<lightandlight>
Is there a low-maintenance way to aggregate multiple derivations into one, so that a single `nix-build` will build many derivations? I want to maintain a list of packages that my CI server should build and uploaded to a binary cache. Also, I'm not going to use hydra.
<colemickens>
lightandlight: nix-build will straight up build a list of attrs
<jackdk>
Does anyone have access to openstack who can help me test #96593 ? Pretty sure the AWS side works, and I don't expect it to break openstack images, but I would like to check. OpenStack's "devstack" scripts have exploded on every VM I've tried (ubuntu 18.04, RHEL8.2). I've tried asking here a few times, and on discourse.nixos.org "PRs in distress thread"; where else should I be looking for help?
<zecnate>
So I keep getting an autoconf `./configure` failure to detect one of my dependencies. I tried setting `NIX_DEBUG=6` on my derivation, and that gives a bunch of good `./configure` diagnostic output, but it doesn't show the specific `conftest.cpp` source file that's failing. Any advice on how to diagnose this issue?
marek has joined #nixos
butcher has joined #nixos
amfl has quit [Ping timeout: 256 seconds]
amfl has joined #nixos
joesventek has quit [Quit: Quit]
joesventek has joined #nixos
<zecnate>
BTW- I am porting another build system to nix, so I have logs and files from the other build system. AFAICT, the dependency directory in the nix store and the legacy build dependency directory have the same layout and files, with binary differences as well as `.cmake` file differences.
<zecnate>
The only cmake differences are that the working build has `# toolset=gcc7` whereas the nix build has `# toolset=gcc9`. Could that be related to the failure?
<euank>
zecnate: I've had luck with `breakpointHook` before and then hopping into the build environment from where it failed to poke around more
<zecnate>
Hm, the `config.log` of the two builds is notably different. The working/legacy build says checking for <foo dependency>, then shows a `g++` command with the specific `-I` option I expect. The nix version shows a commandline that does not have any `-I` options, and then that line is followed by a bunch of info like `extra flags after to
<euank>
do you know where that -I is coming from? pkg-config maybe?
<zecnate>
So what's going on here? Is the nix `g++` a wrapper around the real `g++` that is inserting these extra args that aren't shown on the commandline?
<euank>
those docs also recommend setting `NIX_DEBUG=1` to get more visibility into the wrapper
<zecnate>
For the working-legacy build I'm not sure if it relies on pkg-config. I think it might, but there's also a `config.site` that has the literal `-I`.
<zecnate>
In that build, all of the dependencies dump their outputs into a common directory so there's a single `-I` directory.
<zecnate>
In the nix build, I'm using the `stdenv` generic builder with very few tweaks, mainly `buildInputs` some from nixpkgs and some of my own creation.
<zecnate>
So I have at the top-level `mkDerivation { pname = "top-level-example"; ... buildInputs = [ autoreconfHook ... (import ./my-dependency.nix) ]; }`ndency-derivation)
butcher has quit [Ping timeout: 256 seconds]
<zecnate>
Whoops, ignore the noise after the final `
<euank>
I mean, from just that there's not enough to be sure what's going wrong
<euank>
you have cmake and pkg-config in nativeBuildInputs I assume, right?
<zecnate>
Ok. I'll keep poking at it. I wish I could tell `./configure` to get even more verbose, like if I could see the literal `g++` command and the literal `conftest.cpp` source, I could run that manually and see if that reveals the issue.
<zecnate>
Oh, I have pkg-config, but not cmake. I'll try that!
<zecnate>
I'm not too familiar with cmake. I assumed cmake would be used to build the dependency derivation, but for the top-level is it also necessary for dependency detection?
growpotk- has quit [Ping timeout: 260 seconds]
<euank>
It's not if your project isn't using cmake
<euank>
I couldn't tell from the above if it was or not. I guess since it's automake, it probably isn't
<euank>
You referenced .cmake file differences I wasn't sure about above though.
dstzd has joined #nixos
dstzd has joined #nixos
dstzd has quit [Changing host]
<zecnate>
Huh... no, that fails. That expects the current package to be cmake, but this one isn't.
<zecnate>
Yeah, I think those cmake files are only for cmake apps that consume the dependency, which mine is not.
<euank>
... I'm outta ideas though. If you've got it posted somewhere, I could poke at it, but the best I've got without seeing it is using a breakpoint hook to debug it better.
<zecnate>
Ok, how do I use a breakpoint hook? In my top-level I have no builder, I'm just using the generic stdenv builder.
<zecnate>
Do I need to define my own builder or can I add a more specific derivation attribute somehow?
butcher has joined #nixos
<euank>
just add it to buildInputs, and when a step exits non-zero it should give you instructions to enter into the build container
<euank>
iirc that's all there is to it
justanotheruser has quit [Ping timeout: 272 seconds]
<__red__>
euank: is there a recommended way to work out why builds in nix-shell ./nixpkgs -A packagename work, but the same thing with nix-build fails?
Pidgeotto has joined #nixos
Pidgeotto has quit [Excess Flood]
<__red__>
and advice or resources?
<euank>
with what sort of error?
Pidgeotto has joined #nixos
Pidgeotto has quit [Excess Flood]
<__red__>
it claimed it was unable to locate a file
<__red__>
but the file was there
<__red__>
the file being a python script
<__red__>
and I have yet to find a diffeernce between the one test example and another
Pidgeotto has joined #nixos
Pidgeotto has quit [Excess Flood]
<__red__>
but you telling me that it builds inside a mount namespace gives me more information again
<euank>
I was just recommending a breakpointHook a second ago, and that again seems relevant
<euank>
that lets you run a command to jump into the container it was building in when it failed, and then you can run the command yourself or poke around and look at files
<__red__>
The app I'm trying to package uses a combination of make and cmake
<__red__>
so it's a bit of a bear
waleee-cl has quit [Quit: Connection closed for inactivity]
<__red__>
I wonder if a nix-shell --pure would make a difference in my debugging - as that would rule out another variabel
fresheyeball has joined #nixos
<fresheyeball>
hey out there
<Mic92>
__red__: breakpointHook, see nixpkgs
<fresheyeball>
I have a Haskell + Nix project, and am happy with nix
<fresheyeball>
but I have people who want to use stack instead
<__red__>
bash: /build/source/build/build_libs/gbenchmark-prefix/src/gbenchmark/tools/strip_asm.py: /usr/bin/env: bad interpreter: No such file or directory
<__red__>
bash-4.4$ /usr/bin/env
<__red__>
bash: /usr/bin/env: No such file or directory
<__red__>
I'm guessing all the other tests just had #! env
<__red__>
looking
<euank>
patchShebangs might get that for you
stoile has joined #nixos
ericsagnes has joined #nixos
<__red__>
interestingly - I can't even find python in the environment
<__red__>
even though it's in my buildInputs
spacekookie has quit [Quit: No Ping reply in 60 seconds.]
ninjin_ has quit [Ping timeout: 240 seconds]
<euank>
perhaps it needs to be in nativeBuildInputs instead
spacekookie has joined #nixos
<__red__>
I'm still blurry on the difference to be honest
ninjin_ has joined #nixos
<__red__>
let me try that
<__red__>
but this breakpoint is a lifesaver
<euank>
basically, build tools like compilers and interpreters and the like are nativeBuildInputs, and libraries you link against are buildInputs
<__red__>
got further in the last 10 mins than the last few days
<__red__>
so thank you again <3
<__red__>
ahh
<euank>
it mostly matters when you cross compile. If you're on x86 and you're building for arm, nativeBuildInputs are x86 and buildInputs are the target platform arch, arm
<__red__>
so anything that needs bin is probably nativeBuildInputs
endocrimes has quit [Remote host closed the connection]
<__red__>
bingo
f0x2 has quit [Ping timeout: 240 seconds]
endocrimes has joined #nixos
<euank>
I'm glad it ended up being helpful. I've definitely found it handy a few times
<__red__>
time to put vim in my nativeBuildInputs (temporarily)
<euank>
That and `--keep-failed` together can help quite a bit
f0x2 has joined #nixos
hke has quit [Read error: Connection reset by peer]
Spiney has quit [Ping timeout: 240 seconds]
hke has joined #nixos
zecnate has joined #nixos
caranatar has quit [Ping timeout: 264 seconds]
<__red__>
okay - that's weird
<__red__>
#! env python
<__red__>
can't find env
endformationage has quit [Quit: WeeChat 2.9]
<euank>
`patchShebangs` should do the right thing, though I'm also surprised env isn't included in the stdenv path
caranatar has joined #nixos
<Ke>
do relative paths really work in shebangs
Orbstheorem has quit [Ping timeout: 244 seconds]
la-jesystani has joined #nixos
<hsngrmpf[m]>
How can I force a dependency into a derivation? I have a python package that depends on kerberos. But even after adding kerberos to its build inputs, kerberos doesn't end up in the final derivation.
<hart111>
I don't know if I want to know the answer to that Ke
<euank>
truly with binfmt_misc anything is possible :P
caranatar has quit [Ping timeout: 260 seconds]
<euank>
hsngrmpf[m]: does python use ffi to open kerberos? Is that why there's no proper reference?
<zecnate>
I get a bunch of results for `grep -r '/usr/bin/file' /nix/store/lf8b1c9m3px0bmsi4zwhcj89w38vcq7a-libtool-2.4.6/`. Isn't this a bug, because builders should be referring to a nix-store path for `file`, right?
<Ke>
heh
<zecnate>
When my builder runs `./configure` there are occasional file-not-found errors for /usr/bin/file.
<zecnate>
And of course autotools & sh helpfully continue running after those failures in some weird inconsistent state, so I have no idea what impact that has.
<zecnate>
Should I try to manually patch my own `configure` after it is generated by autoconf to replace all references to a nix store path?
<zecnate>
I already have `file` in `buildInputs`, so I think `sed -i 's,/usr/bin/file,file,g' ./configure` may do the trick... Feels very tenuous, though.
* zecnate
searches for tickets about this.
<clever>
zecnate: that feels like something patchShebangs should already support
<clever>
though, only if its in a #!
<zecnate>
clever: To me it seems like the `libtool` derivation should fix it once and for all at the source, rather than every downstream consumer having to patch their generated configure scripts.
<zecnate>
No, these aren't in `#!`, they are all over scattered throughout m4 macros used by autoconf.
<zecnate>
Like the macro source literally has stuff like "case `/usr/bin/file conftest.$ac_objext` in" ...
<zecnate>
I see 28 matches from grep.
<clever>
ah, id use sed then
<__red__>
so can I get patchSheBangs to recurse?
<__red__>
and do all the things
<__red__>
all the examples I see are single file in install
<zecnate>
But rather than use sed in my derivation, I'm tempted to make a derivation that takes the libtool source as input and patches the output...
<drakonis>
mpd seems to be defaulting to jack
<__red__>
whereas I need it to postPatch :-(
<euank>
It can run on a directory too, in which it recurses for +x files.
<zecnate>
srhb: Thanks. Then is there a better reference doc?
<srhb>
zecnate: I think most of them are documented in the Nix manual :)
zupo_ has joined #nixos
<srhb>
zecnate: And yes, that is an unfortunate but historical collission that's probably hard to do away with now.
polarfire has quit [Quit: WeeChat 2.7.1]
lsix has joined #nixos
<zecnate>
I think one thing that keeps tripping me up on documentation is how some functions/language docs are in the nix manual and the rest is in nixpkgs, and then some more language info is on the wiki.
<zecnate>
I wish there were a single reference that was all about the nix expr language, then one all about all of things like `builtins` and `nixpkgs.lib`…
<srhb>
zecnate: Yes, most users of nixpkgs find that confusing. It's less confusing if you're primarily a nix-but-not-nixpkgs user, as you're less likely to see "nixpkgs" as the main thing.
<zecnate>
If I had the time I'd take a stab at it.
<{^_^}>
[nixos-homepage] @github-actions[bot] pushed commit from GitHub Actions to master « Update flake.lock and blogs.xml [ci skip] »: https://git.io/JUz89
karantan has joined #nixos
<zecnate>
How large is the nix-but-not-nixpkgs userbase, though? I would suspect that's rare unless there're popular use cases I'm not aware of (which could easily be true).
<srhb>
zecnate: Not large for sure, but it does exist. :)
<srhb>
(I've worked for some)
<srhb>
zecnate: So yes, we may be sub-optimizing by delimiting the manuals in the "technically correct but largely obscure" way :-P
<zecnate>
Thanks for the context.
<srhb>
Welcome. You're definitely not the first person to have this issue.
<zecnate>
If there are any efforts to refactor docs, I can be a guinea pig, since I'm fairly new to nix. ;-)
<eyJhb>
I guess we have all been there, or at least most of us have been
<srhb>
zecnate: That's excellent :) I'm not aware of any current efforts, but I imagine they must be coming with Nix 3.X in particular.
<zecnate>
Hm, so I'm trying to use `builtins.fetchGit ./../../..` but I get a result with `.rev` set to a long string of all 0's. That should be the root of a git clone, so that relative path has a `.git` inside of it. Is that the wrong way to use this?
<eyJhb>
FetchGit wants a src, rev, sha
<eyJhb>
You can try to look in nixpkgs and see how it is used there. Generally nixpkgs repo is a good resource ;p
<zecnate>
"args can be a URL, in which case the HEAD of the repo at that URL is fetched."
<srhb>
zecnate: Which version of Nix?
<zecnate>
Nothing in the docs say I can pass a path, but @clever suggested as much yesterday.
<zecnate>
Uh... how can I tell which version?
<srhb>
zecnate: nix --verison
<srhb>
... spelled correctly ;)
<zecnate>
$ nix --versionnix (Nix) 2.3.7
<zecnate>
Sorry about the missing newline.
saschagrunert has joined #nixos
<srhb>
zecnate: Can't seem to reproduce that.
civodul has joined #nixos
est31 has quit [Remote host closed the connection]
<Mic92>
hsngr
<Mic92>
hsngrmpf[m]: link to the source?
est31 has joined #nixos
cr4y1_ has quit [Read error: Connection reset by peer]
azure3 has joined #nixos
azure2 has quit [Ping timeout: 240 seconds]
cr4y1_ has joined #nixos
cfricke has joined #nixos
cr4y1_ has quit [Read error: Connection reset by peer]
cr4y1_ has joined #nixos
orivej has quit [Ping timeout: 265 seconds]
Pidgeotto has joined #nixos
Pidgeotto has quit [Excess Flood]
nixuser has quit [Read error: Connection reset by peer]
Orbstheorem has joined #nixos
Pidgeotto has joined #nixos
Pidgeotto has quit [Excess Flood]
Pidgeotto has joined #nixos
Pidgeotto has quit [Excess Flood]
bahamas has quit [Ping timeout: 265 seconds]
mmohammadi9812 has joined #nixos
mmohammadi98124 has joined #nixos
<zecnate>
Are there expressions for joining/extending path values? I often want to do something like `let basedir = ./../../../; thingone = basedir / "thing1"; thingtwo = basedir / "thing2"; in …`
<zecnate>
> ./.
azure3 has quit [Read error: Connection reset by peer]
<{^_^}>
/var/lib/nixbot/state/nixpkgs
mmohammadi9812 has quit [Read error: Connection reset by peer]
mmohammadi98124 is now known as mmohammadi9812
<cole-h>
> let basedir = "/root/some/path"; thingone = basedir + "/thing1"; in thingone
<zecnate>
> (./.) / "foo"
<{^_^}>
"/root/some/path/thing1"
<{^_^}>
value is a string while a float was expected, at (string):324:1
<zecnate>
cole-h: Thanks! So the result is still a path? That will lead to two different paths that get inserted into the store separately?
<V>
yeah, your best bet is path1 + "/${path2}"
azure3 has joined #nixos
<cole-h>
Literally just string concatenation. And if you need to turn it back into a path for whatever reason, `builtins.path { path = thingone; }`
<cole-h>
No, the result is a string
<cole-h>
> ./. == "${./.}"
<{^_^}>
access to path '/var/lib/nixbot/state/nixpkgs' is forbidden in restricted mode
<cole-h>
heh
<cole-h>
> ./. == "/var/lib/nixbot/state/nixpkgs"
<{^_^}>
false
<zecnate>
Oh... hrm. I'm a bit confused. So for example, I have a bunch of patch files in a directory, so they need to be inserted as inputs prior to building.
<zecnate>
It's fine if they were all inserted as one entry in a directory, or if they were separate entries, shouldn't matter.
<zecnate>
I'll just try it and see how it works. ;-)
<zecnate>
I guess what I want to know is if path values themselves are special and get inserted into the store as inputs, or if it's only when they appear in specific attributes of a derivation?
bahamas has joined #nixos
zecnate has quit [Remote host closed the connection]
cole-h has quit [Quit: Goodbye]
zupo_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<NobbZ[m]>
What they point to gets copied into the store @zecnate
zupo has joined #nixos
azure4 has joined #nixos
<{^_^}>
[nixpkgs] @HaoZeke opened pull request #98449 → doxyrest: init at master → https://git.io/JUzR2
Izorkin has quit [Ping timeout: 272 seconds]
Darkmatter66 has joined #nixos
Darkmatter66_ has quit [Ping timeout: 256 seconds]
azure3 has quit [Ping timeout: 272 seconds]
azure4 has quit [Ping timeout: 265 seconds]
azure4 has joined #nixos
<{^_^}>
[nix] @regnat opened pull request #4048 → Disable `FORTIFY_SOURCE` when compiling without optims → https://git.io/JUzRb
cyphase has quit [Ping timeout: 240 seconds]
mir100 has quit [Ping timeout: 272 seconds]
mmohammadi9812 has quit [Ping timeout: 256 seconds]
<eyJhb>
What was the command to check, why you have a path in your store?
Sanchayan has joined #nixos
supercoven has joined #nixos
supercoven has quit [Max SendQ exceeded]
supercoven has joined #nixos
supercoven has quit [Max SendQ exceeded]
supercoven has joined #nixos
<srhb>
eyJhb: why-depends?
<{^_^}>
[patchelf] @eytanaim opened pull request #243 → Corruption with program headers → https://git.io/JUz0P
<srhb>
eyJhb: Also helpful to use nix-store -q --tree etc.
<eyJhb>
,locate bin why-depends
<eyJhb>
?
<srhb>
eyJhb: nix why-depends
<{^_^}>
Couldn't find in any packages
<eyJhb>
Ahh, thanks :)
<eyJhb>
Hoping Docker is gone now that I removed all previous generations however!
hyper_ch2 has quit [Ping timeout: 258 seconds]
hyper_ch2 has joined #nixos
nagayasu has joined #nixos
eyJhb has joined #nixos
eyJhb has joined #nixos
eyJhb has quit [Changing host]
nagayasu has quit [Remote host closed the connection]
<hsngrmpf[m]>
Thanks. But in my case LD_DEBUG=libs doesn't really reveal additional info about the problem with libQt5DBus.so.5.
SanchayanM has joined #nixos
<hsngrmpf[m]>
Not sure what `undefined symbol: _ZN14QMetaCallEventC2EttPFvP7QObjectN11QMetaObject4CallEiPPvEPKS0_iiPiS5_P10QSemaphore, version Qt_5_PRIVATE_API)` really tells me. Something is missing for sure. but autoPatchElfHook didn't raise any errors.
mmohammadi98124 has joined #nixos
rain1 has left #nixos ["WeeChat 1.6"]
mmohammadi9812 has quit [Read error: Connection reset by peer]
mmohammadi98124 is now known as mmohammadi9812
<{^_^}>
[nix] @edolstra merged pull request #4041 → Serialize SandboxMode enum to string for JSON → https://git.io/JUEEk
<maralorn>
Would it be a good idea to convert services.postfix.virtual to a attrsOf (either str (nonEmptyListOf str))? Right now it's lines. But it is used to generate a map file from e-mail addresses to one or multiple mailboxes or other e-mail addresses.
<Ke>
is there some service module for nix-index
<Ke>
as in something to create global db and keep it up to date
werner291 has quit [Remote host closed the connection]
<l-as>
I am asking because the recent style change uses some colors which are very hard to read for me, and I assume that it's the same for some other colorblind people
<Baughn>
Since the last update of nixos, nginx is failing to read files in /home.
<Baughn>
It's getting 'permission denied' from fopen. Which is curious, as the files are world-readable, and the directory chain is a+x.
<grobi>
will network-manager-aplet be installed by default?
<srhb>
grobi: I don't recall :)
supercoven__ has joined #nixos
etalecohomology has quit [Quit: Lost terminal]
<srhb>
grobi: Following the option definition, it looks like "no"
supercoven_ has quit [Ping timeout: 272 seconds]
supercoven has quit [Ping timeout: 256 seconds]
<Ke>
does nix have complement operation
<Ke>
bitwise
<l-as>
I don't think so, and I don't think it would make sense to have one
<l-as>
What are you trying to do? Wouldn't it make sense just to use arrays?
Mateon1 has joined #nixos
<grobi>
srhb: would this be a correct line to enable network for a user: " extraGroups = [ "wheel" ][ "networkmanager" ]; # Enable ‘sudo’ 'network' for the user." ?
supercoven has joined #nixos
<{^_^}>
[nixpkgs] @fpletz pushed commit from @r-ryantm to release-20.03 « coturn: 4.5.1.2 -> 4.5.1.3 »: https://git.io/JUzaS
<V>
[ "wheel" "networkmanager" ]
<grobi>
V ah thanks
<V>
I don't know what group NetworkManager needs (is it uppercased?) but you write lists like that
supercoven__ has quit [Ping timeout: 260 seconds]
<grobi>
do i need to do this for the root user as well?
<NobbZ[m]>
Indeed one of the few things with NetworkManager that are lowercase…
<etu>
grobi: Root in general has access to everything
<V>
ah, it's lowercase
<grobi>
thanks
<grobi>
" extraGroups = [ "wheel" "networkmanager" ]; # Enable ‘sudo’ for the user." like this?
ckauhaus has joined #nixos
<V>
well, it enables both sudo and NetworkManager, but otherwise yes
<grobi>
this sounds, that you would prefer, to do this apart?
<V>
extraGroups = [
<V>
];
<V>
"networkmanager" # enable NetworkManager
<V>
"wheel" # enable sudo
<V>
etc
<V>
do what you want though
<grobi>
:)
<V>
there's no right or wrong here
supercoven has quit [Max SendQ exceeded]
supercoven has joined #nixos
spudly- has joined #nixos
<grobi>
is ti possible, to "allow" to insttall nonfree-packages during installation-process of nixos, or do i need to do this after its installation?
srhm has joined #nixos
spudly has quit [Ping timeout: 272 seconds]
spudly- is now known as spudly
supercoven has quit [Max SendQ exceeded]
supercoven has joined #nixos
<{^_^}>
[nixos-homepage] @github-actions[bot] pushed commit from GitHub Actions to master « Update flake.lock and blogs.xml [ci skip] »: https://git.io/JUzVy
<srhb>
grobi: nixos-install _may_ allow NIXPKGS_ALLOW_UNFREE=1 -- not sure
<LnL>
"install" uses the configuration.nix just like any other switch, does it not?
<V>
grobi: have you actually tried that?
<V>
I don't see why the normal methods wouldn't work
<{^_^}>
[nixpkgs] @mmahut merged pull request #98445 → Removing myself from maintainers lists of blockchains → https://git.io/JUzla
<maralorn>
example_3 generates a nix-shell via a default.nix which uses mkShell. Wouldn‘t it be better to use shell.nix when using mkShell?
<karantan>
I'm trying to upload nixos 20.03 image to digital ocean but it only supports "gz, bz2, vmdk, vhdx, qcow, qcow2, vdi, raw, img, xz"
<karantan>
there are some instructions how to use virtualbox and create a new image then export it ...
<karantan>
is there any online resource that already has nixos in one of those formats?
<marek>
ey all, can anyone try to run nix-build inside this repo https://github.com/mmahut/yoroi-ergo-backend ? I'm having an odd stack size error, but I think it might be my setup, please.
lb5tr has quit [Ping timeout: 264 seconds]
<V>
maralorn: I'd prefer shell.nix at least
ghasshee has joined #nixos
evanjs has quit [Read error: Connection reset by peer]
alexherbo2 has joined #nixos
<srk>
marek: running
<maralorn>
V: k, gonna make a PR for that. Let's see how it's received.
evanjs has joined #nixos
<ghasshee>
Hi, Does anyone know how to run GHC 8.8.1 on NixOS-20.03 ?
bahamas has quit [Ping timeout: 246 seconds]
<V>
generally I have a package.nix (callable with callPackage, like other things in nixpkgs), default.nix (imports nixpkgs & calls the package with that), and shell.nix (for the dev shell for use w/ lorri/etc)
<V>
usually just a shell.nix though
evanjs has quit [Read error: Connection reset by peer]
evanjs has joined #nixos
<grobi>
got this error-mssg: - You can not use networking.networkmanager with networking.wireless. Except if you mark some interfaces as <literal>unmanaged</literal> by NetworkManager.
<srhb>
grobi: networking.wireless is (mostly) for when you want a declarative setup _without_ networkmanager. If you want netowrkmanager, you can get rid of networking.wireless.
<grobi>
thus i can comment that line out?
<V>
yes
<grobi>
thanks
<{^_^}>
[nixos-homepage] @maralorn opened pull request #556 → Fix various typos or formulations in screen casts → https://git.io/JUzrV
<grobi>
now installation is running
<ghasshee>
Or, is it the better way to use some package manager to handle Haskell rather than depneding on nixpkgs-wrapped pkgs?
<{^_^}>
[nixos-homepage] @maralorn opened pull request #557 → Replace default.nix by shell.nix in screen cast for nix-shell → https://git.io/JUzry
<grobi>
^>^°.°^<^
<maralorn>
ghasshee: I‘d always opt fo the nixpkgs ghc.
<maralorn>
ghasshee: Have you tried pkgs.haskell.compiler.ghc882? For some reason there is no 8.8.1, but is that important to you?
tom39291_ has quit [Quit: WeeChat 2.7.1]
knupfer has joined #nixos
<ghasshee>
maralon: No, not important. I would like to just try type operator (.)
<ghasshee>
Oh, I found that. haskell.compiler have the argument !
evanjs has quit [Read error: Connection reset by peer]
evanjs has joined #nixos
<ghasshee>
maralon: I forgot the name ... would you mind telling me what was the term haskell.compiler.ghc882.withPackages ?
<ghasshee>
maralorn: Oops, excuse me.
<{^_^}>
[nix] @edolstra merged pull request #4038 → Add a nix.conf option for allowing a symlinked store → https://git.io/JU09v
knupfer has quit [Read error: Connection reset by peer]
<maralorn>
ghasshee: btw. what I do is using nixos-stable but pull haskellPackages from unstable. That way I always get the latest and greatest from the Haskell ecosystem.
<ghasshee>
maralorn: Thank you for that information! I will try that.
<maralorn>
And since 8.8.4 is the default on 20.09 and master right now. It will probably be better supported and you get more cache hits.
bahamas has joined #nixos
etalecohomology has joined #nixos
supercoven has quit [Ping timeout: 240 seconds]
<ghasshee>
maralorn: Thanks, I will try 20.09 now! ( I was just using 19.03 yesterday )
n9nes- has joined #nixos
quinn has quit [Ping timeout: 246 seconds]
n9nes has quit [Ping timeout: 272 seconds]
<etalecohomology>
hi, i am new to nixos and try to compile a haskell package (-version) with pkg-config dependencies which is *not* yet in nixos/stackage
<etalecohomology>
pkg-config and nixos seems to be quite complicated
<etalecohomology>
i always get pkg-config errors in the nix-shell, although the cairo.dev package in buildInputs should provide the .pc files
n9nes- has quit [Ping timeout: 258 seconds]
Sanchayan has joined #nixos
zupo has joined #nixos
civodul has quit [Quit: ERC (IRC client for Emacs 27.1)]
<maralorn>
etalecohomology: Can you post your error message? Does something in your nix-shell fail or the generation of the nix-shell? If gi-cairo fails to build, that might be because that package you create is not affected by the buildInputs to your mkShell.
quinn has joined #nixos
<etalecohomology>
maralorn: yes, gi-cairo fails to but, because it cannot find the cairo-gobject via pkg-config
<etalecohomology>
build
zupo has quit [Ping timeout: 272 seconds]
Pidgeotto has joined #nixos
Pidgeotto has quit [Excess Flood]
meh` has quit [Read error: Connection reset by peer]
zupo_ has joined #nixos
<etalecohomology>
how to provide cairo.dev and pkg-config as a prerequisite before gi-cairo is build
Pidgeotto has joined #nixos
Pidgeotto has quit [Excess Flood]
<etalecohomology>
I through the order might matter, but this doesn't seem to be the case
Pidgeotto has joined #nixos
Pidgeotto has quit [Excess Flood]
<etalecohomology>
compileBuildDriverPhase
<etalecohomology>
Package cairo-gobject was not found in the pkg-config search path.
rprije has quit [Ping timeout: 260 seconds]
supercoven has joined #nixos
<{^_^}>
[nixpkgs] @cdepillabout opened pull request #98459 → tests.haskell-shellFor: get compiling again → https://git.io/JUzKv
supercoven_ has joined #nixos
<maralorn>
etalecohomology: Is ther a specific reason you are not using the gi-cairo from nixpkgs?
meh` has joined #nixos
<etalecohomology>
yes, i have my own package hlabyrinth, which depends on 1.0.24 of gi-cairo
<etalecohomology>
I try to compile my own package on nixos
supercoven has quit [Ping timeout: 260 seconds]
<etalecohomology>
but I started with something simpler ...
<etalecohomology>
when writing packages yourself, you are often bound to the newest versions
grobi has joined #nixos
<grobi>
ah shit..
<maralorn>
etalecohomology: We use gi-cairo 1.0.24 in nixpkgs.
<maralorn>
I personally made the override a few weeks agou.
n9nes- has joined #nixos
<etalecohomology>
but how does it work with hackage?
<etalecohomology>
ok
<maralorn>
So maybe just upgrade to recent unstable?
n9nes has quit [Ping timeout: 260 seconds]
<grobi>
i failed to give in correct psswd during installation an have been prompted to " nixos-enter --root '/mnt' " in the new system..
<etalecohomology>
but how can I change my shell.nix to correctly install pkgconfig and cairo before building gi-cairo?
<etalecohomology>
the same will happen for my package, which is definively not in nixos
<etalecohomology>
so i have to solve this problem anyhow
<grobi>
but after starting i could not find the possibility to write tha into aproprate terminal
<grobi>
how????
<maralorn>
etalecohomology: At the point where you bind "gi-cairo = self.callHackage "gi-cairo" "1.0.24" {};" you can use pkgs.haskell.lib.overrideCabal to override the inputs of the package.
waleee-cl has joined #nixos
zupo_ has quit [Ping timeout: 256 seconds]
<grobi>
i'm now on the installation media again
<maralorn>
gi-cairo does some pkg-config magic. You can see the derivation we use in nixpkgs if you look into pkgs/development/haskell-modules/hackage-packages.nix and search for the string "gi-cairo" =. (It's around line 10000, the file is to big to send you a github link, sorry.)
<maralorn>
grobi: Yes, I think that's what that hint meant.
mariatsji has quit []
<etalecohomology>
ah ok, so I do overrideCabal and then libraryPkgConfigDepends or so
<maralorn>
etalecohomology: Yes
<etalecohomology>
what is exactly the problem if I use some hackage package and he complains about not finding some json file when unpacking the .tar
<etalecohomology>
happens often to me
<maralorn>
etalecohomology: That means the database with all hackage packages in nixpkgs is older than that package version.
<maralorn>
etalecohomology: You can circumvent that with callHackageDirect
<grobi>
maralorn but what to do now???
<etalecohomology>
maralorn: thanks! ... i will try to get it working
<maralorn>
grobi: You mount your system under /mnt again like you did during installation.
<grobi>
ah, and then that command?
<maralorn>
grobi: Then you can either run nixos-install again. Or you use the above command and then run passwd to set the root password.
<maralorn>
Although I am not completely sure about the second approach.
<grobi>
no please no new installation of all?
hyper_ch2 has quit [Ping timeout: 272 seconds]
hyper_ch2 has joined #nixos
<grobi>
the latter sonds better
meh` has quit [Ping timeout: 260 seconds]
<maralorn>
grobi: It wouldn‘t reinstall everything. Imagine it like a nixos-rebuild it will only download anything missing to the nix store.
<maralorn>
So running nixos-install twice will be much faster the second time.
<maralorn>
etalecohomology: Two hints though: 1. gi-cairo is tightly coupled with a lot of other gi-* packages. So I fear just overriding that one to a newer version might give you build errors (but not sure). 2. If the nixpkgs checkout you use is new enough there might also be gi-cairo_1_0_24. (The logic is on nixpkgs master always the latest hackage version is available via the packagename_version attribute.)
grobi has quit [Remote host closed the connection]
<clever>
default also depends on osgearth, so youll only get that if you use pkgs.nix
nikivi has joined #nixos
nikivi has quit [Remote host closed the connection]
nikivi has joined #nixos
karantan has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
ris has quit [Ping timeout: 256 seconds]
<{^_^}>
[nixpkgs] @mpoquet opened pull request #98464 → i3nator: init at 1.2.0 → https://git.io/JUz1Z
evanjs has quit [Read error: Connection reset by peer]
evanjs has joined #nixos
<eyJhb>
Thanks!
<eyJhb>
I remember why it did not work last time, or rather, it did not work last time :(
n9nes has joined #nixos
evanjs has quit [Excess Flood]
evanjs has joined #nixos
<{^_^}>
[nixos-homepage] @github-actions[bot] pushed commit from GitHub Actions to master « Update flake.lock and blogs.xml [ci skip] »: https://git.io/JUz1H
n9nes- has quit [Ping timeout: 265 seconds]
domogled has quit [Ping timeout: 258 seconds]
karantan has joined #nixos
domogled has joined #nixos
<eyJhb>
Easy way to get around this on a build? `/nix/store/6ffp6xbr6j2ryqw2dafbbvv55ggq16iv-bash-4.4-p23/bin/bash: /build/omnetpp-5.6.2/bin/opp_makemake: /usr/bin/env: bad interpreter: No such file or directory`
<clever>
eyJhb: postPatch = "patchShebangs bin";
<{^_^}>
[nixpkgs] @teto opened pull request #98465 → Tree sitter update → https://git.io/JUzMJ
tilcreator_ has joined #nixos
<eyJhb>
Can I just point it at a dir?
karantan has quit [Client Quit]
<clever>
eyJhb: yes
<{^_^}>
[nixpkgs] @stigtsp opened pull request #98466 → perl-cross: fix . being included in INC → https://git.io/JUzMq
<clever>
26 | # warning "The GEOS C++ API is unstable, please use the C API instead"
<clever>
27 | # warning "HINT: #include geos_c.h"
<clever>
but if its the same verison on both, it should be fine...
<clever>
eyJhb: what is the output to a successful build?
alp has quit [Ping timeout: 272 seconds]
<eyJhb>
clever: the other machine is a Ubuntu machine, so I might not have communicated that well enougt
<eyJhb>
I will try to see the version on Ubuntu
<clever>
eyJhb: thats why i said "on both"
<eyJhb>
That uses 3.8.0
<clever>
and on ubuntu it builds?
<eyJhb>
So I might have to downgrade to that then
<eyJhb>
Yeah
<clever>
simplest option is to use the same nixpkgs
<clever>
first, do `nix eval -f '<nixpkgs>' lib.version` on the ubuntu machine
etalecohomology has quit [Ping timeout: 272 seconds]
<eyJhb>
Didn't specify this as well, ubuntu is non-nix, just plain ubuntu with build-essential! So sorry clever :( Should describe the setup a little better
<eyJhb>
So there is no Nixpkgs on it
<clever>
ahh
<eyJhb>
So I should just find i nixpkgs commit where geos is 3.8.0
<clever>
then youll need to either find a 3.8.0 geos in nixpkgs history, or override geos to 3.8.0
<clever>
or fix osgEarth to work on 3.8.1
<eyJhb>
Commit e80d29221325f056f47ef4bfc0a11772b65cba54 is 3.8.0
<eyJhb>
Is there a easy way to do a build with nixpkgs on that commit?
<clever>
then add `-I nixpkgs=https://github.com/nixos/nixpkgs/archive/e80d29221325f056f47ef4bfc0a11772b65cba54.tar.gz`
<clever>
and it will use that version for <nixpkgs>
<clever>
you can also edit the pkgs.nix that is doing the callPackage, to fetch it with fetchTarball
<clever>
Taneb: i forget the exact option, but you need to use something like: qemu.networkingOptions = [ "-net" "user,hostfwd=tcp:127.0.0.2:2222-:22" "-net" "nic" ];
<clever>
Taneb: you can also build it more easily with `nix-build '<nixpkgs/nixos>' --arg configuration ./configuration.nix -A vm`
cr4y1_ has quit [Ping timeout: 256 seconds]
<{^_^}>
[nixpkgs] @hakuch opened pull request #98472 → WIP: ocaml: Export mkOcamlPackages again (and a question) → https://git.io/JUzSz
<flythief>
hi, I build llvm and clang in nixos, but I found that using clang++ to compile source file will get error: can't find crt1.o: No such file or directory.(can't find crti.o, crtbegin.o -lgcc_s...) I googled it, someone said that installing gcc-multilib may work. But I'm in nixos. can someone give me some ideas? thanks firstly.
<flythief>
Thanks, I checked this link before. but I want to use clang++ which built by my custom compilation, this link can work only when I use it in system installation? @symphorien[m]
<symphorien[m]>
I think you are better off pacakaging your version of clang and llvm in nixpkgs properly
<symphorien[m]>
in a local clone of nixpkgs, I mean
<flythief>
I found that clang++ in system installation will get crt1.o, crti.o .... etc. but clang++ from my compilation will failed to find them. you mean I should config nix package for my custom compilation?
ckauhaus has quit [Quit: WeeChat 2.7.1]
butcher_ has joined #nixos
<symphorien[m]>
yes. the clang provided by nixpkgs is wrapped to inject a few flags that you can see with `export NIX_DEBUG=1`, but the complexity is very high, and I think if your fork of clang is not too strange, it's probably easier to just edit nixpkgs to use it
<ghasshee>
maralorn: I am sorry I accidentarily have sent the previous message. By the way, I could not override haskell.packages.ghc882.ghc.withPackages, do you know how to ?
<flythief>
ok, I'm not so familiar with nix config, but I'll try it. maybe it's a chance to know it. thanks. : ) @symphorien[m]
n9nes has joined #nixos
<ghasshee>
I am now back on 20.03 since sagemath compilation failed on 20.09 .
<ToxicFrog>
flythief: this might be a place where overlays are useful, too
zakame has joined #nixos
<maralorn>
ghasshee: What do you mean with "override"?
n9nes- has quit [Ping timeout: 256 seconds]
<maralorn>
An example would be `pkgs.haskell.packages.ghc882.ghc.withPackages (ps: [ps.cabal-install ps.hlint ps.lens])`
zupo has joined #nixos
<ghasshee>
maralorn: some of my haskell packages were overriden since it depends on some old softwares.
<flythief>
@ToxicFrog well, cool place. :)
<maralorn>
ghasshee: Ah, I see.
n9nes- has joined #nixos
<maralorn>
ghasshee: Try that one `(pkgs.haskell.packages.ghc882.override (self: super: {...})).ghc.withPackages (ps: [...])`
sputny has quit [Quit: sputny]
n9nes has quit [Ping timeout: 260 seconds]
cfricke has quit [Quit: WeeChat 2.9]
karantan has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
cr4y1_ has joined #nixos
pineapples has joined #nixos
ATROCK has joined #nixos
daemon1024 has joined #nixos
cr4y1_ has quit [Read error: Connection reset by peer]
cr4y1 has quit [Ping timeout: 240 seconds]
pineapples has quit [Client Quit]
pineapples has joined #nixos
pineapples has left #nixos [#nixos]
pineapples has joined #nixos
cr4y1_ has joined #nixos
n9nes has joined #nixos
<{^_^}>
[nixpkgs] @illiusdope opened pull request #98474 → t.py: init at 2020-04-11 → https://git.io/JUzHw
<ghasshee>
maralorn: it seems to return error " error: value is a function while a set was expected"
cr4y1_ has quit [Read error: Connection reset by peer]
<maralorn>
Uh
n9nes- has quit [Ping timeout: 256 seconds]
<clever>
ghasshee: you may need to wrap it with another () if its inside a list
damjan has quit [Quit: bye]
damjan has joined #nixos
<maralorn>
ghasshee: It needs to be "extend" instead of override.
<maralorn>
I always confuse those …
<maralorn>
Not even sure what override does at that point …
alp has joined #nixos
<{^_^}>
[nixpkgs] @hakuch closed pull request #98472 → WIP: ocaml: Export mkOcamlPackages again (and a question) → https://git.io/JUzSz
erasmas has joined #nixos
<ghasshee>
maralorn: Thank you, 'extend' seems to have solved and I encountered another problem.
<clever>
maralorn: extend takes an overlay, and you can keep chaining it, so you can do `(pkgs.haskell.packages.ghc882.extend overlay1).extend overlay2`
<clever>
maralorn: override changes the args callPackage is supplying, one of which is overrides, so you can do `pkgs.haskell.packages.ghc882.override { overrides = overlay; }`
mallox has quit [Quit: WeeChat 2.9]
<clever>
maralorn: but .override cant chain, so if you .override again, you replace the old overrides, and loose that overlay
veleiro has joined #nixos
<maralorn>
Ah, okay.
etalecohomology has joined #nixos
<clever>
maralorn: there is a way to do it with .override (old: { overrides ... }) but then you have to merge the overlays yourself, and it gets ugly
<clever>
the main downside to extend, is that it deletes override
domogled has quit [Ping timeout: 258 seconds]
<ghasshee>
clever: how to overwrap with () ? I might have three overrided packages but, it does not seem to be written as a list .
<clever>
ghasshee: just stick another ( at the front, and a ) at the very end
<ghasshee>
clever: ah, I see what you mean. Thanks. I have a single function for now.
sputny has quit [Remote host closed the connection]
<ghasshee>
clever: I was using ugly way, override (old: { ... } )
mir100 has joined #nixos
mallox has joined #nixos
maljub01 has joined #nixos
<maljub01>
Hi, I'm wondering if it's possible to run docker from within a NixOS container. I tried to enable boot.enableContainers but I still get permission errors
maljub01 has quit [Read error: Connection reset by peer]
maljub01_ has joined #nixos
euandreh has quit [Remote host closed the connection]
euandreh has joined #nixos
maljub01 has joined #nixos
knupfer has quit [Ping timeout: 260 seconds]
pineapples has quit [Quit: pineapples]
maljub01_ has quit [Ping timeout: 256 seconds]
maljub01 has quit [Client Quit]
Pidgeotto has joined #nixos
Pidgeotto has quit [Excess Flood]
meh` has quit [Ping timeout: 272 seconds]
Pidgeotto has joined #nixos
Pidgeotto has quit [Excess Flood]
Pidgeotto has joined #nixos
Pidgeotto has quit [Excess Flood]
knupfer has joined #nixos
maljub01 has joined #nixos
etalecohomology has quit [Quit: Lost terminal]
<ghasshee>
clever: You are right! Indeed wrapping also solved the same problem, Thank you. ( And I encountered compilation error, which did not occur in ghc865 .. ummm)
<lordcirth__>
maljub01, nixos uses nspawn, so this may be helpful
<maljub01>
lordcirth__ thank you, I'll give that a go
lb5tr has joined #nixos
domogled has joined #nixos
zebrag has joined #nixos
<Taneb>
clever: earlier were you thinking of the QEMU_NET_OPTS environment variable? I'm trying that and it feels like it's *almost* working but I'm getting "connection reset" errors
<clever>
Taneb: that happens if qemu cant connect to the guest, fire up tcpdump within the guest to see why
ransom has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
rajivr has quit [Quit: Connection closed for inactivity]
<{^_^}>
[nixpkgs] @OPNA2608 opened pull request #98479 → megapixels: init at 0.8.0 → https://git.io/JUzNJ
<mcaju>
Hi, I'm trying to get this PR merged https://github.com/NixOS/nixpkgs/pull/94049 I'm currently stuck because I don't know if I should add vulkan-loader to RPATH or to propagateBuildInputs? It's dlopen'ed by vulkaninfo
zupo_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
maljub01 has quit [Remote host closed the connection]
fayyyy1 has quit [Remote host closed the connection]
<AWizzArd>
When I want to install one single unfree program via nix-env, do I still have to edit my ~/.config/nixpkgs/config.nix so that it will include `{ allowUnfree = true; }`? Or can I pass this in as a one-time command line arg?
<cole-h>
You could try `--arg config.allowUnfree true`
<cole-h>
If that doesn't work, you'll need to edit your config.nix.
bahamas has joined #nixos
<AWizzArd>
cole-h: okay, thx
<Henson>
does anybody here use procmail on NixOS? I'm trying to make a .procmailrc file do some mail forwarding, and it's not working and I keep getting a "/bin/sh: : No such file or directory" error an am wondering if there's a variable in the .procmailrc file I need to set to make it work properly with NixOS. I've tried the SHELL variable already but it didn't do anything.
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
hke has quit [Read error: Connection reset by peer]
<worldofpeace>
selfsymmetric-mu: lol, welcome xorg. Though it could be driver specific issue, and we did probably update the X stack
Mateon1 has quit [Read error: Connection reset by peer]
ris has joined #nixos
Mateon1 has joined #nixos
<cole-h>
clever: Darn, I was hoping I could be smart and just use the attr path :(
<clever>
cole-h: --arg is a lot more complicated then it seems
<roconnor>
cole-h: I'm guessing nixpkgs.overlays = [ (self: super: { gcc = self.gcc8; }) ]; might accomplish what I desire
<cole-h>
Maybe, but it would probably be better to do `stdenv = self.gcc8Stdenv`...
<roconnor>
hmm
<selfsymmetric-mu>
worldofpeace: Yeah, I should really record a video at some point. I suspect it's not rendering, but input.
<selfsymmetric-mu>
I'll install simplescreenrecorder and give it a shot.
<roconnor>
cole-h: he he. That doesn't work because gcc8Stdenv is a modified version of stdenv.
<cole-h>
Ah
<roconnor>
infinite recursion.
<roconnor>
I mean I could use super.
<roconnor>
but maybe setting gcc default version is more in line what I am thinking.
<simpson>
roconnor: I looked for like 10min but couldn't find exactly where the Linux stdenv chooses GCC versions. `getCompatibleTools` seemed like the most promising thing I haven't explored.
<selfsymmetric-mu>
What's a good video pastbin for this?
<simpson>
It should be something simple, right? Because surely somebody sometimes bumps some single line to change the default GCC on Linux.
<roconnor>
I think I went through the same process before giving up and asking here.
<jasom>
So I was experimenting with NIS on nixos, and I'd like to talk abuot nsswitch.nix right now that file needs to be changed for every single module that wants to change nsswitch. I was thinking we could do something like nsswitch.passwd.<module>.priority and then have nsswitch.conf build the files array in priority order, omitting anything with e.g. priority = 0. Then this file could remain
<jasom>
unchanged when ading different configuration options, plus people could re-order. Right now the order is hard-coded for passwd to "files sss ldap mymachines cache_oslogin oslogin systemd" and a user can freely omit or include items on that list but not rearrange.
mmohammadi9812 has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @jbaum98 opened pull request #98489 → Add all Iosevka build plans, including Aile, Etoile, and Sparkle → https://git.io/JUzp6
<vpfeiffer[m]>
Any advice for a noobie packaging a dependency for the first time?
arahael has joined #nixos
<tomberek>
vpfeiffer[m] Nix Pills, start simple, don't try with a large proprietary system with multiple parts until you have more experience
bahamas has joined #nixos
__Sander__ has joined #nixos
knupfer has quit [Quit: knupfer]
knupfer has joined #nixos
<samueldr>
nix-pills help to understand the unlyings of Nixpkgs packaging, but I'm not entirely convinced it's the best first steps guide... though there is not really any for now
<samueldr>
I guess the first steps would be getting comfy with nix's syntax
<__red__>
So I am packaging something which uses make and cmake. One of the makefile targets calls two cmake targets. Once does a whole bunch of detection and copies some python scripts into the /build directory - the second executes them. Problem I have is that when I sheBang them before the target, the cmake un'sheBangs' them.
<__red__>
so I guess I have to somehow separate out that makefile target into two separate makefile targets with a shebang between them?
<__red__>
(I tried inserting the shebang into the Makefile, but make doesn't seem to be able to execute shebang
<{^_^}>
[nixpkgs] @risicle opened pull request #98498 → cvc3: fix build against bison 3.7 → https://git.io/JUgfh
zupo_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
zupo has joined #nixos
<jasom>
I have a working prototype for just the passwd entry of my suggestion for nsswitch.nix; I'll cleanup and put up for people to demolish^H^H^H^H^H^H^H^H comment on tonight
<{^_^}>
[nixpkgs] @zowoq pushed commit from @saschagrunert to master « podman: 2.0.6 -> 2.1.0 »: https://git.io/JUgJC
knupfer has quit [Ping timeout: 244 seconds]
lordcirth__ has joined #nixos
lordcirth_ has quit [Ping timeout: 244 seconds]
meh` has joined #nixos
<{^_^}>
[nixos-homepage] @garbas pushed to fix-553 « Bring back search boxes on the learn page »: https://git.io/JUgJ2
<{^_^}>
[nixos-homepage] @garbas opened pull request #564 → Bring back search boxes on the learn page → https://git.io/JUgJa
<jasom>
speaking of which, is it possible to just see the output of etc (or even better a single file in etc) when changing nixos (instead of rebuilding all the manuals and restarting service &c.)?
<cransom>
building the system (like nixos-rebuild build) will leave you a result sym link which is the top level of the system. so result/etc/ would be the upcoming/built files
<{^_^}>
[nixos-homepage] @garbas merged pull request #564 → Bring back search boxes on the learn page → https://git.io/JUgJa
<{^_^}>
[nixos-homepage] @garbas pushed 2 commits to master: https://git.io/JUgJ6
<{^_^}>
[nixos-homepage] @garbas pushed 0 commits to fix-553: https://git.io/JUgJi
superherointj has quit [Quit: Leaving]
bahamas has joined #nixos
bahamas has joined #nixos
bahamas has quit [Changing host]
<jasom>
cransom: that at least lets me test without enabling various services on my current system. Still rebuilds the manuals though every time I change a configuration option
azure4 has quit [Ping timeout: 272 seconds]
<{^_^}>
[nixpkgs] @doronbehar opened pull request #98499 → qrupdate: refactor, and assert blas && lapack compatibility → https://git.io/JUgJQ
bahamas has quit [Ping timeout: 256 seconds]
azure4 has joined #nixos
<jlv>
Can I set an environment variable to persist through the phases of a derivation? When I set a variable in `preConfigure`, it was no longer set in `configurePhase`.
jumper149 has quit [Quit: WeeChat 2.9]
<{^_^}>
[nixpkgs] @zowoq merged pull request #98353 → bettercap: build on darwin → https://git.io/JUuGn
<{^_^}>
[nixpkgs] @zowoq pushed to master « bettercap: build on darwin »: https://git.io/JUgJp
<musteresel>
Hi; I'm trying to globally override a package (in /etc/nixos/configuration.nix) but am not getting it right somehow: I want to use a newer version of the VirtualBox Guest Additions, and as such tried to overwrite it both in nixpkgs.config.packageOverrides as well as when specifying boot.kernelPackages. During nixos-rebuild switch it DOES build the newer guest additions, but it still uses the old ones (verified using nix-store --query
<musteresel>
--referrers PATH)
Mirrexagon has quit [Ping timeout: 240 seconds]
puffnfresh_ has joined #nixos
niksnut has quit [Ping timeout: 246 seconds]
<jlv>
Wow. The NixOS website is looking very nice.
cole-h has quit [Quit: Goodbye]
cole-h has joined #nixos
orivej has joined #nixos
Mirrexagon has joined #nixos
Rusty1 has joined #nixos
grobi has joined #nixos
<evanjs>
any way to produce a list of filenames and paths from a list of absolute paths?
<evanjs>
which I'm hoping will solve my problem lol
<stigo>
yeah, gj on the new web design
tomberek has quit [Remote host closed the connection]
<evanjs>
stigo: I was so confused opening the manual the other day lol
<evanjs>
"Where am I"
philr has joined #nixos
<{^_^}>
[nixpkgs] @stigtsp opened pull request #98505 → [20.09] perl-cross: fix . being included in INC → https://git.io/JUgIe
<musteresel>
I don't get it ... the config value (in the nixos modules) is evaluated as fixed point, no?
Mateon2 has joined #nixos
Mateon1 has quit [Ping timeout: 240 seconds]
Mateon2 is now known as Mateon1
<jasom>
ah, Janne Heß beat me to refactoring nsswitch, I don't like their solution quite as much as mine though.
<das_j>
jasom: What's yours?
<das_j>
(also I didn't?)
mbrgm_ has joined #nixos
mbrgm has quit [Ping timeout: 272 seconds]
mbrgm_ is now known as mbrgm
<jasom>
das_j: each module for each entry has a priority for ordering (including files and systemd) so the configuration can customize orderings of modules
lordcirth_ has joined #nixos
<jasom>
das_j: I see there is this mkOrder used which I was not aware of; if say the sssd configuration provides a mkOrder, can the user override it in their configuration?
<das_j>
they can `mkForce` and then they'll have to write all entries themselves
<jasom>
right, which is what I was trying to avoid
ericsagnes has quit [Ping timeout: 260 seconds]
<das_j>
oof
<das_j>
feel free to ping me on that PR
<abathur>
evanjs: in Nix, I guess?
<jasom>
das_j: I'm putting a PR together in a few hours, I noticed your change after I rebased to master and got conflicts
<jasom>
das_j: my idea is roughly: `nsswitch.passwd.foo = { priority = mkDefault 123; };` would add a "foo" entry to passwd with ordering of 123, and then a user can do `nsswitch.passwd.foo.priority=456` and reorder just that one entry
alp has joined #nixos
gueorgui has joined #nixos
* jasom
has learned to pull before trying out some changes; the conflicting change was 6 months old