<samueldr>
oh, it's added intentionally to the docbook source, I now understand
<gchristensen>
yeah
<samueldr>
I thought it was something generated, with e.g. xsl or anything making it "tangible" in the output
<gchristensen>
ah, no
<samueldr>
no?
<gchristensen>
it isn't generated
<samueldr>
the reply was so speedy I thought this was supposed to come before my message :)
<gchristensen>
:)
<gchristensen>
talking to the author of DocBook: The Definitive Guide, he suggested inventing our own tag and doing pre-processing but that seems like a big step. bug we could try it?
<samueldr>
I feel it's probably needed not only for this
<oldandwise>
i'm doing `make`, and missing term.h (http://termbin.com/6o0u). How do get this header?. I tried `nix-env -i ncurses.dev` but not working
<clever>
,library oldandwise
<{^_^}>
oldandwise: Don't install libraries through nix-env or systemPackages, use nix-shell instead. See https://nixos.wiki/wiki/FAQ/Libraries for details.
delroth has joined #nixos
slack1256 has quit [Remote host closed the connection]
slack1256 has joined #nixos
<oldandwise>
ok.. now did `nix-shell -p ncurses.dev` and new error is missing -ltermcap
<clever>
oldandwise: you want `nix-shell -p ncurses` i believe
<oldandwise>
did `nix-shell -p ncurses` .. still same error missing -ltermcap
<{^_^}>
[nixpkgs] @magnetophon opened pull request #48362 → spectmorph: init at 0.4.1 → https://git.io/fxR8M
graphene has quit [Remote host closed the connection]
rotaerk has quit [Ping timeout: 252 seconds]
graphene has joined #nixos
Dedalo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
clacke has joined #nixos
<clacke>
I am trying to join #nixos from matrix. How do I talk to NickServ on pierrebeauchamp.com?
<adisbladis>
clacke: Wrong url? ;)
<clacke>
When I go to #freenode_#nixos:matrix.org it redirects to #nixos:pierrebeauchamp.com, so someone created that alias
rotaerk has joined #nixos
<clacke>
But hm, now I see that the connection is actually mediated by appservice-irc on matrix.org
<clacke>
So at this point it should Just Work -- I've managed to join other channels.
<clacke>
I thought there was something quirky about #nixos, but if it's using the same bridge as everyone else, I won't bother the channel, I'll just keep trying.
<clacke>
aliases on matrix can be really confusing sometimes
fractal has quit [Ping timeout: 244 seconds]
zenhoth has joined #nixos
<adisbladis>
One of many reasons I stopped using the bridges
drakonis_ has quit [Read error: Connection reset by peer]
<clacke>
Is anyone on Matrix and can tell me the cryptohash name of the room?
<clacke>
(room settings: "This room's internal ID is !asdfqwerfoobar1234:matrix.org
dmc has quit [Quit: WeeChat 2.2]
<{^_^}>
[nixpkgs] @bgamari opened pull request #48377 → Cross fixes for 18.09 → https://git.io/fxRuW
<clacke>
Never mind, I'm actually almost-in the room, and it's !AinLFXQRxTuqNpXyXk:matrix.org. I'm just not able to join or leave, just lurk.
<clacke>
sorry to have bothered you all, this will probably resolve itself at some arbitrary point in the future when some thing or other restarts
dmc has joined #nixos
<{^_^}>
[nixpkgs] @bgamari opened pull request #48378 → [WIP] Port switch-to-configuration script to Python → https://git.io/fxRuu
<sphalerit>
clacke: I see you in the matrix room
<zenhoth>
Got a question about configuration. If a package ships with /etc and expects to read it, nixos doesn't seem to link it anywhere, so is the package expected to access it via ../etc?
<Myrl-saki>
zenhoth: "If a package ships with /etc"?
<Myrl-saki>
I'm guessing you mean example configurations?
<zenhoth>
Files under /etc
<zenhoth>
Well no, actual configuration that it reads at runtime
<Myrl-saki>
zenhoth: A lot(citation needed) of packages accept a configuration path override.
<zenhoth>
Right, except you'd need to update that every time you update the package
<zenhoth>
Because of the nix store path containing the hash
<zenhoth>
I'm trying to avoid using user dirs for the moment, because you need a store path in the config (it ships with several executables that it needs to be able to find as part of normal operation)
<Myrl-saki>
zenhoth: The package does not need rebuilding.
<Myrl-saki>
zenhoth: But the service does.
<zenhoth>
Those two are different? Interesting.
<zenhoth>
Anyway, not entirely relevant :P
<Myrl-saki>
zenhoth: Can you tell me more about your exact problem?
<zenhoth>
i3blocks.conf needs a path to the root of a bunch of executables that come with i3blocks
<zenhoth>
This is in /nix/store
<Myrl-saki>
Ohh lol
<zenhoth>
The default config for i3blocks got magicked to contain this path
<Myrl-saki>
That does sound pretty complicated.
<zenhoth>
So I'm trying to get it to use the default for now
<zenhoth>
Yeah, it's somewhat unfortunate
<Myrl-saki>
zenhoth: FWIW, why not just make a custom i3blocks.conf?
<zenhoth>
Because I'm trying to avoid having to manually update the nix store path every time I update the package
Peaker has joined #nixos
<zenhoth>
If I could get configuration.nix to render this config to my home directory for me, that would be great, but that seems to be unsupported
<Peaker>
why does nix-build take ~30 sec to start doing anything (after 5 sec or so it prints warning: "dumping very large path ..", and takes much longer)
<Ralith>
zenhoth: so just do the thing every other service with a config file does?
<zenhoth>
What is this thing? (new to nixos)
<Ralith>
or if the config file doesn't actually have any interesting configuring to do, you can just generate it during installHase
<Ralith>
installPhase*
<zenhoth>
It does. It comes with some basic default config that is otherwise not defaulted to
<Ralith>
Myrl-saki just explained and linked some examples above
<Peaker>
does cabal2nix support embedded C code in packages? manually looking at the output of 'cabal2nix' seems not to
<zenhoth>
Right, it's not a service though
<zenhoth>
It gets executed by i3
<zenhoth>
And I can change how in the i3 config, but again, that's manually written by me in my home directory
<Peaker>
does cabal2nix have any other output besides stdout when given a package dir? or should that be enough to build the derivation?
<Myrl-saki>
Have to run an errand, will check back once done.
<clacke>
Thanks, spharelit. I can't type there though. I have a big "[you have been kicked because err_needreggednick]", with a "Rejoin" and a "Forget room", neither of which works ... But I am able to lurk and see the conversation. I even got a notification for your message.
<clacke>
(I am able to join other rooms that required a regged nick)
<clacke[m]>
typing from mobile client
<clacke[m]>
oh wow ok, maybe I just need to kill the browser session
clacke has quit []
<Peaker>
how do I change a cabal2nix derivation to pass -I of a directory (IOW: how to make cabal2nix respect the 'include-dirs' field?)
clacke[m] has left #nixos ["User left"]
clacke[m] has joined #nixos
<clacke[m]>
Would you look at that, it helped. Again, sorry everyone for the noise.
<{^_^}>
[nixpkgs] @FRidh pushed commit from @r-ryantm to master « tvheadend: 4.2.6 -> 4.2.7 »: https://git.io/fxR2e
Lears has quit [Ping timeout: 246 seconds]
oldandwise has joined #nixos
<oldandwise>
`convert` is part of imagemagick? I have `gm`, but could not find `convert`
capisce has quit [Ping timeout: 264 seconds]
<sphalerite>
oldandwise: convert is part of imagemagick, and graphicsmagick has convert as a subcommand of gm
<sphalerite>
oldandwise: so you can either use imagemagick, or use `gm convert` instead of `convert`
<oldandwise>
i see it now.. got mixed up 'imagemagick' with 'graphicsmagick'
<etu>
oldandwise: That's one of the reasons I prefer graphicsmagick. They don't litter the path with generic names like "convert" :p
<oldandwise>
Can i do a nix-shell of nixpkgs/pkgs/applications/graphics/imgcat/default.nix? Just want to examine more the environment inside a *.nix definition
marusich has quit [Quit: Leaving]
<oldandwise>
ie, want to see where a var is pointing to...${ncurses} for example as defined in default.nix
<sphalerite>
nix-shell '<nixpkgs>' -A imgcat
init_6 has quit []
<sphalerite>
but that probably won't really get you what you want exactly
<oldandwise>
so, aside from `ldd`, how else can i know where in store are my dependencies taken from?
<cyris212>
I want to install intellij ultimate from the unstable channel. So I've added the channel, however the new expressions don't show up when using `nix search -u`
<sphalerite>
cyris212: yeah channels are a bit broken UX-wise. If you just want to install a single package, you can do something like nix-env -f channel:nixpkgs-unstable -iA idea.idea-community
<sphalerite>
(without doing anything with nix-channel)
<cyris212>
sphalerite: Seems to work. Thank you!
<sphalerite>
cyris212: to update it you'll need to rerun the same command, because nix-env's UX is also a bit broken
<sphalerite>
,imperative
<{^_^}>
nix-env has multiple drawbacks as an imperative package manager. nix-env -u will sometimes upgrade to the wrong thing; the outputs to install are very finicky to override; and packages that have been removed or are otherwise unavailable will remain in your profile without any warnings. Consider using a ,declarative setup instead.
<symphorien>
cyris212: to search, use nix search -f "<thechannel>" -u
siers has quit [Ping timeout: 252 seconds]
<sphalerite>
or better --no-cache since that won't confuse future search results
<symphorien>
oh I did not know about this flag
<sphalerite>
… yeah another thing that's kind of broken UX-wise >_<
<cyris212>
sphalerite: nix-env -f channel:nixpkgs-unstable -iA idea.idea-community works. However if you swap "community" with "ultimate" it does nothing and exits with rc 0.
<sphalerite>
cyris212: but basically: ultimate is nonfree and you need to allow that explicitly. The error message that's supposed to show up should show if you do nix build -f channel:nixpkgs-unstable idea.idea-ultimate
<cyris212>
sphalerite: well i've already set `nixpkgs.config.allowUnfree = true;` in my configuration.nix. Should that be enough?
<sphalerite>
,unfree
<cyris212>
(I could not find a flag or something in the nix-env man page)
<sphalerite>
there are two different nixpkgs configs that take effect — one for the system, which you've set, and a _separate_ one for users
<mokasin>
But I get 'The option `containers.foobar.unstable' defined in `<unknown-file>' does not exist.'
<srhb>
mokasin: unstable is simply a nixpkgs checkout (presumably, your channel named unstable) -- it doesn't somehow provide a sub-namespaced module system for you, so unstable.foo is morally equivalent to pkgs.foo -- that is, just the package set.
<srhb>
mokasin: Assuming you just want to override the package version for nextcloud, what you want to look into is overlays
<mokasin>
srhb, yes, I named the channel unstable. I must admit, I only understood half of what your saying ;)
simukis has joined #nixos
<mokasin>
What is a checkout?
<srhb>
OK, fair enough
<srhb>
Essentially just the files in the nixpkgs repository (at some version)
<mokasin>
What do you mean by 'morally equivalent'? Why does unstable.foo does not refer to a checkout from the unstable channel?
<srhb>
Am I right to assume the nextcloud service doesn't exist on the stable branches?
<srhb>
mokasin: It does, but it doesn't provide the nixos modules in the way that you seem to expect
<mokasin>
An old version does, but a newer and much more complete version has recently been merged and not yet backported
drakonis has joined #nixos
<srhb>
mokasin: When you say unstable.services.nextcloud, that's essentially pkgs.services.nextcloud -- (where pkgs is from the unstable channel) but there's no such thing in pkgs. It's a NixOS module
<srhb>
So, you'll need to do two things..
<srhb>
1) import the new module and 2) override the package versions using overlays.
<srhb>
Let me find you some links..
<mokasin>
srhb, that would be nice. I'm a nix noob and I find the documentation not exactly lacking but very distributed. It is hard to find intro level guides
<srhb>
mokasin: Yep, it's hard to find your way around at first :)
fendor has joined #nixos
<mokasin>
srhb, okay, I fiddle around a bit. Thanks so far!
<srhb>
mokasin: So on a high level, importing the module from unstable will make the services.nextcloud options available to you, while overlaying the packages will provide the correct versions of the _packages_ the module uses.
jasongrossman has joined #nixos
<srhb>
mokasin: Sure. Feel free to ask away if you need guidance.
offlinehacker has joined #nixos
<mokasin>
Hmm, is there a glossar somewhere?
<srhb>
mokasin: No, but there really ought to be... :-)
<srhb>
(A NixOS module is the thing that provides the configuration options you use in configuration.nix, and turns them into a full system)
<srhb>
mokasin: (An overlay is a modification to a nixpkgs package set)
<mokasin>
and a package is, well, an app for example?
<srhb>
Package is surprisingly hard to define in nix. But yes, something like that. In this case I'm thinking of the nextcloud "software" specifically
<srhb>
We loosely speak about "packages" as if they were the same thing you'd know from eg. Ubuntu or Arch..
<srhb>
That's usually good enough to get pretty far.
<mokasin>
Okay, so I basically was only loading the module from unstable. But without the fitting pkg, this does not make sense. Therefore I need an overlay to override the pkg. Right?
<srhb>
mokasin: Almost. You were referring to something that doesn't make sense. pkgs, and hence your "unstable" name doesn't actually expose any of the modules directly. So pkgs.services makes no sense, and therefore unstable.services doesn't either.
<mokasin>
So that the module makes a configuration for the package of the same checkout?
<srhb>
First: You need to have the module in order to even be able to refer to its options.
elgoosy has quit [Remote host closed the connection]
<srhb>
that will look something like `imports = [ "${unstable}/nixos/modules/.../path-to-nextcloud" ];`
<srhb>
mokasin: Then, you need an overlay to provide the packages that module uses. That will look like the nixpkgs.overlays line I wrote above.
<mokasin>
This is what you meant by, there is no namespace, I can access directly I suppose
<srhb>
Right, once you've imported it it will just be called services.nextcloud
<mokasin>
I'm not right there, but a few pieces fell into the right places in my mind ;)
<gchristensen>
make the dir ~/.config/nixpkgs/overlays/ and put your overlay at ~/.config/nixpkgs/overlays/myoverlay.nix, or ~/.config/nixpkgs/overlays/myoverlay/default.nix
<mokasin>
gchristensen, it strikes me a bit strange, if the all the configuration sits in /etc/nixos, but overlays in the roots home directory. What is the reason behind that?
<mokasin>
I'd expect, that I can define it somewhere in /etc/nixos/configuration.nix
<gchristensen>
ooh for nixos!
jasongrossman has joined #nixos
<mokasin>
yes :)
<gchristensen>
On a NixOS system the value of the nixpkgs.overlays option, if present, is passed to the system Nixpkgs directly as an argument. Note that this does not affect the overlays for non-NixOS operations (e.g. nix-env), which are looked up independently. ----- add: nixpkgs.overlays = [(import ./my-overlay)];
<mokasin>
gchristensen, okay. But what I don't get, what puts nixpkgs into the namespace?
<infinisil>
gchristensen: Well it's hard to tell how often platforms is used to indicate it being broken, but I guess that's another reason for using `broken = ...' :)
<{^_^}>
[nixpkgs] @Infinisil merged pull request #48401 → clisp: Mark as broken for aarch64 → https://git.io/fxRXT
<hodapp>
okay, I now see that cura_stable and cura are different packages o_O
<hodapp>
can anyone on 18.09 reproduce my issue with cura_stable, in which it repeats "freeglut (foo): glutBitmapWidth: bitmap font 0x00000000 not found. Make sure you're not passing a stroke font." to the console when you run 'cura'?
fendor has quit [Read error: Connection reset by peer]
<hodapp>
I don't know yet if one must have settings already present, or if it doesn't happen on first startup because a first-time setup wizard appears
EarlDeLaWarr has joined #nixos
<grabb0id>
Hi trying to get my AMD card to run in X. I'm trying the amdgpu-pro driver but so far no luck... getting this: "(WW) glamor0: Failed to get GLSL" version followed by "(EE) AMDGPU(0): Failed to initialize glamor." Then the screen start flickering and keyboard becomes mildly unresponsive.. anyone have a clue?
<grabb0id>
It's an RX Vega 56 btw. amdgpu-pro is the only one supporting this is that correct?
<arianvp>
I've been running 4.18 this week (just updated). I have occasional freezes though
<arianvp>
so not totally stable yet
<arianvp>
but YMMV
<grabb0id>
arianvp: so if I select 4.18 kernel, should I remove the items in services.xserver.videoDrivers?
<arianvp>
yes
<jasongrossman>
weeel: I've only tried running cron jobs as root. I'll try to replicate your problem on my machine.
<jasongrossman>
weel: ^
simukis has quit [Quit: simukis]
<weel>
jasongrossman: thanks a ton. One thing it might be is that the setbackground script is added to my PATH from within .bashrc
<grabb0id>
arianvp: so then it will revert to the default I suppose which is [ "ati" "cirrus" "intel" "vesa" "vmware" "modesetting" ]. Should it pick modesetting do you think?
<weel>
jasongrossman: could that cause some issues? If so, where should I be placing the setbackground script?
<arianvp>
hmmm it depends. I had some issues with modesetting on my RX570
<jasongrossman>
weel: Ah, well I don't think there's any way for NixOS to pick up that path, but why don't you just give the full path in the cron line?
<grabb0id>
I'll just exhaust all possibilities. Thx :)
<jasongrossman>
weel: "0 * * * * theo /path/to/setbackground"
<weel>
jasongrossman: oh actually that's a really simple solution, i'll try it now
<jasongrossman>
:-) Please report back.
<thefloweringash>
what does `setbackground` do? Does it need to talk to your display server?
<arianvp>
grabb0id: having the "ati" option there is probably mandatory
<arianvp>
that triggers using the free amd driver
<arianvp>
bit of a weird name. I mnow
<arianvp>
know*
<weel>
thefloweringash: yeah, it's some basic imagemagick stuff followed by a feh --bg-tile call
<thefloweringash>
In that case, I don't think cron is what you want. Maybe try a systemd user unit?
<weel>
jasongrossman: does running nixos-rebuild switch also restart the cron service if I've changed the cron config?
<jasongrossman>
weel: Yes.
<weel>
thefloweringash: why would cron not be acceptable here? It's a much simpler, easy to use service than a systemd unit
<jasongrossman>
weel: Although NixOS doesn't give a complete guarantee that restarting services will work. Sometimes you have to reboot. But I doubt you'll have to reboot in this case.
<zenhoth>
Is "nixos" as a channel name special in some way, or just a random convention?
<thefloweringash>
I think systemd will handle the display variable for you
<weel>
jasongrossman: Ah, ok the cron is running successfully, it's the script that's not working - thefloweringash was right, the feh --bg-tile call isn't working
<jasongrossman>
weel: Hm.
<weel>
thefloweringash: I was under the impression that by running the cron job as my user that it should work just fine, why is this not the case?
<thefloweringash>
Have to make some assumptions about your display environment, but typically the `DISPLAY` variable gets set and/or inherited from the login manager, and tells apps which X server to talk to. You might be able to make assumptions about the value of this variable, and encode that in your crontab like `env DISPLAY=:0 /path/to/setbackground`.
<arianvp>
zenhoth: it's convention.
<arianvp>
whatever nixpkgs=/<path-to-channel-here> points to is used as the <nixpkgs> path inside NIX_PATH
<weel>
thefloweringash: ah thanks, I'll give that a try. I'm only using one monitor and it's on display 0 so this should work
<thefloweringash>
The same might apply for `XAUTHORITY` (?)
grabb0id has quit [Ping timeout: 256 seconds]
<arianvp>
afaim
<arianvp>
afaik*
drakonis has joined #nixos
<zenhoth>
When you say nixpkgs=/<path>, where is that?
<arianvp>
echo $NIX_PATH
<arianvp>
it's inside the NIX_PATH environment variable
<zenhoth>
Ok, so if I add another channel under some non-"nixos" name, nix-env and friends will still use "nixos" by default because that's what NIX_PATH has configured?
<infinisil>
But then it's not really pinning, because each project just gets the version you have in your overlays
<infinisil>
And not a project-specific one
<infinisil>
Others won't be able to reproduce it
<mpickering>
ok, I also add my pins to NUR
mokasin has quit [Remote host closed the connection]
<weel>
bit of a long shot, but I've also got a problem with pulseaudio and crackling audio - does anyone have any experience with using an audio interface on NixOS?
<jasongrossman>
weel: \o/
<jasongrossman>
thefloweringash++
<{^_^}>
thefloweringash's karma got increased to 1
<weel>
jasongrossman++
<{^_^}>
jasongrossman's karma got increased to 3
<weel>
thefloweringash++
<{^_^}>
thefloweringash's karma got increased to 2
<infinisil>
mpickering: Still doesn't give you reproducibility. Checking out a specific git version of your project would just use whatever nur pin the user has at the time
<mpickering>
ah the pins should be immutable
<mpickering>
that's a fair point though, no way to enforce that
<{^_^}>
[nixpkgs] @peti pushed 7 commits to haskell-updates: https://git.io/fxR7T
<infinisil>
mpickering: The best thing to do imo is to have a standard pin format for your projects, such that you can write a script to update them all at once while running a compilation/tests to make sure the new version works for all of them
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « cabal2nix: update overrides for latest yaml version »: https://git.io/fxR7q
<{^_^}>
[nixpkgs] @peti pushed 2 commits to haskell-updates: https://git.io/fxR7Z
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « git-annex: update sha256 hash for the new 6.20181011 version »: https://git.io/fxR78
weel has quit [Quit: leaving]
<{^_^}>
[nixpkgs] @peti pushed 2 commits to haskell-updates: https://git.io/fxR72
<das_j>
How do I properly package vim syntax highlightings?
weel has joined #nixos
<mpickering>
I considered doing that but it doesn't seem like that's how the agda mode works
<weel>
one last request - has anyone got any wireless peripherals working with NixOS?
<weel>
I've got a wireless mouse I'd like to use and the wiki makes it sound easy enough (hardware.bluetooth.enable = true, then pair the device) but it doesn't appear to work
<infinisil>
mpickering: What do you mean?
<weel>
the mouse has a usb receiver I've plugged in, but pairing it is not easy
<infinisil>
That way you don't have to specify the cedille binary in your config and it works well by default
<mkroehnert>
Hi everyone, I have question. Been running the previous Nixos release in unstable mode for a long time now and most of the things worked quite well.
<mkroehnert>
With the latest nixos-rebuil boot --upgrade however, I get asked at the boot prompt for a luks password for /dev/sda2 which does not have any password set.
<mkroehnert>
The initial setup with using luks was done during the initial Nixos installation and worked without any issues up to now.
<mkroehnert>
Does anybody have an idea on how this could be fixed? Googling unfortunately didn't help so far.
jbgi has joined #nixos
<samueldr>
mkroehnert: you do have the previous generations still?
<mkroehnert>
yes
<mkroehnert>
the last working one is two to three months old
<samueldr>
can you tell me the commit ID of the previous working and current? it may help pin-point or reject changes
<mkroehnert>
let me check on how to get the commit ID
<samueldr>
IIRC by default they are listed in the name of the generation in the bootlaoder
<samueldr>
(there are other ways also, don't know them off the top of my head)
<samueldr>
with grub I can grep `menuentry /boot/grub/grub.cfg` to list them
<mkroehnert>
the working one is listed as nixos-19.03pre154487.0a7e258012b
<samueldr>
thanks
<mkroehnert>
via nix-info
<samueldr>
I can probably assume the non-working is the current nixos-unstable release I guess
<mkroehnert>
it was generated at 2018-08-04
<mkroehnert>
samueldr: yes
<mkroehnert>
but I also tried a month ago with the same problem but didn't get a chance to ask until now
<samueldr>
mkroehnert: how is your luks configured?
<samueldr>
(I'm looking at commits between 0a7e258012b and now touching luks)
<mkroehnert>
I was never asked for a password before
<samueldr>
hmm...
<mkroehnert>
and checking the git log of my nixos/configuration.nix I never touched this line after the initial installation
<mkroehnert>
this was 2016-06-18
<samueldr>
the only bit that has me dumbfounded is that I didn't think a luks volume could work without a key file or a passphrase, but apparently it can
<mkroehnert>
after reading up on luks after this issue I am also wondering how the setup had worked before
<samueldr>
mkroehnert: this will sound dumb, but have you tried pressing enter? I really don't know if it would help though
<samueldr>
and additionally, I have assumed you are using `nixos-unstable`, are you? I haven't checked nixpkgs-unstable (which shouldn't be used to build a nixos system)
shabius has joined #nixos
<mkroehnert>
samueldr: yes, I actually tried that but the prompt reappears
<samueldr>
"good"... at least good to know it isn't a workaround
kai_w has joined #nixos
<mkroehnert>
and I am using the nixos-unstable channel
drakonis_ has joined #nixos
<mkroehnert>
strangely, nix-channel --list does not return anything
<samueldr>
root's channels are used to build the system
<samueldr>
so a dash of sudo may be required
<mpickering>
infinisil: I updated the PR now with your suggestion
<samueldr>
hmm, a bit annoying, I can't guess at any commit between the ID you gave earlier and current nixos-unstable that would cause such an issue :/
<mkroehnert>
strange
<samueldr>
the only commit(s) touching luks in stage-1 are for UUID
<samueldr>
though there is a commit that changes the indentation around that bit of stage-1
<mkroehnert>
would it make sense to try regenerating configuration.nix and see what it produces?
<{^_^}>
[nixpkgs] @c0bw3b closed pull request #24163 → riak: adding Darwin support → https://git.io/vyAH8
<samueldr>
not sure it really would
<samueldr>
but if you have yours in a git repo, and properly backed-up, maybe do it anyways?
<samueldr>
and compare
aleph- has joined #nixos
aszlig has quit [Quit: Kerneling down for reboot NOW.]
aszlig has joined #nixos
drakonis has quit [Ping timeout: 244 seconds]
kiloreux_ has quit [Remote host closed the connection]
kiloreux has quit [Read error: Connection reset by peer]
<infinisil>
mpickering: Yeah the ofborg failure is a bit odd
<samueldr>
mkroehnert: i I were you I wouldn't, but do you remember how you setup the luks partition initially?
drakonis has joined #nixos
<samueldr>
if I were*
<infinisil>
mpickering: glibcLocales seems to be null
<mpickering>
right but it works on both our machines?
<infinisil>
mpickering: Why is that needed anyways?
<infinisil>
ofborg might fail on a different arch I suspect
<mpickering>
It's something you need when running a Haskell executable in the builder
<mpickering>
You need the same stuff when using hakyll for example
<infinisil>
Along with a check for glibc, maybe that's what we need, hmm
<mkroehnert>
samueldr: so regenerating the configuration does not produce anything with luks and I am sure that I do not need luks right now, since this is for encryption
<mkroehnert>
samueldr: would it be safe to just remove the luks part in the configuration.nix?
<mpickering>
infinisil: or check if the platform == linux?
<mpickering>
I suppose the check you suggested is more precise
<infinisil>
mpickering: The stdenv.buildPlatform.libc == "glibc" check seems to be used a couple times in pkgs/development/haskell-modules
<mpickering>
why is the attribute null anyway if it's not supported? Shouldn't it just not be defined?
mkoenig has quit [Quit: Lost terminal]
<infinisil>
Not sure
<samueldr>
mkroehnert: check with `cryptsetup status NixOS` when booted in your working system
<samueldr>
a working luks will output something starting with: /dev/mapper/LUKS-DUFFMAN-HOME is active and is in use.
mkoenig has joined #nixos
<samueldr>
and having a bunch of infos following
<samueldr>
and if it isn't used, I guess it would output nothing, with exit code 4
<mkroehnert>
it says /dev/mapper/NixOS is inactive
<samueldr>
hmm, inactive is probably different than "not existing"
<samueldr>
mkroehnert: what's mounted on / ? `mount | grep 'on / '` could be used to check
<samueldr>
alternatively, this I think will list all mounted luks devices `lsblk /dev/mapper/*`
<samueldr>
(with "control: not a block device" being printed, which isn't an issue)
<mkroehnert>
there is only docker-8.2
<samueldr>
well, your intuition to remove the luks part might be right
<mkroehnert>
and /dev/sda2 is mounted on /
<samueldr>
yeah, so it's not in use it seems
<mkroehnert>
okay, I'll give it a try
<samueldr>
something must have changed in how it fails to open the luks device when it isn't one
<mkroehnert>
apparently :-)
<mkroehnert>
samueldr: that did the trick. sorry for not trying this out on my own first.
<samueldr>
no worries, I can see how it's worrying not being able to boot :)
<mkroehnert>
with nixos you have at least the option to rollback and use an older configuration :-)
<{^_^}>
[nixpkgs] @Infinisil opened pull request #48427 → fwupd: Fix daemon config not being found → https://git.io/fxRjt
<das_j>
What am I supposed to do with "error: selector 'x86_64-linux' matches no derivations" when running nixos-install?
<{^_^}>
[nixpkgs] @jeroendehaas opened pull request #48428 → fmt: init at 5.2.1 → https://git.io/fxRjY
<{^_^}>
[nixpkgs] @marsam opened pull request #48429 → pythonPackages.nvchecker: init at 1.1 → https://git.io/fxRjC
<{^_^}>
[nixpkgs] @Infinisil opened pull request #48430 → [18.09] fwupd: Fix daemon config not being found → https://git.io/fxRjR
mokasin has joined #nixos
<mokasin>
Hey. Do nixos containers share their users with the host system?
<joko>
mokasin: no. I think the containers share /nix/store with the host, but apart from that it is a different system
<mokasin>
joko, hmm, okay. I'm trying to use a module (and package) from unstable, which apparently should create a user, but there seems an issue with that -.-
<mokasin>
I have trouble debugging, because I cannot start the container because of this
<joko>
mokasin: could you share some of the configuration?
<{^_^}>
[nixpkgs] @dasJ opened pull request #48431 → nixos/nixos-install: Unset system → https://git.io/fx0ek
<mokasin>
the error is from a nextcloud-setup.service, "chown: invalid group: ‘nextcloud:nginx’"
<mokasin>
maybe I'm using the package wrong, could be. But maybe I somehow also need nginx from unstable, because it is a modified dependency as far as I can see from the PR
* mokasin
is a nix n00b
<joko>
mokasin: no worries, we all have to start from somewhere :D
<{^_^}>
#11106 (by nbp, 2 years ago, merged): Add a way to pin a NixOS version within the module system.
<mokasin>
joko, Okay, thanks for the input, now I have a good idea what to do. You were very helpful. Have a nice day or evening or whatever is your timezone.
<bigvalen>
Is there any way to debug systemd not starting services properly ? Seems it's not starting 11 services on startup, before it tries to start logind/sshd. So I can't login to look around. If I start in single user mode, then do 'systemctl default', it works fine.
<mokasin>
bigvalen, you can check your logs
<mokasin>
with journalctl -b -1 for example
<mokasin>
with journalctl -u some.service, you can filter for the failing services
Mr_Keyser_Soze has quit [Ping timeout: 245 seconds]
<dhess>
Ugh, just deployed the latest nixos-unstable and somehow the eth0 bring-up script on one of my build servers is an empty file, so it (obviously) won't come up
<dhess>
anyone else seen something like that? The same Nix store path on my NixOps deployment machine is fine
lawlesseel has quit [Remote host closed the connection]
<dhess>
infinisil: cool thanks. I was able to repair one obviously broken path with nix-store --repair-path
das_j has quit [Quit: "Bye!";]
<dhess>
now the whole thing. This is going to take awhile :\
<clever>
dhess: nix-store --verify --check-contents will scan for corruption
<mpickering>
infinisil++ Thank you for saving me from the PR graveyard
<infinisil>
Np :P
<dhess>
clever: yep running that now :)
wykurz has joined #nixos
<dhess>
In the interest of speeding this up, is there a tool that will start with a given closure and nix-store --verify its dependencies, recursively?
<dhess>
Is it systemConfig in /run/current-system/activate?
<infinisil>
dhess: Just /run/current-system
orivej has joined #nixos
<dhess>
infinisil: yeah looks like those point to the same thing.
shabius has quit [Remote host closed the connection]
nDuff has joined #nixos
shabius has joined #nixos
shabius has quit [Max SendQ exceeded]
o1lo01ol1o has quit [Remote host closed the connection]
shabius has joined #nixos
<{^_^}>
[nixpkgs] @lschuermann opened pull request #48434 → libusb-compat: 0.1.5 -> 0.1.7 && change libusb source to GitHub → https://git.io/fx0Tf
shabius has quit [Max SendQ exceeded]
o1lo01ol1o has joined #nixos
<dhess>
oh the full check just finished. That wasn't so bad
shabius has joined #nixos
<dhess>
that's damn cool. Yet another advantage of using NixOS
<drakonis_>
so it is
elgoosy has quit [Remote host closed the connection]
<drakonis_>
but navigating the whole thing to make use of it is the hard part
<drakonis_>
using it for complex tasks
o1lo01ol1o has quit [Remote host closed the connection]
goibhniu has joined #nixos
Fare has joined #nixos
graphene has quit [Remote host closed the connection]
<Fare>
Upgraded to 18.09, but my stumpwm config doesn't work anymore :-(
graphene has joined #nixos
<Fare>
what window manager is in these days?
capisce has joined #nixos
<lassulus>
i3, xmonad?
<maurer>
I use Xmonad, but really, don't pick your window manager based on what's "in", pick something that does what you want
<infinisil>
Well, I picked xmonad because it's configured using Haskell, and I like Haskell :P
<infinisil>
I hear lots of people using i3, which means it's also going to be decently maintained and you won't run into a lot of problems
<nDuff>
Are there best practices around building derivations that use uncontrolled/impure/changing upstream inputs? I'd like to trigger a rebuild whenever an upstream resource changes, while caching downstream builds that depend on it when it doesn't. I suppose I could write code that polls that resource and generates/updates a checksum for its current contents -- is that the best practice?
<makefu>
nDuff: there is fetchTarball and fetchgit, both do not demand sha256sums. downstream builds are always stored unless manually gc'ed. however i personally would mirror the sources and manually trigger updates if required
<nDuff>
It looks like the __impure code that edolstra wrote could address my use case (storing the results content-addressed), but I'm not yet clear if that's actually been merged yet or is dev-tree code for a future release of nix.
<Fare>
I picked stumpwm because it's written in Common Lisp, and I like Common Lisp.
<Fare>
But after the current breakage, well, sigh.
symphorien has joined #nixos
<Fare>
Long ago, I used scwm, written in Scheme, but it was leaking memory and dying after a few days.
<infinisil>
I've got something similar with my status bar
<Fare>
I'm wondering whether to clean up stumpwm or use something else.
<infinisil>
Currently using xmobar, which has some problem that makes it restart randomly once every hour, or repeatedly like 10 times in a minute
<infinisil>
I tried to switch to taffybar, but that uses 100% CPU for unknown reason after running for a while
<manveru>
infinisil: you like polybar?
<infinisil>
Never tried it
<manveru>
got a neat home-manager module :)
<infinisil>
Ahh I think I saw that
<manveru>
using it for a few months now without issues
<infinisil>
I wanted to use taffybar because it's also configured using Haskell
<infinisil>
But I guess I might give polybar a try too soon :)
<manveru>
fixed-output derivation produced path '/nix/store/bmh2s8hlc3jmllajarl1f9f3y38mvc9x-0a3755c1799d3a4dc1875d4c59c7c568a64c8456.patch' with sha256 hash '1flnxgjjrrlzdk5klx47l8y8rg1vmzdd9vhzxxq6wiwyvd4bd92r' instead of the expected hash '0bizaf2yf93hwkrrjcl3fhawyhmw9dzq9pc283dxmmpxqvvif5xg'
<clever>
searched nixpkgs for the gitrev in the filename
<clever>
its using fetchurl, which is the wrong way to get patches
<clever>
patch files on sites like that can have minor formating changes when software updates
<clever>
and that breaks fetchurl's hash
<{^_^}>
[nixpkgs] @dywedir pushed commit from @tycho01 to release-18.09 « skypeforlinux: 8.28.0.41 -> 8.32.0.44 (#48391) »: https://git.io/fx0qo
<clever>
fetchpatch normalizes a patch to prevent that problem
<manveru>
ah, ok
<manveru>
still not sure why it didn't just fetch network-manager from the cache
<elvishjerricco>
clever: How perfect is fetchpatch's normalization? Does it guarantee that two semantically identical patches will be normalized to the exact same format, or does it just try its best?
<clever>
elvishjerricco: i believe it will guarantee that two semantically identical patches will be normalized to the exact same format
<elvishjerricco>
Awesome :)
<clever>
the last case i saw fail, had some trailing garbage after the patch, that would have just been stripped out
<clever>
the 2.18.1 in manveru's last link is that kind of trailing garbage
<{^_^}>
[nixpkgs] @dywedir pushed commit from @tycho01 to release-18.09 « skypeforlinux: 8.28.0.41 -> 8.32.0.44 (#48391) »: https://git.io/fx0q9
<Dezgeg>
no, it doesn't and can't
<Dezgeg>
consider e.g. two consecutive identical lines where patch 1 deletes the 1st of them and patch 2 deletes the second of them
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<clever>
that would be 2 patch files, right? and fetchpatch only works on a single patch
<Dezgeg>
no but there you have two semantically identical patches that won't normalize to the same thing
Ariakenom has quit [Read error: Connection reset by peer]
<kyren>
Hi, I have a workflow with nixos-rebuild that works pretty well by building on a powerful desktop machine and deploying to remote servers, and I'd like to modify this slightly so that the system results are kept as symlinks on the build machine so that they're not garbage collected automatically. I think in order to do this cleanly I have to understand what nixos-rebuild is doing under the hood so that I can build the system with e.g. nix-build and then
<kyren>
deploy the result separately, but I don't know how to do that yet
<kyren>
if somebody could point me in the right direction for docs that would be great
<clever>
kyren: behind the scenes, its just using nix-env --profile --set
<kyren>
so it's like nix-build, nix-copy-closure, then nix-env --profile --set
<clever>
nix-env --profile /nix/var/nix/profiles/per-user/clever/profile-name --set -f '<nixpkgs/nixos>' -A system -I nixos-config=./machine1/configuration.nix
<clever>
the -f and -A select a derivation from a file
<clever>
the --set tells nix-env to just replace the whole profile, rather then trying to do the normal mutable adding things onto a dir
<clever>
and --profile changes the profile it uses
<clever>
but you may just want to use nixops at that point
<kyren>
but I'd be using the system profile right
<clever>
for the target machines, yeah, you would use the system profile
<clever>
but for the build machine that holds many nixos builds, you would use a custom profile
<kyren>
yeah I could just use nixops, I mean what I'm doing now is 99.9% perfect, I just want the output symlink as well
<kyren>
I could also just invoke nixos-rebuild twice
<kyren>
it just seems dirty
<kyren>
this is what I'm doing now, there's just no option for a symlink output.. I could go full nixops though: http://ix.io/1mF6
<kyren>
but I'm kinda also just trying to understand it
<clever>
does that update the system profile on the remote host?
<kyren>
if you know an easy way of doing that that doesn't require so much deep tinkering that would be great, but I think I have enough breadcrumbs to figure it out the hard way as well
<kyren>
yeah it does
<clever>
i dont think there is any way to make symlinks locally with nixos-rebuild
<clever>
you would have to just unwrap it some and use a nix-env like i gave above
<kyren>
yeah okay, that's fine
<clever>
and then nix-copy-closure yourself
<clever>
and then finish reinventing nixops
<kyren>
hahaha
<kyren>
but nixops will do this for me right
<clever>
yep
<jasongrossman>
Fare: some of us still use stumpwm. I wonder what your problem is. Want to post your stumpwmrc?
<clever>
the enable rollback flag will make it keep profiles locally
<kyren>
so, how do you use nixops if not every machine is on the same channel
<kyren>
is that super gross
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<clever>
[clever@system76:~]$ ls -l /nix/var/nix/profiles/per-user/clever/nixops/ca55b021-87f5-11e8-b6f5-024225fad90e/nas
<clever>
kyren: this is the last build of my nas
<jasongrossman>
(I would actually prefer to use exwm, but can't get it to work - I can get it to run, but some of the most basic examples from its documentation don't work.)
<clever>
# nixops modify -d house deployments/house.nix -I nixpkgs=https://github.com/nixos/nixpkgs/archive/dae9cf6106d.tar.gz
<clever>
kyren: when you run nixops modify or create, you can pass normal -I flags
<clever>
and it will rememeber them (check nixops info to see them)
<kyren>
oh okay.. oh I think I get what you're saying
<kyren>
don't define all the machines in the same configuration
<clever>
there is also an option to specify it in the nix, but it harms performance
<kyren>
how does it harm performance? you mean build performance?
<clever>
nix eval performance
<clever>
it uses builtins.scopedImport to mess with the search path to change <nixpkgs>, and that disables the cache that `import` has
<kyren>
right okay, I didn't even really think about not using the same config though, I would just use it to manage my local desktop which is on unstable so it's not like I have a hugely compelling reason to use multiple channels in one nixops config
<kyren>
but yeah, down this road is reinventing nixops, got it
<clever>
i run my laptop from nixos-unstable, and them my nixops deployment (router+nas) is also unstable, but a rev i picked and locked down
graphene has quit [Remote host closed the connection]
<clever>
so even if the laptop updates, the router doesnt, and there is no unplanned breakage
graphene has joined #nixos
<kyren>
makes sense, I'll just look into using nixops instead then, thanks!