<scoates>
I've been using nix on my mac for a bit, to replace homebrew, and I'll admit I still don't quite grok it. (-: But I've had a little help here and there and that has been enough. Tonight I'm pushing my limits a bit.
<scoates>
how far down the rathole will I need to go? (hoping it's something simple that I just don't quite understand yet)
<v0|d>
scoates: try python3.withPackages instead of python.withPackages
<scoates>
that gives `error: attribute 'python3' missing, at /Users/sean/.config/nixpkgs/overlays/personal.nix:33:12` ; where line 33 is `(availablePlugins.python3.withPackages (pythonPackages: [`
<samueldr>
see, `availablePlugins` comes from the function parameters, { availablePlugins, ... }
<samueldr>
and the python attribute is set on it in the weechat/default.nix file
<samueldr>
that python attribute's withPackage attribute uses python from pythonPackages
<samueldr>
it's not the most obvious file :)
<scoates>
with the above, it tells me `undefined variable 'python3Packages'` ; and I'll be honest: I'm not following. I think I probably need to do a bunch more reading before it's useful for me to keep asking questions.
<scoates>
was hoping there was an easy trick (-:
<samueldr>
drop the three everywhere
<samueldr>
oh
<samueldr>
duh
<samueldr>
no
<fusion809>
Is there a reason why there's an option to autologin using SDDM, but there's no option for specifying the desktop to autologin to? Like the /etc/sddm.conf file does include the option to specify the desktop autologin to.
<samueldr>
pythonPackages = python3Packages; <- this needs to come from super, since it's an overlay
<fusion809>
So why doesn't /etc/nixos/configuration.nix have such an option
<samueldr>
you don't have `python3Packages` in scope :)
Diagon has joined #nixos
<mojjoo>
trying to build a custom nixos iso, follwed the instructions on https://nixos.org/nixos/manual/index.html#sec-building-cd ... I'm getting this: `error: attribute 'isoImage' in selection path 'config.system.build.isoImage' not found`
<scoates>
brb (the pain of using the package you're building (-: )
<scoates>
build failed. but looks like it's close. I'll poke at it a bit when I return (I was just summoned for dinner stuff).
<scoates>
thanks very much!
jrolfs has quit [Ping timeout: 250 seconds]
ng0 has quit [Quit: Alexa, when is the end of world?]
pkx3__ has joined #nixos
Diagon has quit [Quit: Leaving]
<yl[m]>
anyone figured out a way to avoid getting killed on Travis during nix-build?
<gchristensen>
scoates: you can build it without exiting the program, of course
pkx3 has quit [Read error: Connection reset by peer]
Hotkeys has joined #nixos
mojjoo has quit [Ping timeout: 240 seconds]
<scoates>
Yeah; I expected it to be quicker. (-;
<{^_^}>
[nixpkgs] @thoughtpolice merged pull request #51306 → nixos/cockroachdb: create new service → https://git.io/fponX
<{^_^}>
[nixpkgs] @thoughtpolice pushed commit from @jbboehr to master « nixos/cockroachdb: create new service »: https://git.io/fpojR
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fpKeo
hellrazor has quit [Ping timeout: 245 seconds]
Ariakenom has quit [Quit: Leaving]
freeman42x]NixOS has quit [Quit: Leaving]
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
jperras has joined #nixos
fusion809 has quit [Remote host closed the connection]
fragamus has joined #nixos
fragamus has quit [Read error: Connection reset by peer]
jperras has quit [Ping timeout: 246 seconds]
jrolfs has quit [Ping timeout: 250 seconds]
init_6 has joined #nixos
rtjure has joined #nixos
jperras has joined #nixos
alex`` has quit [Ping timeout: 245 seconds]
betaboon has quit [Quit: WeeChat 2.2]
mmlb has quit [Read error: Connection reset by peer]
s2mitrov has joined #nixos
mmlb has joined #nixos
virus_dave has joined #nixos
jrolfs has joined #nixos
<s2mitrov>
Anyone have experience installing nixos in a laptop with a pre-existing windows install? I've shrunk the windows partition (120GB free, unallocated) and created a boot US and all that, but like 10 seconds into the install process i get a kernel panic
virus_dave has quit [Client Quit]
<s2mitrov>
the final error message says 'sustemd: error while loading shared libraries: /nix/store/.../libcap.so.2: cannot read file data: Input/output error', but before it there's a bunch of 'SQUASHFS error: Unable to read page, block 121f12a4, size 40f30'
acarrico has joined #nixos
<ottidmes>
s2mitrov: should that not be unrelated to having Windows? The origin of the partition should not matter
<ottidmes>
seems more like a problem with your install medium than your target partition
<clever>
s2mitrov: Input/output error sounds more like a corrupt FS, did you umount properly when the install was done?
<samueldr>
squashfs makes me think it's the installer media
<s2mitrov>
the install never gets to any point where I can interact with it at all
jrolfs has quit [Ping timeout: 245 seconds]
<clever>
s2mitrov: is it failing to boot the installer, or did you finish the install, and then it fails booting the hdd?
<samueldr>
s2mitrov: did you eject or sync after `dd` (or powered-off sanely)
<s2mitrov>
it runs the first phase fine, and then pretty much immediately after phase 2 it spits out a couple pages of that squashfs stuff followed by the final error
<ottidmes>
s2mitrov: you mean at the installer, right?
<s2mitrov>
failing to boot the installer, i guess? I get to the screen where I have the option of like 'Install nixos 18.09, Install nixos 18.09 (nomodeset), ...' and select the plain old 'install nixos 18.09'
<ottidmes>
could you try and remake the install media, are you using an USB stick?
<s2mitrov>
yeah
<ottidmes>
s2mitrov: maybe check the checksum of your downloaded iso too, just in case that somehow got corrupted
<ottidmes>
I would recreate your USB stick with: sudo dd bs=4M oflag=sync status=progress if=/path/to/nixos.iso of=/dev/of/usb
<s2mitrov>
i don't have access to a linux machine at the moment
<s2mitrov>
just windows unfortunately :/
<s2mitrov>
but i did just mess around with the bios settings
orivej has quit [Ping timeout: 250 seconds]
<s2mitrov>
it was set to boot both uefi + legacy, wth legacy first so i changed that to just uefi
<samueldr>
the way it fails definitely smells like either a bad "burn" or a bad usb stick
<s2mitrov>
still kernel panic but the error is different now
<samueldr>
how different?
<ottidmes>
yeah, so I would just try one of the many windows burn iso to usb tools that other distros point to like Ubuntu
<s2mitrov>
sched: Unexpected reschedule of offline CPU#2! lol
<ottidmes>
s2mitrov: how new is your setup?
<ottidmes>
your computer
<s2mitrov>
it's an x1 carbon from like last year or something
<samueldr>
ottidmes: that may be problematic as unetbootin is no good with nixos :/
jrolfs has joined #nixos
<ottidmes>
samueldr: ah, I always remember trying a bunch until one worked, never taking that route now that I am way more familiar with Linux
<s2mitrov>
on an unrelated note, 0/10 would not recommend this laptop to anyone
<samueldr>
(but unetbootin wouldn't look like the issues here)
<samueldr>
(and the fault is partly ours, our installer iso assumes full control of the device)
<ottidmes>
but with a year old laptop, kernel 4.14 should be fine I guess
<samueldr>
I have no idea what those means; if they try to operate at the filesystem level it may break some assumptions from nixos
<ottidmes>
s2mitrov: its probably possible from Windows, but a stupid but effective way would be, take another USB stick, install Ubuntu or whatever distro that does work with those burn tools on Windows, and burn NixOS on the other USB stick from there, but if others have anything less roundabout, please use that instead
<samueldr>
though it does detect that it's a UEFI bootable image, which probably is fine
<samueldr>
I believe the most popular recent dd-like tool for windows is etcher
<scoates>
hmm. and will probably run into other scripts. I might be inviting too much pain, here.
<s2mitrov>
i ran badblocks on the usb stick, says there's one corrupted block
<s2mitrov>
:"^)
<samueldr>
:( physical makes digital hard
<s2mitrov>
i think i might have another kickin around somewhere, or if not ill just run down to walmart or whatever and pick one up the things cost like nothin these days
graphene has quit [Remote host closed the connection]
graphene has joined #nixos
selfsymmetric-mu has joined #nixos
<selfsymmetric-mu>
,locate cask
<{^_^}>
Found in packages: oh-my-zsh
Supersonic has quit [Ping timeout: 252 seconds]
<selfsymmetric-mu>
Anybody here use Cask for Emacs package development? Because the `cask` in `nixpkgs` appears to install `cask-package-toolset`, not `cask` itself.
jperras has quit [Ping timeout: 246 seconds]
<selfsymmetric-mu>
So I get `cask badge`, for example, but not `cask init`.
<s2mitrov>
god damn does it feel good to be off windows
<s2mitrov>
what an absolute dumpster fire of an operating system
<Ashy>
haha
Izorkin has quit [Ping timeout: 244 seconds]
<betaboon>
s2mitrov: after all dumpster fires have the possibility of keeping some people warm who need it the most in winter. i doubt windows can provide that ...
endformationage has joined #nixos
graphene has quit [Remote host closed the connection]
graphene has joined #nixos
Growpotkin has joined #nixos
<Growpotkin>
hey does anybody know why weechat's services are missing from the NixOS Options in the newer versions?
<Growpotkin>
They're listed on nixos.org (no idea which version the search page is querying from) but they're not available on my machines
<Ralith>
nixos.org documentation is from nixos-unstable, iirc
<Ralith>
so, it is probably more recent than your machine
<samueldr>
nixos.org only lists thinkgs for current-stable
<Ralith>
oh, huh
<Ralith>
that makes more sense
<Growpotkin>
oh I hadn't imagined nixos-unstable would be missing things from the stable branch. I'll add it back to my channel list.
<Growpotkin>
thanks
<samueldr>
growpotkin: let me check something
<samueldr>
I think the options are there, but differently built
<Growpotkin>
bizarre. I just updated and its still missing.
<samueldr>
yeah, I wasn't entirely sure, but the change is in both 18.09 and unstable
<samueldr>
which option exacly?
<Growpotkin>
i just want to enable the daemon
jrolfs has quit [Ping timeout: 246 seconds]
<samueldr>
what does `nixos-version` say?
<Growpotkin>
okay I gotta refresh my memory on how to add channels properly. Should I be doing that in my configuration.nix? Right now I just point my nixPath to "nixpkgs=/nix/var/.../per-user/root/channels/nixos/nixpkgs"
<Growpotkin>
I've got 18.03 impala. Which makes no sense because my configuration.nix says "19.03"
<Growpotkin>
in my conf i have : system.statVersion="19.03";
<samueldr>
did you touch stateVersion?
<samueldr>
,stateVersion growpotkin
<{^_^}>
growpotkin: Setting stateVersion to the latest release doesn't update anything and can only break your setup at best. If you want to regardless, Ctrl-F for "stateVersion" in https://nixos.org/nixos/manual/release-notes.html to see things that need to be manually upgraded with the new value
<samueldr>
yeah, it's a confusing setting
<Growpotkin>
thanks I'll give that a shot
<samueldr>
it smells and looks like a duck^W update toggle, but isn't
<Growpotkin>
all this time I thought changing that "updated" my environment lol
<samueldr>
nah :/ only changes some conditionals for stateful things like databases
acarrico has quit [Ping timeout: 246 seconds]
<Growpotkin>
alright I'm gonna updrade this bad boy. Cross your fingers that It doesn't brick my machine lol
<samueldr>
chances are quite low
cyounkins has quit [Remote host closed the connection]
<Growpotkin>
wow. I hope this magically fixes the variety of broken haskell junk thats been forcing me to use ghc-822 :)
metastance has joined #nixos
sigmundv__ has quit [Read error: Connection reset by peer]
<mojjoo>
trying to build a custom nixos iso, follwed the instructions on https://nixos.org/nixos/manual/index.html#sec-building-cd ... I'm getting this: `error: attribute 'isoImage' in selection path 'config.system.build.isoImage' not found`
Rusty1 has quit [Quit: Konversation terminated!]
mmlb has quit [Read error: Connection reset by peer]
<scott>
if a package installs some stuff into an alternate bin/ directory, what's the right way to get those binaries added to PATH when added to an environment? just symlink them all into $out/bin? at what stage?
jrolfs has quit [Ping timeout: 246 seconds]
hellrazor has joined #nixos
<scott>
i tried making symlinks in postInstall but they didn't show up... i'm not sure what happened
<scott>
the default installPhase does fine in putting everything in $out, but some of the binaries go in $out/bin and some go in $out/share/perl6/bin (this is rakudo star, the perl6 compiler)
<scott>
the official instructions don't discuss typical global installation, but suggest adding both those paths to PATH in .bashrc for a local install
<gerschtli>
can you show me the nix expression?
<gerschtli>
maybe you can add instructions to copy $out/share/perl6/bin/* to $out/bin to the installPhase
<ptotter[m]>
eyjhb: typical advantage: only having one thing to manage
<ptotter[m]>
but OTOH people usually have something in their vimrc
<ptotter[m]>
I'm planning to move from debian to nixos, a couple of questions, if I may:
<ptotter[m]>
I run many services using docker
<eyjhb>
ptotter[m]: precisely. I still have my vimrc, and I having it NixOS specific, doesn't seem that nice :/ So I think I will just add ycm, so I get all the dependencies etc.
mojjoo has quit [Quit: Page closed]
<ptotter[m]>
since it's a convenient way to separate code from data and I get updates from the service authors as docker images
<ptotter[m]>
is there a recommended place to store data? I just made a /srv mountpoint and mount volumes to the container, since that's what I did with debian
<scott>
gerschtli: i have a new problem though, which is that $out/share/perl6/bin is simply empty... so i need to undo a couple changes and see when that happened
<ptotter[m]>
and I also like to manage the services using systemd
<ptotter[m]>
can I just put units into /etc/systemd/system or is there some preferred way to handle extra systemd units in nixos?
<eyjhb>
ptotter[m]: I would guess making a custom .nix file for each service, or something :)
<gerschtli>
scott: im building the package, maybe i come up with an idea
<ptotter[m]>
eyjhb: sounds like another layer of indirection
<ptotter[m]>
I'm actually considering just doing away with the systemd layer and have just docker manage the containers
<ptotter[m]>
but then systemctl --failed and stuff like that don't give a unified view
<scott>
gerschtli: oh dang, i had a typo, the path needs to be $out/share/perl6/site/bin, not $out/share/perl/site/bin
jtojnar has joined #nixos
mmlb has quit [Read error: Connection reset by peer]
<gerschtli>
scott: oh :D okay, so its working now?
<scott>
hopefully, i'm running the build right now
mmlb has joined #nixos
thc202 has joined #nixos
metastance has quit [Quit: WeeChat 1.4]
<ptotter[m]>
also, I guess there's no point in backing up anything under /nixos?
<gerschtli>
ptotter[m]: systemd units and mounts can all be configured via config options in /etc/nixos/configuration.nix
<gerschtli>
scott: yep, fixing the path works like a charm
<scott>
yeah my build just finished
<scott>
i guess it's just one of those days where i have to spend 30 minutes figuring out i made a typo =)
<ptotter[m]>
gerschtli: mounts were handled by nixos-generate-config
<gerschtli>
you need to be glad that it only took 30 minutes.. :D
<ptotter[m]>
gerschtli: do you have any pointers for the systemd configs?
iyzsong-x has joined #nixos
<gerschtli>
ptotter[m]: static mounts like what you would write in /etc/fstab, i meant systemd mount units
<ptotter[m]>
ah
<gerschtli>
services can be configured via the systemd.services.* options
<bigvalen>
I've an odd problem, where my nixos box boots, but systemd can't start a bunch of stuff. I can't even get in on serial console, it's hanging that early in the boot process.
<bigvalen>
If I start in single user mode, 'systemctl default' boots the box normally.
<Mic92>
does your bootloader also has serial support?
teehemka_ is now known as teehemkay[away]
teehemkay[away] has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
init_6 has quit []
cyphase has quit [Quit: cyphase.com]
<{^_^}>
[nixpkgs] @Gerschtli opened pull request #51379 → nixos/nm-applet: add nm-applet program → https://git.io/fpKYE
<{^_^}>
[nixpkgs] @Mic92 pushed to release-18.09 « networkmanager-applet: set default icon theme »: https://git.io/fpKOB
alex`` has joined #nixos
__monty__ has joined #nixos
<__monty__>
The mac os installer left me without channels. I added nixos-18.09-darwin as `nixpkgs`. Now I get this warning: warning: name collision in input Nix expressions, skipping '/Users/toonn/.nix-defexpr/channels_root/nixpkgs'
ris has joined #nixos
hotfuzz has joined #nixos
<__monty__>
Is there a way to find out what channel I should be using? Or, preferably, a way to get nixos-18.09-darwin to work?
<__monty__>
Am I even doing it right? I ran sudo nix-channel --add on a multi-user install.
mmlb has quit [Read error: Connection reset by peer]
<clever>
__monty__: sudo sets the wrong $HOME on mac
<clever>
__monty__: you must use `sudo -i` then `nix-channel --list`
Ariakenom has joined #nixos
<ris>
thoughtpolice: is afl-qemu working for you as packaged in 18.09? afl 2.52b's intended qemu version is 2.10.0, and the package uses 2.3.0 - the result is that afl passes it the QEMU_LOG=nochain env var which qemu 2.3.0 doesn't yet understand, causing it to fail
<__monty__>
clever: Hmm, I first did sudo nix-channel --remove, then sudo -i, nix-channel --add then --update. Still get the same warning though : /
<__monty__>
You were right though, sudo --add just created a .nix-channels in my user's $HOME.
<clever>
__monty__: you must also run --update after --remove
orivej has joined #nixos
jrolfs has joined #nixos
Radivarig has joined #nixos
<__monty__>
clever: Uhm, but I did after the new --add. Do I need to --remove again, then --update, --add, --update?
<clever>
each user has its own set of channels, and --update is needed on the right user, after any --add or --remove, to re-sync things
<clever>
and its best to just have a single channel on root most of the time, or root wont be able to use nix-env
<__monty__>
clever: I've run --update both for root and my user but I still get the warning : /
vidbina has joined #nixos
<__monty__>
Is the stuff in ~/.nix-defexpr/channels_root/nixpkgs somehow a channel added by the macos installer but not under the control of nix-channel?
<clever>
__monty__: channels_root is roots channels, its managed by nix-channel ran as root
<clever>
but there is also a bug, if no channels are setup, --update wont do anything, so there is no way to remove the last channel
<__monty__>
clever: Where do I find the paths I should remove manually?
<clever>
__monty__: the target of _channels or channels_root, will be another symlink called channels, deleting that and its own target, channels-1-link should remove it entirely
<__monty__>
clever: Do I need to remove .nix-defexpr/channels_root as well? I get another warning now: warning: Nix search path entry '/nix/var/nix/profiles/per-user/root/channels' does not exist, ignoring
<{^_^}>
[nixpkgs] @Mic92 opened pull request #51382 → Networkmanager: rename service to match upstream → https://git.io/fpK3Y
<clever>
__monty__: thats because the default NIX_PATH expects a single channel on root, which you have now deleted
<clever>
you either need to fix NIX_PATH, or make root the only user with a channel, as it was after the install finished
<__monty__>
clever: But nix-channel --list didn't list anything, not as root, not as my user.
<__monty__>
sudo -i; nix-channel --list now lists the channel I added. So root should have exactly *one* channel now, right?
<clever>
yeah
<__monty__>
I'm lost, I no longer understand the problem.
<clever>
the non-root user should have zero channels
<clever>
so you will want to get rid of the link under .nix-defexpr/channels, not channels_root
jrolfs has quit [Ping timeout: 240 seconds]
revtintin has joined #nixos
<__monty__>
clever: Gotcha.
<__monty__>
Now, when I want to update my channel, I have to sudo -i first every time?
<clever>
yeah
<clever>
or fix the silly defaults in /etc/sudoers
<ottidmes>
clever: any idea what might go wrong with this config, I do get IPs from the range, I do see devices talking to the it and getting responses, and the static IPs work great, its just that dynamicly assigned IPs from the range don't work (e.g. smartphone): https://gist.github.com/ottidmes/5e27131af28464dfc815d86b22658e74 I probably am going to try and explicitly assigning some domain-name-servers other than the
<ottidmes>
router itself
<clever>
on nearly every os sudo exists on, it sets the right $HOME when you `sudo foo`
<__monty__>
Also, should I even be on nixos-18.09-darwin, or should I just stick to nixpkgs-unstable?
<{^_^}>
clever's karma got increased to 53, clever's karma got increased to 54, clever's karma got increased to 55, clever's karma got increased to 56, clever's karma got increased to 57, clever's karma got increased to 58, clever's karma got increased to 59, clever's karma got increased to 60, clever's karma got increased to 61, clever's karma got increased to 62
<__monty__>
I'd kiss ya, but karma'll have to do.
<clever>
lol :)
<clever>
ottidmes: is .0.1 running a proper nameserver? what does and doesnt work? can the phone ping an IP on the lan? can it ping an IP on the internet?, is it failing dns?
<ottidmes>
alright, I got permission to break the internet connection by the other internet users here :P so now I can test that :P
<__monty__>
Should nix-env -i still work for my user? I get `error: attribute 'nixpkgs'... not found`.
<ottidmes>
its not running a proper nameserver that I know, it should just delegate it to the provider's nameservers
<clever>
ottidmes: if you run `dig google.com @192.168.0.1` does it get a reply?
<clever>
__monty__: it should work, what does `ls -l ~/.nix-defexpr/*/* output?
<clever>
__monty__: your channels_root is missing now
<{^_^}>
[nixpkgs] @M-Gregoire opened pull request #51383 → network-manager: rename systemd service back to match upstream → https://git.io/fpK37
<__monty__>
clever: Can I just rm .nix-defexpr/channels_root; ln -s /nix/var/... .nix-defexpr/channels_root?
<clever>
__monty__: the one in /nix isnt called .nix-defexpr
<ottidmes>
clever: still works after disabling the router's dhcp and starting the dhcp on the server with the domain-name-servers changes explicitly to the DNS of the provider
<clever>
__monty__: ls -lh /nix/var/nix/profiles/per-user/root/channels ; sudo -i and cat ~/.nix-channels
<ottidmes>
clever: I never considered that a possiblity, so my server is then like those other nameservers? First I am going to test my DNS setup (which did not work due to the DHCP mistake)
teehemka_ has joined #nixos
<clever>
ottidmes: yeah
teehemka_ has quit [Client Quit]
<ottidmes>
clever: then I am definitely going to do that :D
teehemkay has joined #nixos
<__monty__>
clever: `lrwxr-xr-x 1 root wheel 15B Dec 2 13:49 /nix/var/nix/profiles/per-user/root/channels -> channels-9-link` and `https://nixos.org/channels/nixpkgs-18.09-darwin nixpkgs` respectively.
<clever>
__monty__: ls -lh /nix/var/nix/profiles/per-user/root/channels/
<__monty__>
clever: As root `lrwxr-xr-x 1 root wheel 60B Dec 2 13:37 /nix/var/nix/profiles/per-user/root/channels/ -> /nix/store/1qjgg5c5mq8id4a8x0gwyhr3xnkzk5r3-user-environment`
<clever>
__monty__: and `ls -l /nix/store/1qjgg5c5mq8id4a8x0gwyhr3xnkzk5r3-user-environment` ?
<__monty__>
Ah, no such file or dir : /
<clever>
__monty__: you need to run nix-channel --update, as root
<__monty__>
Same result.
<clever>
you may have also deleted something in /nix/store by accident
<clever>
and nix doesnt like that
<clever>
try nix-store --verify --repair
orivej has quit [Ping timeout: 250 seconds]
Radivarig has quit [Quit: Leaving]
Radivarig has joined #nixos
<__monty__>
As root, right? Just to be sure.
<clever>
yeah
<__monty__>
Hmm, next problem. Now, as my user, nix-env -iA finishes instantly and nix-env -q doesn't list anything.
<clever>
__monty__: -q is normal, you just dont have anything installed yet
<clever>
-iA, what package are you trying to install?
<__monty__>
clever: No no, I had a bunch of stuff installed.
<__monty__>
For weeks I didn't realize I didn't actually have a channel. I just figured nix-channel --update was finishing so fast because there where no updates because I wasn't on unstable anymore.
<clever>
__monty__: you may have deleted the wrong thing, did you delete profile or .nix-profile at any point?
Thra11 has joined #nixos
<clever>
you likely always had a channel, on root
<clever>
but couldnt see it because of sudo being wonky
<ottidmes>
clever: about the caching DNS, the main benefit would be speed, right?
<clever>
ottidmes: yeah
<__monty__>
clever: I probably did yes, `rm -r /nix/var/nix/profiles/per-user/toonn/*` I erroneously assumed it was just `channels` and a bunch of `channel-N-link`s.
<clever>
__monty__: that also deleted all generations of your nix-env, so thats toast
<clever>
nix-env will recreate it as you -iA things
<__monty__>
I guess the cat's out of the bag, now I can openly admit I'm an idiot.
<clever>
just blame apple for having wonky sudo settings :P
<clever>
none of this would have happened if `sudo nix-channel --list` showed a channel!
steshaw has quit [Quit: Connection closed for inactivity]
<__monty__>
True dat.
<__monty__>
While I'm at it I'm gonna go more declarative. Not quite nixos-darwin but at least a more complete config.nix.
<__monty__>
Any ssh implementations I should look at over openssh?
palo2 is now known as palo
<clever>
if you give nix-env an attrset, via nix-env -iA nixpkgs.mystuff, and setting mystuff = { inherit (pkgs) hello; }; in config.nix
<clever>
then nix-env will install every package in the set
<clever>
and unlike the other buildEnv ideas, it will be seperate entries in `nix-env -q` and you can still add/remove a single package if you want to deviate and go imperative for a day
<ottidmes>
clever: stupid Samba is in the way :P if I setup dnsmasq on my laptop it works great, I can dig from my desktop and get the right responses, but on my server I have samba running, which has its own dns server, and a dns forwarder option, but its not working with me (I now see when listing the ports, that dnsmasq has bound to 0.0.0.0 while I specified 127.0.0.1, so it restarted faster probably)
<{^_^}>
[nixpkgs] @symphorien opened pull request #51385 → pyflame: init at 1.6.7 → https://git.io/fpKsb
<ottidmes>
clever: yep, thats it, I expected listen-address=::1,127.0.0.1 of dnsmasq to make it well, bound to 127.0.0.1 rather than 0.0.0.0, but it bounds to 0.0.0.0, but then queries will fail, since the option apparantly does work in the sense that it only listens to local requests :P
<__monty__>
clever: Good suggestion, good point about the flexibility, but I like the clutterless approach of buildEnv.
<ottidmes>
if I want multiple things that open the same port, how can I make one visible to the outside and the other working just internally? By default the both try to bind to 0.0.0.0:53, I thought I could just have one do that and the internal use 127.0.0.1:53, but this is not allowed
ThatDocsLady has joined #nixos
<jw358>
should i be able to install nix-prefetch-github on non-nixos via `nix-env -i nix-prefetch-github`?
mmlb has quit [Read error: Connection reset by peer]
<ottidmes>
Baughn: Samba needs to run on port 53, not configurable, its dns forwarder also does not support specifying ports on the nameservers, but I read I can specify the binding ip address, so I guess I will bind to the external facing IP for samba, and 127.0.0.1 for dnsmasq
<feep>
and that is definitely off, so it seems like bumblebeed just failed to rebuild? ...
<feep>
I don't know how to fix this.
<feep>
is it because I'm on unstable kernel?
<schmittlauch[m]>
When does a channel switch to the latest hydra-evalutaion, and when doesn't it? Or more precisely, why hasn't the channel yet switched to https://hydra.nixos.org/eval/1492768#tabs-now-fail
<feep>
ie. maybe bumblebee tries to depend on stable kernel and gets a different version?
<clever>
feep: if it did, a reboot would clear it up
<schmittlauch[m]>
clever: Some tests only fail due to the VM not responding. Are they tried again then or only after the next nixpkgs commit initiates another build?
<feep>
okay, I tried nixos-rebuild post reboot and it did nothing
<feep>
clever: how do I use that?
<clever>
schmittlauch[m]: if somebody doesnt restart it manually, it will get ran again when something in nixpkgs changes, because most tests include a full copy of nixpkgs
iyzsong-x has quit [Ping timeout: 252 seconds]
<clever>
feep: run `nix why-depends /run/current-system /nix/store/foo` on the "wrong" nvidia version
<feep>
will do
<feep>
'/nix/store/k3dmawbvimd0jk59gw9rbhlzg8z5yy1n-nixos-system-nop-18.09.1446.db561c90ef3' does not depend on '/nix/store/088gr5v648kj84h2h0jh82hqlnb0zyv4-opengl-drivers'
<liminal18>
clever: removed the bash script still works thanks!
<feep>
how come there is a pkgs_i686 but no unstable_i686?
<clever>
feep: thats tricky, you may need to just unstable.pkgsi686_linux.linuxPackages_latest.nvidia_x11.out
<feep>
oh kay
<clever>
feep: pkgsi686Linux exists in all package sets
<clever>
and just forces a 32bit version of that set
<feep>
thanks
<feep>
I tried to stick that in boot.kernelPackages_i686 but that option does not exist
<feep>
I thought I could just define it :(
<feep>
where in configuration.nix can I stick a random variable`?
<feep>
*?
<clever>
a let block
<clever>
let foo = bar; in { ... }
<feep>
oh are those visible from imports?
<clever>
before the main attrset it returns
<clever>
nope
<feep>
yeah that's kinda the problem
<feep>
since I need it in hardware-configuration.nix
<clever>
nothing says it has to be in hardware-configuration.nix
<feep>
truuue
<feep>
I mean, it is kind of a flag that starts with hardware.
<feep>
so it feels like it *should* be there?
<clever>
if you make any changes to hardware-configuration.nix, then your defeating the point of keeping it a seperate file
<feep>
oh!
<feep>
that makes sense
<clever>
its a seperate file, so you can re-run nixos-generate-config, to remake it, to match the current hardware
<feep>
I get it~
<clever>
which will destroy your changes
<clever>
and at that point, it might as well be merged into configuration.nix
<feep>
:nod:
<clever>
and just never have nixos-generate-config help you again
<ottidmes>
is it possible to add another IP address to my server that is not covered by 0.0.0.0?
<clever>
0.0.0.0 is the wildcard, it always covers all IP's
<feep>
clever: do I need to reboot? nixos-rebuild switch still does basically nothing
<clever>
feep: possibly
<feep>
bleh, bbiab
feep has quit [Remote host closed the connection]
jperras has quit [Ping timeout: 272 seconds]
<{^_^}>
[nixpkgs] @arianvp opened pull request #51393 → nixos/containers: Add assertion for container name length → https://git.io/fpKZh
<ottidmes>
clever: I prefered dnsmasq over bind, but it seems I have no choice in the matter, unless I can change the IP onto which samba binds its DNS server, I got no workaround, but it does give me the option do use a bind backend rather than the basic internal DNS with forward, so bind it is
<{^_^}>
[nixpkgs] @FRidh pushed 5 commits to python-unstable: https://git.io/fpKn4
<ottidmes>
__monty__: thats true, but samba will use bind for its own DNS needs, so it will be actually bind that binds to 0.0.0.0:53 (from my understanding, will actually be trying it later today)
<feep>
oh god damn, I forgot to remove the config from hardware-configuration.nix
<feep>
... how did that even work
<feep>
wait, configuration.nix should definitely have overridden it.
<__monty__>
ottidmes: Oh, ok. Can't you have it defer to dnsmasq the same way?
<feep>
yeah since it worked at all, it has to have overridden it.
<feep>
and nixos-rebuild does nothing, yeah okay.
<clever>
feep: most list based options merge
<feep>
yeah but removing it did not prompt switch to reload bumblebeed
<clever>
feep: have you tried just running nixos-unstable for the entire system yet?
ThatDocsLady has quit [Ping timeout: 252 seconds]
<feep>
I specifically don't want that though...
<feep>
hm, on reflection if I'm going back on nvidia-drivers, I don't really need an unstable kernel.
<feep>
trying
elibrokeit has quit [Quit: A random quit message]
feep has quit [Remote host closed the connection]
pie__ has joined #nixos
elibrokeit has joined #nixos
pie___ has quit [Ping timeout: 268 seconds]
<{^_^}>
[nixpkgs] @basvandijk pushed to master « elasticsearch: make sure to patchelf the x-pack-ml binaries »: https://git.io/fpKnp
<liminal18>
anyone ever gotten the libcef error with steam? I have libcef in my buuldpath
<{^_^}>
[nixpkgs] @basvandijk pushed 2 commits to release-18.09: https://git.io/fpKcl
dramforever has quit [Quit: Page closed]
Anton-Latukha has quit [Quit: Leaving.]
feep has joined #nixos
<feep>
clever: brief update
<feep>
it works now
<feep>
final change was not turning the discrete gpu off during boot, which caused it to drop off the bus
<feep>
hm, well, glxgears still doesn't work
<feep>
does work with optirun, just not primusrun
[Leary] has joined #nixos
Lears has quit [Read error: No route to host]
ixxie has quit [Ping timeout: 250 seconds]
arjen-jonathan has quit [Ping timeout: 250 seconds]
<feep>
well, time to find out if it works with factorio
endformationage has joined #nixos
feep has quit [Disconnected by services]
<eyjhb>
How would I go about appending "displaylink" to "services.xserver.VideoDrivers"?
<eyjhb>
So that I do not overwrite the default value
<simukis>
Is there any purpose in raising `system.nixos.stateVersion`?
<samueldr>
,stateVersion simukis
<{^_^}>
simukis: Setting stateVersion to the latest release doesn't update anything and can only break your setup at best. If you want to regardless, Ctrl-F for "stateVersion" in https://nixos.org/nixos/manual/release-notes.html to see things that need to be manually upgraded with the new value
<samueldr>
it's basically used to know where to find stateful things when stateful things are updated to be at a different place
<{^_^}>
[nixpkgs] @markuskowa merged pull request #51329 → Favor HTTPS URLs - the GNU edition → https://git.io/fpoyt
<arianvp>
(e.g. we changed the location of postgresql data dir some releases back)
<samueldr>
so it could have been named system.nixos.whereWillMyStatefulThingsBe = "18.03";
<simukis>
okay, that’s fine. Is there any danger in changing the variable from deprecated `system.nixos.stateVersion` to whatever the new variable is named according to nixos-rebuild?
<ottidmes>
__monty__: I wish, that is what I am trying, but I can only specify either their internal DNS that allows a forwards for undefined entries and Bind as the backend, and the forward does not allow you to specify a port for the nameserver
emily has quit [Ping timeout: 252 seconds]
emily has joined #nixos
philippD has joined #nixos
<{^_^}>
[nixpkgs] @M-Gregoire closed pull request #51383 → network-manager: rename systemd service back to match upstream → https://git.io/fpK37
<ottidmes>
and I want to always use the CLI one when I am inside a shell, I don't think it would fallback in that situation normally, it whould just have used the GUI one
<eyjhb>
Anybody have experiences with displaylink and NixOS?
<{^_^}>
[nixpkgs] @Mic92 pushed to release-18.09 « Revert "nixos/luksroot: Check whether the device already exists" »: https://git.io/fpKBR
<ottidmes>
from the answers about loopback devices I get that I can add additional static ip addresses, but its not like I can then refer to it with say eth:0 or eth:1?
<ottidmes>
infinisil: what do you mean?
<infinisil>
ottidmes: Every new NixOS option increases evaluation time for all NixOS systems
<infinisil>
And I've been going around telling people not to add slightly redundant options because of that
<ottidmes>
infinisil: but rather than trying to limit the options, would it not be more worthwile to see if that can be optimized?
<infinisil>
But there's so many new services added all the time, this is probably not worth it
<infinisil>
ottidmes: Yeah, I've been thinking about doing that, but it's a lot harder of course
<ottidmes>
infinisil: harder implementation wise, but I would say that trying to limit the options added is way harder :P
arjen-jonathan has joined #nixos
<infinisil>
Well, if we could get some better support for reusable options (e.g. something like systemd.services.<name>.openFirewall) then every service wouldn't need to add such options itself
<ottidmes>
infinisil: that would be a good suggestion indeed
<infinisil>
Same with options like `service.<name>.{user,group,extraOptions,extraConfig,enable,package}
nschoe has joined #nixos
<adisbladis>
lucus16: The expression in the vgo2nix repo is broken it should have the attribute `allowGoReference = true`
<adisbladis>
I'm pushing a fix now
hotfuzz_ has joined #nixos
arjen-jonathan has quit [Ping timeout: 250 seconds]
hotfuzz has quit [Ping timeout: 250 seconds]
mmlb has quit [Read error: Connection reset by peer]
<rhalff>
seems a bit like all-packages.nix is causing unncessary naming inconsistencies. caused by name already containing the version number instead of letting the versioned named being derived from name and version "${name}-${version}"
graphene has quit [Remote host closed the connection]
<ottidmes>
rhalff: its a backwards compatible thing I believe, in theory you could just take the current all-packages.nix and automatically rename them in some consistent way based on the derivation name and version, and the whole of nixpkgs could probably be updated to reflect this as well in an automatic away, but then you still have all the users that will have broken configurations the moment they update to this new
<ottidmes>
situation
lo_mlatu has quit [Quit: Connection closed for inactivity]
graphene has joined #nixos
vidbina has joined #nixos
feep_ is now known as feep
jrolfs has joined #nixos
<dmj`>
I have an nginx setup reversing to a hydra server and am trying to enable ACME certs from lets encrypt
<ottidmes>
I resolved my Samba DNS forwarding to dnsmasq issue by using the dummy kernel module, would love to know if there is another way to get a loopback devices though
<{^_^}>
[nixpkgs] @roberth opened pull request #51401 → all-packages.nix: Alias self to res, deprecating self → https://git.io/fpKEz
<makefu>
dmj`: i prepared the same thing for the congress. is there any issue regarding that?
<dmj`>
Everytime I try to start the service I get a cryptic error message about the certificate, return OpenSSL.crypto.dump_certificate(self.typ, data.wrapped).strip()
ng0 has quit [Remote host closed the connection]
<dmj`>
makefu: I'm glad nixos is making it to the halls of congress
<makefu>
like last year but are asking for much more space this time
<dmj`>
makefu: raise Error('Challenge validation has failed, see error log.')
ng0 has joined #nixos
<dmj`>
makefu: if you give a mouse a cookie.. it will ask for a glass of milk
<makefu>
last year we essentially took over the place of the gentoo people right of us (and some of the space of void)
<dmj`>
makefu: nice
pie___ has joined #nixos
<dmj`>
makefu: so any ideas
pie__ has quit [Remote host closed the connection]
<dmj`>
makefu: weird part is that I have other servers that were able to acquire certs like this
<makefu>
sounds like your dns stuff does not work in your favor right now
<nyanloutre[m]>
I am trying to configure a Logitech G920 wheel to be used with Steam games
<nyanloutre[m]>
Do you see any reason to not be detected in games ?
<nyanloutre[m]>
I can see it with lsusb and the force feedback tool (fftest) is working
<das_j>
nyanloutre[m]: Steam is run in a fhs environment
<das_j>
I'm not sure if it needs files which are not in the fhsenv
lsyoyom has joined #nixos
<nyanloutre[m]>
with steam-run I can see all devices in /dev/input
<nyanloutre[m]>
but maybe it's missing something else
<fusion809>
So, when there's a maintainer for a nixpkg, does one have to wait for the maintainer to approve PRs on said package? Or can it be approved by anyone with write access to the nixpkgs repo? I'm asking as I created a PR and based on the maintainer's GitHub history I suspect I'll be waiting a while for a reply.
<gchristensen>
we don't usually block on maintainers
<fusion809>
Sorry, I don't quite follow what you mean. Do you mean no, there's no need to wait for the maintainer?
<{^_^}>
[nixpkgs] @kalbasit pushed 7 commits to release-18.09: https://git.io/fpKgI
fragamus has quit [Read error: Connection reset by peer]
jrolfs has joined #nixos
<ottidmes>
if you run NetworkManager, can you also specify which DHCP server to use, I ask because I want my laptop use my own DHCP server, but split the available IPs between my DHCP server and that of the router, so that if the server is down, you can at least still get a new IP on e.g. your phone
fragamus has joined #nixos
ixxie has quit [Ping timeout: 240 seconds]
<ottidmes>
another solution would have been to set a slowdown, which is apparantly possible in dhcpd, but its not available as an option of my router (which is the fallback)
sigmundv_ has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @matthewbauer pushed 9 commits to master: https://git.io/fpKgw
<symphorien>
You can't choose a dhcp server by design of the protocol, can you ?
fendor has quit [Ping timeout: 268 seconds]
<gchristensen>
you don't need a dhcp server to get IPs in a private network without dhcp. devices will self-assign 169.254.x.x IPs to facilitate communication
<ottidmes>
symphorien: one way at least is that you can ignore responses from specific DHCP servers
<infinisil>
gchristensen: Just a couple days ago I've been reeeally confused by this and thought somebody was messing with my network. It's weird when suddenly you have such an address without knowing that it's a non-public space
<infinisil>
Found out later with whois
<{^_^}>
[nixpkgs] @matthewbauer opened pull request #51408 → gnumake: disable subsecond mtime on darwin → https://git.io/fpKgp
<symphorien>
ottidmes: right. Then maybe do it with a firewall ?
<ottidmes>
I want a static IP address for my laptop when on my home network, and otherwise request one
<gchristensen>
if that is truly your goal, you could set that in networkmanager
<symphorien>
You can create several wired connections: one with static ip and one with dhcp
<ottidmes>
I would have liked to have it all configured in one place
<symphorien>
But I doubt you can autodectect which one to use
<gchristensen>
I'm not sure what you mean.
<symphorien>
Connections are plain text files in /etc so you can manage them like any configuration file
<gchristensen>
if you want a static ip while on your home network, that is trivial - especially with wifi
vidbina has joined #nixos
<ottidmes>
I will try NetworkManager for now, hopefully now it works, it did not for me in the past
<Ralith>
usually the best behavior is to have your home DHCP server cache MAC addrs and reissue the same address to the same host
<Ralith>
and also get DHCP-based DNS set up so you mostly don't need to think about IPs in the first place
<Ralith>
then all machines on the network get nice behavior without any local configuration
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<ottidmes>
Ralith: I would prefer that, first trying static NetworkManager just to see if it works as a plan B, but definitely liking yours for plan A, I already have dnsmasq in place that should be able to do that, but I do need to run a second instance, because I had to employ some tricks to make it work with Samba's DNS server
<Ralith>
...samba has a DNS server?
<ottidmes>
Ralith: it has one builtin, and it needs to be there for it to properly work (at least the features I am using)
<Ralith>
weird.
vidbina has quit [Ping timeout: 250 seconds]
Supersonic has quit [Quit: Bye]
<ottidmes>
Ralith: plan A wont work, that assumes 1 DHCP server
Supersonic has joined #nixos
<Ralith>
shame
aanderse has joined #nixos
jrolfs has quit [Ping timeout: 246 seconds]
hamishmack has joined #nixos
JosW has joined #nixos
arjen-jonathan has quit [Quit: WeeChat 2.3]
simendsjo has quit [Quit: ERC (IRC client for Emacs 26.1)]
<{^_^}>
[nixpkgs] @pbogdan opened pull request #51409 → nixos/cockroachdb: supply defaultText for the package option → https://git.io/fpKV5
Hotkeys has quit []
<{^_^}>
[nixpkgs] @dotlambda pushed to python-unstable « python.pkgs.forbiddenfruit: disable tests because they are missing in tarball »: https://git.io/fpKVh
metastance has quit [Quit: WeeChat 1.4]
<{^_^}>
[nixpkgs] @grahamc pushed to master « Revert "flent/http-getter: broken on clang" »: https://git.io/fpKwG
<{^_^}>
[nixpkgs] @grahamc pushed to master « Revert "hugs: broken on clang" »: https://git.io/fpKwR
<{^_^}>
[nixpkgs] @grahamc pushed to master « Revert "i3-lock-color: broken on clang" »: https://git.io/fpKwV
<{^_^}>
[nixpkgs] @Mic92 closed pull request #17828 → Add cmakeBuildType and dontAddCmakeBuildType variables to be able to make Debug builds. → https://git.io/v6iXS
<coconnor>
j_r0m: yes. By default nix-build will look in the current directory
<coconnor>
nix-build /nixpkgs -A pkgs.ns-3
graphene has joined #nixos
<coconnor>
uses the nix-build form where the directory is explicit
feep has quit [Remote host closed the connection]
feep has joined #nixos
jtojnar has quit [Quit: jtojnar]
jackdk has joined #nixos
<ottidmes>
Ashy: it has not been backported to 18.09, so if you want to install it under 18.09, you can do multiple things, the simplest would probably be to copy the azcopy directory, and move it to your configuration.nix parent directory, and add it to your systemPackages via: (pkgs.callPackage ./azcopy { })
<ottidmes>
Ashy: even better would be to add it to your own nixpkgs overlay
Rusty1 has joined #nixos
<{^_^}>
[nixos-hardware] @Mic92 merged pull request #83 → add basic support for Lenovo Thinkpad T440s → https://git.io/fx113
<{^_^}>
[nixos-hardware] @Mic92 pushed commit from @schmittlauch to master « add basic support for Lenovo Thinkpad T440s (#83) »: https://git.io/fpK6O
<ottidmes>
Ashy: Or you could add the nixos-unstable channel to nix-channel
<{^_^}>
[nixpkgs] @Mic92 pushed to master « Revert "btrfs-progs: fix on cross compilation" »: https://git.io/fpK6B
jrolfs has quit [Ping timeout: 246 seconds]
<hlolli>
I'm making a nix expression with qmake, I'm trying to set a flag to write the source file to $out: qmakeFlags = ["INSTALL_DIR=\$out"]; but I get: ln -sf ut/bin/CsoundQt-d-py-cs6 /ut/bin/csoundqt. Any other tricks to reference $out or should I double escape it?