lord| has quit [Read error: Connection reset by peer]
lord| has joined #nixos
lordcirth_ has quit [Ping timeout: 240 seconds]
<aminechikhaoui>
gchristensen I remember you once looked at the wireguard systemd services issue where they don't start on boot correctly ? do you remember that and did you ever figure what was the problem
mbrgm_ has joined #nixos
mbrgm has quit [Ping timeout: 256 seconds]
mbrgm_ is now known as mbrgm
<kalbasit>
is there a way to get the list of all the files that made up the modules in a lib.evalModules?
<cole-h>
I wish fontconfig 2.13.92 would merge so I could stop having to rebuild qt every time I update my nixpkgs :(
<colemickens>
lovesegfault if I have an nvidia 1080 but my cpu/mobo have integrated, is the nvidia worth using with the patches you mentioned weeks back?
<colemickens>
if I'm going to be locked in for weeks and hardware shipments are all delayed, I'm going to dual boot my gaming pc.
bvdw has quit [Read error: Connection reset by peer]
bvdw has joined #nixos
<ottidmes>
is $PYTHON supposed to be set when python is added as a build input? I am trying to build a modified samba, I basically only change some hardcoded flags from the current master samba package, still it fails for me in ./configure due to $PYTHON being empty.
<ottidmes>
to be sure it was not any of my changes I just tried the package file as-is from master, and it still fails with the same issue, guess I could try to change to the master nixpkgs commit and see if it then builds, or else a failing package was committed to master
<ottidmes>
phfew, it does work with Nixpkgs on the master commit, so $PYTHON is being set somehow between my 20.03 commit and the master commit that last changed samba
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/JvPeQ
nek0 has joined #nixos
orivej has quit [Ping timeout: 240 seconds]
<lovesegfault>
colemickens: with my patches nouveau all the way
<lovesegfault>
it works _flawlessly_
<ottidmes>
lovesegfault: just curious what kind of patches are those, for the simplest dual boot, you don't normally need to patch the other OS, right?
<lovesegfault>
ottidmes: they are nouveau patches for pascal-family nvidia gpus to work
slack1256 has quit [Remote host closed the connection]
<tomberek>
did something recently change with overridePythonAttrs?
<ajs124>
For context: I'm trying to split some options out of a module into another one and these options use modulesPath. which gives me "attribute 'modulesPath' missing, at /nix/store/10381d12gcvpjhgv5pm2p5ppcg5bv7bw-nixos-19.09-small-19.09-small.tar.gz/nixos-19.09-small/lib/modules.nix:163:28"
blaggacao has quit [Read error: Connection reset by peer]
blaggacao has joined #nixos
<ajs124>
as for what that is, you'll have to ask however wrote (this part of) subversion
<ajs124>
s/what/why/
<ajs124>
I also have this other commit on our fork, for multiple concurrent invocations of nix-prefetch, but for my current use-case I'll probably drop nix-prefetch and just use svn and nix-hash.
<ottidmes>
ajs124: Ah cool, I added it to my TODO a few days ago :) "The overlays linking workaround does not work when `nix-prefetch` is called concurrently."
<ajs124>
^^ my implementation isn't great, though. It tends to leave files behind, in certain situations. Some traps/exit hooks could probably solve that.
<ottidmes>
Yeah, thought about that too, but I am thinking of doing a bit of an overhaul some time soon, because it was also brought to my attention that allow-unsafe-native-code-during-evaluation can be problematic for some users
<ottidmes>
So I plan to take a look at these workarounds in the progress and see if I can maybe approach them differently
dycan has quit [Remote host closed the connection]
<ottidmes>
ajs124: BTW the documentation of SVN says it will only do --trust-server-cert with --non-interactive, but wrapInsecureArgBin, removes those arguments, rather than adding them, so I don't see why it would need to remove --non-interactive too
BugeyeD has quit [Remote host closed the connection]
<ajs124>
Hm. Right. Err. I guess I don't know why this works, but it does.
<ajs124>
Yep, just reproduced it. Without it svn throws a "svn: E230001: Server SSL certificate verification failed: issuer is not trusted".
<ajs124>
huh, so every fetchsvn can be mitm'd? nice. maybe we should just fix this in nixpkgs?
b has quit [Quit: Lost terminal]
<gchristensen>
the idea is the pre-determined hash makes the mitm vector moot
<ajs124>
True. Then again, people tend to TOFU a lot
<ottidmes>
and it makes it more future proof, cause certicates will fail over time
<gchristensen>
ajs124: a MITM which impacts you, and me, and ofborg, and hydra is quite a large MITM. that said, yes
<ottidmes>
so I get why it is the default in nixpkgs, but for the prefetcher you want to be able to actually trust your first use, so there it makes sense to want to check
<gchristensen>
+1
<{^_^}>
[nixpkgs] @jtojnar opened pull request #82705 → maintainers/teams: add gnome team → https://git.io/JvPvH
<ottidmes>
I am off to sleep now, I will put those fixes for svn in sometime tomorrow, thanks for the help ajs124!
<ajs124>
gchristensen: sure. Although for my particular use case, it would only need to affect my computer at home and my hydra
<ajs124>
ottidmes: thanks for nix-prefetch ^^ glad I could help
<lovesegfault>
can I override the ffmpeg mpv uses?
<lovesegfault>
cc. gchristensen maybe?
* lovesegfault
finds himself wishing for global USE flags
ottidmes has quit [Ping timeout: 265 seconds]
<tomberek>
lovesegfault: it looks like you can.
<tomberek>
lovesegfault: are you familiar with overlays?
<lovesegfault>
oh, is that what I'd do?
<lovesegfault>
I'd overlay ffmpeg_4 to just _always_ have opengl support?
<ajs124>
you can just mpv.override { ffmpeg_4 = custom-ffmpeg; }
<jojojojojo>
you can also do a global packageOverrides
<{^_^}>
[nixpkgs] @maksteel opened pull request #82706 → wfdb: init at 10.6.2 → https://git.io/JvPvx
<lovesegfault>
jojojojojo: could you show me what that looks like?
<jojojojojo>
uh yeah lemme find the docs
<jojojojojo>
it's not a perfect solution cuz iirc it forces some extra build-from-source
nuncanada2 has quit [Ping timeout: 240 seconds]
<jojojojojo>
the thing ajs124 mentioned is a more targeted technique
<cole-h>
lovesegfault: tbh I'd do what ajs124 suggested and override `ffmpeg_4` to be an overlayed ffmpeg with opengl
<cole-h>
lovesegfault: You could do that. Or you could make a new ffmpeg4 using that override (e.g. `ffmpeg_4_with_opengl`) and then use that in mpv and not require rebuilds of whatever else uses ffmpeg_4
<lovesegfault>
cole-h: I'm rebuilding on that beast of a server
<lovesegfault>
so it's fine
<lovesegfault>
takes like 5 mins
<cole-h>
Yeah, but it's a waste
<kolbycrouch>
If someone has the time. Would they mind looking at my PR #82633? I'm not understanding this check failure what-so-ever.
<cole-h>
Add literally 1 character and prevent unnecessary rebuilds :P
<lovesegfault>
cole-h: Why? I _always_ want ffmpeg to work with opengl
<lovesegfault>
I can't really think of a scenario in my life where I don't want that
<cole-h>
Fair
<lovesegfault>
Maybe in a really low-powered thing that I only use for audio?
iyzsong has joined #nixos
<lovesegfault>
build started
<evils>
kolbycrouch: have you clicked the 'details' link?
<kolbycrouch>
evils: yes, I can't seem to find/understand anything in that log that makes sense to me.
<lovesegfault>
build finishes (cole-h)
<lovesegfault>
all 174 revdeps of ffmpeg_4
<cole-h>
OK, OK, no need to flex :P
<lovesegfault>
i stronk
<evils>
kolbycrouch: not to be rude, but line 2: trace: lib.zip is deprecated, use lib.zipAttrsWith instead; those lines seem fairly straight forward to me
lordcirth__ has quit [Remote host closed the connection]
<cole-h>
Pretty sure that's just borg and not related to his PR. I don't see any `zip` used anywhere except for `libzip`
<evils>
huh, yea i was assuming that libzip was what is being referred to
<cole-h>
libzip != lib.zip :P
<kolbycrouch>
Yea lib.zip isn't used anywhere in this directly at least.
<cole-h>
kolbycrouch: Unrelated, but pretty sure we require `githubId` in the maintainer's list now
<kolbycrouch>
cole-h: Yea I saw that after my PR, but I didn't see anything in the log complaining about it.
<cole-h>
Only people complain about it. IIRC there's an RFC mandating it now, I think
<evils>
the actual error, "this PR does not cleanly list..." i've had that when i forgot to add myself as a maintainer, maybe that's related your maintainer-list change not being a separate commit (which it must be)
<cole-h>
"GitHub handles can change from one user to another, and so we will change the maintainer list to include the GitHub user ID as well as their handle. "
<cole-h>
Please add it :)
<kolbycrouch>
evils: Ahh I see. I did notice after posting my PR of another PR with that prefix "[maintainers]". Maybe I should close the PR and make seperate ones.
<cole-h>
Why would you close it?
<cole-h>
You just need to separate out the changes into logical commits :) `maintainers: add kolbycrouch`, `retroarch: 1.8.1 -> 1.8.4`, etc
<kolbycrouch>
cole-h: Oh, so two different commits are needed not PR's.
<cole-h>
I don't know if it makes sense to separate out the rest of your retroarch changes, but you will indeed need at least 2 commits
<gchristensen>
while evaluating the attribute 'license' at /var/lib/ofborg/checkout/repo/38dca4e3aa6bca43ea96d2fcc04e8229/mr-est/eval-0-gleber.ewr1.nix.ci/pkgs/misc/emulators/retroarch/cores.nix:42:14: is in the trace, and a hint as to where the error is coming from
waleee-cl has quit [Quit: Connection closed for inactivity]
<lovesegfault>
idk what changed but nix search sucks now
<kolbycrouch>
gchristensen: I saw that, but I can't see how that line in that file is a problem. I get no error trying to build it with this tree on multiple machines.
<cole-h>
Because we require a license for all package, IIRC
<cole-h>
packages*
<kolbycrouch>
cole-h: but that line was already in the expression, even before my last successful PR 6 months ago. I just inherits the license from the main retroarch derivation right?
blaggacao has quit [Ping timeout: 246 seconds]
mexisme has quit [Ping timeout: 256 seconds]
blaggacao has joined #nixos
<colemickens>
lovesegfault: I can repro if I unset NIX_PATH and run nix search -u vdpau
<cole-h>
RIP his war on NIX_PATH
<colemickens>
I mean, `NIX_PATH= nix search -u -f ~/code/nixpkgs vdpau` works fine
<colemickens>
(similar to nix-env)
<lovesegfault>
cole-h: I will kill NIX_PATH
<lovesegfault>
it's the root of all evil
<lovesegfault>
🤔
<gchristensen>
NIX_PATH is where software comes from, software is bad. bad is evil. q.e.d.
<colemickens>
haha
<cole-h>
A-ha, kolbycrouch, I think I found the issue.
<gchristensen>
cole-h: those links don't work because it is linking in to a collapsed file :( what line of cores.nix?
<cole-h>
482
<cole-h>
license = "Non-Commercial";
cptchaos83 has joined #nixos
<gchristensen>
ah, okay I'm off to bed!
<cole-h>
Should be `license = "Non-commercial";`, I bet
<cole-h>
o/ gchristensen++
<{^_^}>
gchristensen's karma got increased to 232
<evils>
huh, i was under the impression license could be an arbitrary string
<cole-h>
¯\_(ツ)_/¯
<cole-h>
It's the only weird-looking license string
<kolbycrouch>
cole-h: I see where I did that, why would it fail the PR check but not error out when building? I thought that was just an arbitrary string for unknown licenses.
<cole-h>
¯\_(ツ)_/¯
<cole-h>
Same reason it doesn't error when you make a local derivation that has no metadata at all
<evils>
building it doesn't check the metadata, the bot on the PR does
<cole-h>
Nixpkgs is stricter and requires valid metadata fields
<{^_^}>
[nixpkgs] @hamishmack opened pull request #82708 → openssl: Fix openssl build for x86_64-unknown-linux-musl → https://git.io/JvPfi
<kolbycrouch>
I'm a bit of a git noob, is it OK for me to close the PR and submit a new one without the maintainer change, and fixing the license thing?
<cole-h>
I'm of the opinion that that is a big no-no.
<kolbycrouch>
Isn't making further commits bad? espcially ones that fix mistakes from the previous?
<Squarism>
Hey, how do I make changes to /etc/nixos/configuration.nix actually "install" things?
<kolbycrouch>
I guess I just make a new commit fixing it all, then squash it into the first?
<cole-h>
Yep. Or you could rebase to just edit that first commit
<evils>
kolbycrouch: i'd probably clean up my local repo and then force push to the github branch you made the PR for
<cole-h>
^ (easiest way IMO is to rebase that first commit)
<cole-h>
Maybe case doesn't matter... We'll find out when you push again :P
<lovesegfault>
I h/quit
lovesegfault has quit [Quit: WeeChat 2.7.1]
<Squarism>
or how does /etc/nixos/configuration.nix relate to nix-env install command? Is the latter a way of manipulating the former?
<cole-h>
To install packages system-wide, I believe you add them to the `environment.systemPackages` list
<cole-h>
`nix-env` is totally unrelated to configuration.nix
<Squarism>
cole-h, what do they both represent then?
<cole-h>
`nix-env` is for imperative, user-specific packages, `environment.systemPackages` is for declarative, system-wide packages
<Squarism>
okok
<Squarism>
I read that "nixos-rebuild" should implement updates found in /etc/nixos/configuration.nix. But that command isnt available for me - on a otherwise seemingly working vanilla nix install on ubuntu 18.04
<cole-h>
Read the command again ;)
<cole-h>
*nixos*-rebuild
<Squarism>
oh ok.. so I dont have nixos then i guess but just nix
<cole-h>
Correct. I don't know how normal people manage packages and the like on non-NixOS systems, but I personally use home-manager: https://github.com/rycee/home-manager
<cole-h>
Similar to configuration.nix (just add to `home.packages = []` for declarative package installation), but not the same
<Squarism>
cole-h, thanks. Appreciated
felixfoertsch23 has joined #nixos
felixfoertsch has quit [Ping timeout: 240 seconds]
spacefrogg has quit [Quit: Gone.]
aw has quit [Quit: Quitting.]
<Baughn>
Squarism: Though when you decide to try nixos, we'll be happy to help. :)
<danderson>
hmkay, at least the periodic freezing seems to have been fixed by bypassing the in-display USB hub. I guess laptop->dock->screen->device was a bit too much complexity.
<danderson>
so, the various desktop environments like plasma and mate feel like they're missing a ton of software, compared to a similar install on Arch or Ubuntu. Is there an obvious well-known metapackage I'm missing to make it less bare out of the box?
<danderson>
e.g. on mate, I don't get any way to edit settings, for instance.
bvdw has quit [Read error: Connection reset by peer]
linarcx has joined #nixos
bvdw has joined #nixos
akegalj has joined #nixos
jakobrs has quit [Quit: WeeChat 2.7.1]
<colemickens>
jakobrs what is your goal?
xelxebar has quit [Ping timeout: 240 seconds]
<danderson>
Right. Turns out powertop autotuning is not safe to run on modern Ryzen systems. It's way, way too aggressive.
<danderson>
e.g. it turns on USB autosuspend for HID peripherals, so keyboard/mouse get all janky because USB keeps shoving them into low-power mode
<danderson>
In contrast, TLP is like "oh hey, HID is for interactive input, maybe I won't make it randomly lag 200ms for no reason", like a sensible boy
<the_pumpkin_man[>
I've got a little question : when will NixOS 20.03 be available ? Will it be the 31, or sooner ?
fendor has joined #nixos
<tilpner>
From past releases, don't rely on it releasing this month
ottidmes has joined #nixos
<the_pumpkin_man[>
Oh, but then why is it called YY.MM then ?
veske2 has joined #nixos
<evils>
it's aspirational, and it's usually released the 40th of that month
veske has joined #nixos
<etu>
the_pumpkin_man[: We freeze it that month, and we start testing. But there's no gurantee that testing and fixing is done.
<etu>
the_pumpkin_man[: Helping out with the ZHF testing is greatly appreciated
<the_pumpkin_man[>
<evils "it's aspirational, and it's usua"> oh you don't have the same calendar that I have, ok xD
<LnL>
the kickoff was also delayed a bit by some hydra problems, which won't help
<the_pumpkin_man[>
<etu "the_pumpkin_man: Helping out wit"> Sorry, Im just a newcomer, so can't really help. I was just curious and impatient, because I know on distros like Ubuntu, they are named YY.MM because there is always a release on march and november
o1lo01ol1o has joined #nixos
<the_pumpkin_man[>
Thanks for the infos :)
<the_pumpkin_man[>
And just to know, what is ZHF ?
<etu>
the_pumpkin_man[: Canonical is also a company that have paid engineers
<etu>
the_pumpkin_man[: ZHF is Zero Hydra Failures, it's a list of failed builds that needs to be fixed.
<the_pumpkin_man[>
<etu "the_pumpkin_man: Canonical is al"> and who stole your private data xD but that's another problem
<evils>
well, we use that release model (every 6 months (no LTS')), if we called it 20.04, it'd still be released in the 5th month :P and get confused for another distro
<pkral>
hi all, noob here... is there better way, how to get back into failed build environment, (i.e. when nix-build fails) ? My workflow is like this: run nix-build -K (which left build artifacts in /tmp/nix-build....) and then manually do sudo -s -u nixbld1 && cd /tmp/nix-build.... && source env-vars && cd src I suppose that there must be better way using nix-shell or whatever....
<{^_^}>
[nixpkgs] @misuzu opened pull request #82718 → ext4-fs: fix image creation on armv7l when image is bigger than 2G → https://git.io/JvPqN
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
zupo has joined #nixos
bahamas has quit [Ping timeout: 246 seconds]
kloenk has joined #nixos
<kai_w>
syd: not easily afaik, though you might be able to do it with runCommand. however, keep in mind that any files will also be written to the nix store, which is world readable, so home-manager (and nix in general) are not meant to manage secrets (which is the usual use-case for 600 permissions).
<syd>
I understand the security concern and I'm willing to overlook that for reasons that have to do with my setup
<syd>
I think I got things to work using the `onChange` attribute
<LinuxHackerman>
Taneb: keep in mind that broken RAM can lead to corrupted data on disk as well, so if possible you should shut down your machine and not use it again until the RAM is replaced.
<LinuxHackerman>
Taneb: I hope you have backups ;)
<s1341>
i'm looking for a way to install a cargo package (specifically 'peep': https://github.com/ryochack/peep) from my home.nix. What's the simplest way to do this?
<Taneb>
LinuxHackerman: it's a work machine so I'll probably switch to using another computer while I get the IT person to try and fix it
<Taneb>
And everything I care about is backed up well enough
<Mic92>
s1341: buildRustPackage
kloenk has quit [Ping timeout: 246 seconds]
<s1341>
Mic92: is there an example anywhere of how to do this?
<Yaniel>
nvidia has been awfully stubborn when it comes to collaborating with opensource
<boogiewoogie[m]>
RRedcroft: I'm having trouble setting it up somehow as no display manager lists it. would you mind sharing the relevant part of configuration.nix? :)
<LinuxHackerman>
boogiewoogie: I usually just start it from the text console, i.e. login and then `exec sway
<LinuxHackerman>
`
<RRedcroft>
boogiewoogie[m]: Sure, i think sway isnt supported by default for login managers
<Yaniel>
the whole eglstreams-while-everyone-else-supports-gbm debacle
<adisbladis>
eyJhb: Your expression LGTM
<adisbladis>
eyJhb: He has a tendency to close valid issues without any discussion for one
<RRedcroft>
i would love if either sway/nvidia supported each other, but means stuck having to use i3 and i doubt we'll ever see this resolved
<eyJhb>
Hasn't nvidia said they would start playing nice?
<Yaniel>
many times
<adisbladis>
Nvidia doesn't care. They got eglstreams support in mutter/kwin now which is the majority of linux users
<eyJhb>
Good adisbladis ! I will try it. But currently I am recompiling TexLive... Isn't there a better option regarding testing something like this? A minimal shell?
<adisbladis>
eyJhb: nix-build ./. -A vimPlugins.coc-go ?
bahamas has joined #nixos
<eyJhb>
But that wouldn't allow me to test it using Neovim I assume
<RRedcroft>
didnt they start building a 'new' allocator for everyone to use instead of egl or gbm
<Yaniel>
yes
<Yaniel>
and it has seen like one commit in the past 5 years
<eyJhb>
Well, the thing I didn't change, doesn't work :p Lets see.
<RRedcroft>
lol fuck nvidia
<adisbladis>
RRedcroft: While I agree with you please don't use that language in #nix* channels.
<RRedcroft>
adisbladis: sorry, i'll keep it clean
<adisbladis>
Thanks :)
<Yaniel>
but yeah gnome and kde already gave in, and it's not like there is much heat from nvidia's competition either
<ToxicFrog>
I'm trying to do a from-source airsonic build using mavenix. Mavenix generates the wrong source path for the jars included in the checkout, so I patch mavenix.lock at build time (using readFile/writeFile/replaceStrings) to point to the right place.
<ToxicFrog>
My best guess is that the nix build restricts which store paths are accessible to ones it thinks the build depends on, and for some reason the build of dependency jars doesn't depend on the source...
bahamas has quit [Quit: leaving]
<ottidmes>
,locate bin nsupdate
<{^_^}>
Found in packages: bind.dnsutils
mjrosenb has joined #nixos
jgeerds has quit [Ping timeout: 246 seconds]
helozjisky has quit [Remote host closed the connection]
helozjisky has joined #nixos
<s1341>
Mic92: thanks!
<s1341>
ok. so i am trying to use the broot package for home-manager. I set enableZshIntegration to true, and it does install a br function in my .zshrc, but when trying to run br i get the following message:
<Shell>
ToxicFrog: so Nix does restrict what you can access to what it thinks your dependencies are, and merely reading from a file containing paths to your dependencies doesn’t work - you’ve got to reference their actual derivations in Nix. Some tools (nix-bundle being one of them) depend on a previous behaviour where you could reference arbitrary paths in the Nix store, that no longer works.
<Shell>
I don’t know if Mavenix maybe depends on the previous behaviour too.
<Shell>
ToxicFrog: if you feel like doing more research, the keyword is string contexts.
<bqv[m]>
Question. How is it that I can still use <nixpkgs> even when that channel doesn't actually exist?
<immae>
bqv[m]: <nixpkgs> relates to the content of NIX_PATH=nixpkgs=something
<immae>
it has little to do with channel (except if "something" refers to a channel)
rardiol has joined #nixos
<ottidmes>
I am trying to get Samba in NixOS to work without errors as an Active Directory Domain Controller, it works fine, but I rather fix the remaining error, namely that it fails to update the DNS records.
<ottidmes>
Anybody experience with nsupdate?
<s1341>
NobbZ[m]: yes. the .zshrc is evaluated.
aswanson has joined #nixos
<s1341>
i noticed that there is a difference between the snippet that br tries to install and the one installed by the home-manager package.
<s1341>
but it seems minor.
<ottidmes>
After inspecting samba_dnsupdate with --verbose, I got that dnsupdate was non-existant on the path they checked, so I configured that properly, but now I get: samba_dnsupdate: update failed: REFUSED
<ottidmes>
Now I am not sure if that is just due to the nature of dnsupdate that conflicts with readonly files on NixOS, or that I am still misconfiguring something
<NobbZ[m]>
s1341: How do you know the correct `.zshrc` is evaluated? Anyway, what does `which br` print?
<NobbZ[m]>
Then it seems correctly evaluated, and those "minimal" differences between the actual zshrc content and what br would install are significant and not minimal.
johnny101 has joined #nixos
<andi->
exarkun: ahh, yeah.. It is a bit less straight forward then what it should be
helozjisky has quit [Remote host closed the connection]
helozjisky has joined #nixos
<s1341>
NobbZ[m]: so how do I update the broot pack
<s1341>
package?
<s1341>
do i need to do an overlay?
<s1341>
(i'm very new to nixos)
<NobbZ[m]>
No need to update the package, I think just not doing `ztshIntegragtion`, but using `extraInit` to provide the correct function.
c0c0 has joined #nixos
linarcx has joined #nixos
bvdw has quit [Read error: Connection reset by peer]
bvdw has joined #nixos
dckc has quit [Ping timeout: 268 seconds]
aveltras has joined #nixos
v0|d has joined #nixos
dckc has joined #nixos
Guest85905 is now known as bezaban
lorimer has joined #nixos
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos
<s1341>
ok. Thanks NobbZ[m]
<iceypoi>
is there any way to combine nix-env with overrides? E.g. have something like like `nix-env -i chromium.override { useVaapi = true }`
<iceypoi>
I'm asking because the use case would be a non-nixos system, so I can't refer to configuration.nix
<tilpner>
iceypoi: 1. you can use an overlay to give it a name
<tilpner>
I don't know when you would prefer which
<iceypoi>
the latter gives a name, right?
<tilpner>
Neither of them use overlays, so not in the way I meant in 1.
<iceypoi>
i see
<iceypoi>
ok, cool
<iceypoi>
that should be good enough for now, thanks!
gustavderdrache has joined #nixos
<v0|d>
I use a custom kernel but nix-env -iA nixos.wireguard installs for the channel kernel. how do I say, OK nix, compile module for my kernel?
<srk>
you need to access it via kernelPackages of your kernel
<tilpner>
On NixOS, config.boot.kernelPackages.wireguard
<v0|d>
aha.
<srk>
or enable it via module and forget about it
<v0|d>
is it possible to replace module in modules-list.nix?
<v0|d>
say overriding via imports = [];
<v0|d>
looking for a way to remove it and replace it with a new one.
<tilpner>
Yes, see disabledModules
rogue_koder has joined #nixos
jakobrs has joined #nixos
<v0|d>
tilpner: thnx.
<AWizzArd>
Is a Nix channel a little bit like a specific version of a Linux distribution? If I were on Ubuntu 15.04 and did a "apt install firefox" then I would get a specific version, and (at least today) it is not the newest one anymore.
<AWizzArd>
Is a channel like that? A list of packages with an exact version number?
<tilpner>
Yes, it's similar to that
<exarkun>
On Ubuntu, you don't get an exact version of Firefox because the Ubuntu 15.04 repository receives updates (well, it did at some point in the past, it doesn't anymore).
<exarkun>
And that's like what a nix channel is
<tilpner>
But you can use multiple channels at once, and mix your system together from any channels you like
cosimone has joined #nixos
<AWizzArd>
exarkun: yes, but let’s assume that we look at Ubuntu 15.04 frozen at a specific timestamp. At this moment there is a concrete set of apps and corresponding version numbers.
<AWizzArd>
tilpner: okay I see. And is nix-unstable the only channel that constantly can change unexpectedly, while nixos-15.09 is frozen forever?
<v0|d>
AWizzArd: there is staging, staging-next, unstable, stable etc.
<tilpner>
AWizzArd: None of them should move unexpectedly, you control exactly when they update on your system, and you can undo any updates you want
<AWizzArd>
By “update” I mean that in, say, nixos-15.09 there was a specific version of Firefox, let’s say version 45. It was FF45 from day 1 and still is FF45 today and will be in seven years.
<AWizzArd>
While in unstable today it’s FF72 and tomorrow it could be FF73
<v0|d>
AWizzArd: you can always checkout a specific rev and install according to that rev by nix-build './some-rev-nixpkgs' -A firefox
<tilpner>
Yes, nixos-unstable and nixpkgs-unstable are not expected to stop evolving any time soon
<exarkun>
A channel is precisely a mechanism to allow the nixpkgs revision to change
<AWizzArd>
But the nixos channels that have version numbers: they don’t evolve, right? They stay fixed?
<exarkun>
If you want something that *doesn't* change ever, that thing is a nixpkgs revision
<tilpner>
AWizzArd: They do evolve, e.g. security fixes get backported
<AWizzArd>
exarkun: so a channel is not a nixpkgs revision
<exarkun>
Correct.
<exarkun>
A channel is a delivery mechanism for new nixpkgs revisions.
<AWizzArd>
Okay then what I had in mind above is a nixpkgs revision. That would be the Ubuntu 15.04 version at a specific timestamp.
<tilpner>
You can skip the entire channel mechanism and pin to revisions
<{^_^}>
[nixos-homepage] @garbas closed pull request #178 → Document how to uninstall on MacOS → https://git.io/JvPGv
Rusty1 has joined #nixos
malSet_ has joined #nixos
<exarkun>
AWizzArd: Yes, more or less (the "less" being that you need to ignore some more properties of Debian-based package systems which means that even at a specific timestamp, "Ubuntu 15.04" is maybe an ambigious package set, but I guess you're trying to ignore that level of detail :)
malSet_ has quit [Read error: Connection reset by peer]
malSet has quit [Ping timeout: 250 seconds]
detran has quit [Ping timeout: 256 seconds]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<AWizzArd>
But then I am still not really sure what a channel is and why it is useful.
<AWizzArd>
The revision makes sense. A set of thousands of packages, fixed to a specific version. As long as I work against such a specific revision my builds have a fair chance to be reproducible.
b42 has joined #nixos
<niksnut>
channels are on their way out
<exarkun>
is something replacing them?
<tilpner>
Pinned revisions are necessary for bit-reproducability, but unfortunately there are still plenty of ways your build can become bit-unreproducible
<niksnut>
yes, nix flakes, which are generally git repositories
<gchristensen>
AWizzArd: a channel is a URL to a set of Nix expressions. that URL is like https://nixos.org/channels/nixos-20.03 this URL will redirect to a very specific version of nixpkgs. however, when a new version of nixpkgs is found to be good enough, that URL will point to the new version
<AWizzArd>
tilpner: yes, while bit-unreproducible may occur those revisions bring me at least a bit closer to bit-REproducible.
Neo-- has quit [Ping timeout: 256 seconds]
<exarkun>
I guess I should try to learn about flakes at some point.
<AWizzArd>
gchristensen: So is a channel a bit like a pointer to a revision? And that pointer can change?
<niksnut>
a channel is a tar file with nix expressions in them
<AWizzArd>
The idea is: when a specific channel is created then there is a set of people that try to make everything inside this channel as compatible as possible?
<gchristensen>
worldofpeace: yep.
<niksnut>
in the case of nixpkgs, it comes from a particular revision of the nixpkgs repo but the nix-channel command doesn't know anything about that
<gchristensen>
AWizzArd: yep
<AWizzArd>
So a nix channel is actually what comes closest to a specific Ubuntu version, such as 15.04. While a revision is a bit comparable to all Ubuntu 15.04 packages at a concrete version number, frozen at a point in time.
<niksnut>
flake dispense with the tar file indirection so you can point at a revision of a repo directly, e.g. nix build github:edolstra/dwarffs/47218f1b8f971925241b1b307a1e770a7c220b5e
<{^_^}>
[nixos-homepage] @garbas opened pull request #351 → remove search from the toolbar → https://git.io/JvPGY
<AWizzArd>
If it were technically/magically possible to guarantee that all packages inside a revision are perfectly compatible to each other, then there would be no need for a channel. Is that correct?
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<niksnut>
well, there is no *need* for channels to begin with
<niksnut>
you can just do a git checkout of nixpkgs and use that
<niksnut>
channels mostly exist because back then we didn't want to make Nix have a dependency on some particular version management tool
<MichaelRaskin>
And for channel to have compatibility, you still need _some_ revision to have this internal self-compatibility
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<AWizzArd>
niksnut: yes, I can use a specific revision and not care about a channel. But isn’t the purpose of a channel also to see that when some package doesn’t build and it later gets fixed, to update the channel then to a new revision?
<MichaelRaskin>
This can be achieved just with a branch
<niksnut>
right, for example, the nixos-20.03 branch gets updated from the release-20.03 branch if certain tests succeed
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<niksnut>
and the 20.03 channel corresponds to the 20.03 branch
<{^_^}>
[nixos-homepage] @garbas closed pull request #157 → Note entry point to top level packages. → https://git.io/JvPGa
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<tilpner>
AWizzArd: Security updates will always be necessary
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
tomberek has joined #nixos
helozjisky has quit [Remote host closed the connection]
helozjisky has joined #nixos
<v0|d>
when I install a package via specific overrides for instance nix repl :b somepackage.override {}, then ru nix-env -u, update resets to defaults.
<NobbZ[m]>
Okay, thats different to my situation, where I have 2 users, me and my son.
__red__ has joined #nixos
<morgrimm>
Hey folks - I just created a new profile in a multi-user install, but the new profile doesn't have any of the shell profile stuff (etc/profile.d/nix.sh or nix-daemon.sh) the default profile does, and as a result I don't have access to any of the nix binaries in my shell unless I'm using the default profile :/
<morgrimm>
Is this intended?
cosimone has quit [Quit: Quit.]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
<NobbZ[m]>
morgrimm: How have you created that user? On what Distribution are you?
Neo-- has joined #nixos
Makaveli7 has quit [Quit: WeeChat 2.4]
<{^_^}>
[nixos-homepage] @garbas closed pull request #314 → Add Manual link to menu bar. as layout → https://git.io/Je4cB
malSet has quit [Read error: Connection reset by peer]
<{^_^}>
[nixos-homepage] @garbas closed pull request #313 → add manual link in nixos page → https://git.io/Je8Eh
<morgrimm>
MacOS Catalina - I followed the procedure for creating synthetic volumes/enabling permissions and did a multi-user install. It works fine, but I wanted a different profile, so I used nix-env --switch-profile prefix/profiles/per-user/$USER/profilename, and then installed something since it creates it lazily
<NobbZ[m]>
You might need to manually "`source`" `/etc/profile.d/nix.sh` depending on your host distribution and shell used.
<morgrimm>
But now lilytarball's fish plugin to source the nix binaries doesn't work, or my zsh/bash changes, because they all expect nix.sh in the current profile (the one symlinked to my homedir)
<morgrimm>
I don't have an `etc/profile.d` folder, the install didn't create me one
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<NobbZ[m]>
`/etc`, not `etc`…
Neo-- has quit [Remote host closed the connection]
<morgrimm>
Yes - I don't have a `/etc/profile.d` folder
malSet has joined #nixos
<morgrimm>
the only `nix.sh` that it created for me is in the default profile
<NobbZ[m]>
It was created for me on my Arch Linux, though as far as I've heard, cataline doesn't like stuff in `/` at all, I think differences beween catalina and Arch are far too big to be able to help you here…
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
malSet has quit [Read error: Connection reset by peer]
<{^_^}>
[nixos-homepage] @garbas merged pull request #204 → add link to online package viewer → https://git.io/vx9Wn
<{^_^}>
[nixos-homepage] @garbas pushed commit from @ShalokShalom to master « add link to online package viewer (#204) »: https://git.io/JvPZk
glittershark has joined #nixos
Neo-- has joined #nixos
<morgrimm>
Yeah, no worries. I'll log an issue with it, and see where it goes - I can't find any documentation about nix on MacOS besides the GH issues, so
<morgrimm>
I'm just trying to figure it out as I go
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @bennofs opened pull request #82728 → racket: enable building on aarch64 → https://git.io/JvPZt
malSet has joined #nixos
<manveru>
morgrimm: there's #nix-darwin
malSet has quit [Read error: Connection reset by peer]
<morgrimm>
Oh, thank you! I'll pose the question there
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<exarkun>
I updated the rev I passed to fetchgit and, sensibly, it complained that the sha256 was wrong. I ran nix-prefetch-git to get the correct sha256 and, surprisingly, it was different from the sha256 in the error from the fetchgit call. I put in the sha256 nix-prefetch-git gave me and, surprisingly, fetchgit was happy.
<exarkun>
The possible explanations seems like it would be that the new sha256 already identifies a store object so fetchgit shortcircuited and didn't do anything, I just got the existing store object?
<__red__>
I have this error. Did the format of services.httpd.virtualHosts change?
<__red__>
error: The option value `services.httpd.virtualHosts' in `/etc/nixos/evil-webapps.nix' is not of type `attribute set of submodules'.
<evanjs>
Trying to use pkgsCross.armv7l-hf-multiplatform.callPackage with a gcc49Stdenv, and that all worked great, but, well, yeah, glibc too new
<__red__>
(
<mdash>
exarkun: is it base32 vs base62 or whatever?
<exarkun>
hm
glittershark has quit [Ping timeout: 246 seconds]
<exarkun>
They're both the same length
<{^_^}>
[nixos-homepage] @garbas merged pull request #305 → Enforce HTTPS and TLSv1.2 for curl | sh → https://git.io/Jem29
<{^_^}>
[nixos-homepage] @garbas pushed commit from @yegortimoshenko to master « Enforce HTTPS and TLSv1.2 for curl | sh (#305) »: https://git.io/JvPZG
<exarkun>
nix-prefetch-git tells me "sha256": "11sriz1g5vcfmdfsx0ava2wj30j78zp62l3c8wq2b33c7f7bjf44"
<exarkun>
nix-build tells me it got: sha256:189hd0g3yyp6vqc4xm2cd3ypnj4nibnl6yvm4vjcifz6waw76a2l
<clever>
evanjs: ive not really tried to override parts of the stdenv
glittershark has joined #nixos
detran has joined #nixos
<clever>
evanjs: what is the nix expression that you tried to nix-build?
<exarkun>
I assume there's some command I can use to inspect these objects in the store...
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<clever>
exarkun: if you use `nix-hash` to convert the base32 hashes to base16, you can then query both in the DB
<clever>
exarkun: then youll have the $out for each, and can just run plain `diff -r` on the 2 dirs
* exarkun
tries
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
helozjisky has quit [Remote host closed the connection]
helozjisky has joined #nixos
Ariakenom has joined #nixos
<exarkun>
Ah. They differ by the presence of a `.git` directory.
malSet has joined #nixos
<clever>
exarkun: did you set keepDotGit ?
malSet has quit [Read error: Connection reset by peer]
<exarkun>
and I forgot to pass --leave-dotGit to nix-prefetch-git.
<clever>
yep, that will do it
<evanjs>
hrm I can either ssh into an arm device and build this (which will introduce its own issues) or try to override glibc with an older version. I saw #50329 but that overrides glibc in the entire pkgs expression.
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
Neo-- has quit [Ping timeout: 250 seconds]
malSet has quit [Read error: Connection reset by peer]
<__red__>
I guess there must be some kind of migration guide?
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<clever>
__red__: what are you setting services.httpd.virtualHosts to?
<Ankhers>
I am trying to use autoPatchelfHook in my buildInputs to patch some binaries. I am getting a failure because it is unable to find libstdc++.so.6. I have tried adding gcc and android-studio (because I am trying to get a standalone android NDK toolchain) and it is still failing to find the .so file. Any ideas on how I should fix this?
blaggacao has quit [Read error: Connection reset by peer]
zupo_ has joined #nixos
<clever>
Ankhers: try gcc.cc.lib
blaggacao has joined #nixos
<Ankhers>
I'm giving that a go. Maybe a dumb question, but the NDK uses clang, does it matter that I use gcc's .so files?
<clever>
Ankhers: for clang, you want to instead use clangStdenv
<clever>
and maybe clang.cc.lib
<Ankhers>
Thanks. I will switch to that. But the gcc.cc.lib at least allowed things to compile.
<Ankhers>
And it appears that I am able to compile my application now.
<exarkun>
anyone have any tips for constructing a derivation when you have a source and an expression using that source, and you want a new package that uses the same expression but a modified version of the source?
kleisli has quit [Max SendQ exceeded]
<exarkun>
I guess I can use overrideAttrs to supply a new src
kleisli has joined #nixos
<exarkun>
I'm not so clear on how to create a new, modified source
<exarkun>
... rip the original src out and use it in a new derivation with ... a patch phase, I guess? or maybe I should just inject something into the patch phase of the existing derivation
<clever>
$ cat ~/apps/qemu/shell.nix
<clever>
with import <nixpkgs> {}; qemu
<clever>
exarkun: if you only want nix-shell, even this is enough
<exarkun>
nah, I want to build a package. it's some awkward whitebox/branding shennanigans
<clever>
if you want to add in nix-build, qemu.overrideAtrs (drv: { src = ./.; })
<clever>
or fetchFromGitHub
<evanjs>
Anybody know of a good way to remove specific patches when I'm overriding something?
<exarkun>
does that imply having a complete copy of the whitebox source?
<jakobrs>
I'm not sure if it's a good idea or not so it would be nice if someone could take a look. It's 118 worlds.
<jakobrs>
*words
syd has quit [Remote host closed the connection]
captn3m0 has quit [Read error: Connection reset by peer]
<simpson>
jakobrs: Humble suggestion: $(export NIX_PATH=...) and direnv have solved this sort of issue for me. It might be worth checking out direnv first: https://github.com/direnv/direnv/wiki/Nix
is_null has quit [Ping timeout: 246 seconds]
<jakobrs>
simpson: I didn't realise direnv could help. You're presumably just setting the NIX_PATH locally inside a directory?
<evanjs>
what's the best way to get a filename from a path using nix?
<clever>
evanjs: nix-instantiate --find-file
<clever>
evanjs: if its a <path>
<simpson>
jakobrs: Yeah. That's almost all of what I need. I'm not sure what your workflow is, though, or what you're building.
<evanjs>
clever: what if I need to get it within e.g. builtins.filter?
<simpson>
jakobrs: I think that you should be prepared for a long and intricate discussion if you do decide to push forward. But I have no strong feelings either way.
<evanjs>
did not know builtins.elem was a thing, either :D
is_null has quit [Ping timeout: 258 seconds]
ckauhaus has quit [Quit: WeeChat 2.7.1]
<jakobrs>
{^_^}++
<{^_^}>
{^_^}'s karma got increased to 171
ixxie has joined #nixos
jakobrs has left #nixos ["WeeChat 2.7.1"]
FRidh has quit [Quit: Konversation terminated!]
<{^_^}>
[nixos-homepage] @garbas closed pull request #317 → build(deps): bump eslint-utils from 1.3.1 to 1.4.3 in /packages-explorer → https://git.io/JegRH
<{^_^}>
[nixos-homepage] @garbas pushed 0 commits to dependabot/npm_and_yarn/packages-explorer/eslint-utils-1.4.3: https://git.io/JvPnj
<evanjs>
starting to wonder if just disabling all these patches is going to work if I'm using such an ancient glibc, though...
is_null has joined #nixos
<ottidmes>
I am going through some dependencies and on Debian/Ubuntu they need libjson-perl libparse-yapp-perl, while the package only includes perl, are they present by default?
<steveeJ>
trying to update a VPS to 20.03: I'm getting the following error `error: efibootmgr failed to register the boot entry: No such file or directory.` does someone have an idea what could cause this for 19.09 -> 20.03?
eacameron has joined #nixos
<steveeJ>
the system has `boot.loader.grub.efiSupport = false` set
<Ankhers>
Does anyone know if lorri sets env vars from the shell.nix? I have a shell.nix file that works correctly if I run `nix-shell', but not with direnv/lorri.
rogue_koder has quit [Ping timeout: 240 seconds]
<cole-h>
If you use it with direnv, I think it does
<fresheye1>
how can I switch keyboard layouts?
<evanjs>
I wonder if it's more practical to include the newer glibc version with my app deployment and just patchelf the deps (or just ld?) to the new one in the app's $out/lib folder
<fresheye1>
I set grp:toggle, but I don't get how to switch
<evanjs>
vs using a very old glibc... which I've basically given up on
<s1341>
specifically, i want to install the android ndk r10e.
philr has quit [Ping timeout: 246 seconds]
zupo has joined #nixos
<Ankhers>
Does anyone know much about lorri and direnv? I have the lorri service running. I confirmed the daemon is running with `ps'. But it does not actually seem to build my project when I change my shell.nix file. But if I run `lorri watch', it will build it.
<mikky>
hey, I'm trying to make my external GPU work. An RX580 connected via TB3. Everything seems ok, kernel knows about the card, even Xorg seems to be aware but it always starts on the integrated graphics instead and when I remove the `intel` driver and leave only `amdgpu`, Xorg won't start: `(EE) No devices detected`
<{^_^}>
[nixos-homepage] @garbas merged pull request #288 → Add option to automatically rebuild with entr → https://git.io/fjoNm
<{^_^}>
[nixos-homepage] @garbas pushed commit from @egasimus to master « Add option to automatically rebuild with entr (#288) »: https://git.io/JvPlE
<fresheyeball>
Is there anything I can do to get the apl keyboard layout back?
v88m has joined #nixos
jgeerds has joined #nixos
zupo has quit [Ping timeout: 265 seconds]
zupo has joined #nixos
<s1341>
anyone?
<fresheyeball>
s1341: what have you tried?
CptCaptain has joined #nixos
chloekek has quit [Quit: WeeChat 2.7]
<s1341>
fresheyeball: i can't get it to build that package...
<fresheyeball>
s1341: how are you trying to build it?
<s1341>
i tried adding androidndk-pkgs to my home.nix's home.packages
<s1341>
no luck.
<fresheyeball>
s1341: you are dangerously close to doing more than providing the IRC with a check engine light
<fresheyeball>
what happens? do you get an error? what error? can you share the nix expression?
<s1341>
fresheyeball: hang on...
<fresheyeball>
clever: you knew this one last time. Did they remove the apl keyboard layout?
<fresheyeball>
how can I check?
growpotkin has joined #nixos
<clever>
fresheyeball: cant remember exactly where that was, but if you have an older generating with it in the nix store, try just searching the whole store for a file with apl in the name?
<s1341>
how do i install libm.so?
<clever>
s1341: its part of glibc, so nix-shell will just provide it automatically
<clever>
s1341: though, android-ndk may complicate things, if you need the andriod libm.so
<fresheyeball>
clever: that would be a good skill to learn. How can I do that?
<s1341>
and it says no package found that provides libm.so.6
<fresheyeball>
clever: clever looks like it's there
<clever>
s1341: which libm.so does glibc contain?
<fresheyeball>
I got 2 store paths though
<s1341>
clever: how do i know?
<clever>
s1341: `nix-build '<nixpkgs>' -A glibc` then `ls result/lib`
<s1341>
it has libm.so.6
<s1341>
!
<clever>
s1341: how exactly did you "install" glibc?
<s1341>
clever it's in my home.packages, and i did home-manager switch
<clever>
,libraries s1341
<{^_^}>
s1341: Don't install libraries through nix-env or systemPackages, use nix-shell instead. See https://nixos.wiki/wiki/FAQ/Libraries for details.
<s1341>
oh...
<clever>
s1341: no library you install with home-manager will be visible to anything
<fresheyeball>
clever: ok, so it looks like apl is in there
<fresheyeball>
setxkbmap apl
<fresheyeball>
Error loading new keyboard description
<fresheyeball>
but it's not linked for some reason
<clever>
fresheyeball: strace, what syscall failed, what is it not finding?
mehlon has joined #nixos
<mehlon>
what's a good way to do DNS-over-TLS in nixos?
<fresheyeball>
strace is beyond my depths
<fresheyeball>
lots of (No such file or directory) in here
<s1341>
clever: how exactly do i install glibc with nix-shell
<clever>
s1341: 2020-03-16 13:56:28 < clever> s1341: its part of glibc, so nix-shell will just provide it automatically
<fresheyeball>
s1341: nix-shell -p glibc
<fresheyeball>
will get you a nix shell with it
<clever>
s1341: but when trying to patchelf things, you have to add it to the list of libraries being patched in
<rnhmjoj>
do you know how can i get nix to pick up and install a manpage? specifically it's a cabal haskell project
<{^_^}>
[nixpkgs] @roberth opened pull request #82741 → dockerTools.buildLayeredImage: store all paths passed in final layer (backport) → https://git.io/JvP4r
<cole-h>
Would creating a `man` output and symlinking the manpage into that output work?
sigwinch28 has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<cole-h>
Or... probably copying instead of linking
<{^_^}>
[nixpkgs] @skykanin opened pull request #82742 → osu-lazer: Init at 2020.306.0 → https://git.io/JvP4H
<cole-h>
rnhmjoj: ^ maybe
<rnhmjoj>
cole-h: the thing is i have never written a man page before and i simply don't know where to put the file. i tried following what pandoc does (putting the page in $src/man/pandoc.1) and adding it to the extra-source-files but apparently it's not working
<{^_^}>
[nixpkgs] @roberth merged pull request #82741 → dockerTools.buildLayeredImage: store all paths passed in final layer (backport) → https://git.io/JvP4r
<{^_^}>
[nixpkgs] @roberth pushed 2 commits to release-20.03: https://git.io/JvP4A
Avaq has joined #nixos
snajpa has joined #nixos
alexherbo2 has quit [Ping timeout: 246 seconds]
alexherbo21 is now known as alexherbo2
<sigwinch28>
does anyone here use Vorta (a borg backup frontend) on NixOS? it doesn't appear to be in my channels (I'm tracking nixos 19.09 and unstable).
abathur has joined #nixos
<cole-h>
It hasn't been packaged yet, so that would be why
<sigwinch28>
I'm happy to package it myself, but I was wondering if there would be anyone here who would go "oh, I use it! and here's how..."
_ris has joined #nixos
nDuff has joined #nixos
is_null has quit [Ping timeout: 246 seconds]
cartwright has quit [Remote host closed the connection]
<Avaq>
Hi folks! I'm trying to set up wireguard. First using `networking.wireguard`, and when that didn't work, I switched to `networking.wg-quick`. I've gone over things a hundred times, but whatever I do, I end up in a situation where the wireguard server appears to be running, yet cannot be reached (Connection refused) by anyone.
<Avaq>
It just seems like there's nothing listening n 0.0.0.0:51820, which is confirmed by `lsof -i -P`, but disproven by `netstat -lptu`, which lists udp 0.0.0.0:51820, but with no PID/Program. Furthermore, I can bind another process to 0.0.0.0:51820 without problems, and then communicate with that process over that port.
<Avaq>
My problem is that I have no frame of reference for how a correctly running WireGuard setup should behave. So maybe one of you can help me out? Also I'm not very well versed when it comes to networking, so I might be mixing some things up. For example, I used telnet to try and connect to the wireguard server, but I fear that might be useless as wireguard expects UDP traffic.
is_null has joined #nixos
lovesegfault has joined #nixos
<sigwinch28>
Avaq: I have some experience running Wireguard (both vanilla setups and through nixos)
rogue_koder has quit [Ping timeout: 246 seconds]
rogue_koder has joined #nixos
<cole-h>
Avaq: If you aren't already, have you tried connecting from inside the network the host is on? You might have an issue with port forwarding, etc
<Avaq>
I have the port forwarding on my router working, which I confirmed by disabling wireguard and having something else listen to 51820, and I could in fact reach it from outside the network.
c0c0 has quit [Ping timeout: 246 seconds]
<Avaq>
I also tried in various ways to connect from my localhost to the wireguard server, but I'm not sure what I'm supposed to be looking for.
<misuzu>
Avaq: make sure you are forwarding UDP port
<sigwinch28>
Avaq: could you post the relevant part of your confirurations to a pastebin? make sure to take out any encryption keys and/or public IPs.
<sigwinch28>
the real method of checking whether wireguard is _working_ is to use `wg show`, probably as root
<cole-h>
(I think just `wg` is an alias for `wg show` -- shorter to type])
<misuzu>
Avaq: also you have to use wg client to check if it's working
<sigwinch28>
Wireguard is a bit misleading to troubleshoot because wireguard servers/clients will not respond at all unless the correct keys are configured
<Avaq>
sigwinch28: the `wg show` output appears to be one of a well working setup. Yet I can't manage to reach the server.
<Avaq>
Hold on, I'll create some pastes.
<sigwinch28>
Avaq: check the server. On the client you should see an interface with a public key, and a peer with the server's public key. on the server you should see the opposite.
<Avaq>
But if we forget the forwarding for a second - is there any way I can check that my wireguard server is running and doing what it should by approaching it from localhost somehow?
o1lo01ol1o has joined #nixos
xfix has joined #nixos
<misuzu>
Avaq: is server on your home network or on internet?
<sigwinch28>
if `wg` (or `wg show`) gives output, then the interface is up and running.
<Avaq>
I am currently behind NAT.
<misuzu>
Avaq: and server is?
<Avaq>
The server is my laptop for now. The device I'm working on. :P Sorry
Ariakenom has joined #nixos
<Avaq>
And I'm using my phone (with WireGuard android) to try and connect to my laptop from the outside.
blaggacao has quit [Ping timeout: 246 seconds]
<Avaq>
And the android client logs just show endless failed handshaking attempts.
<misuzu>
Avaq: try connecting from the inside
<sigwinch28>
Avaq: the config you've provided is fine (although I can't speak for the NAT/iptables rules).
<mikky>
hey, the external GPU problem looks more like a dual-gpu problem. The fact that the GPU is external doesn't make a difference. Anyone tried running dual GPU setup? For starters I would be happy to make X use the secondary GPU only, doesn't even have to switch on-the-fly.
blaggacao has joined #nixos
<sigwinch28>
Avaq: it's also possible the configuration on your android device is incorrect. You can check that the interface is working by creating a second interface (say wg1) on the machine with the same settings as your android device.
<{^_^}>
[nixpkgs] @vcunat pushed 8 commits to release-19.09: https://git.io/JvPRk
<lucc>
I need to use the user luc and the group users because they can read the music directory.
<misuzu>
lucc: you probably need to set up pulseaudio output in services.mpd.extraConfig
alexherbo2 has quit [Ping timeout: 265 seconds]
alexherbo2 has joined #nixos
vika_nezrimaya has joined #nixos
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @vcunat pushed commit from @mmilata to staging-19.09 « libxml2: add patch for CVE-2019-20388 »: https://git.io/JvPRW
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
mexisme has joined #nixos
<{^_^}>
[nixpkgs] @vcunat pushed commit from @mmilata to staging-20.03 « libxml2: add patch for CVE-2019-20388 »: https://git.io/JvPRR
malSet has joined #nixos
fling has quit [Read error: Connection reset by peer]
patrl has joined #nixos
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
mexisme has quit [Ping timeout: 246 seconds]
malSet has joined #nixos
Shouou has joined #nixos
<Avaq>
misuzu: sigwinch28: It's working for me now. It was just the firewall blocking UDP traffic. Thank you very much!
is_null has quit [Ping timeout: 250 seconds]
<Avaq>
The port forwarding, key setup, client config, everything else, was already set up correctly.
alexherbo2 has quit [Ping timeout: 250 seconds]
alexherbo2 has joined #nixos
<cole-h>
Pesky firewalls
<Avaq>
I spent hours going over each of those layers.
fling has joined #nixos
<Avaq>
Haha..
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
_ris is now known as ris
<lucc>
misuzu: I tried that before like this:
<lucc>
services.mpd.extraConfig = ''
<lucc>
audio_output {
patrl has quit [Quit: WeeChat 2.7.1]
<lucc>
type "pulse"
<sigwinch28>
me, wanting to package a python package for myself first, then wanting to put it in nixpkgs later: wow what a ballache it is to just have one "default.nix" that works for nix-shell and callpackage
<lucc>
}
<lucc>
name "My MPD PulseAudio Output"
<lucc>
'';
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<ottidmes>
If I want xserver running but just want the terminal login (text mode) like when xserver is disabled, how can I decleratively configure that in NixOS, the guide I am following is for Ubuntu and says to run: sudo systemctl set-default multi-user.target
<softinio[m]>
I recently switch to using home manager and latest 20.03 version on a work laptop
<lucc>
But I get the error: ERROR: Failed to enable output "My MPD PulseAudio Output" (pulse); pa_context_connect() has failed: Connection refused
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<lucc>
I assume that mpd can not find the pulse process.
patrl has joined #nixos
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<softinio[m]>
since then when I nix-shell to a python3 environment which builds a virtualenv inside it still points to my non virtualenv when I do which python
jb55 has joined #nixos
<misuzu>
lucc: that's probably it. you can try `hardware.pulseaudio.systemWide = true;`
<softinio[m]>
also when I try to pip install certain packages I get No module found Cython
Avaq has quit [Quit: Leaving.]
malSet has joined #nixos
<misuzu>
lucc: also please youse pastebin
malSet has quit [Read error: Connection reset by peer]
<softinio[m]>
any ideas on what I may have misconfigured
<softinio[m]>
I did add my nix-profile bin dir to my path in my shell profile
<softinio[m]>
actually I removed that
<lucc>
misuzu: I used to have multi line support in my client but it seems to be broken. Would short code snippets be acceptable here if I fix my client?
alexherbo20 has joined #nixos
Ariakenom has quit [Quit: WeeChat 2.7]
alexherbo2 has quit [Ping timeout: 246 seconds]
alexherbo20 is now known as alexherbo2
<T_S_>
hi. longtime nix user but not a committer. I am working to update the biconductor packages under r-modules. They are way out of date. Seems like it will be just a bunch of tweaks that I should be able to work through but I won't know until I get done. Is there anyone who I should consult prior to making a pull request or even earlier?
<ottidmes>
I probably just want services.xserver.autorun = false;
<ottidmes>
softinio[m]: maybe try --pure with your nix-shell? Although that would not actually fix your problem, might give you insights if it then works
<simpson>
T_S_: Anybody you find listed as maintainers might care.
<lucc>
now if I try the syste wide pulse audio I need to set the group to "audio" but that does seem to work, mpd inside the systemd unit complains: exception: Failed to set group 17: Operation not permitted
<lucc>
Is there any way to modify an implicitly generated systemd service like mpd.service?
<samueldr>
or is it that you are missing a way to get to the manual from the nixos.org homepage?
<samueldr>
(which may be a valid concern)
lucc has quit [Ping timeout: 256 seconds]
<mojjo>
this is kind of a different version. I was looking for the acutally existing secion on 'deployment to digital ocean'
<mojjo>
I don't mind how to get it :)
lucc has joined #nixos
captjakk has joined #nixos
Jackneill has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @turboMaCk opened pull request #82746 → elmPackages.elm-instrument: Init at 0.0.7 → https://git.io/JvP0O
nckx has quit [*.net *.split]
Luker has quit [*.net *.split]
hexagoxel has quit [*.net *.split]
<T_S_>
thanks @simpson
hexagoxel has joined #nixos
Luker has joined #nixos
nckx has joined #nixos
M4cheese[m] has joined #nixos
<lucc>
ok I had to do `systemd.services.mpd.serviceConfig.Group = "audio";` and add "luc" to the "audio" group but now it works.
<lucc>
misuzu: thank you!
mexisme has joined #nixos
<softinio[m]>
@ott
<softinio[m]>
ottidmes: ty that helped me getting it to work
nbg^ has quit [Read error: Connection reset by peer]
LysergicDreams has quit [Ping timeout: 256 seconds]
LysergicDreams has joined #nixos
proaralyst has joined #nixos
gagbo_ has quit [Quit: I'm out !]
<rnhmjoj>
just found out that cabal2nix has a bunch of hardcoded postInstalls for packages with man pages: https://git.io/JvP02. apparently there is no standardised location :(
captjakk has quit [Remote host closed the connection]
captjakk has joined #nixos
<proaralyst>
Hello. I have an unencrypted boot disk, but I'd like to encrypt the backing disks for my raid array. I've set up mdadm on top of dmcrypt but I can't figure out how to get nixos to mount the encrypted disks correctly
kleisli has quit [Remote host closed the connection]
<proaralyst>
If I use the boot.initrd.luks stuff, it looks for the keyfile before mounting / & fails to boot. I can't use fileSystems.path.encrypted directly because there are six disks to one mount point. I've tried using the cryptsetup generator but systemd doesn't autostart the servies.
<proaralyst>
I have narrowed some of this down to the discovery that I have no cryptsetup.target file, but I can't figure out why the systemd package isn't installing it. It's in the systemd source
<ottidmes>
proaralyst: I generally just write my own commands in boot.initrd.preLVMCommands to handle cryptsetup decryption
<proaralyst>
Do those run after the root gets mounted?
linarcx has quit [Quit: WeeChat 2.7.1]
captjakk has quit [Ping timeout: 246 seconds]
<proaralyst>
Because my keyfile is currently in /etc/
linarcx has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<rnhmjoj>
ottidmes: but there is no way to do that from the .cabal only, right? i have would have to add an override in nixpkgs because hackage2nix won't do that automatically
mexisme has quit [Ping timeout: 250 seconds]
tokudan has joined #nixos
<ottidmes>
rnhmjoj: no clue, sorry, my haskell projects have been too simple in terms of cabal config
kraem1 has quit [Quit: outta here]
lucc has quit [Quit: WeeChat 2.7.1]
kraem has joined #nixos
<proaralyst>
ottidmes: looks like I can drop it into postMountCommands, and also drop the fileSystems configuration. Feels kinda sad.
wiml has joined #nixos
<rnhmjoj>
ottimdmes: no worries, thanks for the example
<{^_^}>
[nixpkgs] @jonringer pushed commit from @sjfloat to master « cmt: init at 1.17 »: https://git.io/JvPEN
ixxie has quit [Ping timeout: 246 seconds]
ixxie has joined #nixos
knupfer has quit [Ping timeout: 240 seconds]
knupfer1 has joined #nixos
o1lo01ol1o has joined #nixos
nisby44_ has joined #nixos
mexisme has joined #nixos
knupfer1 is now known as knupfer
gyroninja has quit [Quit: gyroninja]
<{^_^}>
[nixpkgs] @svanderburg pushed to release-20.03 « nodePackages: regenerate with node2nix 1.8.0 »: https://git.io/JvPuC
nuncanada2 has quit [Read error: Connection reset by peer]
nuncanada2 has joined #nixos
<{^_^}>
[nixpkgs] @svanderburg pushed to release-20.03 « base16-builder: regenerate with node2nix 1.8.0 and add supplement.json to fix build »: https://git.io/JvPui
inkbottle has joined #nixos
nisby44_ has quit [Remote host closed the connection]
tomberek has quit [Remote host closed the connection]
<evanjs>
So I'm trying to use an old glibc (2.24) with a newer base nixpkgs. I thought I might be able to grab the gcc from 16.09 (which uses 2.24) but there's no pkgsCross there, no libc_bin, etc, and I'm not sure if what I'm trying to do is feasible
<evanjs>
the big thing that keeps tripping me up is eg error: assertion failed at /nix/store/278h70apn46p6i15qcrid94jxs75kh61-nixos-20.09pre215947.82b54d49066/nixos/pkgs/build-support/cc-wrapper/default.nix:88:1
<evanjs>
which is `assert libc_bin == bintools.libc_bin;`
jreed has quit [Remote host closed the connection]
tsrt^ has joined #nixos
<dieraca>
I'm trying to install a custom version of ocaml (4.06). https://nixos.wiki/wiki/OCaml#Custom_version says I need to add the relevant attribute set, but when I add nixpkgs.ocaml-ng.ocamlPackages_4_06, I get an error. Perhaps the wiki is out of date. Any idea how to do this?
<raboof>
I'm starting out with lxc, and notice https://github.com/nix-community/nixos-generators talks about an 'lxc' command-line tool, while 'nix-shell -p lxc' seems to introduce 'lxc-*' tools. Am I missing something? :)
gustavderdrache has quit [Quit: Leaving.]
captjakk_ has joined #nixos
<captjakk_>
how does one install the broadcom_sta drivers assuming you have wired internet
<captjakk_>
I am brand new to this but have managed to get an install of nixos onto my macbook
<gchristensen>
I'm feeling pretty novice here. I'm starting kresd.socket, and getting kresd.socket: Failed to create listening socket ([::]:53): Address already in use -- but then `netstat -lp` shows nothing listening on 53, and `sudo -i lsof | grep :53` isn't showing anything either. anyone have ideas on what is going on?
<tilpner>
gchristensen: Are you on 20.03?
<gchristensen>
this machine is running 19.09
<raboof>
aaah it's part of lxd apparently
<raboof>
this looks suspiciously docker-esk :D
mehlon has joined #nixos
<mehlon>
whazzup nix
<mehlon>
I want to build a program for i686-linux but I'm on x86_64... how would I do this?
<Ashy>
the monitor arrangement tool doesnt render in the display configuration and you can't arrange multiple external monitors
fresheyeball has quit [Quit: WeeChat 2.6]
<cole-h>
bennofs: A paste of your config would help us help you as well
<evanjs>
okay, think I got everything "sorted", but now I get "/nix/store/q3wx1gab2ysnk5nyvyyg56ana2v4r2ar-glibc-2.24/lib/crti.o: file not recognized: file format not recognized"
<Ashy>
at least i couldnt on a dell precision 3520 with a (lenovo) thunderbolt dock with one displayport monitor and a second hdmi monitor directly into the laptop
<{^_^}>
error: syntax error, unexpected '=', at (string):122:11
<cole-h>
🤔
<bennofs[m]>
that did work in my local nix rpel
<MichaelRaskin>
Ouch, import is defined via scopedImport
<bennofs>
> import == import
<{^_^}>
error: syntax error, unexpected '=', at (string):122:11
<evanjs>
Okay, got past that... now an even sillier error... armv7l-unknown-linux-gnueabihf-c++: command not found
<evanjs>
... but armv7l-unknown-linux-gnueabihf-cpp exists haha
<MichaelRaskin>
> ((import) == (import))
<{^_^}>
false
<bennofs[m]>
wat?
<dirkx>
admin:When you submit a pull request for something in tests -- anything you need to do in terms of updating docs or other places -or is https://github.com/NixOS/nixpkgs/pull/82657 complete enough ?
<{^_^}>
#82657 (by dirkx, 1 day ago, open): new: 4 extra test/examples for the apache redwax modules
<bennofs[m]>
ah, is this due to the context the bot evals the expression in?
fresheyeball has quit [Quit: WeeChat 2.6]
<MichaelRaskin>
I think it is about functions being uncomparable
<cole-h>
dirkx: I can't answer this, but what I would do is inspect other commits that touched tests
<MichaelRaskin>
> let f = x: x; in f == f
<{^_^}>
false
opticnerve has joined #nixos
<bennofs[m]>
MichaelRaskin: ah, no I was wondering about the > import == import giving a syntax error
<bennofs[m]>
MichaelRaskin: do you know if there is any builtin that could be used to differentiate import from "normal", user defined functions?
<MichaelRaskin>
> foo == foo
<{^_^}>
error: syntax error, unexpected '=', at (string):94:8
<MichaelRaskin>
I think it is just the special assignment parser in the bot
<MichaelRaskin>
I think import could be defined on the user side; it happens to be a hardwired definition in C++ code, but the same definition could come from Nix
<bennofs[m]>
right now the type definition for "overlay" only has check = lib.isFunction, I wonder if there is something more appropriate that could catch nixpkgs.overlays = [ import whatever ] ?
<cole-h>
Not without loads of complexity, I'd think
<bennofs[m]>
might also be the case that type checking runs too late anyway to catch this issue, since it should've at least noticed that `whatever` is not a function
<MichaelRaskin>
Well, it might process entries one by one
<MichaelRaskin>
import is a function, so…
<{^_^}>
[nixpkgs] @Enteee opened pull request #82755 → update-users-groups.pl: use atomic write_file → https://git.io/JvP2s
mexisme has joined #nixos
<MichaelRaskin>
bennofs: funny, try applying builtins.functionArgs then builtins.isFunction to import
jgeerds has quit [Remote host closed the connection]
<bennofs>
> builtins.functionArgs import
<{^_^}>
'functionArgs' requires a function, at (string):289:1
<bennofs>
> builtins.isFunction import
<{^_^}>
true
cosimone has quit [Quit: Quit.]
CptCaptain has quit [Quit: WeeChat 2.7.1]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<{^_^}>
[nixpkgs] @grahamc pushed to arm-bootstrap « armv{5tel,6l,7l}: update bootstrap tools after #80497 »: https://git.io/JvP2z
knupfer has quit [Quit: knupfer]
<infinisil>
bennofs: MichaelRaskin: x == x doesn't work because it parses it as an assignment
<infinisil>
I should fix that eventually
<infinisil>
> (10 = 10)
<{^_^}>
error: syntax error, unexpected '=', expecting ')', at (string):289:5
<infinisil>
> (10 == 10)
<{^_^}>
true
<infinisil>
Wraping it with () is a workaround for now
<flokli>
huh? shouldn't a `nix-copy-closure --to root@$host /nix/store/*-system-${hostname}-*`, and then executing `…/bin/switch-to-configuration switch` also update the bootloader entries?
o1lo01ol1o has quit [Remote host closed the connection]
<flokli>
I'm on a systemd-boot machine, and for some reason, it still seems to come up with the old config
mexisme has quit [Ping timeout: 250 seconds]
<mehlon>
switch-to-configuration? what's that?
<mehlon>
oh I see. I just found it on my machine
mexisme_ has quit [Ping timeout: 246 seconds]
<mehlon>
must be a bug. it's a perl script and it does mention updating the bootloader
<lovesegfault>
What's the right way to specify the sources in a drv? I thought this would work but it complains: `src = [ ./Cargo.toml ./build.rs ./vendor ./src ];`
<flokli>
tilpner: not sure if I understand
<lovesegfault>
error: cannot coerce a list to a string, at /home/bemeurer/src/chromaprint_sys/Cargo.nix:1333:11
<lovesegfault>
wait
<lovesegfault>
nvm I am silly
<flokli>
tilpner: I mean, I can see morph calls that as well, before invoking bin/switch-to-configuration. But I don't understand why it's not part of the script
<flokli>
and how it fails
<tilpner>
flokli: I'm not sure without checking the sources, but I assume activation builds the bootloader entries from the profile
<tilpner>
flokli: I don't know which script you're talking about
<{^_^}>
[nixpkgs] @matthewbauer pushed 31 commits to staging: https://git.io/JvPa8
nDuff has quit [Quit: zzz]
Jackneill has joined #nixos
<ottidmes>
What is different in nix-shell --pure other than a cleaner environment? I am trying to figure out why svn export does manage to find the cacert bundle with nix-shell -p subversion, but fails with nix-shell -p subversion --pure, I hoped it was due to SSL_CERT_FILE, but it does not seem to be the case
justanotheruser has quit [Ping timeout: 246 seconds]
malSet has quit [Read error: Connection reset by peer]
chloekek has quit [Ping timeout: 246 seconds]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
<dingenskirchen>
ottidmes: nix-shell without --pure basically adds to your PATH and general environment, with --pure it clears it beforehand
nerdmaxx has quit [Quit: WeeChat 2.7.1]
<dingenskirchen>
if you want to figure out where it gets the files, try strace
aveltras has quit [Quit: Connection closed for inactivity]
<das-g[m]>
Can I let one package provide data for another package in a XDG_DATA_DIRS dir?
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
m0rphism has quit [Ping timeout: 265 seconds]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
opticnerve has quit [Quit: Leaving]
malSet has joined #nixos
mehlon has quit [Quit: Leaving]
<{^_^}>
[nixpkgs] @grahamc pushed to arm-bootstrap « armv{5tel,6l,7l}: bootstrap-files: quote URLs for rfc-45 »: https://git.io/JvPVv
malSet has quit [Read error: Connection reset by peer]
malSet has joined #nixos
malSet has quit [Read error: Connection reset by peer]
erasmas has quit [Ping timeout: 250 seconds]
gustavderdrache has joined #nixos
<ottidmes>
dingenskirchen: Thanks, that helped! Always forget about that piece of software, but by checking the #nixos channel log for strace and getting the right command for files, I managed to figure out how it is determined
<dingenskirchen>
glad to have helped :)
mjrosenb has joined #nixos
<{^_^}>
[nixos-homepage] @garbas pushed commit from @samueldr to master « Strip nix store paths from source nixpkgs in packages.json (#352) »: https://git.io/JvPVq
<{^_^}>
[nixos-homepage] @garbas merged pull request #352 → Strip nix store paths from source nixpkgs in packages.json → https://git.io/JvPzF
leomar has joined #nixos
linarcx has quit [Quit: WeeChat 2.7.1]
Jackneill has quit [Remote host closed the connection]
ben_ is now known as ben
fresheyeball has joined #nixos
mjrosenb has quit [Ping timeout: 264 seconds]
codygman has joined #nixos
<ottidmes>
Considering it tries to load /no-cert-file.crt in nix-shell --pure it matches the value of SSL_CERT_FILE, but I tried changing it and exporting it, and I tried prepending it before the command, but it still keeps using the old value
<ottidmes>
Without --pure it loads /nix/store/6566wp1fvzbdy242czysqyg8l7wxn9im-nss-cacert-3.49.2/etc/ssl/certs/ca-bundle.crt, which also matches SSL_CERT_FILE at that point in time
<Ashy>
ottidmes: i'm going the other way, from zfs to btrfs
<ToxicFrog>
Ashy: I used to, it was seriously bad times, ditched it after a few months
leomar has quit [Quit: leomar]
codygman has quit [Client Quit]
codygman has joined #nixos
<Ashy>
ToxicFrog: performance issues or actual data loss issues?
<Ashy>
btrfs seems to be so contentious whenever i ask who's using it and what their experience is
codygman has quit [Client Quit]
<Ashy>
filesystems in general seem incredibly disappointing in 2020 :/
codygman has joined #nixos
<misuzu>
all hopes on bcachefs
o1lo01ol1o has quit [Remote host closed the connection]
<ottidmes>
I can't complain with performance and and data loss on my desktop, have been running the same btrfs for 5 years now, without problem, but I did fail on me once on a server
<Ashy>
misuzu: i should probably chip in some money towards bcachefs development
<cole-h>
bqv[m]: Also lib.traceValSeq might help
<fresheyeball>
ottidmes: any idea why it's broken on nixos?
<ottidmes>
bqv[m]: nix eval '(builtins.fromJSON (builtins.readFile ...))' > file.nix, should do the work, no?
<cole-h>
(if you want to peek into the <CODE> and similar stuff)
<ToxicFrog>
Ashy: massive performance issues (poor throughput plus some individual read/write operations taking 30+ seconds to complete on an SSD). Minor data loss when it started returning ENOSPC when less than half full. The mitigations on the wiki did not help.
Blessjah has quit [Ping timeout: 256 seconds]
fresheyeball has quit [Quit: WeeChat 2.6]
<{^_^}>
[nixpkgs] @veprbl opened pull request #82760 → [ZHF] kicad: don't build on Hydra versions with 3d → https://git.io/JvPw0
<bqv[m]>
ah yes
<bqv[m]>
works
<bqv[m]>
thanks guys
Blessjah has joined #nixos
<Ashy>
ottidmes: wow 53% compression ratio on /nix/store is much better than i expected
<Ashy>
ottidmes: have you tried zstd at all?
<ToxicFrog>
Fortunately I learned the terrible secret of btrfs RAID1 (it's not actually RAID1) from other people rather than encountering it in production.
<Ashy>
ok, new compression algo from facebook (but foss and patent free from what i understand), supposedly better compression ratio and speed than lzo
fresheyeball has joined #nixos
<Ashy>
ToxicFrog: that ENOSPC issue sounds fairly worrying