<fuiltilt>
immae: xorg.libX11.man is the kind of thing I was looking for, but didn't find an explicit reference to in nixpkgs. I can see it has what I want, but including it in a nix-shell doesn't let me do "man XOpenDisplay". Trying it through configuration.nix...
<clever>
fuiltilt: also try referencing the gist i gave, and add in xorg.libX11.man
supersandro2000 has quit [Ping timeout: 246 seconds]
<fuiltilt>
clever: Yeah, I saw that. Thanks for the pointer to "man-pages" especially. Adding that via shell.nix/nix-shell *does* work (eg. "man perror"), even though xorg.libX11.man doesn't. They both work when added via configuration.nix.
<fuiltilt>
immae++
<{^_^}>
immae's karma got increased to 29
<fuiltilt>
clever++
<{^_^}>
clever's karma got increased to 554
<immae>
cool :)
gxt has quit [Ping timeout: 268 seconds]
dadada_ has quit [Quit: WeeChat 2.9]
dadada_ has joined #nixos
nwm_ has joined #nixos
<fuiltilt>
It's still very mysterious. Why does xorg.libX11.man exist even though there's no reference to it in nixpkgs/pkgs/servers/x11/xorg/default.nix? Why does one package work under nix-shell and the other doesn't?
mbrgm_ has joined #nixos
<clever>
fuiltilt: the outputs array contains "man"
<clever>
its a split-output package, where it will have a $out and a $man
<devalot>
Using `virtualisation.oci-containers.containers.<name>.ports` causes Docker to insert a firewall rule that opens the given port to traffic *outside* the host, bypassing the nixos firewall chain. This seems like a big deal to me. I didn't realize that I had an insecure service open to the internet. Is this behavior expected?
supersandro2000 has quit [Ping timeout: 246 seconds]
<energizer>
there are subdirectories in the nix store
<gchristensen>
nwm_: how would it work while following FHS?
thelounge52 has quit [Ping timeout: 276 seconds]
<energizer>
you could reorganize so that /lib got all the /nix/store/*lib/ stuff
<tpw_rules>
you could have a database with the hash and FHS path and some system which monitored what a particular derivation laid down, but then you couldn't have duplicates
<nwm_>
not FHS directly but similar, /api/$triplet/{bin,lib,etc}/$hash
<tpw_rules>
and it would be easy to have a program work but not be monitored by nix leading to the breakage nix is designed to avoid
ahmed_elgabri has joined #nixos
Jd007 has joined #nixos
<gchristensen>
one of the ideas of Nix is that $triplet is a lie, basically -- nowhere near precise enough to be actually useful
<nwm_>
sure it is
<gchristensen>
oh ok
aswanson has joined #nixos
<tpw_rules>
nwm_: you're just swapping around two components of the path it looks like
<nwm_>
it is just a means to store data not the actual visible binaries
<tpw_rules>
like right now it's /nix/store/$hash/{bin,lib,etc}
<gchristensen>
right, but an important piece is the fundamental model of Nix works on all sorts of OSes, and not every OS would let us do that sort of fancy stuff
<nwm_>
isn't there a hard dep on systemd to boot nixos?
<cole-h>
For now.
<gchristensen>
nixos yes but nix works on macos and even windows, and used to work on freebsd
<gchristensen>
and other Linuxes too
<nwm_>
all have RUN_PATH
<nwm_>
~/.local/nixos/api/$triplet...
<nwm_>
and users/groups
<gchristensen>
yeah, I'm still wondering your motivation, because so far I haven't found the suggestions very compelling -- but they might be compelling if we knew the motivation
marcusr has quit [Remote host closed the connection]
<citadelcore>
How can I run nixos-rebuild switch with --impure ?
<citadelcore>
Says it's an invalid parameter
<citadelcore>
I've tried `sudo nix build .#nixosConfigurations.avalon.config.system.build.toplevel --impure` then `sudo ./result/bin/switch-to-configuration switch`, but this does not set up the boot menu entries
aveltras has quit [Quit: Connection closed for inactivity]
thelounge52 has joined #nixos
stiell has joined #nixos
nhs has joined #nixos
stephank has quit [Quit: stephank]
Qubasa has joined #nixos
thelounge52 has quit [Ping timeout: 240 seconds]
stephank has joined #nixos
nhs has quit [Ping timeout: 240 seconds]
totapunk has joined #nixos
totapunk has quit [Client Quit]
stiell has quit [Ping timeout: 246 seconds]
nhs has joined #nixos
nwspk has joined #nixos
mzan has joined #nixos
<yjftsjthsd>
I'm working on a nice little raspberry pi appliance. Is there any way that I can flash the generic image to the card, and then drop my configuration.nix somewhere on the card and have the system automatically read and apply my configuration on first boot? Docs talk about it automatically resizing the partition, but I'm assuming that particular bit
<yjftsjthsd>
is just hardcoded in
nhs has quit [Ping timeout: 258 seconds]
ManiacOfMadness has quit [Ping timeout: 272 seconds]
stiell has joined #nixos
marcusr has quit [Remote host closed the connection]
marcusr has joined #nixos
LilleCarl has quit [Ping timeout: 258 seconds]
LilleCarl has joined #nixos
Jd007 has quit [Quit: Jd007]
marcusr has quit [Read error: Connection reset by peer]
nhs has joined #nixos
Supersonic112 has joined #nixos
Supersonic has quit [Ping timeout: 258 seconds]
Supersonic112 is now known as Supersonic
marcusr has joined #nixos
stiell has quit [Ping timeout: 258 seconds]
nhs has quit [Ping timeout: 240 seconds]
thelounge52 has joined #nixos
nhs has joined #nixos
stiell has joined #nixos
IRCsum has joined #nixos
nhs has quit [Ping timeout: 240 seconds]
thelounge52 has quit [Ping timeout: 240 seconds]
<colemickens>
yjftsjthsd: how will it go about acquiring the things that the new config.nix needs?
<colemickens>
(not saying it's a bad idea, just making sure you've considered this part)
<yjftsjthsd>
The same way that it would if I waited for it to boot, manually typed in a configuration.nix, and ran a rebuild-switch
<colemickens>
fair enough, ok
<colemickens>
yjftsjthsd: I think you'd more or less need to write this yourself.
<colemickens>
grah/amc's packet nixos images do this, you might be able to look there for inspiration, and/or might not need as robust-of-a-solution as was devised for Packet.
<yjftsjthsd>
which is to say... not 100% sure, but something like "boot using the default config, which should include a one-time task to check for /magic/path/configuration.nix and if it exists then once the system is "up" (probably just multi-user.target or w/e) immediately apply that config
<yjftsjthsd>
yeah, okay... I suppose nix has enough tooling that making my own images is doable, I just hoped I wouldn't need to
<colemickens>
"immediately apply that config" = use sed or something to edit /etc/nixos/configuration.nix to include a reference to /path/to/file, and then `nixos-rebuild ... `
<colemickens>
but yeah, that's what I imagined
<yjftsjthsd>
can't you just give nixos-rebuild a file argument?
<colemickens>
uhhh I'm not sure actually, I don't tend to directly use it often
<yjftsjthsd>
I looked at it once because it felt awkward to have /etc/nixos be a git repo directly
* colemickens
really needs to turn back on docs for this machine
<yjftsjthsd>
Anyways. It's a simple-sounding idea that seems useful in general - if it did work and was actually that simple I wonder if I could get it merged into the official images
<colemickens>
yjftsjthsd: there's something else to consider, you'd need to verify if the SD image you create is ready to be used in-place.
<colemickens>
but depending on how the image is generated, it might not have the config or nixpkgs channel copied into the SD card image
<yjftsjthsd>
I just don't understand what you mean by "used in-place"
<yjftsjthsd>
ah
<yjftsjthsd>
wouldn't that be broken out of the box anyways?
nhs has joined #nixos
<colemickens>
I don't want to make things more complicated than they need to be, but I almost imagine a sort of what to specify a different second-stage-2 to handoff to.
<colemickens>
the stage-1 would read from /special/external/FAT32/stage-2-toplevel to determine which stage-2 to load into
<colemickens>
(I haven't really fully thought this through or tried this)
<yjftsjthsd>
I think I follow. I dunno, that *feels* over-complicated to me, but I may be grossly underestimating the problem space:)
<colemickens>
yeah, I don't want to speculate more without experimenting first
<colemickens>
in the SD image, we already add some extra stuff to postBootCommands to (finish) setup of the overlayed nix store and set the current-system to boot into
<colemickens>
It seems like if you had stage-1 networking enabled, you could try reading the text file, nix-store -r $(cat /the/fat32/toplevel), and then set that to /run/current-system instead?
<colemickens>
samueldr: have you ever tried something like this? or this seems like something clever might have done ?
nhs has quit [Ping timeout: 276 seconds]
<samueldr>
no idea, I see it's about provisioning on first boot?
<samueldr>
I think no effort has been made as it's assumed you'd "just generate your own image"
<colemickens>
but I can't be the first one to dream up stage-1-only images
zie has joined #nixos
rail has joined #nixos
<samueldr>
well, target disk mode for mobile nixos doesn't generate a stage2
<samueldr>
but it doesn't use the usual nixos stage-1 either
<samueldr>
:)
orivej has quit [Ping timeout: 240 seconds]
<colemickens>
I want it for my cloud vm images too. I currently just boot the latest image I made, and then immediately apply my latest config, but in reality, I'd like to have a floating pointer somewhere that points to stage-2 and then stage-1 just loads it into the ephemeral disk at boot
<colemickens>
Sort of gets dicey if I start changing mounts and stuff though, maybe that's too much moving parts.
<samueldr>
sounds to me what you want is "not our stage-1, but something else"
<samueldr>
not that there's no value
<samueldr>
after all... I did go for "not our stage-1"
<colemickens>
samueldr: maybe? but I'm not sure it's different enough that I couldn't just throw it in postBootCommands?
<colemickens>
samueldr: or maybe like a stage-1.5
<samueldr>
postBoot happens in stage-2!
<colemickens>
you're probably right :)
<colemickens>
-_-
<samueldr>
it's really not a bad idea to experiment with those things
<samueldr>
it's pretty trivial with NixOS
<samueldr>
(compared to other distros)
<colemickens>
yeah, more of if I'm willing to dedicate the time to trying all of the iterations it would take
<eyJhb>
I am basically just building small dirs for s6, so I could use either one. Runcommand might be easier
cfricke has quit [Quit: WeeChat 3.0]
cfricke has joined #nixos
<cole-h>
You'll need to patch out that `git` call and drop the kemal source where it expects to be found. But I'd expect it's wanting to do much more networking.
<Hail_Spacecake>
what is that fetchFromGithub line doing then, if you can't make network requests?
<eyJhb>
But I somewhat need it to place the dirs inside the container... :/
<Hail_Spacecake>
I'm very new to nix, and I'm trying to write a derivation that will build a piece of software currently on github, so I can use it in nixos
<adisbladis>
Nix allows for networking _if and only if_ the output hash is known ahead of time
<cole-h>
Hail_Spacecake: Assuming you're trying to package something using Crystal, you might have a look at https://danilafe.com/blog/crystal_nix_revisited/. I don't know if it'll help at all, though.
kaivai has joined #nixos
nhs has quit [Ping timeout: 256 seconds]
<Hail_Spacecake>
yeah I'm following that blog post
<cole-h>
To further specify: Nix allows networking iff the output hash is known ahead of time in a way that Nix can understand (e.g. the `sha256` attr of `fetchFromGitHub`)
<Rovanion>
I'm trying to package prometheus-slurm-exporter for Nix but can't for the life of me debug the buildGoPackage procedure. It fails when the build program exits with a non-zero exit, but that is all I can get out of my nix-build invocation, despite --verbose --debug and --keep-failed.
<Rovanion>
When running `nix-shell $NIXPKGS -A prometheus-slurm-exporter` only the genericBuild shell functions seem present, and they fail at a different place it seems.
<djanatyn>
when i use `:b` in `nix repl` i only get the last 10 lines of build output if there is a failure. how do i get more, or see the rest of the output from the previously failed build?
puck has joined #nixos
orivej has quit [Quit: No Ping reply in 180 seconds.]
<hexa->
I have a list of components and I have a function that returns a list packages required for that component. I want to extend checkInputs with that aggregated list of packages.
<hexa->
can someone recommend a function to achieve this? :)
vrinek has quit [Remote host closed the connection]
<maralorn>
gentauro: Yes, you can downgrade. Pick the VB 5.0 from a pinned older version of nixpkgs and override the vb-version in an overlay. That will lead to the module also using the older package.
<maralorn>
I don‘t know though, if there are maybe some changes in the module which might be incompatible with the old VB version.
<gentauro>
maralorn: I ended up restarting nixos and now it works. `NixOS` the new M$ OS? :|
<gentauro>
could somebody update the topic of the channel to: «Have you tried to turn it off and turn it on again?» xD
<maralorn>
I think all linux distribution will require manual intervention after the update of a system service. It’s sad when it doesn‘t work on nixos, but sometimes there is no chance. e.g. with kernel-updates.
<gentauro>
maralorn: well a message like: "Please restart" would be helpfull :)
<gentauro>
no shame in trying to be like M$ xD
<maralorn>
gentauro: That is actually a good point.
<maralorn>
I think we cater to that a little bit with "Not restarting the following services:"
hiro98 has quit [Ping timeout: 265 seconds]
<gentauro>
:)
thelounge52 has joined #nixos
nhs has joined #nixos
alexherbo2 has quit [Ping timeout: 265 seconds]
saschagrunert has quit [Remote host closed the connection]
<attila_lendvai>
my package has been merged into the nixpkgs repo. when/how will it automatically enter the unstable, and then the stable channel? where can i read about the process?
<tpw_rules>
tl;dr the build server automatically rebuilds all the packages in the repo for each commit and makes sure the tests pass and every build succeeds if possible. then once the packages are transferred to the cache server that commit becomes the new version of that channel
<attila_lendvai>
tpw_rules, thank you!
<tpw_rules>
for ending up in stable, that happens when the new stable version is getting ready to be released, or sometimes packages are backported if they contain important fixes
crdb has joined #nixos
<crdb>
Hello
nhs has quit [Ping timeout: 246 seconds]
<crdb>
I was wondering if it's possible to build Docker images with Nix without docker?
<lukegb>
crdb: nixpkgs' dockerTools already doesn't use docker, I think
mmmattyx has joined #nixos
<crdb>
lukegb oh really
<crdb>
Was just reading on the official doc/guide and the assumption is that you have docker installed
<Luker>
Hi. I noted that the last succesful hydra build for raspi4-aarch64 was almost a month ago...any known problems? I have problems with wpa_supplicant and don't know if it's me or the build
<lukegb>
Oh, heh. No, it just outputs an image tarball, you shouldn't need docker installed for that
<attila_lendvai>
tpw_rules, so, standalone additions are usually not backported, right? (thinking of the freshly added services.bee, a network app/service that nothing else depends on)
<tpw_rules>
no that most likely wouldn't be a candidate
<crdb>
lukegb interesting guess I should try this out !
respawn_ has joined #nixos
<tpw_rules>
attila_lendvai: if you want to try the package on your personal system i would download the derivation to a local file and create an overlay
<{^_^}>
[nix] @regnat pushed 3 commits to ca/drvoutputs-commands: https://git.io/JtzrV
<attila_lendvai>
tpw_rules, i'm new to nix, but i'm the author of the package. someone helped me to set up a local overlay to be able to test it locally. now i'm about to extend the app's documentation about nix(os) install instructions, but i'm still too new to all of this...
devalot has left #nixos ["ERC (IRC client for Emacs 27.1)"]
<gchristensen>
anyone have a nice way of setting up vscode-server on a nixos target? Right now I run this nightmare script: find ~/.vscode-server -name node -type f | xargs patchelf --set-interpreter $(patchelf --print-interpreter $(which patchelf)) --set-rpath $(dirname $(ldd $(which patchelf) | grep stdc++ | cut -d' ' -f3))
<jsoo>
Hi nix! I'm a guix user and I have some questions about impurity in my builds. When building a cpp program with cmake somewhere when checking if the cpp compiler works, /bin/sh is not found. What should I do?
<{^_^}>
[nixpkgs] @dotlambda pushed to master « pythonPackages.netifaces: use pythonImportsCheck »: https://git.io/Jtzoo
<supersandro2000>
jsoo: Are you on darwin with sandbox enabled?
<jsoo>
Im trying to use nic for work in a guix system host
lsix1 has quit [Ping timeout: 260 seconds]
<theuni>
hmm. i have a weird issue with crng init.
<theuni>
starting with a netboot image everything is fine
<adisbladis>
jsoo: Ok, then you want patchShebangs
<jsoo>
Ah, how do I do that?
<theuni>
starting (exact same kernel) into an installed system never gets to "crng init done" and blocks the lvm
<adisbladis>
Which will rewrite #!/bin/sh to a store path
<jsoo>
In guix that's done by default
lsix1 has joined #nixos
<theuni>
in the netboot image "crng init done" happens after 13 seconds, but in the real system many things seem to try to read from random but crng init never happens ... even after 60s :/
<jsoo>
Is patchShebangs an executable I run in a hook? A thing to enable?
<theuni>
any pointers? i've seen this issue before but really solving it seems to be a hit-and-miss thing
<jsoo>
updating my nixpkgs pin to 25d09bd also has the same issue
<avaq>
Hi folks! I've created a simple HTTP server with NixOps, using mostly options from 'services.nginx.virtualHosts'. One in particular: enableACME (https://is.gd/xqTe2F), is causing me some trouble.
<avaq>
In particular, it seems to be requesting a certificate for the www subdomain as well as the parent domain, but the DNS for the subdomain is not pointing to my server. It's likely unrelated to Nix, but there's a chance it is.
<avaq>
I've made a writeup here in a support ticket with LetsEncrypt: https://is.gd/gs4fcZ .. But I'm checking if anyone here might know what it is, or if it could be related to Nix.
<s1341>
is there a way to override a variable set in the let section of a derivation?
<s1341>
(from a package depending on that one?
<s1341>
)
ahmed_elgabri has quit [Ping timeout: 258 seconds]
<s1341>
(that in my derivation's let section) but when i try to use compiler_rt where I would use pkgs.llvmPackages_11.compiler-rt, I get 'variable compiler_rt is not defined:w'
<adisbladis>
s1341: Wouldn't you expect that to be set by pkgs.llvmPackages_11 ?
<adisbladis>
(I'm not familiar with this)
akaWolf has quit [Ping timeout: 240 seconds]
dsg has joined #nixos
<jsoo>
from the docs on sandbox-paths, /bin/sh might be mounted depending on how nix was built. how can i make sure that points to the nix/store/.../bin/sh?
<s1341>
adisbladis: maybe...
<s1341>
but in any case, I want to work around this issue. because for me useLLVM IS set.
<s1341>
(because I set it explicitly in a crossSystem).
<jsoo>
hm. looks like the default is not set, that must be the issue.
<jsoo>
adisbladis: thanks a bunch! i filed a bug with guix
domogled has quit [Ping timeout: 256 seconds]
<attila_lendvai>
how come that my changes in /etc/systemd/logind.service don't survive a reboot. after i do another nixos-rebuild switch, it becomes once again what i expect it to be -- until a reboot. any ideas?
mario6221[m] has left #nixos ["User left"]
jiribenes has quit [Quit: jiribenes]
jiribenes has joined #nixos
<jsoo>
s1341: yeah though it is bind-mounted into the build sandbox and i think we're missing the default sandbox-paths
<adisbladis>
jsoo: I felt like I was mostly fumbling, but glad you figured something out :)
<{^_^}>
#111939 (by Avaq, 58 seconds ago, open): The globalRedirect and enableACME options under services.nginx.virtualHosts produce a broken nginx config when used together
zupo has joined #nixos
Qwerky has joined #nixos
KarlJoad has joined #nixos
malook has joined #nixos
zupo has quit [Client Quit]
koldbyte has joined #nixos
<{^_^}>
[nixpkgs] @dotlambda pushed to release-20.09 « checkinstall: mark as insecure »: https://git.io/JtziV
koldbyte has quit [Ping timeout: 276 seconds]
avaq has quit [Quit: Leaving]
<aanderse>
infinisil: with `freeformType` module options can i use `either` to specify an option type? i tried but got an error
<{^_^}>
ofborg#524 (by FRidh, 31 weeks ago, open): Evaluation check does not consider `passthru` and thus `passthru.tests`
zakame has quit [Ping timeout: 240 seconds]
<turion>
So they are actually supposed to be run upon @ofborg eval ..., or @ofborg build ...?
<Taneb>
Is there machinery in nixpkgs for making bootable isos that have various things preinstalled, but no package manager? I think dockerTools does something similar
zebrag has quit [Quit: Konversation terminated!]
zebrag has joined #nixos
<gchristensen>
turion: where is your PR?
arjen-jonathan has quit [Ping timeout: 272 seconds]
<{^_^}>
[nixpkgs] @peti pushed to master « mlt: update to version 6.24.0 »: https://git.io/JtzbB
lsix1 has joined #nixos
turion has quit [Ping timeout: 272 seconds]
mmmattyx has quit [Quit: Connection closed for inactivity]
<theuni>
ok, my issue with the RNG wasn't RNG but a missing disk driver in the initrd hiding behind a lot of RNG output ...
<theuni>
another question: i'm currently looking at the build results for netboot and neither my hydra nor the official one represent the build results properly as links - any idea where to start looking to fix this?
<{^_^}>
hydra#827 (by ajs124, 13 weeks ago, open): add test for nix-support/hydra-build-products edge case
thelounge52 has quit [Ping timeout: 258 seconds]
<ajs124>
yup
actor__ has joined #nixos
<actor__>
hello, has anyone here got experience or resources about installing nixos on system76 laptops (specifically lemur?) I can't get the display to work for the install disc, and I'm also curious about the system76 drivers for once I've got that set up.
simba1 has quit [Ping timeout: 272 seconds]
Qwerky has quit [Remote host closed the connection]
<ajs124>
theuni: considering the age of this issue, it seems like nobody is in a rush to fix it
simba1 has quit [Ping timeout: 260 seconds]
<theuni>
yeah, it's a blocker for me so i'll keep looking at it then
<theuni>
quick question though: would fixing this show the results for older builds or would those builds need to be restarted to make them work again?
simba1 has joined #nixos
<theuni>
i'm not sure whether the meta data is rendered on access or stored in a db.
actor__ has quit [Quit: Ping timeout (120 seconds)]
DanC has quit [Ping timeout: 258 seconds]
footlooseboss has quit [Client Quit]
CMCDragonkai1 has joined #nixos
actor__ has joined #nixos
<ajs124>
theuni: from what I remember about this, file in nix-support is actually empty, so you might actually need to restart the builds. not sure about that, though.
<theuni>
i checked the commit that broke it and there is a small change with how data in the nar is accessed and that results in store paths from different derivations (which the netboot job just references) are now looked up in the nar and obviously not found
<theuni>
anyway, it's fine if no one is actively working on it at the moment, I saw you involved last year, but i'm happy to do it next week. i could be completely wrong, of course ;)
<theuni>
off to bed for now
<ajs124>
did your build run on the same host the hydra runs on or on a remote builder?
DanC has joined #nixos
noudle_ has quit []
<ajs124>
ah, ok. well, feel free to post your findings on that PR or reference it, if you find a fix for it.
CMCDragonkai1 has quit [Remote host closed the connection]
<theuni>
same host
<theuni>
now off for real
<theuni>
could be two separate issues and maybe the second one gets easier when the other half gets fixed
<Mic92>
matthewcroughan: I think one fix could be to build glibc by default with the locale they try to use so that setting LOCALE_ARCHIVE is no longer needed
<matthewcroughan>
is that expressable in a nix-shell today?
<{^_^}>
[nixpkgs] @rycee pushed commit from @r-ryantm to master « gpsprune: 20.1 -> 20.2 »: https://git.io/Jtzpd
<matthewcroughan>
or does this require fundamental changes?
<abathur>
does anyone know a pattern for including/excluding specific files in a local src on invocation? I've painted myself into a bit of a dumb corner :P
actor__ has quit [Quit: Connection closed]
<matthewcroughan>
Mic92: So I'd override the postInstall?
<Mic92>
matthewcroughan: add before old.postInstall
lsix1 has quit [Ping timeout: 272 seconds]
<matthewcroughan>
Mic92: I'm still too new to know what that means
<matthewcroughan>
old.?
<Mic92>
matthewcroughan: (glibc.override
<abathur>
I have a folder for a blog post, and it's got a markdown file, a few nix/sh sources for use in the post, and I added a Makefile + shell script to automate invoking the other files and collecting their output without thinking about the `src = ./.;` also picking up the Makefile and changing the build
<abathur>
I can obviously use something like filterSource to exclude it, but it'll be quoted in the post, and I'd prefer not leaving meta tool-marks that only distract from the point of the post
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<matthewcroughan>
ninteresting.. Mic92
<matthewcroughan>
error: undefined variable 'glibc' at /home/matthew/yocto/yocto.nix:7:11
<Mic92>
matthewcroughan: that's what I mean with self and upser
<Mic92>
*super
<Mic92>
check out the wiki
<matthewcroughan>
what is self/super?
<matthewcroughan>
before/after?
<matthewcroughan>
which is before and which is after?
philr has joined #nixos
<matthewcroughan>
Mic92: looking at the wiki syntax, we no longer use overrideAttrs?
<Mic92>
matthewcroughan: no overrideAttrs is the right tool for this change
<matthewcroughan>
`error: anonymous function at /nix/store/0fa2g5a7rdimq5b8wbwk92rb9lijmggb-source/pkgs/build-support/build-fhs-userenv/env.nix:3:1 called with unexpected argument 'overlays', at /nix/store/0fa2g5a7rdimq5b8wbwk92rb9lijmggb-source/pkgs/build-support/build-fhs-userenv/default.nix:8:9
<matthewcroughan>
(use '--show-trace' to show detailed location information)`
<clever>
matthewcroughan: overrides against glibc arent easy, so its simpler to just edit nixpkgs directly
<matthewcroughan>
I want to prebake in support for `en_US.UTF-8 > ../glibc-2*/localedata/SUPPORTED
<matthewcroughan>
make -j''${NIX_BUILD_CORES:-1} -l''${NIX_BUILD_C`
<matthewcroughan>
whoops, pasted too much sorry
<matthewcroughan>
en_US.UTF-8 is what I want prebaked in, and I change this L82 to be `SUPPORTED-LOCALES=C.UTF-8/UTF-8/en_US.UTF-8` and it fails claiming no such file or dir
tokudan has quit [Remote host closed the connection]
<matthewcroughan>
`C.en_US.UTF-8...[error] character map file `charmaps/en_US.UTF-8' not found: No such file or directory`
SomeoneSerge has joined #nixos
fendor_ has quit [Read error: Connection reset by peer]
<SomeoneSerge>
Hi, I've got a new silly question. Has anyone tried and succeeded running singularity containers with nix-provided singularity on a non-nixos system?
<HedgeMage>
SomeoneSerge: I have not, but if there's something specific you are trying to figure out, I have access to the Singularity dev team, my local nix box, and a pretty big set of Singularity-supporting HTC and HPC clusters...care to fill me in?
<HedgeMage>
*local nixos box
cognemo has quit [Quit: cognemo]
cognemo has joined #nixos
sigmundv has quit [Ping timeout: 240 seconds]
thelounge52 has joined #nixos
erasmas has quit [Quit: leaving]
Darkmatter66 has quit [Ping timeout: 276 seconds]
<ornxka_>
i have patches = [ ./blah.patch ]; but the file in $out is not patched
cseder__ has quit [Ping timeout: 240 seconds]
<clever>
ornxka_: the patch is applied to the src in . before the build starts
<SomeoneSerge>
HedgeMage: well, I only tried so far to `singularity --run` and get a message that that'd require setuid
<SomeoneSerge>
on nixos that's apparently handled by programs.singularity.enable = true
<cole-h>
On non-nixos, you'll have to setuid it yourself, then.
<SomeoneSerge>
but wouldn't that mean modifying attrs of a file in /nix/store?
cseder has joined #nixos
<HedgeMage>
SomeoneSerge: The target system (the one you want to run the container on) may need setuid enabled for a wide array of functions if the version of Singularity is old enough. It was maybe a year ago that production Singularity got the ability to run all functions without setuid privileges, and a little later that it made it into production for all architectures. Add in the delay for that to
<HedgeMage>
propogate for to some of the more laggy repositories, and it's still new to some people.
<HedgeMage>
(I'm guessing a lot of context here, as I'm not entirely sure what you're doing in your container or where you are trying to do it.)
<ornxka_>
im a bit perplexed at this.. my installphase is literally just copying from $src to $out, it says applying patch <mypatch>, patching file <myfile>, but in the source dir (and out dir) the file is unpatched
<lovesegfault>
Is there an easy way to fetch from a GCP bucket when writing a drv?
<ornxka_>
when i do nix-shell and then unpackPhase; patchPhase, it is patched