red[evilred]>
and it took me a little while to not get frustrated with it.
<V>
yeah, none of this is personal; sorry if it feels like that
<thibm>
And, again, you're asking someone here to deal with this feedback instead of you. But, well…
<etu>
NixOS has made me better at git, when it comes to rebasing/etc to rewrite the history, partly due to feedback I've got in the past. Which lead me to do start giving people feedback on this subject as well. :)
<red[evilred]>
oh - that's so true. I've had people here dig me out of more self-inflicted git holes than I care to cound
<red[evilred]>
counT
<red[evilred]>
I remember accidentally committig 5000 or so changes to my branch
<red[evilred]>
and i'm like 'oops'
<red[evilred]>
and someone here walked me through getting all that straighteend out
<red[evilred]>
they invested time in me to make me better
nikivi has joined #nixos
<etu>
This is what positive feedback and learning is about.
<etu>
It's not complaining.
<red[evilred]>
and now - to be clear I'm not a committer, I'm not ready to be. I now see my role as being someone to help carry some of the load with people who are very new
joesventek has joined #nixos
<red[evilred]>
to help them through their first mile so the same people who helped me don't have to help cover that same first mile with people again.
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkHJa
<red[evilred]>
I want to show that their investment in me was a good investment
<red[evilred]>
and I really care about this community
fendor has quit [Ping timeout: 240 seconds]
<red[evilred]>
and that's why I took the time to help that person, and why I'm willing to take time to help you.
<mizukota[m]>
isn't that easy to fix by re-cloning repo in separate directory, checkouting your work at commit you want, copying needed files to separate directory with re-cloned repo and committing it again?
<noonien>
hello folks
<red[evilred]>
No, because I applied it to a PR
<Abdullah>
hello noonien
<red[evilred]>
so I had the fix my remote branch
<noonien>
i'm trying to package koka for nixpkgs, however, it requires `stack`, which appears need to use a network connections. is there a way to package such packages?
<red[evilred]>
(and honestly, at the time - I was still asking help as to how to make branches)
<noonien>
i see there's stackage2nix, but it doesn't seem to work with 20.03
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkHJ5
<etu>
mizukota[m]: And if you don't know how to do it the "right way" you would have to open a new PR... That's really not the right way if there's already a PR where the feedback process has started.
<red[evilred]>
noonien (IRC): we tend to patch their build processes to remove the network dependency
kreyren has joined #nixos
<noonien>
yes, i've done with other languages. but i'm not familiar with haskell/stack
<red[evilred]>
nor I :-(
kreyren has quit [Remote host closed the connection]
<mizukota[m]>
@etu "feedback process" is act of PR rejection, so does it matter if i'll send another PR with same functionality, maybe even closer to not being rejected?
kreyren has joined #nixos
<etu>
mizukota[m]: It's very rare that I see rejected PR's in nixpkgs.
<red[evilred]>
I think this may be the disconnect.
<etu>
mizukota[m]: Requested changes is not rejected.
<etu>
mizukota[m]: Explaining of things is not rejected.
<red[evilred]>
mizukota[m] (IRC): it's people helping you get things in a better state so that the end product for everyone is better.
<thibm>
mizukota[m]: ^ see, a PR for which you apparently helped get merged :)
<red[evilred]>
please try to think of review as someone taking time to help you work on your submission - not some kind of battle :-/
<mizukota[m]>
etu requested changes does not guarantee that human who created PR will do those changes, so technically PR is rejected untill those changes appear or untill timeout that can last for years
<dutchie>
this has reminded me that i have a wip branch to update zig but the builds were failing :(
<dutchie>
maybe i should open a draft pr for it
<etu>
mizukota[m]: In my $dayjob we work with feedback on a regular basis among peers. Where everyone are encouraged to participate to give and get feedback within your team following a pattern of "Things to keep doing, Things to become better at, Things to stop doing". This is to help everyone improve.
<etu>
mizukota[m]: I see these PR process the same way.
<red[evilred]>
exactly etu (IRC) - we can't improve without external feedback
<red[evilred]>
"This is the way"™
<etu>
mizukota[m]: Well, that's why it's important that the person opening the PR actually cares about it and do the work.
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkHTO
<red[evilred]>
I've been "NixOS community adjacent" for many many years - but only really got involved in the last three or so.
<red[evilred]>
and yes, I found my first three or four PRs incredibly frustrating
<{^_^}>
[nixpkgs] @FRidh opened pull request #105036 → nixpkgs manual in CommonMark using sphinx+myst and jupyter-book → https://git.io/JkHTC
<mizukota[m]>
etu on your $dayjob shouldn't money be your courage to handle this "feedback"? I cant accept what you call "feedback" or "criticism" to be something that helps people or software to get better, because i see that "feedback" as a big stream of negative energy towards a human
dstzd has joined #nixos
mallox_ has joined #nixos
<supersandro2000>
red[evilred]: I am living in a nix hotspot.
werner291 has joined #nixos
bgamari_ has joined #nixos
<red[evilred]>
supersandro2000 (IRC): you are a nix hotspot ;-)
<supersandro2000>
I still don't buy half of the claims of nix and mostly use it as a better brew alternative
<V>
supersandro2000: which ones, out of curiosity?
<red[evilred]>
someone revoke the unbeliever's commit access!
<red[evilred]>
<3
bgamari has quit [Ping timeout: 260 seconds]
<red[evilred]>
mizukota[m] (IRC): do you feel the same way about feedback in other aspects of your life?
<mizukota[m]>
i liked nixos because it's the only distro i could build iso for, with kernel loadable driver for my wifi
<red[evilred]>
I'm really glad it's helped you
<mizukota[m]>
red[evilred] yes
<etu>
mizukota[m]: No, I did dread the feedback sessions before we started with them. But they aren't negative. They focus on the good things but they are there to be honest and to help you. I want to be better, if people can help me to point out how to be that, why would I not want to know that?
mallox has quit [Ping timeout: 256 seconds]
<red[evilred]>
Thank you for your honesty there mizukota[m] (IRC) - that took a lot I'm sure.
hcssmith[m] has joined #nixos
<mizukota[m]>
@etu can you show an example of PR rejected with focus on good things and maybe even some motivation to do those "required changes"?
gthm has quit [Ping timeout: 240 seconds]
<red[evilred]>
I'm guessing then that you probably work in a somewhat hostile environment, in a place where you have to watch your back?
<thibm>
mizukota[m]: so there's usually no reaason to reject them. They are merged when the quality is good, maybe after some updates according to feedback
mullein[m] has left #nixos ["User left"]
<thibm>
I mena, there's no reason to worry about being rejected
<red[evilred]>
mizukota[m] (IRC): That's not my experience on Earth at all
<red[evilred]>
not everyone's life is like that
<red[evilred]>
so I think that may be why we see things differently to you
vidbina has joined #nixos
<red[evilred]>
I work with people who go out of their way to help each other
<red[evilred]>
we go out of our way to share credit.
acarrico has joined #nixos
<red[evilred]>
we go out of our way to take time to care for each other.
<red[evilred]>
if someone is stacked with work, we take some of their work for them
<red[evilred]>
and don't go looking for credit for it
<red[evilred]>
I wish that same environment for you
<red[evilred]>
somehow
FRidh has quit [Ping timeout: 264 seconds]
FRidh has joined #nixos
costrouc has quit [Quit: costrouc]
<gchristensen>
the few people banned from the nixpkgs github tracker were banned because they were jerks
<mizukota[m]>
red[evilred] probably we live in different space dimensions of Earth
orivej has quit [Ping timeout: 272 seconds]
<red[evilred]>
sounds like - and I think you probably think most people live in yours and I believe most people live in mine. That's completely normal.
<red[evilred]>
Tell you what mizukota[m] (IRC) - if you email me your patch I'll go ahead and run the PR for you.
<mizukota[m]>
thibm imagine that you spent a week inventing a solution no other people could invent, and you did it and you glad you did. You share it with people for sake of humanity progress, and then your PR is rejected and you need to change that and this to make your code fit to the taste of reviewer to kindly accept your code. This is how I see those PRs, reviews...
<mizukota[m]>
Also those processes turn contribution into a continious process, but I can dissappear or forget my password tomorrow or face another busfactor
<red[evilred]>
It's a package I want uptodate
<red[evilred]>
but I mean this in the most loving way possible - I don't think I, or anyone else here can convince you that we really mean to help in PRs if your life experience of criticism has been so painful for you.
<mizukota[m]>
red[evilred] I don't know your email, but more importantly I do not have email account xD Thaats hard to explain
<red[evilred]>
Okay - can you paste it to me in a msg
<red[evilred]>
(I'll reformat it)
<thibm>
mizukota[m]: I understood your point. The reviers are not to "fit the taste of reviewers" but to improve the project quality
<red[evilred]>
also - didn't I see someone earlier say that they'd been working on a PR for it but had issues getting it to compile?
<mizukota[m]>
thibm what is the difference?
<thibm>
mizukota[m]: for the bus factor, I'd say a PR is better than a message on IRC to a random pastebin that may delete the file at any moment
<thibm>
mizukota[m]: you're trying to participate in a open source project. The project would just collapse if anything is accepted without review.
<thibm>
And, again, PR are not rejected.
<etu>
PR's can be rejected, but it's super rare, they can be forgotten, not worked on, work in progress, etc before they become accepted and merged.
<simpson>
mizukota[m]: The problem might be with the idea of "inventing a solution no other people could invent". Rarely are people like that; instead, invention is usually *contextual*. IOW you bumped Zig because you wanted Zig bumped, not because you were uniquely in a position to bump Zig.
<red[evilred]>
I'm probably telling you things you already know, but things like network tests will almost always fail in sandbox
<dutchie>
not off the top of my head, i'm at work for the next couple of hours so can't look in detail
<red[evilred]>
so I have to comment them out in the ponyc test suite for example
fendor has joined #nixos
<mizukota[m]>
simpson bumping version is not creative, but somehow it could become a source of huge amount of questions and conversations, with some creative thing i fear I wont be able to explain everything at all. For example I've managed to build custom Nixos iso using flakes, so you can do it in reproducible way and then use flakes from booted iso to install OS with minimal download
<dutchie>
iirc it was some permission error, my guess is that it's trying to touch something outside the sandbox
<dutchie>
i was going to start commenting out tests to figure out which one but i didn't get to it yet
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkHqX
<simpson>
mizukota[m]: Fun! I've got lots of private experiments with nixpkgs which will never go upstream; that's just how it is.
<mizukota[m]>
well they will go upstream, iso building expressions are changing to use only relative paths and eventually the official way of building nixos iso will be with use of flakes
<red[evilred]>
musl makes sense given his focus on minimalization
jonatanb has quit [Remote host closed the connection]
<dutchie>
a lot of the zig std library code is translated from musl
<red[evilred]>
dutchie (IRC): - can you open it as a WIP PR please?
<dutchie>
yeah, will do after work
<red[evilred]>
that will then let people know that people are working on it
<red[evilred]>
cool - thanks
<red[evilred]>
and don't stress the squash - that can wait until all the tomfoolery is complete :_)
<mizukota[m]>
dutchie also you get a lot of different libc for many platforms with zig compiler
<red[evilred]>
I wonder how zig integrates with the cross-compiling environments in nix (or maybe it doesn't)
<mizukota[m]>
it does
<mizukota[m]>
zig build can use pkg-config, and pkg-config works with cross-compilation in nix
<red[evilred]>
Yay! that's awesome
<mizukota[m]>
yes, and no black magic with "rustup toolchain download needed-platform" downloading binaries from external servers
alp has quit [Ping timeout: 246 seconds]
eoli3n has left #nixos [#nixos]
<mizukota[m]>
i've tried to make a rust package with cross-compiler for mingw, and i failed
sangoma has quit [Ping timeout: 256 seconds]
<lukegb>
etu: I've been trying to go through the list of open PRs that haven't been marked as stale (as a way of judging whether they've been forgotten about) and just giving them a review in an effort to tackle the backlog from "the other end"
<red[evilred]>
honestly - apart from my security PRs, I've mainly been procrastinating working on the bacula/bareos nmodules
<red[evilred]>
since they are currently both broken
<{^_^}>
[nixpkgs] @Infinisil opened pull request #105042 → release-lib: Don't use tryEval for packagePlatforms → https://git.io/JkH3G
<red[evilred]>
I don't know how many other maintainers have tape robots at home to work on it... so I should probably do something about it.
<lunik1>
I would differentiate between a doom clone which just copies the gameplay and a doom port/emulator which recreates the engine and uses doom assets
ilmu1 has quit [Ping timeout: 264 seconds]
waleee-cl has joined #nixos
<mizukota[m]>
freedm has it's own assets and level design, but it's not on nixpkgs
<daende>
thibm i didn't time it, but this sound awfull reasonable :3
gulplante has quit [Quit: WeeChat 2.7.1]
supersandro2000 has quit [Ping timeout: 265 seconds]
jonatanb has joined #nixos
jonatanb has quit [Remote host closed the connection]
<thibm>
Awful. That moment when checkPhase fails after a long compilation… (They must be a simple way to decouple build/check in two derivations to avoid that)
pushqrdx_ has quit [Remote host closed the connection]
Mic92 has joined #nixos
pushqrdx has joined #nixos
Steelwire has joined #nixos
cosimone has joined #nixos
CcxWrk has quit [Ping timeout: 240 seconds]
<redcedar[m]>
hi folks, quick question: what might be happening in the case where I add a package to my configuration, rebuild successfully, but the package never shows up (its not in /nix/store)?
CcxWrk has joined #nixos
<drozdziak1>
I'm trying to write a shell.nix for a local rustc repo. I'm starting a shell with `nix-shell '<nixpkgs>' -A rustc`. I get a message about `ninja` missing, how do I turn that command into a `shell.nix` where I override to include `ninja`?
<drozdziak1>
(the missing `ninja` message comes from the rustc build, not Nix)
<V>
redcedar[m]: how did you add it to your configuration, and how did you verify it's not in /nix/store?
<V>
(did you make sure that you saved the file, etc)
<{^_^}>
[nixpkgs] @freezeboy opened pull request #105044 → gitAndTools.glab: init at 1.11.1 → https://git.io/JkHWj
<V>
(if you're using sudoedit then it won't sync until you close your editor)
<mizukota[m]>
redcedar it happens when you add package to file that is not used with nixos-rebuild. For example when you have both configuration.nix and flake.nix, you change flake.nix, but nix choses to use configuration.nix
<redcedar[m]>
V: thanks for responding! in this case its an emacs package, which I manage in a separate file. I know my system is reading it because if I make an error in it, nixos-rebuild switch complains. I used `ls /nix/store | grep emacs` to review all the installed emacs packages. I also confirmed its not the kind of package that gets added to /run/current-system/sw/share/emacs.
<redcedar[m]>
<mizukota[m] "redcedar it happens when you add"> thanks for responding! in this case I know my system is reading my emacs configuration because if I make an error in it, nixos-rebuild switch complains
<V>
redcedar[m]: so you're doing this via emacsWithPackages? and that customised emacs is either in your systemPackages or user.packages?
CcxWrk has quit [Read error: Connection reset by peer]
<redcedar[m]>
<V "redcedar: so you're doing this v"> Yes, I use emacsWithPackages in systemPackages
CcxWrk has joined #nixos
<V>
redcedar[m]: I believe you might only see emacs-with-packages and emacs-packages-deps
<V>
I'm not familiar with the emacs support myself
<redcedar[m]>
<redcedar[m] "Yes, I use emacsWithPackages in "> I add and remove packages from this expression all the time; this is the first time adding a package didn't seem to work
<V>
odd
<etu>
lukegb++ (for looking at backlog of prs)
<{^_^}>
lukegb's karma got increased to 4
<lukegb>
Is it possible that the emacs package just happens to end up without "emacs" in its name in the store but is present anyway?
<Steelwire>
Hello, I'm trying to make a netboot system. I'd like to test it by building a VM, but whenever I try to start the VM, the whole system freezes. I'm running NixOS in VirtualBox, so that could also be a part of the issue.
<matthewcroughan_>
Your latest comment makes it seem like there was a solution. But having tried everyone's shells, I can't find one that works.
<Steelwire>
lukegb: Yeah, I had a NixOS 20.03 VM before where this worked. That was ~3 weeks ago or something like that. Now it doesn't work anymore in both a 20.03 and a 20.09 VM. I'd assume something broke, but I'm not sure where to start debugging this issue / how to do so exactly
SanchayanMaity has quit [Quit: SanchayanMaity]
<Steelwire>
It just freezes without any additional output, so it's hard to see where it's going wrong
iH8c0ff33 has quit [Ping timeout: 256 seconds]
<Mic92>
matthewcroughan_: where are you stuck?
<lukegb>
Does Virtualbox have any logging that might indicate if it's getting stuck?
astylian has quit [Remote host closed the connection]
<Mic92>
still the libc stuff?
<simpson>
matthewcroughan_: How big is your Yocto, uh, specification? Recipe? Can't remember their terminology, but if it's not too big, maybe you could build it with nixpkgs instead. (Maybe your Yocto tolerance is higher than mine!)
<redcedar[m]>
lukegb: I checked in the nix package search to see what the name was (and also tried different greps; and then just read the whole list :)
<lukegb>
oh boo :p
<Mic92>
lukegb: sometimes it feels like oracle don't want people to understand their product. They also removed their debugger api at some point.
<Steelwire>
VirtualBox doesn't show any logging. However, I can see that it's trying to do something / something is locked up, there's a load of around 66% constantly (I've just let it run for the last 15 minutes or so)
iH8c0ff33 has joined #nixos
<lukegb>
Mic92: maybe they're worried if people understood it they'd switch to libvirt/VMware Player :p
astylian has quit [Client Quit]
<matthewcroughan_>
simpson: What do you mean? There's no way I can turn all my colleagues over to nix lol
<lukegb>
Steelwire: I don't have any great suggestions - does it lock the system up even with the default resources?
<matthewcroughan_>
The point is to provide an environment to build yocto in.
<Mic92>
matthewcroughan_: That should be totally feasible with nix.
<matthewcroughan_>
Nix can help unfsck the Yocto build environment, just like you can build Dockerfiles with nix.
<matthewcroughan_>
Mic92: That thread shows how unfeasible it is :D
<matthewcroughan_>
You've not read through the issues with locale?
<matthewcroughan_>
There is nobody in that thread that has managed to get it working. Do you read it differently from me?
<raghavsood>
Steelwire: thanks! keep failed was what I was looking for
<matthewcroughan_>
I cannot do a better job of explaining the issue with locale than the guys in that thread already have.
cmk_zzz has joined #nixos
<matthewcroughan_>
To put it simply though, apps like Perl have the same issue. No matter what you do in a nix shell, they are all unable to use UTF-8 locale.
astylian has joined #nixos
cmk_zzz_ has quit [Ping timeout: 240 seconds]
<simpson>
matthewcroughan_: Ah, the calculus changes somewhat when you're being paid to use something else. Wondering who you were thinking of killing; coworkers~?
<FRidh>
sigh...working on sphinx commonmark manual, using pandoc in a nix-build to convert docbook to md if not yet converted. Need a newer sphinx version as well. Obviously that means rebuilding the whole of haskell
<Mic92>
matthewcroughan_: I could have a look at it but I need some example project.
<{^_^}>
[nixpkgs] @gnidorah opened pull request #105047 → ctrlr: init at 6.0.36 → https://git.io/JkHEV
<FRidh>
add glibcLocales e.g. to your environment / nix-shell invocation and set LC_ALL
<FRidh>
ohh fhs
Steelwire has quit [Ping timeout: 246 seconds]
<raboof>
looking at https://hydra.nixos.org/job/nixos/trunk-combined/tested, am I reading that correctly that nixos-unstable did not advance because the tests for `gnome3.geary` timed out? should someone retrigger that build or is that automatic?
<FRidh>
raboof: correct, among other failures. I am restarting them no
<FRidh>
now
acarrico has quit [Ping timeout: 256 seconds]
werner292 has joined #nixos
werner291 has quit [Ping timeout: 272 seconds]
werner292 is now known as werner291
Steelwire has joined #nixos
<Steelwire>
lukegb: Not sure what happened, I lost all connection to the PC I was doing this on when trying to run with default resources. It seemed like it was not locking up, but now I lost access and can't reach the PC anymore. Thanks for the tips at least, I'll have a look later this week when I can physically go there
<red[evilred]>
dutchie (IRC): I cloned your branch to take a peek at.
<lukegb>
Steelwire: eep, well, good luck
<red[evilred]>
I may not get to complete on it
<red[evilred]>
due to family stuff - but taking a look while stuff is cooking
<Steelwire>
so either the network went down there or it really did crash the host
<{^_^}>
[nixpkgs] @ivanbakel opened pull request #105048 → Make logstash packages run pre/postInstall hooks → https://git.io/JkHzf
turlando has quit [Remote host closed the connection]
FRidh has quit [Ping timeout: 240 seconds]
oxalica has quit [Quit: oxalica]
oxalica has joined #nixos
sangoma has joined #nixos
jonatanb has joined #nixos
<red[evilred]>
holy wow dutchie (IRC) - that's a LOT of tests
FRidh has joined #nixos
FRidh has quit [Client Quit]
turlando has joined #nixos
vidbina has quit [Ping timeout: 256 seconds]
aaabbbbbbbbbb has joined #nixos
werner292 has joined #nixos
vidbina has joined #nixos
<aaabbbbbbbbbb>
Hi, does anybody know a way to modif the timeout setting for xscreensaver, I would like to extend the timeout that it takes to lock my screen?
fendor has quit [Remote host closed the connection]
<mindtree[m]>
rogerr: I think I remember seeing a rustSrc field on rustPlatform or one of its attributes, though I don't quite remember where. Is there some path that nvim searches for the rust src? Or does it specifically check a location that rustup typically installs to?
<SomeoneS1>
mgsk: also note the difference between packages pytorch-bin and pytorch&c (the former fetches the .whl, the others are either built w/o cuda, or trigger building many things from scratch)
<averell>
rogerr: i would say they are essentially equivalent, but the wiki one has this one specific rust config ruststable = (nixpkgs.latest.rustChannels.stable.rust.override { extensions = [ "rust-src" "rls-preview" "rust-analysis" "rustfmt-preview" ];}); and then adds some useful libraries to the environment, whereas the github example just emphasizes different ways to use their repo (a file with a
<averell>
date, stable, etc)
iqubic has joined #nixos
iqubic has quit [Remote host closed the connection]
<mgsk>
omggggggggg SomeoneS1 it worked
<mgsk>
i spent HOURS trying to figure this out, going to like page 10 on google
<mgsk>
in future i'll just ask here first :D
<rogerr>
averell what i dont get is why i have to put all my rust in 1 dir like it says?
<rogerr>
i wanna be able to put rust dirs anywhere in my user account
<rogerr>
im installing rust cargo etc using home.packages home manager so it's user wide not system wide
<SomeoneS1>
simpson: uh, I feel like e.g. libcuda is a runtime dependency for pytorch-bin, but we smhw decide we'll let runtime find it based on global system configuration (on /run/opengl-driver/lib) because "we'd need to modprobe hardware-dependent nvidia_uvc anyway"
<simpson>
SomeoneS1: Aha. It's not arbitrary; it's because the nVidia GPU driver replaces *all* of those dynamically-linked libraries, and so we have to maintain them as a unit.
<SomeoneS1>
simpson: and I'm still not familiar with this stuff enough to tell where this boundary between hw-dependent and hw-independent worlds is
<simpson>
Your next question might be why libGL.so.1 is mandated to be dynamically loaded from some particular place, and the answer is that that shared object's linkage and ABI layout is specified as part of GL on X11 and POSIX.
<SomeoneS1>
simpson: wow, I didn't know about the latter, although had some suspicions - thanks
<averell>
i guess that would work, i have a shell.nix that exports RUST_SRC_PATH and that works for rust-analyzer for me
<SomeoneS1>
simpson: still I assume there should be a way for "pytorch-bin" derivation to declare that at runtime it depends on certain things to be provided by the system
<averell>
it's also a bit more convenient to switch versions like that, rather than nixos-rebuild
<simpson>
SomeoneS1: It would be nice. I think, unfortunately, that that ship has sailed for Linux's particular flavor of direct-rendering API. This happens to be a *very* tough problem historically, with users being able to read Somebody Else's Renderbuffer from the GPU.
<SomeoneS1>
simpson: as in, pytorch-bin needs /run/opengl-driver to be present, but we can't type-check if configurtaion provides it
vidbina has quit [Ping timeout: 272 seconds]
<SomeoneS1>
simpson: and runtime error is not informative either (as discussion in the issue proves)
<SomeoneS1>
simpson: Hmm, I rather meant that it's some other kind of "runtime dependency" that nix hypothetically could manage but, to my awareness, doesn't atm
<simpson>
SomeoneS1: While I don't know if we support them, there exist possible configurations where *indirect* *software* GL is available, and in those cases, the whole song and dance would still be done in the standard way. I agree that surely something better ought to be possible, but it's not clear what.
<rogerr>
averell what you set RUST_SRC_PATH to pls?
<simpson>
GL is very much a special case here. For NixOS, we could put in effort and have a special system GL which does nothing except print an informative error message and crash the process which loaded it~
<averell>
RUST_SRC_PATH = "${rust.toolchain}/lib/rustlib/src/rust/library"; but i believe that path changed in some version when they shuffled stuff around. and rust.toolchain can just be rustChannels.stable.rust or similar
kreyren has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkHha
red[evilred] has quit [Quit: Idle timeout reached: 10800s]
kreyren has joined #nixos
<SomeoneS1>
simpson: I see. On the other hand, there are other programs that depend on some kernel modules to be loaded. Like, vpn depending on tun/tap. It's also a runtime dependency, and it's also specific to the a particular build of the kernel.
astylian has quit [Quit: Leaving]
<simpson>
SomeoneS1: Yeah. In general, we need capability-aware kernels (seL4, Genode) rather than Linux if we want to make serious progress here.
<simpson>
Simply because Linux doesn't have the right API for politely indicating that you hard-depend on a particular bit of functionality and must otherwise die.
<simpson>
...Sorry, that's misleading. Linux does have the polite-indication API, but instead what we'd want is that the *user* chooses which functionalities to give to the application, and the application never knows about the ability to ask; it just has or has not.
<averell>
rogerr: i use it, and i have this: ln -sf ${rust.toolchain}/bin/rust-analyzer "$HOME/.config/coc/extensions/coc-rust-analyzer-data/" so it finds it automatically, but I'm not so sure this whole setup is best practice.
astylian has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @Mindavi opened pull request #105074 → clamav: enable unit tests, add libjson-c, disable rar explicitly → https://git.io/JkQe1
<cirno-999>
> You can install a set with extra packages by using something like (texlive.combine { inherit (texlive) scheme-medium xifthen ifmtarg framed paralist titlesec; })
<{^_^}>
error: syntax error, unexpected WITH, expecting ')', at (string):435:23
<cirno-999>
okay, where do I put this o.0
Morfio has quit [Quit: This computer has gone to sleep]
<rogerr>
i tried it but it didn't do auto fmt but i dunno if i did it wrong
<rogerr>
for the "command" is rust-analyzer right?
<averell>
haven't tried it, but let me know how it goes :) i have a let g:rustfmt_autosave = 1 in my vim conf, maybe it's off by default, i don't remember.
<cole-h>
shapr: `nix.binaryCaches` and `nix.binaryCachePublicKeys` for substituters and trusted-public-keys, respectively
<shapr>
ah, thanks!
<shapr>
cole-h: is there an "append" operation? I don't want to overwrite the existing nixos keys
<cole-h>
Lists merge by default
<cole-h>
Actually, depends on how the defaults are specified. Let me check real quick.
<shapr>
thank you!
<cole-h>
Yeah, they merge
<cole-h>
Since the module implementation specifies `nix.binaryCachePublicKeys` to be the key of `cache.nixos.org`, it will merge with whatever caches you add
<cole-h>
Then nix.conf would contain that cache as well as cache.nixos.org
<shapr>
ah, so I don't need to do something like the shell trick I'm used to: export PATH = PATH : $PATH
<cole-h>
Nope
<shapr>
ok, thanks!
* shapr
tries it
codezero has quit [Ping timeout: 265 seconds]
<{^_^}>
[nixpkgs] @jonringer pushed to staging-next « python/hooks/pythonNamespaces: fix __pycache__ being empty, or not existing »: https://git.io/JkQLF
<cole-h>
Since duplicate module definitions merge (in separate files, of course -- Nix doesn't allow duplicate keys in the same file)
codezero has joined #nixos
<shapr>
vaguely related, is there a language server for configuration.nix ? the attribute trees seem arbitrary to me
<shapr>
would be nice to be able to have autocompletion
<cole-h>
There's rnix-lsp and another one I don't recal
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkQtB
<makefu>
shapr: you should be able to `just use` your zsh in nixos
<makefu>
just make sure that the software you are using in your zsh config is installed for the user (users.users.<username>.packages or environment.systemPackges)
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkQq5
AmandaC has joined #nixos
<{^_^}>
[nixpkgs] @ZerataX opened pull request #105087 → srb2: init at 2.28 → https://git.io/JkQqh
diamondbond has quit [Ping timeout: 244 seconds]
<rogerr>
would it ever make sense to put the shell.nix for a nix-shell environment into a different dir than the environment? like running nix-shell with ~/shells/foo.nix for ~/foo
<steveeJ>
has anyone managed to install additional targets for rust? would be nice if I could avoid the mozilla overlay
<rogerr>
that's exactly what im trying to do right now
<rogerr>
seems like best option is to use nix-shell solution
<rogerr>
each project's dir will have a shell.nix in it that sets up rust
<shapr>
for whatever reason, my CPU frequency stays lower in nixos than in ubuntu, but nixos default settings are much better about caching the filesystem. buff/cache is 123766828
<shapr>
I've tried setting cpu frequency governor to performance, but I'm not convinced it's working
Cale has quit [Remote host closed the connection]
sss2 has quit [Quit: Leaving]
<simpson>
I only know how to use the older tools; `nix-shell -p cpufrequtils` and then use `cpufreq-info` to see what your CPUs support.
<shapr>
simpson: ah, thanks!
est31 has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 0 commits to hydra: https://git.io/JkQYl
Cale has joined #nixos
<supersandro2000>
fck. wrong branch
philr_ has joined #nixos
<mrSpec>
Hello! I'm running nixos container and gitea service in it. I'd like to configure my gitea from scratch, removing all previous config. I've commented out the container, called nix-rebuild swtich and then nix-collect-garbage -d. However after uncommenting the container my old user account is in gitea. What am I missing?
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkQOc
magnus1 has quit [Quit: WeeChat 2.9]
cosimone has quit [Remote host closed the connection]
magnus1 has joined #nixos
cosimone has joined #nixos
<theduke>
Out of curiosity: will nixpkgs (eventually) be rewritten to individual flakes?
<samueldr>
I don't think anyone has an answer for that
<samueldr>
but most likely it'll stay the way it is for a while
<theduke>
Is there something like a forum thread or that goes into pros and cons of this? Or is it just too early to tell?
<steveeJ>
rogerr: in my case there's a crate in the workspace which builds Wasm files in its build.rs, so I need the native and the wasm targets in the same toolchain
Ariakenom_ has quit [Quit: Leaving]
<rogerr>
how you gonna solve it?
<steveeJ>
rogerr: currently looks like the two options are using rustup or rust from the mozilla overlay
<{^_^}>
[nixos-search] @ncfavier opened pull request #236 → Fix rendering of default and example values → https://git.io/JkQnB
werner291 has quit [Ping timeout: 272 seconds]
kdlv has quit [Quit: Leaving]
turlando has quit [Ping timeout: 256 seconds]
<shapr>
next silly question, why is my path different in zsh than in bash? now that I've set zsh to my default shell, I can't find some binaries installed via my configuration.nix ?
<{^_^}>
[nixpkgs] @kwohlfahrt opened pull request #105098 → Use structured settings for PAM configuration → https://git.io/JkQcP
<notgne2>
shapr: you need `programs.zsh.enable = true;` or something
<shapr>
er, what does that do?
<notgne2>
for it to populate the `/etc` files for ZSH to give it the NixOS $PATH