<ldlework>
energizer: do you use Xephyr to test your changes?
<energizer>
no
<ldlework>
ah, that's pretty handy
<ldlework>
If you want some help debugging qtile let me know
<energizer>
thanks
<pie_>
energizer: i havent worked on that project in a few weeks
<energizer>
pie_: which qtile and nixpkgs version are you running?
timotheel-f[m] has joined #nixos
xelxebar has quit [Remote host closed the connection]
justanotheruser has quit [Ping timeout: 240 seconds]
xelxebar has joined #nixos
<energizer>
ok current issue: using impermanence/home-manager.nix hm-activate-energizer: fuse: Activating createAndMountPersistentStoragePaths hm-activate-energizer: fuse: mountpoint is not empty
<energizer>
how do i find out which mount point it's talking about?
puck has quit [Ping timeout: 264 seconds]
puck has joined #nixos
<pie_>
energizer: try evaling the appropriate expression in the repl and look at the output?
rprije has joined #nixos
<pie_>
energizer: alternatively, strace?
<pie_>
-e to filter, but check the man page which filter is for exec
<cole-h>
energizer: I'd just look at every mountpoint, tbh
<cole-h>
`ls -al /mntpoint` and look for which one has files in it
<energizer>
they all say, Transport endpoint is not connected (os error 107)
<euank>
ntnkwmkmnw[m]: yes, it is. There's probably a better way if you give a more specific explanation of what you mean by 'change a single package in my local nix store' though
<euank>
ericmoritz: could be worth trying a different nixpkgs commit you know was definitely cached, or debugging if other nix commands use the cache correctly
<euank>
ntnkwmkmnw[m]: oh, I see what you mean; edit a single package in nixpkgs. I think I misunderstood your question. The keyword you might want is an overlay
<ericmoritz>
euank: thanks that pointed me in the right direction
mariatsji has quit [Ping timeout: 272 seconds]
<ntnkwmkmnw[m]>
euank: aaah okay. Will look into that. I'm trying to change a single number for a version change.
<euank>
Ah, yeah, for that an overlay is probably what you want! I can give an example of that if you'd like
jybs- has quit [Ping timeout: 258 seconds]
<euank>
an overlay, or `pkg.overrideAttrs`; both are discussed in the nixpkgs manual. In general, overlays are more powerful and less surprising
<ntnkwmkmnw[m]>
euank: Sure! I keep hearing about them. I've been doing some `overrideAttrs` in a few nix-shells but never in my configuration
jybs- has joined #nixos
ris has quit [Ping timeout: 258 seconds]
<euank>
is this nixos configuration.nix, or config.nix on another distro?
o1lo01ol1o has quit [Ping timeout: 256 seconds]
<ntnkwmkmnw[m]>
nixos configuration.nix
cr4y1 has quit [Ping timeout: 256 seconds]
jtmcx has quit [Quit: WeeChat 2.7.1]
<ntnkwmkmnw[m]>
okay I actually got it by cloning my own copy of nixpkgs, which was something I should probably just suck it up and learn to do anyway
diamondburned has quit [Ping timeout: 240 seconds]
proofofkeags has quit [Quit: Leaving]
diamondburned has joined #nixos
proofofkeags has joined #nixos
rajivr has joined #nixos
<{^_^}>
[nixpkgs] @orivej-nixos pushed commit from @orivej to master « mamba: 1.3 -> 1.4 »: https://git.io/JUWLU
spease has joined #nixos
<spease>
I'm trying to build with crossSystem, but I get a recursion error with cmake in nativeBuildInputs. This seems like it forces me to create two pkgs variables, one for nativeBuildInputs and one for buildInputs. But I don't know how to override the crossSystem attribute for the second pkgs variable. Am I going down the wrong path?
cmk_zzz has joined #nixos
cmk_zzz_ has quit [Ping timeout: 260 seconds]
knerten2 has joined #nixos
<{^_^}>
[nixpkgs] @orivej opened pull request #97728 → seq66: init at 0.90.5 → https://git.io/JUWLu
<ornxka>
nooooooooooo the nixos-unstable channel is building again
<ornxka>
i was so happy with not having to update
domogled has joined #nixos
CyberManifest has quit [Quit: Leaving...]
maxfield has joined #nixos
Spiney has quit [Ping timeout: 258 seconds]
Spiney has joined #nixos
Supersonic112 has joined #nixos
Supersonic has quit [Ping timeout: 272 seconds]
Supersonic112 is now known as Supersonic
Darkmatter66_ has joined #nixos
Darkmatter66 has quit [Ping timeout: 265 seconds]
xelxebar_ has joined #nixos
xelxebar has quit [Ping timeout: 240 seconds]
domogled has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @marsam merged pull request #97557 → [20.03] terraform_0_12: add patch with fix for macos mojave when built with go 1.14 → https://git.io/JUcDv
<{^_^}>
[nixpkgs] @marsam pushed 2 commits to release-20.03: https://git.io/JUWqZ
<zecnate>
Maybe. If I run that `nix-prefetch-url` command, and then run `nix-build ./my-derivation.nix` will the second command skip downloading that same archive?
cransom has joined #nixos
_habnabit has quit [Ping timeout: 240 seconds]
tpw_rules has joined #nixos
devalot has joined #nixos
<immae>
Ah ok got it, no
_habnabit has joined #nixos
<zecnate>
My goals are: (1) ensure the derivations for `pkgs` map to a known hash, and (2, bonus) separate the fetch/store update process in one command from the `nix-build`.
<zecnate>
I think with `pkgs ? import (fetchTarball …)` I am guaranteeing that `pkgs` are based on the specified hash _unless_ the caller somehow provides a different `pkgs` binding, right?
<redmp>
i'm trying to build a project in nix, but running the tests requires that several _dependent packages_ are already installed..
<zecnate>
`nix-build` requires a derivation in `--expr`.
<immae>
well, it is
<zecnate>
`fetchTarball` is?
<immae>
hmm what else would it be?
<zecnate>
I'm not quite sure yet. New to nix. My current understanding is "derivations only come as a result of evaluating `derivation`", but I guess I don't know if `fetchTarball` evaluates to a `derivation` call or not…
<redmp>
how can i pass a package in as a dependency in haskellOverrides? i think i could use that trick to break the circular dependency cycle
<zecnate>
Hrm, when running `nix-prefetch-url` I see this output:
<zecnate>
```
<zecnate>
1vbi3pnyfkxljfxc8aq4kpv9giv3mxqh861k86n002r18hhvbl94 path is '/nix/store/zndz0fk1mw33ryk5bgdy9is9nfx0j6ss-b58ada326aa612ea1e2fb9a53d550999e94f1985.tar.gz' ```
<zecnate>
Why is the hash on the first line different than the store file name?
<colemickens>
the store path name is computed from more things than just the source
proofofkeags has quit [Ping timeout: 258 seconds]
da39a3ee5e6b4b0d has joined #nixos
<zecnate>
Thanks. I'll look into store path names…
endformationage has quit [Quit: WeeChat 2.9]
<redmp>
anybody here know of a good way to break a circular dependency with nix? i'm trying to put this together in my head and failing
<zecnate>
I'm completely new to nix, so caveat emptor. My understanding is that circular references may be impossible, because the dependency derivations are passed as inputs to a derivation, so what order would they be resolved in?
spease has quit [Remote host closed the connection]
<zecnate>
While reviewing the manual though, I got a bit tripped up because the example of creating a derivation requires editing `all-packages.nix`: https://nixos.org/manual/nix/stable/#sec-arguments
<immae>
zecnate: how do you want to "create a package"?
<immae>
If you want to do it locally in your script, then just do `let some_variable = 'your derivation'; in (...)`
mariatsji has quit [Ping timeout: 256 seconds]
<immae>
redmp: the first thing is to obtain information about the circular dependency, did you try --show-trace for instance to see where the icrcular comes in play?
stiell has joined #nixos
<cole-h>
jtojnar++ Thanks for all your work on fontconfig. You're truly a magician.
<{^_^}>
jtojnar's karma got increased to 61
<{^_^}>
[nixpkgs] @treed opened pull request #97734 → acme-client: build with gcc → https://git.io/JUWYw
<{^_^}>
[nixpkgs] @orivej-nixos pushed commit from @orivej to master « seq66: init at 0.90.5 (#97728) »: https://git.io/JUWYX
<redmp>
what's the point of the extra {} argument passed to a lot of packages?
<redmp>
maybe this is only in the haskell development modules..
<clever>
redmp: callPackage takes the path to a package, and a set of overrides for the args on line 1 of the package
<clever>
{} is an empty set of overrides
* clever
heads to bed
mmohammadi9812 has quit [Quit: I quit (╯°□°）╯︵ ┻━┻]
palo1 has joined #nixos
zupo has joined #nixos
palo has quit [Ping timeout: 264 seconds]
palo1 is now known as palo
<redmp>
immae: thanks for your suggestion; the circular dependency isn't in nix.. it's in the package.. it's only expressed through a maze of configuration files.. effectively, they can't run their tests until they install a bunch of dependent packages
<cole-h>
worldofpeace: btw, I have succeeded in getting the build-vm stuff to work with flakes (in my flake.nix, at least -- which requires impurity since I'm importing something from nixpkgs)
<cole-h>
Next step is to try to get it working in nixpkgs itself
mariatsji has joined #nixos
<{^_^}>
[nixpkgs] @orivej opened pull request #97735 → nixos/plasma5: fix build with hardware.bluetooth.enable = true after #97456 → https://git.io/JUWOW
karantan has joined #nixos
<siraben>
NixOS won't let me do echo core >/proc/sys/kernel/core_pattern, what do I do instead?
detran has quit [Ping timeout: 272 seconds]
<siraben>
I want to be able to fuzz using afl
<etu>
siraben: That file is owned by root, so you need to be root user to do that, and putting sudo in front won't do it because that would run echo as sudo and then pipe it as your user.
treed[m] has joined #nixos
zeta_0 has quit [Quit: rcirc on GNU Emacs 27.1]
karantan has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
redmp has quit [Quit: leaving]
<Mic92>
siraben: echo core | sudo tee /proc/sys/kernel/core_pattern
<cole-h>
omg no way it worked. Now to patch nixos-rebuild.sh :D
<colemickens>
? build-vm stuff working with flakes?
<cole-h>
colemickens: Yeah, patching nixpkgs rn
<treed[m]>
So I'm working away at some low hanging fruit I can find for ZHF/Darwin. I've found multiple perlPackages where the failure can be fixed by adding "export LD=$CC" to the derivation, which is already done for several packages. Is it worth opening an issue to pursue some sort of systematic fix for this, or should I just do one-off PRs to fix each broken package? (Or maybe both)
<cole-h>
It shouldn't, because there is no vm attr...
<colemickens>
oh, I mean, I added the module to that system config
<cole-h>
Oh
<growpotkin>
Are DPMS timeouts just "suggestions" passed to a Monitor's driver, or does the X system explicitly tell the monitor to Standby/Suspend/Power-Off based on timeout settings? Mine just seem to either stay on forever when DPMS is disabled, or turn off after 10 minutes with no regard for timeout settings when DPMS is enabled.
<cole-h>
I mean, yeah, you could do that
<cole-h>
But it would be nice to not have to do that ;P
graf_blutwurst has joined #nixos
<colemickens>
oh interesting
<colemickens>
this is neat, thanks for explaining/linking :)
<cole-h>
colemickens: Almost done. Man I am so happy this worked.
o1lo01ol1o has quit [Ping timeout: 240 seconds]
[krow] has joined #nixos
simonpe^^ has joined #nixos
<simonpe^^>
Hi! How would I go about figuring out why two machines produce different store paths for the same derivation?
<simonpe^^>
it's an impure build
<cole-h>
Seems like you answered your question there.
<zecnate>
In my derivation I can add `curl = pkgs.curl` to make it accessible to my builder. But my builder calls make, and make expects just `curl`. Is the standard thing to do to modify `PATH` before calling `make` in the builder script?
<cole-h>
Impure builds won't have the same store path, IIRC.
<simonpe^^>
cole-h: how are the store paths calculated?
mmohammadi9812 has joined #nixos
<cole-h>
simonpe^^: You're better off reading Eelco's thesis for that (it's probably covered in there).
<zecnate>
Does `$stdenv/setup` place things like `make` on the path?
<cole-h>
Use. Anything that needs to be executed during compile time should go into `nativeBuildInputs`
kroh has quit [Ping timeout: 265 seconds]
<cole-h>
s/Use/Yes/
<Jake[m]>
"cole-h" (https://matrix.to/#/@freenode_cole-h:matrix.org): I don't know exactly how, but you can also look at the derivation, a textual representation of all the inputs.
<Jake[m]>
Any inputs that don't match would change the store path
<{^_^}>
[nixpkgs] @maxxk opened pull request #97737 → python3Packages.parse: 1.16.0 -> 1.18.0, python-docx: run behave tests → https://git.io/JUWsZ
Pwnna has quit [Quit: Bye]
<zecnate>
Hrm, seems like curl can't load/find CA certificates…
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Pwnna has joined #nixos
<{^_^}>
[nixpkgs] @cole-h opened pull request #97738 → nixos-rebuild: add flake support for build-vm → https://git.io/JUWsu
<cole-h>
colemickens: ^^ 🎉
<zecnate>
@cole-h Thanks! I just did a search for "nativeBuildInputs" in https://nixos.org/manual/nix/stable/ and got no hits. Where can I learn about this?
<zecnate>
If a channel has a version number and not the string "unstable", then is the largest version the latest "stable" release? https://status.nixos.org/
<zecnate>
Does that mean 20.09 is the latest stable (even though many docs refer to 20.03 as stable) ?
stiell has joined #nixos
<zecnate>
@cole-h: yeah, this nixpkgs contributor manual is perfect. I hadn't found that before. Thanks!
<symphorien[m]>
it's not released yet
<symphorien[m]>
it should at the end of the month
alexherbo2 has joined #nixos
saschagrunert has joined #nixos
alexherbo20 has joined #nixos
stiell has quit [Ping timeout: 265 seconds]
alexherbo2 has quit [Ping timeout: 256 seconds]
alexherbo20 is now known as alexherbo2
sangoma has joined #nixos
stiell has joined #nixos
jb55 has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @ju1m opened pull request #97740 → tor: improve type-checking and hardening → https://git.io/JUWGd
<zecnate>
I'm using `nativeBuildInputs = [ pkgs.curl ]` on the latest 20.09-small release of nixpkgs.
<zecnate>
Has anyone else seen this kind of issue with curl recently? Am I just missing a known setup step?
<treed[m]>
I only use curl on the command line but I haven't had any such issues there at least.
<symphorien[m]>
zecnate: network is inaccessible inside nix builds (there is a sandbox)
<zecnate>
Woah! Ok, first that's awesome, and second is there a way to remove restrictions?
<zecnate>
(What kind of sandbox? I'm running nix inside of Docker. Is there a recursive container?)
abathur has quit [Ping timeout: 240 seconds]
<symphorien[m]>
I think there are ways to escape the sandbox, but you'll spare time by trying to not need network access instead
alp has joined #nixos
<zecnate>
The funny thing is inside my builder I'm running a build system that fetches dependency tarballs, checks their hashes, then builds them into specific directories… sound familiar?
<symphorien[m]>
just linux namespaces
<symphorien[m]>
sorry, afk
abathur has joined #nixos
<treed[m]>
You could fetch the tarballs from the derivation itself as part of srcs
mallox has joined #nixos
<zecnate>
Ok, so like I said, I have this build system that's basically a home-brew version of nix. My project is to replace it with nix itself, but I had hoped to do this in stages, where the first stage just treats the whole thing like a monolithic black box, then _later_ I start to split out dependencies with nix.
<treed[m]>
Mmm
<zecnate>
-but for that to work, the build would need to be able to fetch tarballs. :-/
<treed[m]>
I thought the sandbox was opt-in personally.
<treed[m]>
I turned it on recently for working on stabilizing 20.09
<zecnate>
I'm using `stdenv`. Does that implement the sandbox?
<treed[m]>
Right
<treed[m]>
Do you know ahead of time what all the tarballs are?
<NinjaTrappeur>
The sandbox is directly implemented in the nix builder itself.
<treed[m]>
I think it's the nix build daemon that does the sandboxing
<NinjaTrappeur>
Which build system is trying to fetch tarballs here? Something custom?
<zecnate>
treed: yes. The custom build system fetches them, then verifies them by hash. Also, if they are present in a cache directory it accepts that.
mmohammadi98126 has joined #nixos
<NinjaTrappeur>
right
karantan has joined #nixos
<zecnate>
So if I can figure out how to have nix place them into the appropriate cache directory, that could work.
<treed[m]>
Ah okay
<zecnate>
NinjaTrappeur: yeah, it's a custom build system made with makefiles that has very similar feature goals to nix. :-)
civodul has joined #nixos
<treed[m]>
So what you can do is say like srcs = [ fetchurl { } ... ]
mmohammadi9812 has quit [Ping timeout: 240 seconds]
mmohammadi98126 is now known as mmohammadi9812
<NinjaTrappeur>
So as long as you have a expected output sha256, you can wrap those in fixed output derivation, where network access are allowed.
<treed[m]>
And then the builder will do the fetching itself
<NinjaTrappeur>
Or you can use the builtin fetchurl
<treed[m]>
and provide you the paths in the build environment as $srcs
jb55 has joined #nixos
<treed[m]>
It'll unpack them for you as well if you let unpackPhase happen
buckley310 has quit [Quit: Connection closed for inactivity]
<zecnate>
So `srcs = [ fetchurl { } ]` gets me kind of close, except they all need to be in a specific directory. I think I can specify the directory itself with a make override, but I don't think I can specify each tarball path separately.
<zecnate>
Hm, I guess I could make a derivation that ties those all together with symlinks?
<simpson>
gunix: Reading through some of them, it seems like we patch Go in order to make it behave better on Nix. Disabling busted tests, adding Nix's SSL certs, etc.
<oleks>
I mean, I can't find a dedicated package for it.
alexherbo2 has quit [Ping timeout: 260 seconds]
alexherbo24 is now known as alexherbo2
<simpson>
,locate nslookup
<etu>
oleks: It's part of the bind package
<{^_^}>
Found in packages: zsh, busybox, bashCompletion, dnsutils.dnsutils
<etu>
And dnsutils it seems, nice
<simpson>
`bind.dnsutils` is one of those quirky package sets.
<gunix>
simpson: are there many packages that have patches and are slightly different from upstream?
<simpson>
gunix: We do patch all the time (and have lots of ad-hoc patching actions like running sed on source code), but I would argue that these don't produce slight differences from upstream. Usually we're removing build-time tests which don't work under the Nix build sandbox.
<simpson>
The slight differences we do introduce are likely *desired*; patching Go so that it looks at $NIX_SSL_CERT_FILE when looking for system SSL certs seems like a fix, not a big divergence.
<simpson>
To be more direct: Did something in this patchset break your workflow?
<{^_^}>
[nixpkgs] @endgame opened pull request #97741 → aws-sam-cli: fix version bounds → https://git.io/JUWcs
<gunix>
simpson: no, i was just curious
da39a3ee5e6b4b0d has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
kahuna has quit [Ping timeout: 256 seconds]
kahuna has joined #nixos
<immae>
Hello there! With the hope to ravive nixpart, I started by repackaging blivet to its latest version. You can find the derivation here: https://temp.immae.eu/blivet/blivet/ . However I’m a bit unhappy: I can successfully use it when invoked as `nix-shell -p python3 python3Packages.blivet3` but not with a `python3.withPackages(p: [p.blivet3])`, where I end up with a "cannot find BlockDev" error.
<immae>
What did I miss? (the BlockDev comes from the libblockdev dependency, which seems to be ignored in the later case)
dermetfan has joined #nixos
mallox has quit [Quit: WeeChat 2.9]
fendor has joined #nixos
<Raito_Bezarius>
aszlig: ^
arjen-jonathan has joined #nixos
gaston12 has joined #nixos
<colemickens>
I do not understand how gpg is supposed to work on nixos sometimes
<{^_^}>
[nixpkgs] @tasmo opened pull request #97742 → sunvox-1.9.5d - new hash for source zip → https://git.io/JUWCI
<colemickens>
I specifically know that bad things happen if you try to restart gpg-agent and get in this state, because gpg comamnds will restart the gpg-agent in a non-functional state.
<hyper_ch2>
gpg with pass works just fine
<colemickens>
and yet, I can't see to manually start the sevice.
<colemickens>
hyper_ch2: `systemctl --user stop gpg-agent` and then recover
<colemickens>
hyper_ch2: I have done it before, I can't remember how to do it now
<damjan>
is there any simple text file template support in nix, I do need a simple `if` in a config file, apart from the typical @@ replacements
<gaston12>
Hi guys. I have a problem for a couple of days with my servers newly installed. I try to use IPv6 at most and I would like to have a fix address. I tried to set one up but it disapear after some times (probably because of the privacy extention)
<aszlig>
immae: the nix-shell invocation given that you added the mentioned expressions as an overlay to python3Packages, right?
<gaston12>
Hi guys. I have a problem for a couple of days with my servers newly installed. I try to use IPv6 at most and I would like to have a fix address. I tried to set one up but it disapear after some times (probably because of the privacy extention). Does anyone of you know how to deal with it on servers?
<immae>
aszlig: ah yes right (or in the nixpkgs source tree)
<hyper_ch2>
colemickens, I have nothing special
<hyper_ch2>
it just works
<hyper_ch2>
gpg is in the list of install packages in my configuration.nix
teto has joined #nixos
<colemickens>
hyper_ch2: I think it's probably related to the HM module
<immae>
aszlig: I can get a shell fine, it’s just that I cannot correctly use blivet in the resulting python shell
<aszlig>
immae: ah, okay, the reason *might* be that the necessary environment for gobject-introspection might not be set, IIRC they were using GI for libblockdev
<hyper_ch2>
I don't use home manager
<immae>
aszlig: they do indeed (it’s a dependency) :)
<immae>
How could I "fix" that in the context of a python3.withPackage?
<aszlig>
immae: i'd make sure to patch in the GI_TYPELIB_PATH (IIRC that it was called) into the corresponding library
kahuna is now known as redkahuna
<immae>
aszlig: It’s not easy to "patch in" blivet since it’s meant to be used as a library
<immae>
There is no wrapper I can put there
<immae>
But I get the idea, I’ll try to find a solution thanks :)
thc202 has joined #nixos
<aszlig>
immae: well, maybe not GI_TYPELIB_PATH but find a more direct way to load those bindings
<{^_^}>
[nixpkgs] @galagora opened pull request #97744 → haskell-language-server, stylish-haskell: Fix build on GHC 8.10.* → https://git.io/JUWCB
<immae>
aszlig: yes, I had to do it for PATH already I’ll find out don’t worry ;)
<aszlig>
immae: okay, thanks for reviving that though :-)
<immae>
:)
tmaekawa has joined #nixos
tmaekawa has quit [Client Quit]
davidv7 has joined #nixos
<gaston12>
does some of you already used IPv6 on servers? I disabled tempAddress and fix one but it disappear after a couple of time
<gaston12>
dminuoso cool: so what is the proper way on server? disable privacy extension?
m0rphism has joined #nixos
<gaston12>
and if yes, how to do it on nixos? I tried tempAddress = 0 , the address is not flagged as "temporary" anymore but still disapear after a couple of tie
<gaston12>
and if yes, how to do it on nixos? I tried tempAddress = 0 , the address is not flagged as "temporary" anymore but still disapear after a couple of time
jasongrossman has joined #nixos
<dminuoso>
gaston12: the privacy extension is for mobile clients, such as mobile phones or laptops.
<gaston12>
it is what I have understood, but then what about my second question?
<dminuoso>
Ah hold on
<dminuoso>
Sorry I misremebered.,
<dminuoso>
The privacy extension is to ensure your IP address is sort of randomized to reduce tracking
<dminuoso>
If your server needs reachability, you likely dont want it slaac configured at all
h0m1 has joined #nixos
<dminuoso>
(Since otherwise changing a NIC would change its IP address, so SLAAC is a bad fit)
<dminuoso>
The privacy extension would make it change definitely every time the link flaps
<{^_^}>
[nixpkgs] @NinjaTrappeur opened pull request #97746 → test-driver.py: defaulting keepVmState in Machine init → https://git.io/JUWWs
<dminuoso>
So just statically assign an address with networking.interfaces.<name?>.ipv6.addresses (or run DHCPv6)
<gaston12>
it is what I thought, I should disable it. I was wondering whether it could be possible to add something like an alias which stays and other addresses which changes.
<Dagger>
privacy extensions give you additional addresses, on top of the base SLAAC address
<gaston12>
no it is not enough. I tried to add an address but it disapears after a couple of time.
<siraben>
Anyone want to help with #97685? After writing a script to check, I found hundreds of packages whose descriptions end with a period. Should the fix be one PR or lots of them?
<Dagger>
you don't need them off for a server; just put the base address into DNS rather than one of the temporary addresses
<siraben>
To fit CONTRIBUTING.md*
gaston12 has quit [Remote host closed the connection]
albert[m] has quit [Quit: Idle for 30+ days]
Niobium[m] has quit [Quit: Idle for 30+ days]
lina[m] has quit [Quit: Idle for 30+ days]
gaston12 has joined #nixos
noderunner_[m] has quit [Quit: Idle for 30+ days]
noderunner[m] has quit [Quit: Idle for 30+ days]
<gaston12>
Dagger I don't understand. I added simply a fix IPv6 address. It is flagged as tempororary as the other "public" one and after a couple of time both are flagged as "deprecated" and finaly disapears
<__monty__>
siraben: If you're just removing periods I'd group the commits in one PR. If you're rewriting descriptions to fit the guideline I'd make separate PRs.
<siraben>
__monty__: any way to edit the files automatically using Nix?
<siraben>
I have the list of packages whose descriptions should be edited
superherointj has joined #nixos
<Dagger>
can I see the output of `ip addr`? manual addresses shouldn't end up with the temporary flag
<__monty__>
siraben: I'm not sure what you mean. There's `nix edit somepackage` but you'll want to make your changes in the nixpkgs repo anyway.
<siraben>
__monty__: what is mean is, there's hundreds of packages to change and that would involve navigating to the right file and editing it
<gaston12>
@dagger sory not now. But I will give it a second try, without having tempAddress=0 .
<gaston12>
(I am not at home now)
<siraben>
I guess I could use `nix edit`, then run a shell script or something
<gaston12>
thx for having helped
<__monty__>
siraben: I guess you could in principle use nix to generate new descriptions. `(s: substring 0 (stringLength s - 1) s) mypackage.meta.description`
<__monty__>
But I don't think there's pretty printing for expression so you'd have to splice that new description into the files somehow.
<siraben>
Ok
<__monty__>
siraben: Maybe hnix or the rust library could help.
<{^_^}>
[nixpkgs] @Ma27 pushed to master « wdisplays: fix startup on sway 1.5 »: https://git.io/JUWWd
<siraben>
`ag '(\ )+description(.*)\."'` rather, because there could be `longDescription`
<dutchie>
don't think you need any of the brackets
<NinjaTrappeur>
\s
gaston12 has quit [Ping timeout: 245 seconds]
<{^_^}>
[nixpkgs] @Ma27 pushed commit from @mweinelt to release-20.09 « pythonPackages.supervisor: drop glibc-2.31 patch »: https://git.io/JUWlG
<NinjaTrappeur>
indeed, no need for brackets here
stteevveen has joined #nixos
sigmundv has joined #nixos
<stteevveen>
Hi! I can get <nixpkgs> value inside nix repl with simply <nixpkgs>, and that retruns the path of it. I was wondering how you do it when you are on the command line and not inside nix repl ?
aborsu has joined #nixos
<__monty__>
That regex matches way more than just a short description with a trailing dot though.
<__monty__>
And why not simply use sed to find the matches if you're gonna use sed for editing anyway?
<infinisil>
stteevveen: `nix-instantiate --find-file nixpkgs` is one way
<infinisil>
stteevveen: `nix-instantiate --eval -E '<nixpkgs>'` another
<hyper_ch2>
nixos is different... very different from traditional linux systems
<FloNix>
so, what i want to achieve is a hetzner cloud, running multiple servers as a kubernetes cluster i could use for my dev stuff i do at home (personal) thats why i want to go to hetzner because it's cheaper like aws or azure etc.
<dminuoso>
Any article starting with "you should never" is likely just random FUDding that you would usually find on medium.com
<FloNix>
i read it, some points may be true (i don't know) some others are just whining
<simpson>
Wait, is that article serious? I thought it was satire.
da39a3ee5e6b4b0d has joined #nixos
<simpson>
Like, all of these good things, reworded to pretend that they're bad things, yeah?
<hyper_ch2>
FloNix, nixos can run kubernetes... but I never run kubernetes.... also with nixos, one of the links you must have is: https://search.nixos.org/options --> there you can see the existing modules for easy configuration and setup
<FloNix>
so first question: could nixops spawn a new machine in hetzner cloud like it does with virtualbox?
<eyJhb>
If I have something `let test = 1; in {}`, is it then possible to access the vars in let, outside of the file by importing it? `let myFileWithLet = import letfile.nix; in { myFileWithLet.test}`?
<teto>
eyJhb: no
<eyJhb>
Damn.
<eyJhb>
I guess I will have to make it into some options then, and use that
<eyJhb>
Should be better anayways
cosimone has joined #nixos
graf_blutwurst has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @lheckemann opened pull request #97756 → xva-img: init at 1.4.1 → https://git.io/JUWBe
winny has joined #nixos
<hyper_ch2>
FloNix, second line of that wiki page: "This is not to be confused with Hetzner Cloud, that provides VMs (an example for how to install NixOS there is shown here).
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<michaelpj>
if I build a package with a remote builder which has dependency D, and D is present on both my machine and a substitutor for the builder, will the builder download it from the substitutor or me by preference? it seems like it's getting it from me, and I'd really prefer it to get it from the substitutor...
zupo has joined #nixos
<{^_^}>
[nixpkgs] @ttuegel merged pull request #97735 → nixos/plasma5: fix build with hardware.bluetooth.enable after #97456 → https://git.io/JUWOW
<Raito_Bezarius>
I'd like to add it to my python3.withPackages
<Raito_Bezarius>
But I don't see it sym-linked in the site-packages
mmohammadi9812 has quit [Quit: I quit (╯°□°）╯︵ ┻━┻]
nij has left #nixos [#nixos]
<Raito_Bezarius>
toPythonModule sounds like what I want
seku has joined #nixos
<teto>
Raito_Bezarius: toPyhtonModule just adjusts the `meta` attribute so that the derivation is marked as a python one. If you want to really build it, you still need to package it I think but you can set the `src` to the one of the other package (or a subfolder)
<Raito_Bezarius>
I don't need to (re)build it
<Raito_Bezarius>
I just want it to be discovered in the site packages
<stteevveen>
Hi! I'm reading this post https://discourse.nixos.org/t/difference-between-fetchgit-and-fetchgit/3619 but it still does not ring a bell for me. What are your personnal opinion on when using fetchGit instead of fetchgit. In my particular case it solves me an issue when I have to select a special branch for building, and fetchGit has a ref attribute
<stteevveen>
for indicating the branch to checkout, instead of a rev hash. What is your personnal rule of thumb on choosing fetchGit or fetchgit ?
<NobbZ[m]>
I have a ./result link that already exists, and I want to add it to the GC roots, I hoped nix-store --add-root $(pwd)/result --indirect would do, though it complains "error: no operation specified". How can I make that ./result a GC root when I do not know how it was originally created?
orivej has quit [Ping timeout: 258 seconds]
allgreed has joined #nixos
<freezeboy>
NobbZ[m] if you keep the result file it will not be removed by GC
<freezeboy>
then you can use --out-link / -o outlink
<NobbZ[m]>
To describe the problem in more detail: I'm `home-manager build` for various configs, though `home-manager` doesn't allow to specify the name for the result. So I am trying to get all possible results
alexherbo2 has quit [Ping timeout: 240 seconds]
<freezeboy>
are you sure you can't add the --out-link option ?
<NobbZ[m]>
It does not pass `--out-link`/`-o` through to `nix-instantiate`
<NobbZ[m]>
That doesn't seem to work, it creates `./result` and `./result-2`. for each iteration. I now use good old `ln -s`, it seems to work for my purposes.
da39a3ee5e6b4b0d has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<simonpe^^>
how can I impurely pass an environment variable to a derivation so that it doesn't change the result?
<simonpe^^>
its an impure build and the variable contains a mirror used by the build script, which shouldn't be available on the build agent since it points to itself
<simonpe^^>
and no, this builder cannot be made impure, its nothing I can do about that unfortunately
<clever>
simonpe^^: is it a fixed-output derivation?
<clever>
> hello.src.impureEnvVars
sigmundv has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @freezeboy closed pull request #97760 → cassandra-driver: disable python2.7 as upstream dropped support → https://git.io/JUWuP
noudle has quit []
saschagrunert has joined #nixos
growpotkin has joined #nixos
<gchristensen>
it is really annoying that /run/keys is root:keys and drwxr-x---
saschagrunert has quit [Client Quit]
<flokli>
gchristensen: how so?
<gchristensen>
it means you have to run your process with Group=keys to access the secret
<gchristensen>
but maybe you wanted to use Group for something else, and so now you have to make a choice about copying it elsewhere in a pre-start, or doing something else scary
<gchristensen>
or am I mistaken?
ransom has joined #nixos
<gchristensen>
flokli: I'm not wrong about that, right?
<qyliss>
Can't you have multiple groups?
<flokli>
gchristensen: you can have multiple groups
<evanjs>
I mean it really depends on the type of package. Often times for e.g. GitKraken and Postman, they're already binary packages, so I have to wonder what the real benefits would be
<evanjs>
Otherwise, thanks for bringing that issue to my attention! I did not realize (or forgot, lol) that there was a tracker for that
<{^_^}>
[nixpkgs] @DamienCassou merged pull request #84246 → couchdb: add support for version 3.0.0 → https://git.io/JvbVO
<{^_^}>
[nixpkgs] @DamienCassou pushed 2 commits to master: https://git.io/JUWSE
<evanjs>
I am definitely a "use-whatever-works-best-for-me" person -- e.g. _All_ the JetBrains stuff, GitKraken, Postman, etc
<evanjs>
bah jetbrains are plain unfree. Figures
maier has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @ehmry opened pull request #97785 → nim: remove stdlib indirection in wrapper → https://git.io/JUWSg
da39a3ee5e6b4b0d has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<evanjs>
I guess I'm even further down that hole :D but yeah I would imagine that would be fine. Trying to remember where I've seen mentions of why it isn't like that right now ...
<mkg20001>
Besides that, there are also other issue I'm aware of such as binary packages vanashing without maintainers noticing, like happened a lot with anydesk. So that prob could also solve that..
<mkg20001>
Hrm... Should I just go ahead and make a PR?
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « haskell-dhall: update override for ghc-8.10.2 »: https://git.io/JUWSr
<{^_^}>
[nixpkgs] @ttuegel opened pull request #97786 → plasma-workspace: fix path to qdbus → https://git.io/JUWSX
<evanjs>
I mean it would be just as good to gather thoughts, sure. That way, you could move it to half-done :P
<{^_^}>
[nixpkgs] @ttuegel opened pull request #97787 → nixos/plasma5: fix build with hardware.bluetooth.enable = true → https://git.io/JUWSb
<chreekat[m]>
Using the same store, 'nix copy' eventually works, but it spends 20 seconds timing out trying to do some authentication before giving up and proceeding successfully anyway. Is that also expected?
justache has quit [Remote host closed the connection]
<ajs124>
utf8cpp is only in nixpkgs since #89276, which is from june of this year, so it hasn't made it to any release yet and is only on the unstable channel.
<{^_^}>
[nixpkgs] @JamieMagee opened pull request #97806 → dotnetCorePackages.sdk_5_0: init at version 5.0.100-preview.8.20417.9 → https://git.io/JUlea
civodul has quit [Read error: Connection reset by peer]
NieDzejkob has joined #nixos
mariatsji has joined #nixos
isposdef has joined #nixos
mariatsji has quit [Ping timeout: 240 seconds]
steve-chavez has joined #nixos
__monty__ has quit [Quit: leaving]
<typetetris>
If I do `nix-shell -p haskellPackages.ghcWithPackages'(p: [])' -p zlib` it still complains it can't find libz.so in some projects. What do I need to do, to get that working?