<ottidmes>
off topic, although I am cleaning the shell code of my NixOS configuration, is there any reason not to set -o pipefail globally?
EarlDeLaWarr has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<colemickens>
How can I learn more about libGL and libGLU_combined packages?
<colemickens>
I'm building a new Chromium variant and it's complaining about libGLESv2 at runtime, despite me putting 'libGL' in buildInputs: "Failed to load GLES library: libGLESv2.so.2: cannot open shared object file: No such file or directory"
<colemickens>
I am not sure if there's some additional patching that I need to do or something else?
<clever>
colemickens: libGLU_combined is just a buildenv over 2 other packages
<clever>
colemickens: also, are you changing LD_LIBRARY_PATH at all? it needs a certain value for gl to work on nixos
<colemickens>
LD_LIBRARY_PATH is set to /run/opengl-driver/lib
rgrau has quit [Ping timeout: 272 seconds]
<clever>
yeah, thats right
<colemickens>
but that directory does not contain the GLES so. but according to a nixpkgs Issue I found, that's expected too?
<ottidmes>
I guess I will wrap my global settings in a check for IN_NIX_SHELL to make sure I do prevent unforseen effects of those settings on my builds done in nix-shell vs nix-build
freeman42]NixOS has joined #nixos
jsgrant- has joined #nixos
freeman42x]NixOS has quit [Ping timeout: 246 seconds]
freeman42]NixOS has quit [Ping timeout: 246 seconds]
erasmas has quit [Quit: leaving]
<Shados>
Are there any problems with adding an attribute to the nixpkgs attribute set that lists the overlays used to build it? I have been playing with a local checkout that adds such an attribute (in pkgs/top-level/stage.nix), but I'm wondering if there's a specific reason it hadn't already been done.
LogicAside has joined #nixos
Rusty1 has quit [Quit: Konversation terminated!]
<ottidmes>
Shados: on top of my head, not all overlays are file based
<{^_^}>
[nixpkgs] @lopsided98 opened pull request #51259 → reptyr: enable for Linux on all architectures → https://git.io/fpwXk
<Shados>
I'm not talking about listing the files, I do mean the literal overlay functions
Mateon3 has joined #nixos
graphene has quit [Remote host closed the connection]
Mateon1 has quit [Ping timeout: 245 seconds]
Mateon3 is now known as Mateon1
graphene has joined #nixos
hamishmack has joined #nixos
graphene has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/fpwXw
<ottidmes>
Shados: but you cannot go from runtime values to nix code again, right? I am probably misunderstanding you somewhere
mayhewluke has quit [Ping timeout: 250 seconds]
jperras has quit [Ping timeout: 268 seconds]
ekleog has quit [Quit: back soon]
ekleog has joined #nixos
<dmj`>
Every ghcjs derivation on nixpkgs is broke af
goibhniu has quit [Ping timeout: 246 seconds]
<simpson>
With kubenix, is it possible to get the cluster's name into my configuration? I have global-level (literally, multi-cloud) stuff that needs a cluster-unique name, and I've just been using the name itself.
jsgrant has joined #nixos
<simpson>
Actually, never mind; I just realized that I can do this in a better way.
<samueldr>
any reasons why nix-dev threads are not all at the same location?
<samueldr>
cyounkin_: ^
<simpson>
cyounkin_: Yeah, why are multiple different domains used? +1 BTW, great idea.
ng0 has quit [Quit: Alexa, when is the end of world?]
<cyounkin_>
I plan to run this on a large number of projects, so I'm trying to make minimal changes. In some cases the page exists at article.gmane.org but not comments.gmane.org
<cyounkin_>
If it isn't at gmane.org, I try others in order of preference.
<simpson>
Mm, makes sense. Well, I have no approval powers, but I informally approve.
<samueldr>
I was thinking of only the `nix-dev` ones, which sometimes are at mail-archive.com, and sometimes at marc.info
pkx3__ has joined #nixos
<samueldr>
the other ones, available at article.gmane.org, I think is fine, even better, than going at another random place
<cyounkin_>
samueldr: You can check by hand but if it changed it to be marc.info, it wasn't found at gmane or mail-archive.com
<cyounkin_>
Er I guess it's hard to check by hand.
<samueldr>
yeah, I would have otherwise :)
pkx3 has quit [Read error: Connection reset by peer]
<samueldr>
I'm not even sure how you found the right threads?
<cyounkin_>
It's connecting to the NNTP server to get the message-id, then querying the other sites based on that.
<cyounkin_>
Apparently the NNTP server has threads that are not available through the web interface.
<samueldr>
oh, they still have their archives as NNTP?
<samueldr>
interesting
epicmetal has quit [Quit: epicmetal]
cyounkin_ is now known as cyounkins
<m0rphism>
is there a straight-forward way to to install nvidia-drivers version 396.54? They seem to be required for running SteamVR via Proton (wine fork).
exarkun1 has joined #nixos
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fpwMe
mayhewluke has joined #nixos
<{^_^}>
[nixpkgs] @dtzWill opened pull request #51261 → libapparmor: drop patch, fix build w/musl → https://git.io/fpwMJ
<{^_^}>
[nixpkgs] @matthewbauer merged pull request #50212 → Add "emulator" function to systems → https://git.io/fptgV
<{^_^}>
[nixpkgs] @matthewbauer pushed 3 commits to master: https://git.io/fpwMq
<{^_^}>
[nixpkgs] @matthewbauer pushed 27 commits to master: https://git.io/fpwMs
Rusty1 has joined #nixos
marusich has joined #nixos
marusich has quit [Ping timeout: 268 seconds]
<simpson>
In kubenix in particular, but also more generally, what *are* my options for parameterizing a nix-build? ISTR that there's some environment variables.
cyounkins has quit [Remote host closed the connection]
<simpson>
Oh, I guess that today is the day for --arg.
pie___ has joined #nixos
dramforever has joined #nixos
pie__ has quit [Remote host closed the connection]
<colemickens>
I can't seem to make any progress with the libGLES issue in my chromium build. I've tried libglvndr instead of libGL but it didn't seem to have much effect
<simpson>
Has anybody packaged the newer Azure CLI? I see the "old" one, which seems fine to me, but I can't get it to login and it doesn't appear to support AKS.
<simpson>
I'd like the $(az) command, not $(azure).
<colemickens>
there are some open issues as well, but basically there are problems with how some python things are packaged that prevents easier inclusion in nixpkgs right now
<simpson>
colemickens: Cool, thanks.
orivej has quit [Ping timeout: 268 seconds]
<simpson>
Progress! /nix/store/bwfygfcdvis9wd1c1v51xwnwhw1hx0a0-python3-3.6.6/bin/python: No module named azure.cli
<simpson>
Hmm. But their Docker image doesn't work either. Clearly it's feierabend.
<bbarker_home>
what's the rationale for setupHook - had a few environment variables in a package's setupHook, but they don't seem to be set either when I install the package to my environment, or use it in a nix-shellb
<colemickens>
simpson: yeah, the exact comment I linked to is what I'm using in a shell.nix right now with some success.
<colemickens>
the overlay may be a bit old though, yeah, I guess if you're trying to update it.
<bbarker_home>
actually if I use the package like nix-shell -p foo they are used
<colemickens>
IDK why the CLI people are so averse to packaging it properly...
<bbarker_home>
so basically the setupHook isn't propagated
<colemickens>
I tried running my build with strace, but I didn't see anything obvious.
epicmetal has quit [Read error: Connection reset by peer]
<simpson>
colemickens: More progress! I have working $(az).
<simpson>
I am sorry that I do not have help on your issue. Yours looks way harder.
srk has joined #nixos
jperras has quit [Ping timeout: 246 seconds]
snajpa has joined #nixos
shabius has joined #nixos
mayhewluke has joined #nixos
hamishmack has quit [Ping timeout: 250 seconds]
worldofpeace has quit [Ping timeout: 246 seconds]
dycan has joined #nixos
lostman has joined #nixos
reinhardt has joined #nixos
<lostman>
Hey folks. How can I set the default `stdenv`? I have `default.nix` which is an entry point to a bunch of derivations and overlays and so on. I want to make the default compiler clang-7. What's the idiomatic way to set it up? (some of the derivations already use clang-7 but they're set individually)
vk3wtf has quit [Ping timeout: 252 seconds]
hamishmack has joined #nixos
Rusty1 has quit [Quit: Konversation terminated!]
aszlig has quit [Quit: Kerneling down for reboot NOW.]
aszlig has joined #nixos
graphene has quit [Remote host closed the connection]
graphene has joined #nixos
orivej has joined #nixos
jmeredith has quit [Quit: Connection closed for inactivity]
marusich has quit [Quit: Leaving]
pie__ has joined #nixos
palo1 has joined #nixos
pie___ has quit [Remote host closed the connection]
<sphalerite>
bbarker_home: installing something into a profile generally won't do anything other than making the relevant symlinks in the env path
ahawkins has joined #nixos
<sphalerite>
bbarker_home: and I mean "generally" in the sense of "globally", not "usually" :)
<sphalerite>
bbarker_home: they're used for adding hooks to builds using the package as a dependency
Diagon has joined #nixos
metastance has joined #nixos
aither has joined #nixos
dycan has quit [Ping timeout: 250 seconds]
Diagon has quit [Quit: Leaving]
crmlt has joined #nixos
jackdk has quit [Ping timeout: 250 seconds]
<colemickens>
If I set LD_LIBRARY_PATH to include libglvnd then I can get my Wayland Chromium build to launch.
<colemickens>
I'm not sure why it's necessary to be set. I did find the code where they open the GLES so too, but I can't tell if it needs patching...
<colemickens>
I guess it's because it loads by name, rather than linking or loading by path. Maybe patching is the only solution?
crmlt has quit [Remote host closed the connection]
<toppler>
I'm trying to remote build, and it seems to be getting as "copying dependencies" but then soon dies with "Connection reset by peer". Any help?
<toppler>
Never mind. Seems the user isn't allowed to connect to the nix daemon.
<{^_^}>
#51271 (by colemickens, 1 hour ago, open): `nix copy x` works, `nix copy x --no-recursive` does not
<colemickens>
that I just filed, that shows the same error in another case.
ahawkins has quit [Quit: Connection closed for inactivity]
<colemickens>
I've been using `nix copy --from ssh-ng://... --no-check-sigs` this evening, etc.
<colemickens>
Yeah, this is just a bug.
<colemickens>
I can take the path that it says is not valid, `nix copy` it from `file://...` and then go back to `--all` and it will just print the next path that it "can't" copy.
<colemickens>
:(
<srhb>
colemickens: Can't reproduce it.
<srhb>
Well, I used a signed path, so maybe that's the issue.
<colemickens>
yeah, I did that, and even stripped the sigs, and `nix copy --all` worked.
<Rovanion>
Does anyone happen to have a Nvidia GDX-1 with nixos on?
<colemickens>
(it was the `--all` that I think I'm having issues with, let me see what else I've got wrong)
<srhb>
Ah.
<srhb>
Not all binary caches support --all, but I guess it would be surprising if local file store didn't.
<colemickens>
srhb: I want to try deleting the store path too, I'm getting " cannot delete path, since it is still alive" for the the test package 'hello'.
<colemickens>
oh right ./result, it's always ./result
<srhb>
colemickens: nix-store -q --roots is also helpful
<colemickens>
srhb: I just don't know why I'm getting "error: path '/nix/store/js4x05nsfhb0q4lyapsgbc6ah13126i3-chromium-72.0.3608.4' is not valid" though.
<colemickens>
so it fails, then I copy the path it complains about, repeat the failed command, and it works.
<colemickens>
??
<srhb>
That's after nuking the metacache?
<colemickens>
yeah
<colemickens>
I need to be thorough though. It's likely that I imported the first two manually and then the rest succeeded because they were already in my store.
<srhb>
From here on, try disabling the metacache entirely while testing, with both negative and positive set to zero..
<colemickens>
unless I'm misunderstanding, the diff I have there should be roughly the same thing. Yet I'm getting the weird complaints about the path being invalid with `--from + --all + --no-check-sigs`.
<{^_^}>
[nixpkgs] @dotlambda opened pull request #51275 → nixos/borgbackup: allow paths to be empty → https://git.io/fprq1
<srhb>
colemickens: No idea. I don't understand why any of the shenanigans you're doing is necessary. Maybe if you can create a small reproducible case (ie. not involving chromium :-P) it'll be easier to see what's going on.
<srhb>
Everything seems to work here.
<colemickens>
I mean, nothing in that script relies on chromium
<colemickens>
it just happens to be that I hard coded the chromium store path to avoid the evaluation time each time
<srhb>
Right, but I can't seem to reproduce the case locally. And I'm not about to try modifying chromium to get the same failure :-P
lnikkila has quit [Quit: lnikkila]
lassulus has quit [Ping timeout: 250 seconds]
<{^_^}>
[nixpkgs] @dotlambda pushed 13 commits to python-unstable: https://git.io/fprmi
<{^_^}>
[nixpkgs] @dotlambda merged pull request #51230 → Upgrade and fix few broken python packages → https://git.io/fpVoi
vidbina has joined #nixos
lostman has joined #nixos
oldandwise has joined #nixos
<colemickens>
there are sooo many places 'path %s is not valid' appears in the nix source too
init_6 has joined #nixos
<colemickens>
basically my chromium build is ttrigger something that produces a path that can be `nix copy`'d to a local store, but not from one. :)
<colemickens>
but only when using the --all flag lol
alex`` has joined #nixos
<colemickens>
oh well, I got something that works, and a working Chromium Wayland build. I think I can call it a night.
symphorien has joined #nixos
metastance has quit [Ping timeout: 250 seconds]
alex`` has quit [Ping timeout: 272 seconds]
ghasshee has joined #nixos
ng0 has joined #nixos
<infinisil>
colemickens: Ohh you're using wayland? Does it work well?
<colemickens>
infinisil: daily driver for 2 months. Firefox Nightly w/ Wayland for the past few days.
m0rphism has quit [Quit: WeeChat 2.2]
<infinisil>
:O
<colemickens>
There are lots of little bugs but nothing stopping me from using Sway.
m0rphism has joined #nixos
<colemickens>
I was playing with my new Chromium build and its not quite there either - doesn't handle HiDPI + Wayland properly.
<colemickens>
Also, it triggers a seg fault in Sway somewhere.
<infinisil>
I see
<colemickens>
Everything else GTK3/QT5 is rock solid. Anything using XWayland is solid.
<infinisil>
Very nice, I'll have to give wayland a try too sometime
<colemickens>
infinisil: of the pkgs I have there, I'm not sure any of them have a stable release.
<colemickens>
I think wlroots had a 0.1 to sort of unblock packaging efforts elsewhere.
<infinisil>
I see, nice work! colemickens++
<{^_^}>
colemickens's karma got increased to 6
<colemickens>
plus I _was_ updating them many times a day :)
mayhewluke has joined #nixos
<colemickens>
due to the channels being stuck and meson changing I haven't pushed the latest updates, but once meson 0.48 is everywhere it will be frequent and steady again.
m0rphism has quit [Quit: WeeChat 2.2]
m0rphism has joined #nixos
<{^_^}>
[nixpkgs] @nomeata opened pull request #51276 → Add bisect_ppx and bisect_ppx-ocamlbuild → https://git.io/fpr3c
spacekookie has quit [Quit: No Ping reply in 60 seconds.]
alex`` has joined #nixos
reinhardt has joined #nixos
spacekookie has joined #nixos
<sebmacc>
Hello! I need help, I think I have broken Nix ^_^
<srhb>
cd
<sebmacc>
I had a local copy of the nixpkg-channels repository. I wanted to remove it and rely on the online repository itself. I don't understand what I have done wrong, and I don't know where to look in order to fix the errors I'm getting, namely "file 'nixpkgs/nixos' was not found in the Nix search path" (see https://pastebin.com/mCAHN1xE). Can someone te
<sebmacc>
ll me where to look in order to fix this?
<srhb>
Woops, sorry
__monty__ has joined #nixos
<srhb>
sebmacc: sudo nix-channel --list ? The channel you want to use for your system needs to be there.
<srhb>
sebmacc: Have you manipulated your NIX_PATH manually?
<srhb>
__monty__: Maybe I forget how channels work then.
astronavt has quit [Quit: Leaving]
<sebmacc>
I don't remember having modified this var. It contains "nixos-config=/etc/nixos/configuration.nix:/nix/var/nix/profiles/per-user/root/channels"
<__monty__>
srhb: Probably not. But I'm in the awkward situation of the mac installer not providing me with a channel.
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
<sebmacc>
Maybe I did change it, but either it was a long time ago when I started having a local copy of the repo, or I fiddled with it when trying to fix this
johanot has quit [Quit: WeeChat 2.2]
<srhb>
sebmacc: That looks OK. The channel should be in /nix/var/nix/profiles/per-user/root/channels -- can you look there?
<srhb>
sebmacc: sudo nix-channel --update *might* be all that's missing, but again, I'm rusty with channels.
<srhb>
__monty__: If it's a single user mac install, that might be the reason.
Acou_Bass has quit [Read error: Connection reset by peer]
<sebmacc>
I already used "--update", but, it downloads the channel but the error is still there.
Acou_Bass has joined #nixos
<__monty__>
srhb: Nope, multi-user. Don't have a personal channel nor a channel for root.
<srhb>
__monty__: Um. :)
<sebmacc>
in /nix/var/nix/profiles/per-user/root/channels I have 3 symlinks: binary-caches manifest.nix nixos
Acou_Bass has quit [Client Quit]
<srhb>
sebmacc: It appears to only be looking for nixpkgs, not nixos. I wonder if that changed recently
<srhb>
sebmacc: You could try renaming your channel to nixpkgs, but maybe someone else who knows channels better can step in.
<ottidmes>
__monty__: but is that a problem if you have setup your NIX_PATH with nixpkgs=/path/to/nixpkgs:/path/to/nixpkgs? On NixOS I do have a channel for both root and my user, but only containing manifest.nix with an empty list
<__monty__>
ottidmes: Well, nix-channel --list lists nothing and nix-channel --update finishes instantly, that seems like a problem to me.
<sebmacc>
both channels exist and both channels have been updated pretty recently...
<srhb>
sebmacc: You probably don't want nixpkgs-unstable, it has a lot fewer integrity checks for NixOS itself.
<sebmacc>
okay
<sebmacc>
this is exactly what I was wondering :)
<ottidmes>
__monty__: same for me, but if you manage your own checkout, I dont see the problem
<__monty__>
ottidmes: Well that's not the case.
rawreraw has joined #nixos
lord| has quit [Ping timeout: 268 seconds]
<ottidmes>
__monty__: than its a problem :P
<__monty__>
The installer just didn't bother to set up a channel it seems.
lord| has joined #nixos
<__monty__>
There *is* information about channels in some manifest.nix files though. So I don't really know what the right way to fix it is.
thedavidmeister has joined #nixos
<ottidmes>
__monty__: if doing nix-channel --add yourself works, then you could see if it has effect on those manifest files, if so, it should probably be a matter if adding the channels found in the manifest again, if they are not already reset when you nix-channel --add
<__monty__>
Do I sudo or not? I'm not sure how channels work with a multi-user install. Also, do I use nixpks-unstable or nixos-18.09-darwin?
jperras has joined #nixos
<ottidmes>
__monty__: I am not on MacOS, but from what I have seen about channels, I believe its supposed to be with sudo, and nixos-18.09-darwin is probably as safe bet
crmlt has joined #nixos
diimdeep has quit [Read error: Connection reset by peer]
lostman has quit [Quit: Connection closed for inactivity]
pie__ has quit [Remote host closed the connection]
virus_dave has quit [Client Quit]
<tilpner>
infinisil - Hey, do you have a minute or ten to talk about the structure of "system"? :)
<__monty__>
ottidmes: Hmm, after adding the darwin channel (name nixpkgs) I get: warning: name collision in input Nix expressions, skipping '/Users/toonn/.nix-defexpr/channels_root/nixpkgs'
cyounkins has quit [Remote host closed the connection]
crmlt has quit [Remote host closed the connection]
<tilpner>
locallycompact - I would have guessed glibcLocales, and I wouldn't expect them to be nix-env installable
<locallycompact>
oh, what do I need to do here then?
<locallycompact>
I had tried installing glibLocales
<locallycompact>
it's in that run I think
orivej has joined #nixos
<tilpner>
Let's see what nixos does to install glibcLocales
nkaretnikov has joined #nixos
<tilpner>
Try setting LOCALE_ARCHIVE to the result of ${glibcLocales}/lib/locale/locale-archiv
<Synthetica>
I am refactoring my configuration.nix, and seem to have caught an infinite (`imports`?) loop. Is there a way to go about debugging such a thing without resorting to just placing trace statements everywhere?
<tilpner>
Synthetica - You can try --show-trace and -vvvvv
<{^_^}>
[nixpkgs] @flokli pushed 13 commits to release-18.09: https://git.io/fpr4z
thedavidmeister has quit [Quit: WeeChat 2.3]
<nkaretnikov>
hi! does anyone here use perl with nix-shell? could you show me a sample shell.nix file that'd include a perl library + an upstream program using that perl library such that the program can actually find the library?
<tilpner>
export LOCALE_ARCHIVE=$(nix-build --no-out-link '<nixpkgs>' -A glibcLocales)/lib/locale/locale-archive
<locallycompact>
oh ok let me see thanks
greymalkin has quit [Ping timeout: 250 seconds]
Acou_Bass has quit [Quit: byeeeeeeeeeeeeeee]
Acou_Bass has joined #nixos
eyjhb has joined #nixos
halfbit has joined #nixos
locallycompact has quit [Remote host closed the connection]
<eyjhb>
Currently trying to do a basic NixOS install, just following "2.2.2" with "2.2.3", doing the mount, initial config, specifying grub2 installation place and then doing `nixos-install`, but when I reboot the machine, it seems to see nothing worth booting off.
<eyjhb>
It just gives me the Lenovo F12 boot menu every time, I try to boot off the HDD. I just had Ubuntu on it, which worked... like Ubuntu does, it worked.
<eyjhb>
I really have no clue, what I am doing wrong, expect maybe there is not a partition marked "boot" ?
baconicsynergy has joined #nixos
<etu>
eyjhb: EFI?
<eyjhb>
Currently using legacy boot with secure boot disabled, so I wouldn't think so :/
<symphorien>
as the last lines of nixos-install you should see the output of grub-install, maybe there are explanations here.
<etu>
eyjhb: Ok, so MSDOS or GPT partition table?
kenogo has joined #nixos
<eyjhb>
etu: well, the manual #2.2.2 specifies that a MBR partition table is created using the parted command
<eyjhb>
symphorien: I will try running the install again and see! :)
<etu>
eyjhb: Well then, that should be good at least to not have a mix of things :)
dslegends has joined #nixos
<kenogo>
Hey there, I'm helping with the development of a wayland compositor, and have recently switched to NixOS. I would usually develop in X11, launch the compositor binary in X11, and then use the envvar DISPLAY=wayland-0 to open windows inside that compositor window
<kenogo>
Now, on NixOS, it just says "Can't open display"
m0rphism has quit [Quit: WeeChat 2.2]
m0rphism has joined #nixos
<kenogo>
I can see that a file "wayland-0" is created under /run/users/1000
<eyjhb>
symphorien: nuked ubuntu and ran the commands directly from the manual. So I guess that removed the boot flag..
<ckauhaus>
some R guru around?
demize is now known as SA0DEM
<eyjhb>
gchristensen: yeah... that was .. stupid of me... But I never really worked that much with flags in fdisk, so didn't know it was that easy to see! :/
<ckauhaus>
I'd like to get some feedback on PR #51282
<{^_^}>
[nixpkgs] @expipiplus1 opened pull request #51285 → fped: init at 2017-05-11 → https://git.io/fprVS
kp__ has joined #nixos
illegalprime has quit [Remote host closed the connection]
<ToxicFrog>
Is there an equivalent to builtins.toPath that works on relative paths?
<ToxicFrog>
I can express relative paths in the config but the docs for toPath say it only works on absolute paths.
<avn>
fetchzip able to unpack .tar.gz? Or I need exclusively zip snapshots?
<gchristensen>
zip
dbmikus has joined #nixos
jrolfs has joined #nixos
<ToxicFrog>
Which is not great when I need to dynamically construct a relative path.
<ottidmes>
ToxicFrog: ./. + "${path}"
<ToxicFrog>
That seems to drop the trailing slash -- ${./borg/. + "${name}"} results in an error complaining that it can't find ../services/borgdurandal
<ottidmes>
ToxicFrog: I indeed meant to say: ./. + "/${path}"
<ToxicFrog>
Aah, but with a leading slash on the name, "/${name}", it works
jrolfs has quit [Ping timeout: 244 seconds]
baconicsynergy has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @expipiplus1 opened pull request #51287 → libgpiod: init at 2018-10-07 → https://git.io/fpriI
dbmikus has quit [Ping timeout: 244 seconds]
ThatDocsLady has quit [Ping timeout: 252 seconds]
haslersn has joined #nixos
<haslersn>
Hi, I submitted a nixpkgs issue 10 days ago. Meanwhile the issue doesn't persist anymore. I don't know why and I don't know what changed. Should I close the issue?
m0rphism has quit [Quit: WeeChat 2.2]
arjen-jonathan has quit [Ping timeout: 244 seconds]
m0rphism has joined #nixos
jrolfs has quit [Ping timeout: 268 seconds]
peacememories has joined #nixos
<{^_^}>
[nixpkgs] @dotlambda merged pull request #51275 → nixos/borgbackup: allow paths to be empty → https://git.io/fprq1
<{^_^}>
[nixpkgs] @dotlambda pushed to master « nixos/borgbackup: allow paths to be empty or relative (#51275) »: https://git.io/fpriP
hotfuzz_ is now known as hotfuzz
<ottidmes>
haslersn: considering it being slow could have resulted from many things and it was not some error that occurred, I would think it is safe to close. https://github.com/NixOS/nixpkgs/issues/50854
<Neo-->
haslersn, you could run git bisect to figure out what happened that resolved it if you feel extra curious, but other than that what ottidmes said
dbmikus has joined #nixos
orivej has quit [Ping timeout: 245 seconds]
dbmikus_ has joined #nixos
sigmundv_ has quit [Ping timeout: 250 seconds]
dbmikus has quit [Ping timeout: 245 seconds]
<eyjhb>
Is there any way to speed up `nixos-rebuild switch`, currently building Virtualbox.. And it takes.. Forever.
<{^_^}>
[nixpkgs] @FRidh pushed 105 commits to staging-next: https://git.io/fprPa
__Sander__ has quit [Quit: Konversation terminated!]
Tucky has quit [Quit: WeeChat 2.2]
<gchristensen>
is there a way to execute a program when DPMS activates?
erasmas has joined #nixos
<ToxicFrog>
Oh wow, nixos-unstable is 12 days behind :/
<symphorien>
maybe wrap your screen locker ?
<gchristensen>
symphorien: that is what I want to implement :P
<eyjhb>
Compiling the whole Virtualbox, not just the kernel modules hodapp
<eyjhb>
Think I am on the 2. hour of this.
<eyjhb>
It might be done!
<eyjhb>
No........
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
haslersn has quit [Quit: Page closed]
jperras has quit [Quit: WeeChat 2.2]
<{^_^}>
[nixpkgs] @thoughtpolice merged pull request #51259 → reptyr: enable on 32-bit ARM Linux and x86 FreeBSD → https://git.io/fpwXk
<{^_^}>
[nixpkgs] @thoughtpolice pushed 2 commits to master: https://git.io/fprMQ
Baughn has quit [Quit: ZNC 1.6.2+deb1 - http://znc.in]
jsgrant has quit [Remote host closed the connection]
<samueldr>
I'm not entirely confident, but I believe `virtualbox.enableExtensionPack = true` will cause virtualbox to need a rebuild
<ottidmes>
eyjhb: to see why it is now building virtualbox with your new configuration, this might help: https://gist.github.com/ottidmes/4708d005fa1feb018fbb1fa339cbd071 you can install nix-diff with: haskell.packages.ghc843.nix-diff (or if you are already using another ghc version, you should probably use that one)
m0rphism has quit [Quit: WeeChat 2.2]
m0rphism has joined #nixos
orivej has joined #nixos
<kenogo>
Anyone developing a wayland compositor in NixOS?
<eyjhb>
ottidmes: compltely new install, but I was not sure, wheter nixos builds everything from scratch
zolk3ri has quit [Quit: Lost terminal]
<ottidmes>
eyjhb: if you modify certain core libraries then you can get in that position, or if you are using a very old nixpkgs commit, but neither seem likely with a normal fresh install. Unless of course you are on master / nixpkgs-unstable, or one of the other bleeding edge branches, than you can expect a lot more builds
<avn>
gchristensen, samueldr: thanks, fetchzip with proper url looks more stable
<samueldr>
avn: just realised, you could want fetchFromGitLab
<samueldr>
it's not used exclusively for gitlab.com (the domain parameter defaults to gitlab.com)
<samueldr>
AFAIUI it has the same guarantees as fetchFromGitHub has; the main one being that it hashes the contents of the zip instead of the zip itself
<samueldr>
in github and gitlab it could realistically change; for GitHub it already happened more than once that the generated archives changed
c0bw3b_ has joined #nixos
<avn>
samueldr: anyway is more stable than fetchgit
<avn>
although idk why fetchgit failed with salsa, I use fetchgit (well, modified fetchgitPrivate with deploy key) for 2 years, and haven't any issues
<gchristensen>
fetchgit, indeed, should be stable
asymmetric has quit [Ping timeout: 268 seconds]
<{^_^}>
[nixpkgs] @Mic92 merged pull request #49470 → kernel: enable MLX5_CORE_EN when possible → https://git.io/fxdRi
<{^_^}>
[nixpkgs] @Mic92 merged pull request #51292 → [WIP][backport-18.09]: kernel: enable MLX5_CORE_EN when possible → https://git.io/fpr9v
<{^_^}>
[nixpkgs] @Mic92 pushed 2 commits to release-18.09: https://git.io/fpr5a
<Mic92>
v0|d: yes. but what are you trying to achieve?
<v0|d>
Mic92: so basically im trying to fix libnl
<v0|d>
it has python support default on
<v0|d>
but it requires build=host=x86 target=arm python
<v0|d>
and swig.
<Mic92>
v0|d: interesting, what does it uses python for?
<v0|d>
swig spits some c, that compiled w/ gcc cross but it fails since swig spits a file which employs build headers.
<v0|d>
if I input crossSwig than swig executable cannot be exec'd for obvious reasons.
<v0|d>
I find the abstraction in nixpkgs still insufficient for canadian cross.
<yl[m]>
I'm having this weird issue with this configuration https://gist.github.com/kalbasit/2e4d35043ae8c5924a56970294af395a : `The option `mine.ssh.privateSSHPath' is used but not defined.`. The reason I find it weird is because it's used only if `mine.ssh.enabled` is true but it's not in my example
<v0|d>
maybe I'm wrong and flag if somebody shows me exactly that.
arjen-jonathan has quit [Ping timeout: 268 seconds]
<Mic92>
v0|d: I am not so sure why it would want to build arm packages while compiling buildPackages.
<Mic92>
intuivly I would say this should not be the case
<Mic92>
*intuitively
<v0|d>
no idea:( btw, did nspr and other closd pr go into staging-next?
dbmikus_ has quit [Ping timeout: 246 seconds]
<v0|d>
saw some go into next today, I feel there is still work to be done.
dbmikus_ has joined #nixos
pie__ has joined #nixos
jsgrant_ has joined #nixos
pie___ has quit [Remote host closed the connection]
Lears has quit [Read error: No route to host]
<yl[m]>
weird. Changing line 29 from `programs.ssh = import cfg.privateSSHPath;` to `programs.ssh = if cfg.enable then import cfg.privateSSHPath else {};` worked!
Lears has joined #nixos
freeman42x]NixOS has joined #nixos
<erasmas>
does nix support drop config files? like adding multiple conf files to /etc/nix/conf.d/
<v0|d>
Mic92: fwiw pkgsCross.armv7l-hf-multiplatform.buildPackages.targetPackages.python this gives me a target=host=arm build=x86 python.
<yl[m]>
fair enough, looks like they release consistently
<yl[m]>
catern: can you address my other comment and I'll test it out and merge!
Baughn has quit [Quit: ZNC 1.6.2+deb1 - http://znc.in]
pkx3__ has quit [Read error: Connection reset by peer]
peterzky has joined #nixos
Baughn has joined #nixos
alex`` has quit [Ping timeout: 250 seconds]
bitmapper has quit [Remote host closed the connection]
peterzky has left #nixos ["ERC (IRC client for Emacs 26.1)"]
alex`` has joined #nixos
graphene has quit [Remote host closed the connection]
graphene has joined #nixos
eyjhb has joined #nixos
<eyjhb>
Hey I just enabled i3 using the `service.xserver.windowManager.i3.enable`, but it doesn't seem like it starts on boot, and I need to type `i3`?
<ottidmes>
eyjhb: did you disable xterm?
<lassulus>
eyjhb: have you selected i3 on the top right on the login-screen?
<eyjhb>
ottidmes: I have not.... Might be the problem, 2 secs
<eyjhb>
lassulus: I think it is using it, but I will check xterm first
<eyjhb>
ottidmes: Do I need any desktopmanager at all?
<ottidmes>
eyjhb: nope, I dont use any
<eyjhb>
I need to read up on what it actually handles normally..
<eyjhb>
Rebooting in a sec to see
Baughn has quit [Quit: ZNC 1.6.2+deb1 - http://znc.in]
<ottidmes>
eyjhb: but it might be easier to use one to start out with, because I have to set some settings to make all programs behave like you would expect
<eyjhb>
Hmm.... But doesn't it just use i3?
<eyjhb>
Works like a charm now ottidmes ! Now I just have to try to set it up, give up on my life, hit my head aginst the desk, and repeat
<ottidmes>
eyjhb: I am talking about things like file managers, if you just use some random file manager from a desktop manager while not using the desktop manager itself (e.g. nautilus), certain things of it might not work out of the box, because of things configured by the desktop manager that are now missing
<eyjhb>
Ahh... But stuff like thunderbird, ff, chrome will work fine?
<ottidmes>
eyjhb: they should, yes
<eyjhb>
But if I wanted e.g. Nautilus, then enabled gnome3 would make sense?
<ottidmes>
eyjhb: for basic usage it should work out of the box too
<ottidmes>
eyjhb: so no need
<eyjhb>
Hmmm, fair ! Will try and see how things goes... Got any good tips and tricks to get started using i3(-gaps)?
<ottidmes>
eyjhb: I will make a config snippet of things you might want to include in your NixOS config
<eyjhb>
ottidmes: you will be my hero, if you do that :D
Baughn has joined #nixos
Baughn has quit [Client Quit]
<ottidmes>
eyjhb: I have used i3 for a time, but am now using bspwm, when you start out I can recommend just using the full screen tabbed view to work (and the basic commands i3 / keyboard shortcuts), then at least you can get productive with it
Baughn has joined #nixos
betaboon has quit [Quit: WeeChat 2.2]
<eyjhb>
ottidmes: makes sense! Why bspwm over i3?
betaboon has joined #nixos
<eyjhb>
I also see that I need something to manage multiple displays.. The joys
<ottidmes>
eyjhb: because I love having a programmable window manager, and although its not like that is not possible in i3, they have very similar capabilities, but in bspwm its how almost everything is approached, so it matches that way of using a window manager better.
<eyjhb>
Makes sense. I have a colega who uses it too! - Do you manage multiple displays using arandr?
<eyjhb>
And thanks for the snippet!
<ottidmes>
eyjhb: I used it, but I recently made a script that I now use instead of arandr
<eyjhb>
I know I am going to get screwed so bad, when I need to connect my LP to a projector at the university...
<eyjhb>
Makes sense to do that instead ! ;)
<ottidmes>
eyjhb: but I keep arandr because its still useful to test things and set things without having to go to the command line
ckauhaus has quit [Quit: WeeChat 2.2]
<eyjhb>
Makes sense! Just started using NixOS today, and going in the deep right to begin..
slyfox_ is now known as slyfox
<{^_^}>
[nixpkgs] @primeos pushed to master « monkeysphere: Update the comment why the tests don't run »: https://git.io/fprp5
<betaboon>
eyjhb: have fun then. you're up for quite a pleasing experience :)
<eyjhb>
betaboon: yeah... A lot of days being not that productive I would guess
<betaboon>
i guess that depends on your definition of productive :D
<eyjhb>
I think my group at uni will define it as... Not being very productive... :p
<eyjhb>
Is there any good way, to include a certificate to connect to a wifi in a nix file or something?
<eyjhb>
c0bw3b_: ANy good way of hiding passwords etc. in *.nix files?
<c0bw3b_>
and I guess you should have a `ca_cert="/path/to/eduroam/ca.pem"`
<c0bw3b_>
humm no that's still an open issue
<c0bw3b_>
(dealing with secrets in the Nix store I meant)
<worldofpeace>
jtojnar++
<{^_^}>
jtojnar's karma got increased to 5
<eyjhb>
So basically, the only thing I can do, if I want to have `secrets` in my configurations, are to insert a placeholder, and replace them on clone?
<worldofpeace>
gchristensen: jtojnar karma should be way higher :D
<ottidmes>
eyjhb: there are a few patterns people use, often a secrets nix file is used, and then they use: let secrets = import ./secrets.nix; in ... on top of their configuration
<c0bw3b_>
eyjhb: or read the secret from a flat file (chmod 0400 that file so only root/you can read it)
<gchristensen>
worldofpeace: much, yesz
<c0bw3b_>
yep that's the same idea
<c0bw3b_>
jtojnar++
<{^_^}>
jtojnar's karma got increased to 6
<eyjhb>
Hmm... Does the `secret.nix` define variables availale to the config that imports it? Or are the configuration just located in that?
steshaw has joined #nixos
<c0bw3b_>
it just defines variables
eeva has joined #nixos
<eyjhb>
Eh.. Might be a stupid question, but how might that look?
<clever>
eyjhb: if you impoad load-secrets.nix, and secrets.nix is missing, it returns an empty set of dummy secrets, which allow others to use the config (with the secrets missing) (and hydra too)
<ottidmes>
eyjhb: you can got nuts with what you put in your secrets.nix file, could contain any valid nix, but most people will indeed have just a attrset with some name/value pairs set
<ottidmes>
offtopic and I have seen this problem before (cannot remember the solution), what if you get 'gpg: decryption failed: No secret key' while `gpg --list-secret-keys` returns the proper secret key? (I really should finish porting my passwords over)
<selfsymmetric-mu>
Isn't there normally a bot that helps out with ,locate commands or something? I'm not sure.
<gchristensen>
c0bw3b_: you here?
<c0bw3b_>
gchristensen: yes
<asymmetric>
selfsymmetric-mu: didn't you get a response from {^_^} ?
<selfsymmetric-mu>
asymmetric: Whoops, got hidden on my client. I see it now. Unfortunately this seems like a _different_ package that provides something called mmdc.
<c0bw3b_>
it looks nice, but what am I looking at? :p
<gchristensen>
ofborg's build results
<selfsymmetric-mu>
Anyone have experience configuring binaries that are normally provided by npm or yarn?
<selfsymmetric-mu>
What's the correct NixOS way to install them?
<gchristensen>
should ofborg leave comments saying its build output moved to CheckS?
mayhewluke has quit [Ping timeout: 250 seconds]
<makefu>
asymmetric: okay i will have a look. seems like building the package literally downloads the whole internet and builds every rust package ever written (nix-build -A dapp)
<betaboon>
finally i have a working i3wm-patch that hides borders and titlebar when a window is alone in a workspace :D
<tdeo>
cannot open .so lib for builtin L"nt0_dll.dll": /nix/store/n4c7f4a9n4lj1kvnryzj49inkwcxi82h-wine-wow-3.14-staging/bin/../lib/wine/nt0_dll.dll.so: cannot open shared object file: No such file or directory
<tdeo>
anything i can do about this?
<ottidmes>
I have a laptop with NixOS installed running NetworkManager for wireless, it works, but I cannot get manual ip addresses to work, so no static ip address, I do have a local server running, so I could maybe setup a DHCP server on it to work around that (my router is too basic), or would you just suggest using something other than NetworkManager?
<samueldr>
what's the reason for the static address?
<samueldr>
I don't manage my addresses, and use avahi (aka zeroconf, bonjour) to find the machines
<ottidmes>
samueldr: good point, it does not have to be, I just want to be able to SSH into it from the desktop, and not having to change my IP settings everytime I do something to cause it to change on my laptop
<samueldr>
(but that's not a good answer to your question :))
jtojnar has quit [Quit: jtojnar]
Rusty1 has quit [Quit: Konversation terminated!]
mayhewluke has quit [Ping timeout: 268 seconds]
<ottidmes>
samueldr: guess I will experiment with avahi first, I see that I tried before, I should have commented why I commented the config...
mayhewluke has joined #nixos
lord| has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/fpomu
mayhewluke has quit [Ping timeout: 250 seconds]
mayhewluke has joined #nixos
<selfsymmetric-mu>
Boy, I'm having a rough time installing mermaid.
<selfsymmetric-mu>
I tried `node2nix`, but I can't seem to get myself a `nix-shell` that has both `mermaid.cli` and a `node.js` to run it with.
<selfsymmetric-mu>
`node2nix` made a `default.nix` but to my surprise `nix-shell` doesn't like it. `error: nix-shell requires a single derivation`. Am I using it wrong?
<makefu>
asymmetric: okay as i would have guessed, you only need to add $out/bin to your path before testing
<selfsymmetric-mu>
ottidmes: Huh! What did this do?
<selfsymmetric-mu>
I'm in a shell now. No mermaid though.
<ottidmes>
selfsymmetric-mu: it wants a derivation and you are giving it a package description (the thing callPackage uses to convert it to an drv)
<selfsymmetric-mu>
Hmm.
<selfsymmetric-mu>
Okay, I think I see.
Enzime has joined #nixos
sir_guy_carleton has joined #nixos
<ottidmes>
selfsymmetric-mu: if you check nix-shell manpage, you will see that -p is just a buildEnv, so you can do that here to, modify it to have like pkgs.buildEnv (not sure about arguments) with your other dependencies and the (callPackage ./. {})
<selfsymmetric-mu>
Ohhh. Thank you!
<ottidmes>
selfsymmetric-mu: might be easier to just add your default.nix to an overlay and run whatever name you give it together with the other deps with nix-shell -p, but essentially the nix expression would do the same, but on the fly
<makefu>
asymmetric: in your place i'd call bats in the checkPhase instead of relying on make to do the right thing(tm). like this: http://paste.krebsco.de/TVYEwTzD/+inline . or maybe there is some magic makeFlags or something which helps you adding $out/bin/ to the path
Rusty1 has joined #nixos
<clever>
ottidmes: -p isnt a buildEnv, but a stdenv.mkDerivation
<clever>
if you nix-shell foo.nix -A bar --exclude baz
<clever>
then every input matching the name "baz" will just not be built/downloaded
<ottidmes>
clever: how is it close, I dont see the difference, except for the parentheses, and I literally copied it from the manpage?
<clever>
ottidmes: the parens
<clever>
it allows things like, nix-shell -p 'foo.override { key = "value"; }' bar
<Baughn>
So, sudo doesn't work with systemd.
<Baughn>
We should probably remove it.
<clever>
and the parens keep the list sane
<ottidmes>
clever: ah makes sense
<clever>
also, its no longer pkgs.runCommand
<clever>
(pkgs.runCommandCC or pkgs.runCommand)
<clever>
runCommand is now an alias to runCommandNoCC (it lacks a gcc), but people expect nix-shell to have one, and runCommandCC may not exist (new nix, old nixpkgs)
<ottidmes>
selfsymmetric-mu: I just experimented, and this works too: nix-shell -p putty $(nix-build-call) (where nix-build-call is a shell script doing what I initially showed for nix-shell, but then using nix-build), so then you dont have to tweak the expression with a buildEnv/runCommand
justanotheruser has joined #nixos
<selfsymmetric-mu>
ottidmes: This might be over my head. ^_^
<selfsymmetric-mu>
Right now I'm getting `error: attribute 'nodejs-4_x' missing`.
<clever>
ottidmes: that will probably mostly work, but confuse cross-compiling
<selfsymmetric-mu>
Sandboxing python environments with nix wasn't too bad an experience.
<selfsymmetric-mu>
But node is proving tough.
<ottidmes>
clever: could you eloborate?
<clever>
selfsymmetric-mu: that sounds like a simple case of nodejs 4 having been removed
<selfsymmetric-mu>
Ah, hm.
<clever>
ottidmes: the crosscompile stuff expects the buildInputs to be attrsets, so it can get the native or cross version
<clever>
ottidmes: but your above cmd just gives it a raw path to a finished product
justan0theruser has quit [Ping timeout: 240 seconds]
<selfsymmetric-mu>
I'm changed `4` to `6` and now I have a shell.
<ottidmes>
clever: makes sense, but would that ever be relevant in this use case? cross compiling that is
<selfsymmetric-mu>
I have a shell, but no `mmdc`. Hmm.
<clever>
ottidmes: yeah, you usually arent cross-compiling with raw nix-shell -p
<selfsymmetric-mu>
lol I just want to run mermaid haha
<selfsymmetric-mu>
So much yak shaving.
<ottidmes>
selfsymmetric-mu: if you want help you have to give more, what is mmdc, maybe some dependency missing or should it been installed by your default.nix?
<clever>
selfsymmetric-mu: is it native or raw js?
<selfsymmetric-mu>
Speaking properly, mermaid is a diagram format. `mermaid.cli` is what I'm trying to install. The repo page is here: https://github.com/mermaidjs/mermaid.cli
<selfsymmetric-mu>
It appears to be an executable that is installed as a node module.
<clever>
selfsymmetric-mu: what happens if you just run `npm install .` in that dir?
<selfsymmetric-mu>
Inside the nix shell? I'll try that.
<selfsymmetric-mu>
It's installing something...
<selfsymmetric-mu>
The executable's name to be installed is ostensibly `mmdc`. For example if installed with `yarn` on a conventional OS it appears in `node_modules/.bin/mmdc`.
<clever>
npm would also drop it in that dir
<selfsymmetric-mu>
`npm install .` appears to have installed `mermaid 8.0` but I cannot see and mmdc...
<clever>
ls -lh node_modules/.bin/
<selfsymmetric-mu>
Oh sweet!
<clever>
2018-11-30 19:40:50 < clever> selfsymmetric-mu: is it native or raw js?
<selfsymmetric-mu>
I believe in the end it is raw js.
<selfsymmetric-mu>
That's what is looks like from experiments, anyway...
<selfsymmetric-mu>
So I guess I can just wrap and run it.
<ottidmes>
clever: you had your own DHCP server, right? did you use dnsmasq?
<tilpner>
Ahh, they were updated together previously
<tilpner>
I'll just do that too then
<clever>
ottidmes: plain old bind and dhcpd
c0bw3b_ has quit [Remote host closed the connection]
<ottidmes>
clever: and thats easy to setup? my current configs expect static IPs and I could modify them to use hostnames instead, but I rather use DHCP than installing avahi or similar on every client to get the hostname solution working