<ryantm>
I missed round 1. For the record, I agree with V that we should be letting maintainers have a shot at responding. Not sure how long that should be. I only merge r-ryantm PRs when a maintainer approves.
rb2k has quit [Read error: Connection reset by peer]
rawtaz has joined #nixos
rawtaz has quit [Changing host]
rb2k has joined #nixos
<ryantm>
Or I've checked it myself carefully.
<{^_^}>
[nixpkgs] @grahamc reopened pull request #110060 → doc: add function argument order convention → https://git.io/JtqGJ
<{^_^}>
[nixpkgs] @grahamc merged pull request #110060 → doc: add function argument order convention → https://git.io/JtqGJ
<{^_^}>
[nixpkgs] @grahamc pushed commit from @deviant to master « doc: add function argument order convention (#110060) »: https://git.io/Jtmbf
<energizer>
in a lot of projects i've worked on, branches don't get merged into master if they break anything. but in nixpkgs, dependent packages don't seem to get tested before merging
<matthewcroughan>
Was just wondering how you might do this on bios, not UEFI.
<energizer>
and in nixpkgs those dependent-package tests have to be manually executed
<adisbladis>
matthewcroughan: That's etu
<matthewcroughan>
Specifically regarding the linked section, which talks about the ESP partitions
<matthewcroughan>
etu: :D
<ryantm>
energizer: OfBorg checks the build and a lot of people run nixpkgs-review to test reverse dependencies build
<V>
energizer: that's because the CI is not integrated with github
<adisbladis>
matthewcroughan: He's very likely asleep right now
<gchristensen>
energizer: with 10s of thousands of packages on 4 native platforms with good support, more native platforms with less support, and cross compilation targets, there is going to be a certain amount of just broken to be expected ... but how much is expected?
<V>
there's two levels, ofborg + hydra
<adisbladis>
matthewcroughan: BIOS is easier
<matthewcroughan>
How so?
<adisbladis>
matthewcroughan: You can set up an mdadm mirror raid as your /boot
thc202 has quit [Ping timeout: 260 seconds]
troydm has quit [Quit: What is Hope? That all of your wishes and all of your dreams come true? To turn back time because things were not supposed to happen like that (C) Rau Le Creuset]
<matthewcroughan>
adisbladis: I want ZFS though.
<adisbladis>
matthewcroughan: I meant _onl_ for /boot
<samueldr>
you'd have a non-ZFS boot partition still, no?
<samueldr>
^
troydm has joined #nixos
<matthewcroughan>
Oh, I see what you're saying, wow.
johnny5 has joined #nixos
<matthewcroughan>
I just need a bit of instruction, as I've never done that before.
<adisbladis>
matthewcroughan: That's handled by `nixos-generate-config`
johnnyfive has quit [Ping timeout: 264 seconds]
<matthewcroughan>
adisbladis: it doesn't handle the zfs stuff, so why would it handle mdadm?
<gchristensen>
it sounds like if we continue getting bots like that, ofborg might be absolved of its "build" responsibilities :)
orivej has quit [Ping timeout: 240 seconds]
<rmcgibbo>
I estimate the cost is only about $10-$15 / month to screen 3/4 of all PRs, which I'm happy to donate.
<rmcgibbo>
Obviously increasing the coverage up to 100% of PRs would increase the cost.
<adisbladis>
matthewcroughan: So from the live media (after creating the zfs file system and mounting it on /mnt), roughly: `mdadm --create /dev/md0 --level=mirror --raid-devices=2 /dev/sda1 /dev/sdb1 && mkfs.btrfs /dev/md0 && mount /dev/md0 /mnt/boot`
<adisbladis>
And from that point you go about your installation as normally
<rmcgibbo>
I'm still very new around here, so I don't want to step on anyone's toes though.
<matthewcroughan>
rmcgibbo: damned straight. Where do I send my money?
zebrag has quit [Quit: Konversation terminated!]
<matthewcroughan>
adisbladis: why btrfs?
<matthewcroughan>
Why can't I do pure zfs?
zebrag has joined #nixos
<{^_^}>
[nixops-aws] @tewfik-ghariani opened pull request #128 → Fixing github ci → https://git.io/Jtmbr
<energizer>
rmcgibbo: how does the selection of "which 3/4" work?
<rmcgibbo>
energizer: I estimated the build time for every package currently in nixpkgs by using the build times on the latest hydra evaluation, and the chose a cutoff based on how much I wanted to spend for this pilot project on cloud resources.
<adisbladis>
matthewcroughan: GRUB has very limited zfs support
<adisbladis>
For example it doesn't boot on encrypted datasets
aasg has quit [Quit: Bridge terminating on SIGTERM]
<energizer>
rmcgibbo: do you have an estimate of the cost for 100% of nixpkgs?
zebrag has quit [Client Quit]
<adisbladis>
energizer: I take it gchristensen is on it :)
marcusr has quit [Read error: Connection reset by peer]
<energizer>
overriding checkPhase works
<infinisil>
energizer: doCheck = false should do it
<energizer>
infinisil: afaict it didnt
marcusr has joined #nixos
<ryantm>
V: For what its worth, I watch all the r-ryantm PRs and I haven't seen a bunch of maintainers complaining about supersandro2000's merges. There were a handful maybe a month or two ago, but not much lately.
<V>
I've not personally seen any problems from the automated version bumps (although I maintain it's good to review the actual changelog + maybe new commits + possibly diff if you're paranoid enough)
berberman has joined #nixos
<ryantm>
Agreed. Sadly sometimes I can tell the Maintainers (who approved) didn't even look at the Nixpkgs PR diff, because the bot made some mistake.
<infinisil>
(though regnat's last point there is a good one, this seems to be supported implicitly with ca derivations)
* ryantm
needs to get CVE reporting working for r-ryantm again.
<V>
the common updater script stuff was so restrictive that I had to effectively reimplement a chunk of it
<gchristensen>
my feeling is that "merge all the PRs" isn't really the important goal, but that is somewhat what is being optimised for
<gchristensen>
(or, rather, that is where their incredible drive is pushing them to)
<V>
and even then, I can't help but feel a little spooked at what's effectively just raw find+replace on version numbers+hashes
<V>
CA drvs fix so many things, I genuinely don't understand why they weren't in from the beginning
<gchristensen>
they're complicated :)
<V>
they're not that complicated
<gchristensen>
oh
<infinisil>
V: Source?
<V>
conceptually
<infinisil>
Well duh
<infinisil>
The implementation is the hard part!
<gchristensen>
I'm off to bed
<V>
there's multiple ways to solve the self-reference problem, though
<V>
Nix is using is a fairly decent approach; but if your system is managed using bind mounts/overlays/etc you can just not have store paths be visible to consumers at all
growpotk- has quit [Ping timeout: 265 seconds]
<V>
I think there was another solution or two I came up with, plus one like what Nix is using but differing in the details, but I don't remember them off the top of my head
<V>
but no, implementation isn't that hard either
<V>
it's basically just a fancy templating system
<KarlJoad>
As far as video-manipulation libraries go, is ffmpeg or libav preferred?
<V>
oh boy
<V>
just go with ffmpeg?
<V>
I feel that's the more commonly used one
<KarlJoad>
Fair enough.
EMayej has joined #nixos
<V>
I wonder how the feature parity is these days; last time I checked they were relatively similar
* V
types ffmpeg vs libav 2021 into her favourite search engine
<K0kada>
I found bump interesting because: there is no update actually (it changed the "filename" but since it is a Dropbox link what matters is the UID of the download, that didn't change)
<ryantm>
Ah crap the new hash format :(
<K0kada>
Yep
<K0kada>
The new hash format made the bot believes that the update is correct
<K0kada>
But I still trying to understand where the bot came with this new release
<ryantm>
I forgot the new format would affect how it detects if the hash changed.
<K0kada>
I mean, there is no update script or anything
<energizer>
why is IFD banned, or something like that
kini has joined #nixos
<bqv>
energizer: I'm guessing something about IFD in ci? But hydra does allow it if you patch that hardcoded restrict-eval out
<energizer>
,ifd
<{^_^}>
import-from-derivation (IFD) is when you evaluate nix from a derivation result, for example `import (pkgs.writeText "n" "1 + 1")` will evaluate to 2. This is sometimes problematic because it requires evaluating some, building some, and then evaluating the build result. It has been described as "such a nice footgun."
<{^_^}>
signalapp/Signal-Desktop#4513 (by eppfel, 18 weeks ago, open): Database startup error: Error: SQLITE_NOTADB: file is not a database
<bqv>
dunno. i never used it that long, i can't remember why. i preferred jami so switched to that
<Nickli>
last i checked i couldn't find anything that said it was decentrlaized
<bqv>
Nickli: oh, perhaps i'm wrong then
EMayej63 has joined #nixos
kwork has joined #nixos
<bqv>
Nickli: yeah, looks like you're right, that'll be why i didn't like it that much
<Nickli>
and was it sessions that have you spend some crypto coin for 1k messages?
EMayej63 has quit [Client Quit]
<bqv>
no i don't think so
<DigitalKiwi>
oh it's a cryptocoin app?
<Nickli>
i didn't look into that one too much. There's also huschat
phfas[m] has joined #nixos
<Nickli>
*hushchat
<DigitalKiwi>
By default, all Session clients use a Session file server run by the Loki Foundation for attachmentsending and storage. Since attachments are not considered a core feature of Session, this design is inkeeping with Session’s design principles. The file server is fully open-source, with setup instructionsprovided so that users are able to set up their own file server [21]. Users are able to specify inthe Session client which file
<DigitalKiwi>
server they want to use for attachment sending functionality. This isimportant both for providing users with choice and control, and ensuring the continued usefulnessand functionality of Session if the Loki Foundation were no longer able to maintain the defaultSession file server.
<Nickli>
but on the positive note, Session doesnt need a phone number to work
<bqv>
i dunno, either way, jami's working fine and dandy for me and meets literally all my needs
<Nickli>
(which it shows to everyone)
EMayej has quit [Ping timeout: 248 seconds]
<bqv>
only downside is that it's not packaged on nixpkgs, but i just use guix for it
<DigitalKiwi>
bqv: it's not in nixpkgs? :(
<DigitalKiwi>
oh you said that lol
<bqv>
people tried. even i tried. it's a pain
<bqv>
guix manages it effortlessly though, somehow
<Nickli>
maybe they sacrificed their first corp
<DigitalKiwi>
guix isn't in nixpkgs :(
<Nickli>
DigitalKiwi > but does sseesions store messages as files?
<DigitalKiwi>
idk
c3hou has quit [Quit: Ping timeout (120 seconds)]
ch88 has joined #nixos
rmcgibbo has quit [Quit: Ping timeout (120 seconds)]
<bqv>
in a properly federated system, if you take down one node, it shouldn't have much effect. in matrix, if you take down matrix.org, basically everything shits the bread, because they've made zero attempt to not make everyone lean heavily into matrix.org, and parts of the protocol actually depend on matrix.org's existence
NinjaTrappeur has joined #nixos
<bqv>
boy i want matrix to succeed, but they're way too keen on matrix being a Product, not a Protocol
<notgne2[m]>
> and parts of the protocol actually depend on matrix.org's existence
<{^_^}>
undefined variable 'parts' at (string):471:5
<bqv>
notgne2[m]: sigs
<notgne2[m]>
I haven't heard of that before, that's really weird
<bqv>
you can't verify the sig of a matrix node without checking it against matrix.org
<bqv>
or, well, you can, but you have to edit your node config and add it to trustedServers
<bqv>
which nobody does, obviously, so the protocol all but depends on matrix.org
<notgne2[m]>
oh, that kinda sucks really
<notgne2[m]>
I also noticed they made little effort to work towards Matrix being usable for non-human communication
<bqv>
matrix is annoyingly successful for something with such terrible direction
<notgne2[m]>
which is weird given they have advertised it as usable in that way many times
<DigitalKiwi>
what does your dog need secure chat for
<notgne2[m]>
I was thinking more along the lines of cars, lightbulbs, and IoT things, but if my pet birds could use it to talk to aliens I also would not complain
quinn has joined #nixos
vonfry has left #nixos ["ERC (IRC client for Emacs 27.1.90)"]
<Nickli>
i would say Matrix is still early on its usability, they just gor E2EE usable and next up is Spaces and message Threading (also a bunch of minor stuff)
<Nickli>
well scratch minor, a lot of stuff
<DigitalKiwi>
do they still make e2ee hard to enable
<DigitalKiwi>
... and not on by default
<bqv>
annoyingly, there's also no popular alternative implementations of the matrix protocol either, as i understand it because the protocol's a mess and keeps changing in violently breaking ways, the e2ee draft especially
amfl_ has joined #nixos
<bqv>
back when i cared to use matrix, i used the construct homeserver (c++ impl), but it doesn't have e2ee cause the dev's struggling enough to keep up with the base protocol
<notgne2[m]>
DigitalKiwi: to be fair, their e2ee is super buggy and always ruins stuff, so I can see why it's not the default, but that really is their problem to deal with
<Nickli>
bqv > there was a point when the protocol was in massive flux, afaik should be more sane now
<DigitalKiwi>
and when i registered it was the worst registration process i've ever had maybe only second to discord
amfl has quit [Ping timeout: 240 seconds]
amfl_ is now known as amfl
<bqv>
Nickli: sane is the last thing it is. from the excerpts i've seen, parts of it don't actually make sense and can only be understood in the context of the reference implementation, which is genuinely ridiculous
<DigitalKiwi>
that was a while ago though
<Nickli>
bqv > when was that?
<bqv>
a few months ago..
c3hou has left #nixos ["Left to pet cats"]
<DigitalKiwi>
the real test of matrix will be when elon musk tweets `use matrix` if their infrastructre holds up
c4droid has left #nixos ["ERC (IRC client for Emacs 27.1)"]
<DigitalKiwi>
notgne2[m]: that's just what i want in an intended to be secure protocl is lots of bugs ;D
<notgne2[m]>
the default homeserver can't hold up even without that
zebrag has quit [Quit: Konversation terminated!]
Rusty1 has quit [Quit: WeeChat 2.3]
poscat has joined #nixos
<bqv>
my genuine opinion w.r.t messaging is that making something federated is currently the only way to be decentralized and actually successful. that said, it's messaging, so i really don't care if it doesn't have critical mass if it's fully distributed, which is why i'm quite happy with jami. i only talk to one person anyway so i just got her to get jami
xinau has joined #nixos
<etu>
matthewcroughan: If you don't want to mess around with mdadm I guess that the grub module should to be able to mirror non EFI partitions as well.
<xinau>
haven't found something like that in the description of the channel.
<bqv>
xinau: it's common sense and politeness
<xinau>
will repost it using such a service.
philr has quit [Ping timeout: 260 seconds]
<xinau>
oh. fuck now i know. what you mean. for got that enter equals a new message :( i'm so sorry
<bqv>
that's ok
<DigitalKiwi>
hey at least it's easier to find logs on search engines when the text is in the channel and not a link :P
<bqv>
mm i mean my suggestion was gonna be hardbin.com
<bqv>
which i guess is worse or better, depending on your perspective
palo1 has joined #nixos
philr has joined #nixos
waleee-cl has quit [Quit: Connection closed for inactivity]
palo has quit [Ping timeout: 265 seconds]
palo1 is now known as palo
Pebs has joined #nixos
<xinau>
ok. next attempt. i've got an error i can't really decipher for my nix-build. i'm unable to compute the correct hash for the file using shasum or nix-hash maybe someone could help me
<c3hou>
What command are you using to get the hash?
<DigitalKiwi>
,tofu
<{^_^}>
To get a sha256 hash of a new source, you can use the Trust On First Use model: use probably-wrong hash (for example: 0000000000000000000000000000000000000000000000000000), then replace it with the correct hash Nix expected. For inserting 52 0's in vim: <esc>52i0<esc>
<c3hou>
Ah yes that. `nix-hash --type sha256 --flat --base32 [file]` also works
<DigitalKiwi>
are lol free memes memes that are not funny
sangoma has joined #nixos
<bqv>
no memes are funny
<bqv>
but these are cool pics
<xinau>
c3hou i already tried that but it didn't work. neither for the zip content nor for the zip itself
<bqv>
i'm glad i clicked
<DigitalKiwi>
click it or ticket
<c3hou>
@xinau I get the correct hash when I run that command though
<c3hou>
`nix-hash --flat --type sha256 --base32 terraform_0.14.5_linux_amd64.zip` gives me `0bri1z2in8syjs2grw0j9waxx69wq2qmfk9fhwqjwxdpc1wg9698`
<xinau>
yeah me too, but when I insert it in my sha256 in the build function it still returns the same error
<xinau>
@c3hou it feels like i overlook something very simple.
<c3hou>
Your error message here: https://paste.ee/p/Le9eW#s=0&l=14 says that it expects hash 0bri1z2in8syjs2grw0j9waxx69wq2qmfk9fhwqjwxdpc1wg9698
<c3hou>
Which is the hash output by nix-hash, but your `fetchzip` function has `sha256 = "2899f47860b7752e31872e4d57b1c03c99de154f12f0fc84965e231bc50f312f";` in the paste. Can you double check if you've changed it to the correct one (and saved the file before trying the build etc etc)
sarcasticadmin has joined #nixos
<xinau>
i fixed the output in the function to `0bri1z2in8syjs2grw0j9waxx69wq2qmfk9fhwqjwxdpc1wg9698`. but I'm getting an error that the hash got in the build is `sha256:1jryzphphb98llynl5crp5v1anjf1nakq4v4vc2ix9s83slcgcyz`
<energizer>
if i use `(import (fetchTarball nixpkgsUrl) {}).writeText "hello" "hello world"` and nix-build it, is that nixpkgs supposed to get a gc root?
<bqv>
no
<energizer>
what are the rules for that stuff?
<{^_^}>
[nixpkgs] @github-actions[bot] pushed 25 commits to staging-next: https://git.io/JtYWm
<{^_^}>
[nixpkgs] @github-actions[bot] pushed 26 commits to staging: https://git.io/JtYWY
<bqv>
the result won't depend on nixpkgs so it's not a dep of a root
<bqv>
nixpkgs will be cached, probably, because of nix magic, but that's not to do with the nix store
Unode has quit [Quit: Off it goes]
<DigitalKiwi>
idk nix 2.4 downloads 20MB every time i do anything...
<energizer>
i'm trying out an experiment of gcing very frequently but it's getting annoying because it keeps downloading a pinned nixpkgs
Unode has joined #nixos
<energizer>
or at least, i think that's what it's doing
<energizer>
it says unpacking [nixpkgs url] and nethogs says it's working hard
<bqv>
lol if you want stuff to not get redownloaded, pin it and add it as a root yourself? or make it a transitive dependency of something you know will be a root
<bqv>
or better yet, use flakes
<DigitalKiwi>
i'm surprised you didn't mention guix
<energizer>
too many people saying critical things about flakes, i'm worried
<energizer>
so probably gonna wait until the ratio improves
<bqv>
DigitalKiwi: no guix is probably worse for that
<DigitalKiwi>
*shocked*
<c3hou>
@xinau whoops I didn't realize it was `fetchzip` and not `fetchurl`, my bad
<bqv>
hm, just realised matthewcroug*an has tried to talk to me a few times but i've had him on ignore for the last two months
<ishan40>
gdm is not showing a wayland session but I have `displayManager.gdm.wayland = true` in my configuration. I am also using the open source nvidia driver.
<bqv>
transmission has some crazy ass ETAs going on
<bqv>
flipping between 6 minutes and 4 days
<bqv>
that's not cool
<DigitalKiwi>
adisbladis: oh
<adisbladis>
zfsUnstable may mean unstable again at some point if there is some wortwhile new feature
<DigitalKiwi>
nixops-digitalocean-unstable-_-kiwi
teto has joined #nixos
graf_blutwurst has quit [Client Quit]
malook has joined #nixos
<Bene>
Hello, im want to create a docker container from an angluar project. Normaly i would just create a docker file with
<Bene>
FROM nginx
<Bene>
COPY ./ /usr/share/nginx/html
<Bene>
EXPOSE 80
<Bene>
How would i build the same docker container with nixos? I did find node2nix, but i did not get it to work and i did not find a guide for creating a docker container.
<Bene>
Thanks for your help.
graf_blutwurst has joined #nixos
<kunrooted[m]>
using someone's config as a starting point is a suicide?
<kunrooted[m]>
just a quick question since I've decided to reimplement something to my personal needs lmao
<bqv>
depends on how complex it is
<adisbladis>
Starting from my config would _not_ be a good idea for example ;)
<kunrooted[m]>
but I don't want to ctrl-c ctrl-v but rather literally rewrite it and keep things which I need and things I don't, during the install it will break anyway if I fuck up somewhere
<kunrooted[m]>
* but I don't want to ctrl-c ctrl-v but rather literally rewrite it and keep things which I need and things I don't, during the install it will break anyway if I f*ck up somewhere
<bqv>
damn that's snazzy.
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JtYBE
<adisbladis>
Bene: Did you look at dockerTools.buildLayeredImage
<bqv>
hlissner: good sh*t
graf_blutwurst has quit [Client Quit]
<DigitalKiwi>
hlissner uses nixos but there's no doom-emacs-nix(?)
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JtYB2
<kunrooted[m]>
sure, gonna give it a shot
<DigitalKiwi>
adisbladis: i guess i meant that hlissner doesn't make one (i saw that one the other day and then read the readme for nix-straight.el and forgot about it lol)
<bqv>
kunrooted[m]: that repo is actually his literal dotfiles repo converted to nixos, i think if you base your stuff on it i'd squash some of the history :p
<Bene>
adisbladis: Hey, thx. Yes i did lock at that. But the documentation is very minimal and i did not find out how to use it.
<ashkitten>
supersandro2000: if anything, given that i'm a maintainer of this package and have verified that it runs, you would be fine to merge it, yet you only leave a useless comment validating what is already known.
<{^_^}>
[cabal2nix] @peti pushed to ci « Drop build instructions for stack. »: https://git.io/JtY0j
tazjin has quit [Quit: Byeee]
<supersandro2000>
ashkitten: to verify that the hash is correct and you did not leave in any old substituteInplace in. Also FYI Darwin support is broken. I am going to add you to the blocklist for such reports.
<{^_^}>
[cabal2nix] @peti pushed to ci « haskell-ci: add debug code »: https://git.io/JtYEE
<ashkitten>
supersandro2000: you didn't build it for darwin though.
<ashkitten>
you built it for x86_64
<ashkitten>
x86_64-linux*
<supersandro2000>
I did
<supersandro2000>
1 package marked as broken and skipped: dolphinEmuMaster
<supersandro2000>
but I stopped posting such comments
<supersandro2000>
because they are noise
<ashkitten>
posting build results is noise unless there is a problem with the build
<ashkitten>
unless your results conflict with ofborg's, it's needless
<supersandro2000>
ashkitten: for amd64 if the PR only contains one package you are right. Sadly ofborg does not fail loudly so I or anyone else could easily miss the grey status. I added you to the list to not receive such reports anymore.
<ehmry>
is there a way to never generate or install any bash/fish-completions, ever
<ashkitten>
supersandro2000: if you have an issue with ofborg, that's something that can potentially be fixed, not license for you to replicate a part of nixpkgs infrastructure by yourself
Bene has quit [Quit: Ping timeout (120 seconds)]
ManiacOfMadness has joined #nixos
<adisbladis>
So much this ^
rmcgibbo has quit [Quit: Connection closed]
<supersandro2000>
ashkitten: also I would appreciate it if you could give me constructive criticism instead. Something along the lines like if I could dive into a way to not post reports that are duplicated with ofborg would be really appreciated
<{^_^}>
[nixpkgs] @vcunat pushed to master « libiscsi: work around i686-linux build problems »: https://git.io/JtYzU
<supersandro2000>
ashkitten: I am not replicating ofborg. Actually I heavily rely on it to label the amounts of rebuilds. The problem it has is that builds take a longer time, you can't get notified if they are done, they fail silently, logs disappear after some time and you can't view them anymore and for darwin it usually takes a very long time so that anything is
<supersandro2000>
build.
<ashkitten>
supersandro2000: what about my criticism was not constructive? i told you why what you are doing is bad and i told you an alternate approach that would be better
<supersandro2000>
I also started a tampermonkey script to get informed about your individual likings to receive less notifications
<ashkitten>
you should help improve our existing infrastructure instead of taking it into your own hands
<supersandro2000>
you started that my work is useless and pointless. Not really how it should start in my opinion.
sarcasticadmin has quit [Read error: Connection reset by peer]
<{^_^}>
[nixos-homepage] @garbas merged pull request #663 → Experiment with different fonts → https://git.io/JLyFH
<{^_^}>
[nixos-homepage] @garbas pushed 3 commits to master: https://git.io/JtYz4
<{^_^}>
[nixos-homepage] @garbas pushed 0 commits to experiment-with-different-fonts: https://git.io/JtYzB
ManiacOfMadness has quit [Remote host closed the connection]
<adisbladis>
This was constructive
Graypup_ has joined #nixos
<ashkitten>
i said the build result messages are pointless because they replicate our existing infrastructure in a noisier and more intrusive way
ManiacOfMadness has joined #nixos
<supersandro2000>
if you think so. I have a different view and would like to stop this discussion here.
<ashkitten>
so you're asking for constructive criticism and not being receptive when you receive it
<ashkitten>
which makes me feel like your request is not in good faith
<ehmry>
knowing about build failures is useful, getting notifications that something builds can be annoying if you have already tested it
<energizer>
the criticsm couldve been friendlier tbh
<ashkitten>
friendly is not the same as constructive
<ashkitten>
and i agree, i could have been less harsh
<supersandro2000>
I told you that I try to find a way to not post replicates with ofborg.
kueckieben[m] has joined #nixos
<ashkitten>
supersandro2000: one thing that would be truly helpful would be if you tested execution of the program
<energizer>
are there a lot of programs whose tests pass but are actually broken?
<ashkitten>
nextcloud-desktop is currently broken in nixpkgs
<ashkitten>
or nextcloud-client, whatever it's called
<ashkitten>
i've got a local overlay that updates it to a newer version, haven't gotten around to pring that
<ashkitten>
and "are there a lot" isn't really valid in package management. it's not about how often you're right, it's about following procedure for the cases where you're not.
<energizer>
the ui opens for me (nixos-unstable), what's the failure mode?
<ashkitten>
it was some bug with the cache going from 3.0.2 to 3.0.3
<ashkitten>
i don't remember right now
<ashkitten>
point is, it built fine, but broke on people's machines
<{^_^}>
[nixpkgs] @zimbatm pushed commit from @max-wittig to master « gitlab-runner: 13.7.0 -> 13.8.0 (#110280) »: https://git.io/JtY2L
<bqv>
adisbladis: lmao
<adisbladis>
bqv: ?
_d0t has quit [Ping timeout: 240 seconds]
<bqv>
constructive
<adisbladis>
Ah, right
<bqv>
supersandro2000: fwiw, i don't care that you've made your infra, i don't think that's an issue at all, and if it helps i'm definitely for it. but yes, the inconsequential notifications are annoying
<adisbladis>
It's much better to collaborate on testing infra than having random contributors operate some random piece of a testing rube goldberg machine
<adisbladis>
Because that's where we're heading if this continues
<bqv>
nixpkgs is a community project, the community should be allowed to do things. specifically, i feel innovation is stifled if every effort has to be funnelled into a modification of an existing system. e.g. i'm constantly waiting for the next iteration of nix-likes, because while we can all modify nix, i don't think it's realistic to see nix modified into something optimal
<bqv>
so yeah, i'm quite happy when i see people breaking off and doing their own thing. obviously, if it's shit, it won't go anywhere, but if it happens to be good, magic happens
malook has quit [Quit: malook]
<energizer>
eg the proliferation of deployment tools is probably a good thing during early exploration phase
<bqv>
absolutely
<eyJhb>
Can anyone run `docker run --rm -it -v $(pwd):/src alpine` for me, and tell me if it works + docker version?
lsix1 has quit [Read error: Connection reset by peer]
rauno has quit [Remote host closed the connection]
graf_blutwurst has quit [Ping timeout: 260 seconds]
<c3hou>
Since I'm using flakes I have `nix.package = pkgs.nixFlakes` and `nix.extraOptions = "experimental-features = nix-command flakes` in my NixOS config. I've found on occassion that I want to run nix 2.3 (e.g. to build older projects), which I've managed to do using `nix-shell -p nix`, but is there a way to remove the `experimental-features` options using a command line option? I've found `--option` to *add* options, but haven't found
<c3hou>
I realized `nix-build --option experimental-features ''` does produce a warning, but it doesn't seem to affect proper operation after. Thanks dutchie!
<ross`>
I just spent about 8 hours reading Nix documentation and surfing examples. With more power comes more opinionation and diversity. I'm slightly overwhelmed with the number of User Repositories. Would it be possible for someone to recommend 3 examples that I might want to focus on that are more modern / up to date, multi system, and fleshed out (containing / organizing multiple examples of pretty much every
<ross`>
disctinct type of thing that might be present as part of a modular configuration?
<ehmry>
ross`: I would just start with something simple that you want to build and go from there, nix code isn't easy to read
ManiacOfMadness has quit [Ping timeout: 256 seconds]
<ross`>
ehmry: I'm looking for examples of organization more than anything else so that I can try to understand why certain methods were used. That way I can inspire a health approach to my own organization without realizing too far down the road that I should restructure things.
<infinisil>
bqv: Well NixOS generates multiple by default
<infinisil>
I also have two, an ed25519 one, and an rsa one
<bqv>
Heh, fair
<infinisil>
But I'm not sure why two would be required, other than preferring one type over the other (in which case you'd only specify that one)
fendor has quit [Ping timeout: 264 seconds]
<bqv>
Either way, why force users to pick one
<infinisil>
Well the module *does* become a bit more complicated, for what could be no benefit at all
ahmedelgabri has quit [Ping timeout: 264 seconds]
<bqv>
In my mind its less complicated. I sat there a few minutes trying to work out which key to use. Decided on modifying it so I didn't have to decide :p
<infinisil>
Hmm I see, that's a good point
<bqv>
But fair
<infinisil>
Another reason for using all keys: ssh-keyscan also shows all of them
<infinisil>
E.g. run `ssh-keyscan infinisil.com`, it shows both rsa and ed25519
<bqv>
Indeed
<immae>
Are there guidelines to submit a new package that is written as a flake? Should we use compatibility layers (the one from edolstra), or should it be converted to a "regular" one first, or...?
<bqv>
immae: submit where..? flakes can just go on github
<infinisil>
bqv: Alright I'm convinced, though I think the type of the hostkeys should probably rather be `coercedTo str singleton (listOf str)`
<bqv>
infinisil: ha, fair!
<immae>
bqv: sorry. submit on the official nixpkgs repo
<infinisil>
Then it's backwards compatible too :)
<infinisil>
bqv: Feel free to PR then!
<bqv>
Oh good lord, I need to work out if I can do that from a subtree
<infinisil>
Let's use `coercedTo str (s: { "" = s; }) (attrsOf str)` then
<bqv>
Heh, ok
<immae>
ok :p
<infinisil>
Hmm, or maybe `s: { default = s; }`
<infinisil>
That doesn't merge well though, darnit
evernite has joined #nixos
<{^_^}>
[nixpkgs] @github-actions[bot] pushed 59 commits to staging-next: https://git.io/JtYXX
<{^_^}>
[nixpkgs] @github-actions[bot] pushed 60 commits to staging: https://git.io/JtYX1
<chvp>
did anyone else's gtk theming break recently? (on nixos-unstable-small)
respawn_ has joined #nixos
meh` has joined #nixos
tazjin has joined #nixos
<bqv>
infinisil: heh it worked, I can subtree push to my local checkout and cherrypick the updater
<bqv>
s/r$/s
<eyJhb>
I am trying to do some module (yet again), but I can't really spot the error I am making here - https://termbin.com/fgwu , I get the error `error: The option value `nodes.srtutti.configuration.systemd.services' in `/home/eyjhb/projects/nixos/nixus-setup/channels/nixpkgs-test/nixos/modules/services/misc/mautrix.nix' is not of type `attribute set of submodules'`
<bqv>
chvp: actually, yes, if recently = last week or so
<aanderse>
Mic92: i'm having an issue with nix_direnv (i just started using it for the first time) - i install as directed onto nixos system level, but once i add "source /run/current-system/sw/share/nix-direnv/direnvrc" to ~/.direnvrc the shellHook in my shell.nix stops working - i don't see any errors or anything indicating a problem :(
<bqv>
eyJhb: I spy nixus :D
<chvp>
bqv: recently = today for me
<eyJhb>
bqv: Noooo. I could NEVER :D (yeah, use it to manage everything :p )
<c3hou>
I've been trying to package some proprietary printer drivers (https://git.io/JtYo4) and managed to get it recognized by CUPS, but the printer doesn't budge despite CUPS saying "Job completed." Does anyone have any idea what I could do to try to debug this?
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JtYyS
alexarice[m] has joined #nixos
<exarkun>
Does nixos work on lower spec systems? I put 20.03 onto a t520 a few days ago. It felt a little sluggish but it was working. But just now I tried `nix search cd rip` which ran with no output for 2 or 3 minutes and then the system wedged.
<exarkun>
Alright. I'll try with some swap next time. Multi-minute `nix search` w/ <=4GB physical ram seems like it may be practically unusable though..
<V>
<DigitalKiwi> guix isn't in nixpkgs :( ← I have a WIP branch that adds proper support for guile, and would make packaging guix significantly more feasible
<V>
it's just a matter of finding time and motivation to finish it
<bqv>
Nix experts: can I set nix options using env vars?
<bqv>
I guess I can use NIX_CONF_DIR actually
<V>
bqv: my experience with flakes so far has been being unable to use debugfs b/c it's locked behind some interface that isn't stable and can't be used from existing setups
<V>
so I have zero interest in it
<bqv>
Seems slightly shortsighted
<V>
flakes definitely aren't anywhere near production-ready AFAICT, I see really bad bugs in the impl on a fairly regular basis
medvid has quit [Ping timeout: 256 seconds]
<V>
"legacy" nix at least works most of the time :)
hoijui has quit [Ping timeout: 264 seconds]
<V>
if I want pinning, there's niv
hoijui_ has quit [Ping timeout: 264 seconds]
lsix1 has quit [Ping timeout: 264 seconds]
medvid has joined #nixos
lsix1 has joined #nixos
xeronull has joined #nixos
nwspk has quit [Ping timeout: 260 seconds]
qp12[m] is now known as p12[m]
<edef>
V: flakes are helping us becoming mainstream in the sense that we're now doing Google's "the old thing is deprecated and the new thing is not ready yet" :p
<niksnut>
V: what do you mean with debugfs? I don't see the connection with flakes...
<V>
niksnut: it was replaced with a flake forever ago
<V>
IIRC you have to use an old commit
<niksnut>
you mean dwarffs?
<V>
oh, that
<immae>
V: you can always include a flake from a legacy one
xantoz has quit [Ping timeout: 246 seconds]
<immae>
the reverse is more complicated, but this way is fine
<niksnut>
okay, but that's not actually a problem with flakes
<V>
immae: I tried all of the suggestions in the issue, and then gave up
<niksnut>
just a problem with *not* using flakes ;-)
<V>
immae: I don't usually use the wiki b/c it's normally so outdated that the information it holds is wrong
<immae>
This one is up to date
<V>
how am I to know that?
<edef>
niksnut: any second system that does not provide for gradual migration of existing large deployments will never be deployed, and be correspondingly full of undiscovered bugs
<immae>
I have the same case as you less than two weeks ago and it went all fine ;)
<V>
it's been cast as "not official" for so long
<niksnut>
edef: flakes support gradual migration just fine
<niksnut>
it's just that the dwarffs repo doesn't do that
<V>
the fact that it ended up being less effort for me to trawl through the assembly w/ objdump and gdb's instruction view than to get this to work (which I still have not yet) should say a lot
<V>
"this" being debugging things
<V>
niksnut: the dwarffs readme tells you to import a nonexistent module.nix
<immae>
V: I don’t know what to tell you, I was completely new to flake and was able to write a flake and use it in my legacy project in less than a day
<V>
immae: I'm not trying to write flakes, though. I'm trying to use a module that's only available as a flake
<niksnut>
V: strange argument. If you don't use the available mechanism for installing something, then it's not surprising that it's hard to install.
<edef>
my entry path is "i have been using Nix just fine for the past half-decade, i am just trying to install another piece of Eelcoware that appears to be for Nix users"
ktosiek has quit [Quit: Ping timeout (120 seconds)]
<V>
niksnut: the argument that the documentation is wrong?
<edef>
i think that last time i tried to install dwarffs i just forked it and removed the flake curft
<edef>
*cruft
nwspk has joined #nixos
ktosiek has joined #nixos
<immae>
V: yes, it’s the same, apart from the fact that the flake was writen by myself "for fun" you’re in the same case. You should give a try at the section about legacy systems in the wiki page I linked...
<V>
if I'm a new NixOS user and I try to install dwarffs, I'm goign to end up on your readme, follow the instructions, and then be confused why it's broken
<V>
and then maybe go in the issues, follow the instructions there, find it still doesn't work, and then probably give up
<{^_^}>
[nixos-homepage] @github-actions[bot] pushed commit from GitHub Actions to master « Update flake.lock and blogs.xml [ci skip] »: https://git.io/JtYdq
<edef>
that's kind of the crux of it
srk has quit [Remote host closed the connection]
cransom_ is now known as cransom
srk has joined #nixos
<V>
immae: maybe I'll take a look at this if I actually get the time and energy to add more New Technology to my system. as it stands, that's not happening anytime soon
<etu>
niksnut: I have a question regarding if you don't use flakes... Say that I have a system that don't use flakes, but somehow I happen to have a /etc/nixos/flake.nix for other systems that use flakes. Why does this break the non-flake system? :)
<etu>
waaait, we have appgate-sdp in nixpkgs? Since when? I need that garbage and attempted to package it more than once.
xantoz has quit [Ping timeout: 240 seconds]
<V>
immae: I've spent hours trying to get this to work in the past, and this was when I actually had energy to do that kind of thing still
<immae>
V: ok, it’s up to you, I won’t force you to do anything :)
xantoz has joined #nixos
<V>
idk, is user experience actually taken into account when these things are developed?
meck has joined #nixos
<cransom>
i'm still a bit hung up on making a change to a repo that invalidates the documents on how to use it, and it's not fixed, or addressed
respawn__ has quit [Quit: Leaving]
xeronull has quit [Quit: Connection closed]
<immae>
V: It’s nothing new that lack of documentation is a big flaw of nixos. I believe that as long as noone is dedicated to that kind of work your best doc is the source..
mcaju_ has joined #nixos
<edef>
the grand promise of flakes was always "user experience"
<edef>
perhaps i forgot to read between the lines, i must've missed the "[blaming] user[s for their] experience"
EmoSpice has quit [Ping timeout: 240 seconds]
<V>
immae: there's a bit of a difference between "the docs are a bit thin here, we acknowledge that this sucks, you need to look through the source code to figure some stuff out" and "this software has been changed in a backwards-compatible manner to make use of a technology that isn't even released yet, without the maintainer even commenting on the issue where people are asking for a stopgap or migration path"
<V>
backwards-incompatible*
<sterni>
the major issue with flakes is that they are an _experimental_ feature which is not treated like one which means flakes essentially is something taht is not ready for release but already used as if it were released
sangoma has quit [Quit: WeeChat 3.0]
<V>
yeah, that
<niksnut>
V: again, that's an issue with dwarffs, not flakes
<gchristensen>
sterni++
<{^_^}>
sterni's karma got increased to 1
TonyTheLion has joined #nixos
<sterni>
at this point it wouldn't even be possible to say flakes doesn't cut it because it would already break dozens of projects using it
ahmedelgabri has joined #nixos
<c3hou>
Looks like dwarffs's readme just got updated too lol
<V>
lol
<niksnut>
exactly
<niksnut>
problem solved
<sterni>
also i don't think it's particularly good to be running around telling people to use this and that if it says „Note: this program is EXPERIMENTAL and subject to change.“
<cransom>
(startling similar story to the nix commands)
<sterni>
indeed yeah
<sterni>
either these kinds of things cease to be experimental meaning we can't improve them anymore
<V>
niksnut: hey, at least it's not pretending to be usable to the vast majority of people using non-flake NixOS systems now
<sterni>
or we ruin the experience for users in the future by breaking stuff for them
<evernite>
I'm having trouble using `pkgs.python3Packages.buildPythonApplication` to build a simple command-line utility. I keep getting a `ModuleNotFoundError` for a dependency I have defined in `propagatedBuildInputs`. Any help would be much appreciated.
<evernite>
c3hou: djanatyn: Yes I am also able to run `nix-build` successfully, but when I try to run `./result/bin/load` I get an "ModuleNotFoundError: No module named 'joblib'"
zupo has joined #nixos
<djanatyn>
i'm not able to run your code because you didn't include the implementation, i get different errors
<tad-lispy[m]>
Hey, folks! I'm learning Nix and want to apply it to web development. Here is my playground: <https://gitlab.com/tad-lispy/nix-for-web-development-playground>. It works in principle, but every time I build it all node dependencies are getting processed and it takes about a minute. Same for `nix-shell`. Is there a way to speed it up?
<lordcirth>
I noticed that the time to do a no-op rebuild of my system decreased greatly when moving to flakes. from ~15s to like, 2s.
Jd007 has joined #nixos
<tad-lispy[m]>
ronthecookie: well, that's my question I guess.
<sterni>
tad-lispy[m]: the core problem is evaluation time of the generated sets of node2nix which always need to be evaluated when you run nix-shell or nix-build
<sterni>
tad-lispy[m]: you can use lorri which can do these evaluations in the background for you and updates your environment with direnv
<tad-lispy[m]>
Really? From the output it seems like it's operations on the files, not evaluation of nix.
<lordcirth>
Current nix has to re-create all .drvs, then build any remaining. With flakes, even the drv builds can be skipped if the source hasn't changed, since there is no other input.
<sterni>
what c3hou says
<c3hou>
and what sterni says
<sterni>
tad-lispy[m]: lorri also keeps the built outputs around
rnhmjoj_ has quit [Remote host closed the connection]
rb2k has quit [Ping timeout: 240 seconds]
<tad-lispy[m]>
But what do I put in `shell.nix` then?
<tad-lispy[m]>
If I just `lorri init` then it doesn't have all the node stuff.
<tad-lispy[m]>
I get the env I want when I do `nix-shell default.nix`. Only it's slow.
<tad-lispy[m]>
And it doesn't integrate with my ZSH like Lorri does. It gives me a plain bash.
rb2k has joined #nixos
<c3hou>
I haven't personally worked on projects without a `shell.nix` yet, but since `nix-shell` works by reading your `default.nix` lorri should be able to too
<c3hou>
Try deleting the `shell.nix` and using lorri's `.envrc`, start a lorri daemon in another terminal with `lorri daemon` then run `direnv allow`
<c3hou>
I see. My mistake, lorri doesn't try to get the default shell from default.nix.
<tad-lispy[m]>
But the direction is good.
<tad-lispy[m]>
It's super fast now to `cd ..` and `cd -`.
rb2k has quit [Read error: Connection reset by peer]
<tad-lispy[m]>
And I have my beloved zsh :D
sangoma has joined #nixos
<chrispickard>
if you are using home-manager it has a module for a lorri service which runs lorri as a systemd user service which is very nice
<c3hou>
I'm not sure the correct way to get the shell for a `mkDerivation`, but for now instead of a symbolic link you might also try having `import ./.` in your `shell.nix` instead
KarlJoad has joined #nixos
<chrispickard>
I forgot lorri even needed a daemon because I've been using it for so long
<c3hou>
tad-lispy: also np, glad it works for you!
<KarlJoad>
I need to test a piece of software, and I need an arduino. Is there a way to simulate/emulate/virtualize one on NixOS?
ManiacOfMadness has quit [Ping timeout: 272 seconds]
<{^_^}>
[nixpkgs] @holymonson opened pull request #110357 → darwin.CoreFoundation: update to swift newest version → https://git.io/JtYxw
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
kalbasit has joined #nixos
<bbigras>
If I set an overlay in my configuration.nix with `nixpkgs.overlays = [];`, will the overlay be working when I switch to this config the first time or does the overlay need to be installed first and will work the next time?
<{^_^}>
[nixpkgs] @fabaff opened pull request #110363 → python3Packages.vsure: init at 1.6.1 → https://git.io/JtOea
<infinisil>
bbigras: Works from the first switch
<bbigras>
infinisil: thanks!
<infinisil>
Though it only applies to the nixos config, not to general nix commands
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JtOvC
<abbe>
if I want to host local copies of nixpkgs, and nixos manual, which derivations should I refer to ? e.g. for hosting nix manual in my local web server, I'm referring to "${config.nix.package.doc}/.../"
<pbb>
abbe: so basically you need to either use flakes or import (pkgs.path + "/nixos/release-small.nix") and (pkgs.path + "/pkgs/top-level/release.nix")
<abbe>
okay
<kalipso>
hello! how can i crosscompile a flake for aarch64 using "nix build"? I tried "nix build --arg crossSystem '{ config = "aarch64-linux"; }' .#flakeName" but it still seems to build for x86_64-linux
<pbb>
nix-build "<nixpkgs/nixos/release-small.nix>" -A nixos.manual.x86_64-linux
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JtOf7
<pbb>
abbe: those commands I sent don't produce the correct outputs, but maybe you can find the correct commands by looking at nixpkgs/flake.nix and nixos-homepage/flake.nix
Darkmatter66 has joined #nixos
EmoSpice has quit [Ping timeout: 240 seconds]
<abbe>
yep, I'm going through the stuff, I just need some path which I can embed in my nginx configuration :)
endformationage has joined #nixos
<abbe>
i wonder if i can utilize nixos-homepage and its binary cache
<abbe>
and then reference to it
kalipso has quit [Quit: Ping timeout (120 seconds)]
chrispickard has quit [Quit: Ping timeout (120 seconds)]
chrispickard has joined #nixos
kalipso has joined #nixos
logand has joined #nixos
rmcgibbo has joined #nixos
ahmedelgabri has quit [Ping timeout: 260 seconds]
philipp[m] has joined #nixos
hoijui has quit [Ping timeout: 272 seconds]
hoijui_ has quit [Ping timeout: 272 seconds]
kalipso has quit [Client Quit]
simba1 has joined #nixos
<{^_^}>
[nixpkgs] @fabaff opened pull request #110368 → python3Packages.python-wink: init at 1.10.5 → https://git.io/JtOJS
Darkmatter66 has quit [Ping timeout: 246 seconds]
vidbina has quit [Ping timeout: 256 seconds]
ahmedelgabri has joined #nixos
rajivr has quit [Quit: Connection closed for inactivity]
<philipp[m]>
I'm trying to enable gst support in nheko. Can somebody explain to me why it isn't enough to add ` (with gst_all_1; [ gstreamer gst-plugins-base gst-plugins-good gst-plugins-bad ])` to buildInputs?
<dmj`>
Hey everybody, I'm trying to get libredirect.so to rewrite filepaths in syscalls of a binary. I have an isolated test here that compiles some haskell code, adds the libredirect.so shared object, and runs strace on it. The file in question should be read and printed to stdout from the nix store (its path having been rewritten by libredirect.so). Instead, according to strace, the libredirect.so is loaded, but the path is not
<dmj`>
rewritten.
<dmj`>
Here is libredirect getting loaded openat(AT_FDCWD, "/nix/store/2a2hvasykj69sih114jbd1dnyxs46590-libredirect-0/lib/libredirect.so", O_RDONLY|O_CLOEXEC) = 3
<{^_^}>
[nixpkgs] @svanderburg opened pull request #110381 → Update all skaware packages (s6, s6-rc, execline etc) → https://git.io/JtOYK
<rmcgibbo>
I'm connected via webchat.freenode.net with this IRC in a firefox tab. Whenever I am away from this tab for an extended period of time, it appears that I disconnect from the channel.
<rmcgibbo>
Can anyone recommend a different IRC client that they like or recommend? I'm new to IRC (I use the, shall we say, "proprietary alternatives") in other capacities, so I haven't explored other IRC clients yet.
<V>
rmcgibbo: this is a general problem with IRC clients, you need a bouncer (either hosted by someone else, or running on a server of your own) to have a persisted connection
<V>
Popular choices include (any client of your choice) + ZNC, weechat + weechat-relay/weechat-android/glowingbear, Quassel, just running weechat/irssi in tmux on a server, the lounge
<bbigras>
rmcgibbo: you can use matrix
<bbigras>
it's simple and you can have it on your phone too
<bbigras>
just need a matrix account. and join the channel using freenode
<V>
Matrix is terrible
<bbigras>
I don't think so
<V>
Ask the freenode staff
<V>
You'll get a different answer :p
<bbigras>
whatever
<rmcgibbo>
I'd prefer not to have to run any of my own infrastructure. I'll take a look at the options that ya'll pointed to.
<{^_^}>
[nixpkgs] @ck3d opened pull request #110383 → rust-bindgen: Resolve "/usr/bin/env bash" to fix usage in Nix build → https://git.io/JtOOI
<V>
You don't have to see mass join/quit spam whenever the bridge goes down
<bbigras>
matrix is a lot simpler than your suggested solution
<V>
I suggested several solutions, some of them very simple
<cransom>
i like matrix, but i don't use it as an irc bouncer.
<V>
I'd suggest using irccloud if you don't want to host anything. You have to pay, but it's one of the best native solutions there are
<Nickli>
if you don't mind there's always The Lounge
<jess>
hello
<jess>
did someone say freenode staff
<V>
If you use matrix you're not really experiencing IRC for what it is, and rather using a matrix chat room which happens to have messages from IRC routed into it
<V>
jess: hi yes
<V>
I hear you love matrix
<jess>
MATRIX SUCKS SO BAD
<V>
<3
<jess>
:3
<jess>
they did deploy a change for us today though, so i'm inclined to be grateful to the bridge operators
<rmcgibbo>
Is "TheLounge" something where there's a nix community server that people share, or does everyone run their own server?
<Nickli>
it's a web client for IRC
<V>
It's a self hosted thing. I don't know if nix-community has something like this
rmcgibbo[m] has joined #nixos
<V>
Maybe go ask in #nix-community, I believe they have servers for stuff
<rmcgibbo[m]>
I've tried the matrix thing, since it took all of 5 seconds to sign up and sign in. If it seems terrible to me, I'll move onto the next suggestions.
<rmcgibbo[m]>
Thanks for the help!
<Nickli>
guess it was 4,5seconds to long
simba1 has quit [Ping timeout: 260 seconds]
logand has quit [Ping timeout: 246 seconds]
cstrahan has quit [*.net *.split]
risson has quit [*.net *.split]
jtle has quit [*.net *.split]
teehemkay has quit [*.net *.split]
sdier has quit [*.net *.split]
ncl3 has quit [*.net *.split]
higherorder has quit [*.net *.split]
feepo has quit [*.net *.split]
bgupta has quit [*.net *.split]
jfhbrook has quit [*.net *.split]
rodarmor has quit [*.net *.split]
nand0p has quit [*.net *.split]
johanot has quit [*.net *.split]
TheNumb has quit [*.net *.split]
sephii has quit [*.net *.split]
TheNumb has joined #nixos
ncl3_ has joined #nixos
nand0p has joined #nixos
teehemkay has joined #nixos
johanot has joined #nixos
sephii has joined #nixos
sdier has joined #nixos
jfhbrook has joined #nixos
cstrahan has joined #nixos
rodarmor has joined #nixos
higherorder has joined #nixos
risson has joined #nixos
asymmetric has joined #nixos
asymmetric has quit [Changing host]
feepo has joined #nixos
<joepie91>
<V> If you use matrix you're not really experiencing IRC for what it is, [...]
bgupta has joined #nixos
lsix1 has quit [Ping timeout: 240 seconds]
<asymmetric>
does the new nixUnstable have something like `nix-build -K`?
<joepie91>
that seems a little disingenuous. what exactly is missing there?
<V>
The actual UI of element is fine
simba1 has joined #nixos
<asymmetric>
i anm trying to inspect the logs of a failed rust build
<joepie91>
like, yes, the reliability/performance issues are known, but that seems to have very little to do with "experiencing IRC for what it is"
<{^_^}>
[nixpkgs] @dasJ opened pull request #110384 → procmail: Drop the package → https://git.io/JtOOa
evernite has quit [Ping timeout: 256 seconds]
<V>
joepie91: there's also how it deals with multiline messages
<V>
If your line is too long, it gets turned into link and has to be clicked on to view it
<V>
Into a link*
<asymmetric>
ah, `nix-build -L`
<rmcgibbo[m]>
asymmetric: I think I use `--print-build-logs` for that, or `nix log {path to failed .drv}`.
<joepie91>
V: yes, as is customary on IRC for very long pastes, you use a pastebin
<V>
joepie91: no, if you write a single line that is too long
<asymmetric>
rmcgibbo[m]: right, `nix build -L` is the shorthand
<V>
There's a difference here
<joepie91>
V: it certainly doesn't do that for me, so I'm not sure what that's based on
<joepie91>
long lines just get cut up into multiple messages, as per message limit
<asymmetric>
nix log xyz.drv is great!
<asymmetric>
rmcgibbo[m]++
<{^_^}>
rmcgibbo[m]'s karma got increased to 1
not-joepie91 has joined #nixos
<V>
joepie91: I'd not be surprised if you've never seen the "[person] has sent a very long message. Click [link] to view it." since it's only visible to people actually on IRC
<joepie91>
V: I run an IRC client in parallel for other reasons, and so I am well aware when it sends that. it happens when people sent many-line messages, not when they send long lines.
<joepie91>
send*
* colemickens
looks at which room we're in
* colemickens
points at #nixos-chat
<joepie91>
V: plus I use a different bridge from most people so I actually get all of the links from people who use the matrix.org bridge.
<joepie91>
ah, right.
<joepie91>
let's move there.
<V>
Nah, I have work to do
fresheyeball has joined #nixos
<fresheyeball>
is there a way to get a reference to an arguement set?
<{^_^}>
[nixpkgs] @holymonson opened pull request #110385 → darwin: drop packages maloader and opencflite → https://git.io/JtOOh
ManiacOfMadness` has quit [Ping timeout: 240 seconds]
<joepie91>
fresheyeball: as in, get the full set of arguments for a function that looks like `{ foo, bar }:`?
<fresheyeball>
yes
<joepie91>
fresheyeball: in that case, `{ foo, bar }@allArgs:`
<joepie91>
where `allArgs` contains the full set
<joepie91>
IIRC
<joepie91>
(or alternatively just `allArgs:` if you don't care for unpacking it at all)
<dmj`>
I think libredirect is only for buildtime, not runtime ...
<matthewcroughan>
You mention that it would be simpler, that you'd just use mdadm.
<matthewcroughan>
Because I have such buggy UEFI that I literally can't even install via UEFI on my Motherboard.
<matthewcroughan>
dmesg would report that UEFI Runtime has been disabled
<{^_^}>
[nixpkgs] @jonringer opened pull request #110390 → vimPlugins.vim-hcl: init at 2017-10-10 → https://git.io/JtOnC
<etu>
matthewcroughan: So mdadm have different types of metadata, different kinds stores the metadata in different places of the partition. "either at the end (for 1.0), at the start (for 1.1) or 4K from the start (for 1.2). '1' is equivalent to '1.2'"
jgarte[m] is now known as jgart[m]
<etu>
matthewcroughan: So this makes a difference for BIOS to be able to detect the boot partition
<etu>
matthewcroughan: Because if there's raid metadata in the beginning of the partition, well, then bios won't care :)
<etu>
matthewcroughan: I've had motherboards like that, I usually figured it out. I had all kinds of weird issues (including ones that doesn't support fat32, only fat16) or ones that only accepts the Microsoft bootloader path :p
marcusr_ has joined #nixos
buddy2003 has joined #nixos
marcusr_ has quit [Remote host closed the connection]
lsix1 has joined #nixos
ManiacOfMadness` has quit [Ping timeout: 265 seconds]
costrouc has joined #nixos
<matthewcroughan>
etu: interestingly, the boot iso for nixos works and boots in EFI mode
<matthewcroughan>
though I can't write to UEFI via NixOS' installer.
<matthewcroughan>
once booted, the OS can't write anything to UEFI
<matthewcroughan>
etu: is there an example guide of using mdadm to mirror the boot, as you've suggested? Since I want to set up what you've done, but on BIOS instead ;D
<{^_^}>
[nixpkgs] @vbgl pushed commit from @sternenseemann to master « ocamlPackages.mirage-crypto*: 0.8.8 -> 0.8.10 »: https://git.io/JtOnx
hoijui_ has quit [Ping timeout: 265 seconds]
hoijui has quit [Ping timeout: 265 seconds]
<etu>
matthewcroughan: Something like this... kinda: https://alchemycs.com/2010/08/mirror-ubuntu-boot-drive/ but with some modifications. So you need to partition both your disks to be identical, with sda1 and sdb1 for boot and sda2 and sdb2 for zfs.
<bbigras>
is boot mirror possible with systemd-boot?
marcusr_ has joined #nixos
<matthewcroughan>
On bios?
<matthewcroughan>
I wasn't aware you could use systemd-boot on bios
drr3w has joined #nixos
<etu>
matthewcroughan: Then you kinda want something like this: "mdadm --create /dev/md1 --level=1.0 --raid-disks=2 /dev/sda1 /dev/sdb1" to create that raid, mkfs.ext4 /dev/md1 to create a filesystem to have on /boot
<etu>
bbigras: I don't think so
<bbigras>
matthewcroughan: no with uefi
<bbigras>
etu: thanks
<etu>
bbigras: It can be done, grub doesn't really support it either, it's more a hack in the nixos module to install it to two places. So the same thing is probably possible to implement for systemd-boot as well.
<matthewcroughan>
Yeah systemd-boot has no mirror functionality, at least in nix's exposed options
<bbigras>
etu: uh. TIL. thanks
<matthewcroughan>
etu: so how much different is this bios method from your uefi method?
<matthewcroughan>
I just replace the esp partitions with mdadm?
<etu>
matthewcroughan: Pretty much yeah, and then no mirrored boots in grub
ManiacOfMadness` has joined #nixos
<matthewcroughan>
oh, how come there'd be no mirrored boot option in grub?
<etu>
matthewcroughan: Because you're having a raid to mirror the partition between the disks with mdadm :)
samhza has quit [Remote host closed the connection]
<matthewcroughan>
and what is actually happening in the UEFI + ZFS + mirroredBoot option?
<matthewcroughan>
I thought ZFS was syncing that.
EmoSpice has joined #nixos
<etu>
I don't use ZFS on the boot partition, and EFI partitions can't be raided, which is why the mirroredBoots option for the grub module is a key part :)
samhza has joined #nixos
<matthewcroughan>
hmm, when you say it's a hack, what is it doing exactly?
<etu>
It copies the kernels and installs grub to more than one location
<etu>
So it registers as different boot options in UEFI
cfricke has joined #nixos
<matthewcroughan>
and how does it not fail to sync? That doesn't seem robust to me.
<matthewcroughan>
or is it literally just doing the equivalent of grub-install /dev/sda && grub-install /dev/sdb ?
<etu>
I usually don't have any issues with installing my bootloaders :p
<matthewcroughan>
if you enable "EFI Optimized Boot" in the BIOS, it'll actually halt the CPU instead of just crashing. LOL.
justanotheruser has joined #nixos
<noonien>
anyone got an example of how to best package a shell script with dependencies?
<energizer>
i want to have some better monitoring for my nix builds, such as what built, when, how long did it take. should i run hydra locally for this?
<matthewcroughan>
etu: to confirm, I want to make a 512M Boot partition at the start of each drive, right?
<etu>
matthewcroughan: yeah
<matthewcroughan>
etu: what about the "storage partition"
<etu>
matthewcroughan: I don't know all the flags and exactly what they do :D You want a partition at the start and a partition at the rest of the disk
abathur has quit [Quit: abathur]
cfricke has quit [Quit: WeeChat 3.0]
<{^_^}>
[nixpkgs] @viric opened pull request #110393 → dduper: init at 0.04 → https://git.io/JtOWv
<rmcgibbo[m]>
@noonien: What I've normally done is to make a simple derivation that copies the script into $out/bin and then calls wrapProgam to set the path to `"${lib.makeBinPath buildInputs}"`
<noonien>
i see, does nixpkgs not have a helper for this by any chance?
concept2 has quit [Ping timeout: 246 seconds]
concept24 is now known as concept2
<rmcgibbo[m]>
Perhaps the fancier way is to use buildInputs = [ makeWrapper ]; and makeWrapperArgs = [ --prefix $PATH ${stdenv.lib.makeBinPath buildInputsOrWhateverElse }"`
<rmcgibbo[m]>
* Perhaps the fancier way is to use `buildInputs = [ makeWrapper ];` and `makeWrapperArgs = [ --prefix $PATH ${stdenv.lib.makeBinPath buildInputsOrWhateverElse }"`
<rmcgibbo[m]>
* Perhaps the fancier way is to use `buildInputs = [ makeWrapper ];` and `makeWrapperArgs = [ "--prefix $PATH : ${stdenv.lib.makeBinPath buildInputsOrWhateverElse }"]`
<{^_^}>
[nixpkgs] @raboof opened pull request #110394 → node-packages: updates without further changes → https://git.io/JtOWZ
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<matthewcroughan>
etu: does the "partition type" of these things matter?
<rmcgibbo[m]>
@viric: If the project is being actively maintained, you could politely ask the maintainer if he or she would add a license?
<samueldr>
viric: in this case, look at the sub-projects, two of them have a license
<samueldr>
project descriptions, here Cargo.toml, often have license info
heyitsrama has quit [Remote host closed the connection]
<samueldr>
especially if they're meant to be public
<etu>
matthewcroughan: Not really, the /boot ones should probably be marked as boot or something. To be honest it was years ago I had a non-UEFI system
<etu>
matthewcroughan: almost, the metadata 1.2 won't work
<samueldr>
I only knew of the general ecosystem-packages license field
<viric>
samueldr: ah ok thank you!
<etu>
matthewcroughan: it needs to be 1.0 for mdadm
<samueldr>
(I had to figure out license info for a couple mrbgems that way)
<matthewcroughan>
Note: this array has metadata at the start and may not be suitable as a boot device. If you plan to store '/boot' on this device please ensure that your boot-loader understands md/v1.x metadata, or use --metadata=0.90
<matthewcroughan>
It did say this etu
<etu>
matthewcroughan: ah, 0.90 should be fine then :)
<matthewcroughan>
oh god, do I have to reboot? lmfao
<etu>
matthewcroughan: your pastebin seems to indicate that it's 1.2 so it needs to be changed regardless :)
<matthewcroughan>
mdadm: /dev/sdb1 is not suitable for this array. │···········
<matthewcroughan>
resource busy :D
<etu>
mdadm has a --stop flag
nwm_ has joined #nixos
<matthewcroughan>
mdadm: /dev/sda1 does not appear to be an md device │···········
<matthewcroughan>
So sorry about the tmux cruft at the end of my messages lol
<matthewcroughan>
`mdadm --stop --scan ` worked
dandart has joined #nixos
<matthewcroughan>
etu: if I did mkfs.ext4 /dev/md1, and my bios doesn't support ext4, aren't I screwed?
<matthewcroughan>
shouldn't I use fat32 to be safe?
ixxie has joined #nixos
<etu>
matthewcroughan: That's the thing with boot partitions in BIOS, bios doesn't have to support the filesystem
<etu>
matthewcroughan: Because the partition leaves some space at the beginning where the bootloader installs itself
nwm has quit [Ping timeout: 260 seconds]
<etu>
matthewcroughan: Which is why the metadata has to be at the end ;)
<matthewcroughan>
it's a privilege to talk to you, putting it in such simple terms.
<matthewcroughan>
<3 etu
<{^_^}>
etu's karma got increased to 30
<etu>
Happy to help :)
<matthewcroughan>
etu: so, just to be careful, I should totally mkfs.fat on this instead, right?
<adisbladis>
matthewcroughan: Nah, ext4 and btrfs are both supported by grub
sshow has quit [Remote host closed the connection]
<matthewcroughan>
yeah but my bios may not support booting from ext4, right? adisbladis
<etu>
matthewcroughan: In the past I always used ext2 or even ext3 on /boot, ext4 is also fine, since the bootloader is outside of the filesystem to read the filesystem to then boot it's fine
<adisbladis>
matthewcroughan: Your bios reads a grub stub from the first 512 bytes of the disk
cole-h has joined #nixos
<adisbladis>
Your bios doesn't need to support _any_ file system for this to happen
<matthewcroughan>
/dev/md1 contains a vfat file system
<matthewcroughan>
why would /dev/md1 contain anything?
<etu>
To be safe it's probably better to go with ext4 :)
<eyJhb>
If anyone else wonders at some point, one can do systemd.services."name".serviceConfig.ExecStartPre = "+"+(pkgs.writeShellScript "test" ''something''); :)
<adisbladis>
You can certainly use vfat, though there is no point to it
<adisbladis>
I'd default to btrfs, but ext4 is also a fine choice for a boot volume
<matthewcroughan>
adisbladis: out of curiosity, why would you use btrfs?
<adisbladis>
matthewcroughan: Because I like CoW file systems
<adisbladis>
I don't trust anything else
dev_mohe has joined #nixos
<dev_mohe>
Hi, does ` nix-build nixos -I nixos-config=nixos/modules/installer/cd-dvd/sd-image-aarch64.nix -A config.system.build.sdImage --argstr system aarch64-linux -j 1 --cores 1 --show-trace` fail for anybody else on master?
<dev_mohe>
with `ErroSysError~executing '/nix/store/ksfgyji1gnn5lzjgb03knlz7kav1wj8i-bash-4.4-p23/bin/bash': No such file or directory`
<matthewcroughan>
adisbladis: I'm bewildered by partitioning, really. This is my first real venture into it. It's kind of amusing how many different ways there are of doing things.
<adisbladis>
matthewcroughan: Once it clicks it's really not that hard :) You'll get there!
Zoom[m] has left #nixos ["User left"]
davidv7 has quit [Read error: Connection reset by peer]
TonyTheL1on has quit [Ping timeout: 240 seconds]
<energizer>
if i `nix-build --builders ssh://somewhere`, and then my connection goes down, will the remote machine continue building?
<matthewcroughan>
etu: what do you think about the ESP bit?
<matthewcroughan>
You mount two partitions, into /mnt/boot and /mnt/boot-fallback, in the md1 case, do I just mount /dev/md1 into /mnt/boot?
<matthewcroughan>
I kinda like this more, feels right to only manage one device like this.
<etu>
matthewcroughan: No ESP bit, but there's a bootable flag on legacy boot
<matthewcroughan>
? do I need to set something?
<etu>
matthewcroughan: And yes, you mount /dev/md1 to /mnt/boot and no fallback device
<matthewcroughan>
I already set bootable via cfdisk
<etu>
That's good enough :)
<etu>
Then I think it should work! But I haven't tested that exact setup for ages, and not with nixos :)
<matthewcroughan>
I think I'll make a blog post similar to yours about doing this, you can put it in the footer, or perhaps reference it in the part where you state mdraid would be simpler.
<aswanson>
how does one cajole nix into making a state directory under /var/lib when defining a new systemd service? I've looked at a couple examples in nixpkgs and don't see an explicit step where the directory is created
dandart has quit [Ping timeout: 264 seconds]
<dev_mohe>
aswanson: I think it's called systemd.tmpfiles
<dev_mohe>
* systemd.tmpfiles.rules
<aswanson>
dev_mohe: I need stuff in this directory to persist though, this is for a dedicated game server so this folder will hold stuff like game config and saves
<dev_mohe>
aswanson: You should get my answer confirmed by somebody else but I think the name is misleading and it actually keeps the data
<{^_^}>
[nixpkgs] @fabaff opened pull request #110399 → python3Packages.avion: init at 0.10 → https://git.io/JtOBz
<energizer>
why does this point to the nix executable and not a binary called nix-shell? % realpath $(which nix-shell) -> /nix/store/2wrsfc7nlhlbvffibwn6shdfja7vcjy5-nix-2.3.10/bin/nix
simba1 has quit [Ping timeout: 265 seconds]
simba1 has joined #nixos
zupo_ has joined #nixos
zupo has quit [Ping timeout: 246 seconds]
<dev_mohe>
energizer: likely because you only need one binary then. A binary can get it's called name by args[0] anyways
civodul has quit [Quit: ERC (IRC client for Emacs 27.1)]
derpadmin has joined #nixos
ManiacOfMadness has joined #nixos
ManiacOfMadness` has quit [Ping timeout: 265 seconds]
mmmattyx has quit [Quit: Connection closed for inactivity]
<fuzzypixelz>
how can I specify the plasma5 package(s) to be used?
<fuzzypixelz>
I want to have it from unstable
<davidak[m]>
i would like to help merge pull requests. how can i get the needed rights? i have created 91 PRs myself and reviewed 121, so i have some experience, but some more information on what to focus or even having a mentor would be great
<matthewcroughan>
clever: do we have to change time.timeZone = "Europe/London"; to time.timeZone = "Britain/London"; now that brexit is done?
<clever>
lol
yornik is now known as kinroy
orivej has quit [Ping timeout: 240 seconds]
<matthewcroughan>
etu: what do I do about You must set the option ‘boot.loader.grub.devices’ or 'boot.loader.grub.mirroredBoots' to make the system bootable.
<matthewcroughan>
I'm amazed that it protected me from a reboot like that.
<matthewcroughan>
nix is so awesome
<matthewcroughan>
worth all the pain
<matthewcroughan>
Does it still have to be boot.loader.grub.device = "nodev"; ?
<matthewcroughan>
or is the device md1 in this case?
<matthewcroughan>
ah god, I can't figure it out lol
KarlJoad has joined #nixos
<matthewcroughan>
warning: File system `ext2' doesn't support embedding.
<matthewcroughan>
/nix/store/a6azwq1lyrpqclcyxsj48dg1s5nyhhnf-install-grub.pl: installation of GRUB on /dev/disk/by-id/ata-WDC_WDS240G2G0A-00JH30_204656464910-part1 failed: No such file or directory
<matthewcroughan>
but indeed there is such a path
cole-h has quit [Quit: Goodbye]
<KarlJoad>
Has anyone ever virtualized/simulated an Arduino on NixOS?
<V>
KarlJoad: that's like the third time you've asked that, so: it won't be any different to any other linux distro, so just look for an emulator for linux
cole-h has joined #nixos
<V>
if it's not packaged, that's easy enough to fix
<ekleog>
Anyone here would know the new process for entering a nix-shell for development with gstreamer available, eg. for gst-launch? Before I had a mkShell with gstreamer and gst-plugins-{base,good} and it allowed me to eg. gst-launch-1.0 videotestsrc ! autovideosink ; but it looks like in the recent months on unstable the procedure changed?
<DigitalKiwi>
Copilot is a runtime verification framework written in Haskell. It allows the user to write programs in a simple but powerful way using a stream-based approach.
<DigitalKiwi>
Programs can be interpreted for testing, or translated into C99 code to be incorporated in a project, or as a standalone application. The C99 backend output is constant in memory and time, making it suitable for systems with hard realtime requirements.
<KarlJoad>
DigitalKiwi: Not really. I don't need to program an arduino. I want to virtualize/simulate one so that another package can use it.