<{^_^}>
[nixpkgs] @matthewbauer merged pull request #40280 → Cleanup of HandBrake related patches → https://git.io/vp1fE
<{^_^}>
[nixpkgs] @matthewbauer pushed 6 commits to master: https://git.io/vpQ5s
<{^_^}>
→ f154aa32 by @Anton-Latukha: handbrake: updated notes
<{^_^}>
→ 104c019a by @Anton-Latukha: a52dec: rm old HandBrake patches
<{^_^}>
→ 3d879e0f by @Anton-Latukha: libdvdnav: rm old HandBrake pathes
jperras has quit [Ping timeout: 255 seconds]
jackdk has joined #nixos
<gchristensen>
Dezgeg: I'm hitting issues where the squashfs file in PXE images are corrupt... "SQUASHFS error: squashfs_read_data failed to read block 0x126be48f" so then I started looking for the squashfs file , and I can't even find it in the initrd! do you know where it is?
<Dezgeg>
maybe it gets deleted at some point
<gchristensen>
I've just extracted the initrd directly, it seems just absent
<Dezgeg>
maybe it's one of those concatenated initrds of multiple parts
<gchristensen>
oohh
<Dezgeg>
so you need to pull them apart... somehow
<{^_^}>
[nixpkgs] @Ericson2314 opened pull request #40530 → MinGW, Hurd: Clean ups and crossConfig removal → https://git.io/vpQF0
<boomshroom>
Ya, that would syntax error.
<tertle||eltret>
i tried to move things aroudn and add some {} but not working
<boomshroom>
tertle||eltret: configuration.nix specifically is supposed to evaluate to a function that takes a set and returns a set.
<boomshroom>
tertle||eltret: It looks like you're trying to use it like a more traditional configuration syntax.
<tertle||eltret>
ohhh yeah
<tertle||eltret>
i didnt know it was code
<boomshroom>
`{ config, pkgs, ... }` is a pattern if you're familiar with functional programming. It goes in place of a function argument to automatically extract its elements.
iqubic has joined #nixos
<tertle||eltret>
ive tried functional and objective but I never buckled down and learned it
<boomshroom>
In Nix, a function is declared by `arg: body`. In this case `arg` is replaced by the set pattern you already have, it just needs a ':' after it to make it a function.
jperras has joined #nixos
<boomshroom>
Everything else if valid field declarations, but the fields aren't going anywhere, so they need to be wrapped in `{ ... }` braces to put them in a set to return from the function.
<tertle||eltret>
so { config, pkgs, ... }:
<boomshroom>
A useful thing to note is that Nix has no regard for whitespace other than whether or not it's there.
<tertle||eltret>
oh ok, thats cool
<boomshroom>
tertle||eltret: Bingo. 1 problem solved, only one more to go.
<boomshroom>
You can put Nix code all on a single line if you really wanted.
<boomshroom>
One thing that I didn't count earlier is I'm not sure if `<demo>` is a valid identifier.
<boomshroom>
> let <demo> = "Test"; in <demo>
<{^_^}>
error: syntax error, unexpected SPATH, at (string):37:5
<boomshroom>
Now I am sure it's not a valid identifier. You'd want to replace it with something else.
<boomshroom>
tertle||eltret: In that case, you can use arbitrary strings as identifiers, but you'd have to wrap it in quotes.
<boomshroom>
> let test = { <demo> = "Test"; }; in test.<demo>
<{^_^}>
error: syntax error, unexpected SPATH, at (string):37:14
<boomshroom>
> let test = { "<demo>" = "Test"; }; in test."<demo>"
<{^_^}>
"Test"
<{^_^}>
[nixpkgs] @Ericson2314 opened pull request #40532 → MinGW, Hurd: Clean ups and crossConfig removal → https://git.io/vpQbm
<boomshroom>
tertle||eltret: Assuming, the ... is just replacing other stuff, the `{ config, pkgs, ... }:` is present at the top, and the username is in quotes, it should be fine.
<{^_^}>
[nixpart] @aszlig pushed 2 commits to release-0.4.x: https://git.io/vpQbc
<{^_^}>
→ dabde329 by Thomas Bach: solves #4 – "nix expression for btrfs subvolume is not correct"
<tertle||eltret>
whoops that first part was supposed to go in chat
OneHandedPenguin has joined #nixos
<boomshroom>
tertle||eltret: `{ config, pkgs, }` is expecting something after the comma. Without the comma, it wouldn't work because it would be passed a set with too many attributes. `...` takes the place of one of the parameters to allow sets with more than was just listed.
<boomshroom>
tertle||eltret: Oh! If "demo" is your username, then you just use `users.users.demo.packages`.
OneHandedPenguin has quit [Client Quit]
<boomshroom>
I thought you wanted "<demo>", which would have required `users.users."<demo>".packages`
<boomshroom>
BRB.
boomshroom is now known as boomshroom-away
johann__ has quit [Ping timeout: 240 seconds]
<tertle||eltret>
ok
<tertle||eltret>
error: anonymous function now at { config, pkgs }: according to error message
<boomshroom>
tertle||eltret: That should contain the current filesystem structure to regenerate on boot. It needs to be imported into configuration.nix, but nixos-generate-config generates a configuration.nix that imports it, but only if you don't already have one.
camsbury has joined #nixos
camsbury has quit [Remote host closed the connection]
camsbury has joined #nixos
camsbury has quit [Client Quit]
blankhart has joined #nixos
camsbury has joined #nixos
<boomshroom>
It seems as though citra can't find GL. Starts with a bunch of `libGL error: unable to load driver` messages and when trying to run a ROM, I get `Unrecognized OpenGL version` and it hangs.
camsbury has quit [Remote host closed the connection]
<samueldr>
battery issues when dealing with both the switch os and linux
Mateon1 has quit [Ping timeout: 250 seconds]
Mateon3 is now known as Mateon1
camsbury has quit [Remote host closed the connection]
<samueldr>
ah, tertle||eltret if I understand correctly, it's on *one drive*, right?
<samueldr>
(virtualbox image?)
<tertle||eltret>
yeah
<samueldr>
then use .device = "/dev/sd___";
<samueldr>
you're hitting an assertion where one drive *needs* the options for mirrored boots too
<tertle||eltret>
how do i know if its sda or sda1?
<andreabedini>
how can I know the reason why nix building a derivation rather than downloading it?
<samueldr>
/dev/sda as it's for the bootloader on legacy
<gchristensen>
Dezgeg: corruption, and the hash is a mismatch!
<angerman>
anyone got a sample of a derivation that takes $src, runs configure, make, and put it into $out? I wonder if tar/gziping src first beforemoving it into out would be a good idea?
<samueldr>
with legacy boot (default for virtualbox) the bootloader needs to be installed to the drive and not to a partition (overly simplified)
<boomshroom>
andreabedini: Usually it's because the exact configuration doesn't exist on the binary cache.
<angerman>
Yes this is a rather esotheric issue I'm dealing with, but I need a partially built source tree as an input for further derivations.
iqubic has quit [Remote host closed the connection]
<andreabedini>
boomshroom: thanks, the package I'm talking about is nixpkgs.purescript, which seems to be recompiled every time (I mean, every time I accidentally garbage collect)
<boomshroom>
angerman: I'm pretty sure that if you don't give stdenv.mkDerivation any build instructions, it will by default run configure && make && make install.
<andreabedini>
there's no particular configuration I am changing
<angerman>
boomshroom: right, I just don't want make install; and I want the source diretory as the output of the derivation.
<angerman>
boomshroom: think about splitting configure && make, and make install into two different derivations.
chessai has quit [Remote host closed the connection]
<andreabedini>
boomshroom: no problem, I just don't know where to look
smallville7123 has quit [Quit: rosa]
<andreabedini>
boomshroom: thanks
<boomshroom>
andreabedini: No problem!
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
drakonis has joined #nixos
iqubic has joined #nixos
drakonis_ has joined #nixos
<iqubic>
I have a pair of unrelated questions that I would like to ask.
<iqubic>
First, is there a key combination that shuts down the current window manager and returns me to the login in screen?
<iqubic>
I was working on a text document, and something happened and the next thing I saw was the WM shutting down and LightDM starting up. What may have caused that?
<iqubic>
I lost a bit of work because the document I was typing up didn't get saved.
<andrewrk>
I have a fresh nixos install but I think I don't have the DNS set up correctly, can I get some help?
<andrewrk>
curl curlmyip.org: could not resolve host: curlmyip.org
<iqubic>
Profpatsch: I didn't run any commands like that.
<andrewrk>
I have: networking.nameservers = [ "1.0.0.1" "8.8.8.8" "8.8.4.4" ]
<Profpatsch>
iqubic: Did you do a rebuild switch in the meantime?
<iqubic>
No. I didn't run any commands at all.
<Profpatsch>
That happened to me recently. Normally the display service should never restart, but I think there was a hiccup in some recent update.
<samueldr>
iqubic: for the first question, with most WM-centric setup, if the WM itself quits or crashes, it will do as you saw
<iqubic>
One moment I was editing a file in emacs, and the next moment I was back at the login manager.
<iqubic>
samueldr: I'll look into that.
<samueldr>
if the process that the display manager started exits, the display manager assumes (rightly?) that the session is finished
<samueldr>
and in most WM-centric setups will, the display manager will start the window manager directly
stumble has quit [Ping timeout: 256 seconds]
<samueldr>
(I'm using an xsession script which ends in a busy `sleep 24h` loop, it's *that* bash script that has to quit for the session to end)
<iqubic>
samueldr: That seems to be what happened here.
<andrewrk>
ok I think it's just 2 addresses
<iqubic>
I'm not doing that. I don't want to do that.
<andrewrk>
seems to be working with [ "1.1.1.1" "1.0.0.1" ]
<iqubic>
Now, onto my second question, unrelated to the first.
<iqubic>
I dual boot Windows 10 and NixOS. I have set-up a partiton that can be has read and write access from both Windows 10 and NixOS. "/mnt/shared" NixOS let's me move files to this directory just as much as I'd like. However I use NixOS (zsh really) to move files out of that directory.
<iqubic>
mv /mnt/shared/foobar ~/folder yeilds the following error:
smallville7123 has joined #nixos
<iqubic>
Removing old name: Read-only file system, /mnt/shared/foobar
<samueldr>
NTFS?
nckx has quit [Quit: Updating my GNU GuixSD server — gnu.org/s/guix]
<iqubic>
Yes NTFS.
<iqubic>
I can post the relevent section of my configuration.nix if you want.
<samueldr>
ah, then journalctl -b0 (no need for less, it will use $PAGER) and check around that time
<samueldr>
maybe it then describes the reason
<iqubic>
And I also have this it cuts out.
spear2 has joined #nixos
<colemickens>
oh, is it just that it was uncleanly unmounted? that will cause ntfs-3g to mount ro
<samueldr>
though, we now know: 1) it's ntfs-3g 2) it is knowingly mounted read-only
<colemickens>
`ntfsfix` might be your friend.
<samueldr>
colemickens: that was a hunch I had (iirc fastboot doesn't put the clean bit)
<samueldr>
not knowing the ntfs ecosystem with linux, I'd read the manual for the command and check if it is declared safe
<samueldr>
I'm not sure if there are any public specs for ntfs
nckx has joined #nixos
<samueldr>
(better be safe than losing data!)
srl295 has quit [Quit: Connection closed for inactivity]
smallville7123_ has joined #nixos
drakonis_ has quit [Quit: Leaving]
smallville7123 has quit [Read error: Connection reset by peer]
<iqubic>
From journalctl:
<iqubic>
The disk contains an unclean file system (0,0). Metadata kept in Windows cache, refused to mount. Falling back to read-only mount because the NTFS partition is in an unsafe state. Please resume and shutdown Windows fully (no hibernation or fast restarting.)
<boomshroom>
I'm working on an init system that uses Nix as the configuration syntax. One problem is that it needs nixpkgs to evaluate the file.
<iqubic>
Well, I suppose that might solve my issue.
maingo has quit [Quit: Connection closed for inactivity]
<samueldr>
well, now you know the reason, now why is it doing that?
<samueldr>
:)
<iqubic>
See the only reason I use Windows at all is for gaming and printing. If I could get cups to work I wouldn't need to share data between OSes.
<boomshroom>
Ugg... printing...
<iqubic>
I know. I tried to get it working last night, and it didn't work.
lord| has joined #nixos
<iqubic>
BRB rebooting.
iqubic has quit [Remote host closed the connection]
<samueldr>
I side-stepped the issue by having a networked printer, luckily enough it works :/
smallville7123_ has quit [Quit: rosa]
<samueldr>
works with google cloud print, so win some lose some
<samueldr>
(also works with the headache inducing linux drivers)
<samueldr>
(they hardcoded the paper dimension in their cups filter!!!)
spear2 has quit [Remote host closed the connection]
<{^_^}>
→ 29b62e07 by @Ericson2314: pam: Remove crossAttrs
<{^_^}>
→ fd5ca417 by @Ericson2314: Merge pull request #40537 from obsidiansystems/linux-pam-cross
<{^_^}>
[nixpkgs] @Ericson2314 pushed 32 commits to staging: https://git.io/vpQj0
<{^_^}>
→ a7ed44cc by @rycee: lib: make use of visible variable in doRename
<{^_^}>
→ 42d11fea by @couchemar: pythonPackages.isort: 4.2.5 -> 4.3.4
<{^_^}>
→ 7a64b072 by @bcdarwin: bicgl: init at 2017-09-10
<{^_^}>
[nixpkgs] @Ericson2314 pushed 6 commits to staging-18.03: https://git.io/vpQj2
<{^_^}>
→ 9be11822 by @Ericson2314: misc mingw: Get rid of crossConfig and clean up
<{^_^}>
→ 84868b8b by @Ericson2314: misc hurd stuff: Get rid of crossConfig
<{^_^}>
→ 7a8f0b3b by @Ericson2314: Merge pull request #40532 from obsidiansystems/mingw-cleanup
iqubic has joined #nixos
<iqubic>
Well, I'm back.
smallville7123 has joined #nixos
<iqubic>
And I was right. Some stupid Windows update had decided to turn fast boot BACK ON.
<samueldr>
:S
<iqubic>
But I turned it off and now everything works the way I want it to.
<samueldr>
glad to hear
johann__ has quit [Ping timeout: 276 seconds]
<{^_^}>
[nixpkgs] @Ericson2314 pushed 2 commits to staging: https://git.io/vpQji
<{^_^}>
→ d7160f39 by @Ericson2314: Merge branch 'linux-pam-cross' into HEAD
<{^_^}>
→ 0fc21a36 by @Ericson2314: Merge commit 'd7160f39bd46e8ee86e95cbaf7a8f3d5685ab30c' into staging
<{^_^}>
[nixpkgs] @Ericson2314 pushed 2 commits to staging-18.03: https://git.io/vpQjX
<{^_^}>
→ 7394e5b1 by @Ericson2314: Merge commit 'd7160f39bd46e8ee86e95cbaf7a8f3d5685ab30c' into staging-18.03
<{^_^}>
→ d7160f39 by @Ericson2314: Merge branch 'linux-pam-cross' into HEAD
hph^ has joined #nixos
<teto1>
would be cool if someone could look at networkmanager dsipatcher exhasutive hook support https://github.com/NixOS/nixpkgs/pull/39142 (26 days without feedback ~20 lines PR)
<iqubic>
Now, I could work on setting up printing, but I don't really want to. The drivers for my current Brother printer are not included in any of the common driver packages listed on the NixOS wiki, nor can they be found by searching the list of packages on the NixOS website
<wilornel>
Hi #nixos! I need to install a Ruby gem from source. I am using bundix with Gemfile right now for my project and I've never done this before. I am not sure if the bundix or the nixos documentation explains how to install these gems from source
<boomshroom>
iqubic: You mean Nix -> C -> Asm -> Binary? Or Nix -> LLVM -> Asm -> binary? The main reason would be because Haskell has similar semantics to Nix.
MercurialAlchemi has joined #nixos
asuryawanshi has joined #nixos
<adelbertc>
given its in the hnix project,i assume by "compile into haskell" they mean parse the Nix expressions in nixpkgs into the Haskell representation of the Nix AST
<adelbertc>
hnix is an implementation of Nix the language in haskell
<boomshroom>
If nobody minds me asking, I know Guix can hook into Nix's daemon, but is it possible for Nix to use Guix's daemon?
sanscoeur has quit [Remote host closed the connection]
<wilornel>
if I run nix-shell, I get a new shell, right? I was wondering.. is there any way to programatically do something akin to cat myscript.sh :: `echo "this is inside myscript.sh"; nix-shell default.nix; echo "this is inside nix-shell"; ` ? As you can see, I am programatically invoking the nix-shell and them running "echo" inside there.
<drakonis>
hmm yes?
<wilornel>
I figured out I can use `--run`
Ross has quit [Ping timeout: 256 seconds]
<adisbladis>
wilornel: You could also use nix-shell as your shebang
<johnw>
you can think of it as self = this scope, super = parent scope
<adisbladis>
The section "How do I use overlays?" addresses self/super
<iqubic>
But still use super for function calls?!?! How does that work?!?!
<iqubic>
adisbladis: I'm actually watching pierron's talk about that right now. It's linked at the top of the page.
<johnw>
iqubic: in actual practice, you don't need to worry too much
<johnw>
just try anything and see if it works
<iqubic>
Why not?
<johnw>
I think I'm breaking like 5 of his rules in my config
<iqubic>
Oh... How?
smallville7123 has quit [Quit: rosa]
<johnw>
i use recursive attrset -- because habit
<johnw>
actually, that might be the only one :)
<johnw>
(after reading through the list again)
inabsentia has joined #nixos
<iqubic>
But would your overlays work without user rec?
<iqubic>
s/user/using/
<johnw>
yeah, I'd just have to use self instead
<iqubic>
self is the same thing as rec?
<johnw>
it's the difference between rec { ... } and fix (self: { ...}) {}
<johnw>
not quite the same
<iqubic>
But doesn't nix automatically call fix for you there.
<johnw>
because with rec, you don't have a "name" for the attrset
MercurialAlchemi has quit [Ping timeout: 240 seconds]
<johnw>
when you make an overlay, your function is evaluated within a fix
<adisbladis>
iqubic: From the article I sent you:
<adisbladis>
self is the result of the fix point calculation. Use it to access packages which could be modified somewhere else in the overlay stack.
<adisbladis>
super one overlay down in the stack (and base nixpkgs for the first overlay). Use it to access the package recipes you want to customize, and for library functions.
<johnw>
I suppose it's safer to use super unless you absolutely need self
<inabsentia>
Hi, I'm trying to change from Arch to NixOs and am getting the error: `mv: cannot move '/boot' to '/boot.bak': Device or resource busy` when I enter the command: `sudo mv -v /boot /boot.bak && sudo /nix/var/nix/profiles/system/bin/switch-to-configuration boot`. Can anyone help with this?
<iqubic>
What's the base case of the overlay fix recursion? What prevents it from becoming an infinite loop?
<johnw>
iqubic: think of it as a chain of composed functions
<johnw>
the "base case" is nixpkgs itself
<iqubic>
I kinda get that.
Have-Quick has joined #nixos
<iqubic>
But really, it's one large loop, with something in nixpkgs not actually depending on self, right?
<iqubic>
Why are you defining new packages in apps.nix, instead of as stand alone derivations with their own default.nix?
<johnw>
laziness
<johnw>
the end result would be identnical
Rusty1_ has quit [Quit: Konversation terminated!]
<{^_^}>
[nixpkgs] @peterhoeg opened pull request #40540 → i7z: do not build the GUI by default → https://git.io/vp7Jx
<iqubic>
What does the haskell.nix overlay even do?
<johnw>
10-haskell.nix?
<iqubic>
Yeah.
<johnw>
it defines my local packages, my overrides for each GHC version, and the environments I build containing those packages
jD91mZM2 has joined #nixos
<johnw>
then there is config/haskell.nix, which lists all the packages to be included (rather than how they are defined or overridden)
Have-Quick has quit [Quit: Have-Quick]
<{^_^}>
[nixpkgs] @peterhoeg opened pull request #40541 → check-{ups,nwc}-health: update to release versions → https://git.io/vp7Uq
bgamari has quit [Ping timeout: 268 seconds]
Have-Quick has joined #nixos
<inabsentia>
Hi, I'm trying to change from Arch to NixOs and am getting the error: `mv: cannot move '/boot' to '/boot.bak': Device or resource busy` when I enter the command: `sudo mv -v /boot /boot.bak && sudo /nix/var/nix/profiles/system/bin/switch-to-configuration boot`. Can anyone help with this?
johann__ has joined #nixos
<boomshroom>
inabsentia: Is /boot on another partion? The EFI boot partition perhaps?
<inabsentia>
yes it is
<boomshroom>
inabsentia: That would explain it. It's hard to move rename a folder if something is mounted there.
<boomshroom>
Try `cp -r` instead.
MercurialAlchemi has joined #nixos
<inabsentia>
cp -r /boot /boot.bak && .......?
johann__ has quit [Ping timeout: 256 seconds]
robstr has joined #nixos
<boomshroom>
inabsentia: Something like that. You're just trying to back-up the partition first. Since you can't just remove it, you'd have to copy it instead.
<boomshroom>
right?
<boomshroom>
Has anyone tried building something with Rust that depends on libbacktrace? It seems to assume that /usr/bin/file exists.
Guanin_ has quit [Ping timeout: 256 seconds]
Guanin_ has joined #nixos
<jD91mZM2>
boomshroom: I'm pretty sure I have, although I haven't gotten that error. Interesting.
<boomshroom>
/home/angelo/.cargo/registry/src/github.com-1ecc6299db9ec823/backtrace-sys-0.1.16/src/libbacktrace/configure: line 6897: /usr/bin/file: No such file or directory
kreetx has joined #nixos
<jD91mZM2>
Does `cargo clean` help perhaps?
* jD91mZM2
goes to get breakfast
<inabsentia>
boomshroom The cp -r worked. But how do I satisfy this requirement? - "Finally, move the /boot directory of your current distribution out of the way". Sorry, finding this step confusing.
<boomshroom>
jD91mZM2: I'm using cargo install. It's morning for you? It's evening here. Are you on the other side of the atlantic?
<boomshroom>
inabsentia: I never did that. :P When installing Nix, I just pretended Arch didn't exist and now I'm pretty sure I can dual boot. Should probably check if I actually can.
<johnw>
iqubic: cool, just watched the video; I had misunderstood the scope of 'self'
<johnw>
i didn't realize it was the self of the aggregate, rather than the components, of the composition
Have-Quick has quit [Quit: Have-Quick]
<inabsentia>
I'm trying to install over it since my Arch setup died in my arms
<boomshroom>
inabsentia: You could probably do an rm -r /boot/*, but that seems dangerous.
<boomshroom>
The fact that the directory itself isn't moving means it should be fine.
<inabsentia>
boomshroom: Wouldn't that be the same as what the mv was accomplishing? But cp and delete instead?
<boomshroom>
inabsentia: Basically, but it leaves the partition itself intact an unmoved.
<boomshroom>
inabsentia: You're not deleting /boot, you're deleting everything inside of /boot
<boomshroom>
inabsentia: reboot into what? The live disk? If you've already backed it up, an alternative solution would be to delete and recreate the partition itself, but the partition needs to be mounted during instalation.
yastero has joined #nixos
kini has quit [Remote host closed the connection]
<inabsentia>
boomshroom: I'm using a live Arch usb to get access to my existing Arch install and using the Nixos lustrate install method to try to install over it
<boomshroom>
inabsentia: Oo, cool! I haven't used NixOS lustrate, but it sounds cool.
<boomshroom>
Had I known about it, I could have used it since the partition my instalation is on used to belong to Ubuntu before it got borked.
<boomshroom>
Funny story: Ubuntu was broken, but technically usable, so I was able to use it to backup my Arch packages since it was really broken and then reinstall Arch.
<etu>
jD91mZM2: We shouldn't put swedish nonsense in here, just announced myself :p
<jD91mZM2>
boomshroom: See how it asks "Is `make` not installed"? Maybe /usr/bin/file is just a warning and make is the real error? That'd make sense because I haven't seen any errors like that before
<jD91mZM2>
etu: Mmmhm. So you're Swedish as well?
<inabsentia>
boomshroom: so I copied /boot instead of moving, then unmounted and removed /boot which was clearly wrong. What should I have done otherwise?
<etu>
jD91mZM2: yeah, if you're in Stockholm you can come by our computer club on mondays, we're at least two people who tend to talk nix or emacs :)
Havvy has quit [Read error: Connection reset by peer]
Havvy has joined #nixos
fractal_ has joined #nixos
<boomshroom>
inabsentia: `cp /boot /boot.bak && rm -r /boot/*` or `umount /boot && fdisk /dev/sda && mount /dev/sd?` with other stuff during fdisk.
<boomshroom>
Or any partition editor.
<jD91mZM2>
etu: Sadly not at the time, but thanks :)
* adisbladis
is swedish too
<boomshroom>
jD91mZM2: Seems like make really was the issue.
jackdk has quit [Ping timeout: 268 seconds]
<jD91mZM2>
adisbladis: If we continue at this rate we can soon take over this IRC channel :P
<adisbladis>
jD91mZM2: I'm not in sweden though :P So maybe I don't really count
<jD91mZM2>
adisbladis: I feel betrayed /s
<inabsentia>
boomshroom: getting Dell SupportAssist saying No bootable devices were found. Think it's gone horribly wrong.
<jD91mZM2>
inabsentia: Did you re-run nixos-install after cleaning /boot?
<boomshroom>
inabsentia: You deleted the boot partition. yay. You still have the live usb?
<inabsentia>
Yeah, going into live usb now
kini has joined #nixos
<inabsentia>
jD91mZM2: trying to do the lustrate install method, which doesn't mention running nixos-install
jmeredith_ has quit [Quit: Connection closed for inactivity]
<boomshroom>
inabsentia: nixos-install is for installing to a separate partition. In your case switch-to-generation boot should be fine as long as the boot partition is mounted.
<inabsentia>
I'll try again
xy2_ has joined #nixos
<boomshroom>
I got the tools built, now rust can't find the target description, despite one of the new tools being built specifically to fix that problem! That said, I don't think that's a Nix problem.
<teto1>
I would like to present nixOS to my colleagues (very high-level overview, mostly for awareness I doubt anyone will install it). Is there a set of public slides I could use ?
<jD91mZM2>
inabsentia: So what are you running instead? nixos-rebuild?
<jD91mZM2>
inabsentia: If yes you need --install-bootloader
<boomshroom>
teto1: One thing I found that made me gasp, even after using NixOS for a while, was distributed builds. If you're doing a demostration, made sure to have a second machine and show how Nix can transparantly distribute builds between multiple machines.
griff_ has joined #nixos
broccoli has joined #nixos
<jD91mZM2>
boomshroom: I've never really tried that. What's the advantages of using distributed builds vs the official binary caches?
pxc has joined #nixos
pjan has quit [Remote host closed the connection]
phreedom has quit [Remote host closed the connection]
phreedom has joined #nixos
<boomshroom>
jD91mZM2: The fact that they can work for any package rather than the ones cached. The downsides are they take longer since they're not cached, and they only work if the systems have the same platform ("x86_64-darwin", "aarch64-linux", etc.).
<boomshroom>
jD91mZM2: That said, you can spoof thwe localSystem to force a system to build packages remotely.
broccoli has quit [Ping timeout: 256 seconds]
<jD91mZM2>
boomshroom: Aren't all packages cached? Or does hydra skip any?
<boomshroom>
jD91mZM2: Hydra doesn't have any of the packages I have that aren't in nixpkgs. It also doesn't have your NixOS system.
<jD91mZM2>
boomshroom: Oh so you can send custom overlays as well?
pxc has quit [Ping timeout: 276 seconds]
bgamari has joined #nixos
<boomshroom>
jD91mZM2: If you have a derivation for a certain system, you can send the derivation to a machine that supports that system. (I didn't mention how an i686-linux machine can distribute to an x86_64-linux machine.)
<jD91mZM2>
boomshroom: Awesome, thanks :)
<boomshroom>
In case you're wondering, I have infact had my /etc filesystem built remotely before. It was mind-blowing.
<adisbladis>
I'm always using distributed builds at home :)
<adisbladis>
jD91mZM2: Also if you are using packages which do not have tests they dont block the channels
<adisbladis>
So some packages may not be built by hydra yet
otwieracz has joined #nixos
xAFFE has joined #nixos
<otwieracz>
Hi!
<adisbladis>
Also any override is likely to invalidate that cache
<adisbladis>
otwieracz: o/
<etu>
Meh, Cython för aarch64 is still not in the binary cache :(
<adisbladis>
etu: Nice umlauts
<etu>
adisbladis: huh?
<adisbladis>
"för"
<etu>
oh :D
<adisbladis>
etu: I'm planning an aarch64 build farm at home
Robbiedobbie has quit [Quit: ZNC 1.6.6 - http://znc.in]
<adisbladis>
Just not sure which board to get
<boomshroom>
adisbladis: Are you on weechat? Because it's just displaying as ? for me.
<adisbladis>
Preferably something cheap
<etu>
same word but with dots on :)
<adisbladis>
boomshroom: Using erc
<otwieracz>
I wanted to ask about one of my use cases to see if it's maybe already addressed somehow in NixOS
<boomshroom>
adisbladis: Does a Nintendo Switch work? :P
<adisbladis>
otwieracz: Shoot
<adisbladis>
boomshroom: Well :P It's a tad expensive :P
<adisbladis>
I want to find the most powerful cheap chinese board possible
<etu>
NixOS on a Switch would be a sweet sweet tablet though
<adisbladis>
etu: Hmm I do have a switch...
<otwieracz>
I am working on software development. More or less I've got few tools: emacs, common lisp, ruby, rbenv, fpm, etc, etc. Those tools, and build process of our software, requires standard directory structure to be available.
<adisbladis>
etu: But I also have my substantially more powerful x86 tablet
kreetx has quit [Ping timeout: 256 seconds]
<adisbladis>
otwieracz: Ok. You can use buildFHSUserEnv to create something that looks like an FHS system
<adisbladis>
otwieracz: I'm assuming you are talking about a nix-shell based workflow
<otwieracz>
And yes, I ended up working in `nix-shell` all the time.
<adisbladis>
otwieracz: That's still the case if you are relying on FHS paths.
<otwieracz>
But for some reasons it was clearly terryfying experience
<otwieracz>
Because I had to start emacs in this nix-shell
<teto1>
boomshroom: not sure that will excite them . I am looking for a 5mn prez, not sure about a demo.
<otwieracz>
And from this emacs I was not able to access „system-wide” things.
<adisbladis>
otwieracz: Yeah.. If you are relying on FHS paths this can be quite messy
<adisbladis>
otwieracz: Can you fix your build to not have hard coded FHS paths?
<otwieracz>
No. This is specific to Common Lisp FFI system.
<adisbladis>
I think MichaelRaskin might be able to chime in on this
pjan has joined #nixos
hyper_ch2 has joined #nixos
xy2_ has quit [Ping timeout: 256 seconds]
<otwieracz>
and one more question - does nix support package versions?
jackdk has joined #nixos
<adisbladis>
otwieracz: What do you mean by that?
<adisbladis>
otwieracz: Packages can happily live side-by-side under different attribute names
<otwieracz>
I mean, let's assume I need different version of one package, equvalent to apt install foo=version
<jD91mZM2>
otwieracz: Some packages are split into multiple with multiple versions, such as openssl_1_1_0. But no syntax like openssl@1.1, no
<otwieracz>
OK. This answers my question.
<adisbladis>
otwieracz: Also you can always override what nixpkgs provides in each respective nix-shell env
smallville7123 has joined #nixos
<boomshroom>
otwieracz: nix-env lets you specify the version for packages where multiple versions are supported, or you can pin a package or the entire repository to an arbitrary commit.
maingo has joined #nixos
<otwieracz>
How to pin a package to arbitrary commit?
<boomshroom>
otwieracz: Short answer: black magic. Longer answer: pkg.overrideAttrs (old: {src = fetchFromGitHub {...};})
<etu>
otwieracz: That's easy by specifying the src attribute of a package to be at a specific commit :)
<otwieracz>
but by `specific commit` you mean in package source repository or in nixpkgs repository, at the time it was in this specific version?
<boomshroom>
Nix, stop trying to run cross-compiled versions of software!
<etu>
You can override packages in your nix-shell definition for example
<boomshroom>
/nix/store/qwg88zvpp8wwaknzgk0jgmf6q9ckzpi5-stdenv-linux/setup: line 1224: /nix/store/3spkj6vhrd891y1fh9wm5gbdalf088v2-coreutils-8.29-aarch64-unknown-linux-gnu/bin/mkdir: cannot execute binary file: Exec format error
<etu>
So you don't have to put it in nixpkgs
griff_ has quit [Quit: griff_]
<otwieracz>
But if the build routine differs, package won't build. Because Nix will try to build it using new routine, with older source code for example.
<otwieracz>
Am I correct?
<otwieracz>
Because I am not pinning to specific commit of nix expression for this package - I am only overriding source files which will be used to build it.
<etu>
Nix will compile things that aren't in the binary cache, and if you done overrides it's very likely that you will compile that package and everything that depends on that override.
<adisbladis>
otwieracz: Or like a lot of people in the nix community does for nix-shell: http://ix.io/1amr/nix
smallville7123 has quit [Quit: rosa]
<adisbladis>
otwieracz: Yes. Correct.
<adisbladis>
otwieracz: But you can combine these approaches, pin nixpkgs and pin more fine-grained in your own overrides
<boomshroom>
Why do so many packages try to run cross-compiled tools when cross compiling?
<boomshroom>
This is not the first time I've run into this problem.
<adisbladis>
boomshroom: Hardly surprising. Cross-compiling nixpkgs is relatively new and a lot of packages do not have the buildInputs/nativeBuildInputs separation
<adisbladis>
Also the package authors may not have thought about cross compiling
<{^_^}>
→ d2f7acf3 by @andersk: pytorch: 0.3.1 -> 0.4.0
<{^_^}>
→ 640b6359 by @nlewo: Merge pull request #40453 from andersk/pytorch
<boomshroom>
adisbladis: I will say the nativeBuildInputs is dubius as it's unclear that it uses packages for the local system especially when the whole repo is set to cross-compile.
Ariakenom has joined #nixos
<boomshroom>
FYI, this is GCC that's acting like this. In case you're wondering, I don't want a GCC that runs on ARM, I want a GCC that targets ARM.
smallville7123 has joined #nixos
<inabsentia>
Still getting `perl: warning: Setting locale failed` when running switch-to-configuration boot. Made sure my locale was generated and selected in locale.conf. Any ideas?
jD91mZM2 has quit [Quit: WeeChat 2.0]
vaninwagen has joined #nixos
johann__ has joined #nixos
<MichaelRaskin>
otwieracz: well, you could just set LD_LIBRARY_PATH, you could start just the Lisp process in nix-shell, start SWANK and connect from Emacs, or you could set CFFI library search path — uiop can run programs and collect output without extra libraries, and then (push (truename library-path) cffi:*foreign-library-directories*)
johann__ has quit [Ping timeout: 276 seconds]
winsom``` has joined #nixos
stumble has joined #nixos
winsom`` has quit [Ping timeout: 268 seconds]
johann__ has joined #nixos
inabsentia has left #nixos [#nixos]
civodul has joined #nixos
MercurialAlchemi has quit [Ping timeout: 240 seconds]
smallville7123 has quit [Quit: rosa]
MP2E has quit [Remote host closed the connection]
MercurialAlchemi has joined #nixos
MP2E has joined #nixos
deepfire` has joined #nixos
Arcaelyx_ has joined #nixos
ryantm has quit [Ping timeout: 250 seconds]
Arcaelyx has quit [Ping timeout: 265 seconds]
<boomshroom>
Good night guys! I will in fact be going to sleep soon.
winem_ has joined #nixos
boomshroom has quit [Quit: WeeChat 2.0]
stumble has quit [Quit: ERC (IRC client for Emacs 25.3.1)]
MichaelRaskin has quit [Quit: MichaelRaskin]
Tucky has joined #nixos
viric has joined #nixos
thc202 has joined #nixos
ckauhaus has joined #nixos
unacceptable has quit [Quit: ZNC 1.6.4 - http://znc.in]
jD91mZM2 has joined #nixos
unacceptable has joined #nixos
jackdk has quit [Ping timeout: 256 seconds]
juri2mol has joined #nixos
lonokhov has joined #nixos
pxc has joined #nixos
<Myrl-saki>
What does nixos-containers use?
<adisbladis>
Myrl-saki: systemd-nspawn
Havvy has quit [Read error: Connection reset by peer]
juri2mol has quit [Quit: My MacBook Air has gone to sleep. ZZZzzz…]
<Myrl-saki>
adisbladis: Thanks.
pxc has quit [Ping timeout: 248 seconds]
Havvy has joined #nixos
<Myrl-saki>
Is there a way to set other options on imperative container management?
<Myrl-saki>
Seems like you can only pass a config file.
<goibhniu>
you use a configuration.nix inside the container
<goibhniu>
although imperative containers seem to be broken at the moment
<Myrl-saki>
I'm reading on the shell script to see how they work with tis.
<goibhniu>
ah right, AFAIK there's no way to tweak those settings for imperative containers
<Myrl-saki>
Oh okay, thanks.
MP2E has quit [Remote host closed the connection]
juri2mol has joined #nixos
winem_ has quit [Ping timeout: 264 seconds]
winem__ has joined #nixos
unacceptable has quit [Quit: ZNC 1.6.4 - http://znc.in]
BigMassive has joined #nixos
unacceptable has joined #nixos
nuncanada has joined #nixos
<jD91mZM2>
Is there a way to change channel for a specific service/module?
<jD91mZM2>
I know I can use specific packages from unstable, but can I make the NetworkManager service use unstable and then use new configuration options from that service?
<{^_^}>
[nixpkgs] @jfrankenau opened pull request #40544 → cura: fix invisible sidebar and install missing materials → https://git.io/vp7CC
bebehei has joined #nixos
bebehei has quit [Remote host closed the connection]
markus1189 has joined #nixos
sigmundv_ has joined #nixos
juri2mol has quit [Ping timeout: 248 seconds]
<robstr>
How to override properly haskell dependencies ? https://gist.github.com/rsoeldner/3af089f9cd44633cfcb5bbec6ea01d68 , `nix-shell -A shells.ghc` is picking up the correct version from the `pkgs/` directory, but now I'm missing basic dependencies like `aeson` which worked before this `override` section (i think due to ONLY supporting packages from _pkgs/_). I found
<mg->
Recently my nixos has started turning off the screen if im inactive for like 1 minute. Is there a setting to turn this off?
<xnaveira[m]>
I am trying to install a custom ssl certifacte, i write in my configuration.nix: security.pki.certificateFiles = [ "path/to/cert" ]; but when building i get cat "path/to/cert" Permission denied. The cert is world readable
lewo has quit [Ping timeout: 240 seconds]
patrl has joined #nixos
patrl has quit [Client Quit]
tehnix[m] has joined #nixos
<LnL>
try sudo -u nixbld1 ls path/to/cert
winem__ has quit [Ping timeout: 256 seconds]
winem_ has joined #nixos
<pierron>
johnw: Nixpkgs is a big loop over all stages + overlays. Every overlay is called with the result of Nixpkgs as self argument.
<adisbladis>
infinisil: I have 32G. The amount of RAM is not the issue :/
Arcaelyx has quit [Read error: Connection reset by peer]
ericsagnes has quit [Ping timeout: 256 seconds]
<jD91mZM2>
But have you tried creating a 512TiB swapfile? No? closed as wontfix
<adisbladis>
:D
<infinisil>
Hmm it was weird though, i added 8GB swap to make it work but in the end only like 5MB got used but it worked
<adisbladis>
srhb: nix 2.0.2 craps out on a rebuild of 557 packages (even with GC_MAXIMUM_HEAP_SIZE=20G)
<jD91mZM2>
Nix just really likes swapfiles :^)
<infinisil>
(I said it worked twice hah)
Ariakenom has quit [Ping timeout: 260 seconds]
<adisbladis>
jD91mZM2: Creating swap files on zfs is not a good idea
smallville7123 has quit [Quit: rosa]
<jD91mZM2>
ooo a zfs user
<adisbladis>
jD91mZM2: There are dozens of us!
<Lears>
It might have some weird limitations. A while back when I was writing ML code in python, the process needed to double its requested RAM just to fork. I ended up adding a 20GB swap partition just to stop OOM crashes, even though barely any of that was actually used.
maccampus has joined #nixos
Arcaelyx has joined #nixos
<jD91mZM2>
sphalerite: You know rust, right? Let's rewrite Nix in rust! (/jk)
<jD91mZM2>
adisbladis: Pretend you're making a commercial for why everyone should use zfs. Why should I use zfs?
smallville7123 has joined #nixos
<adisbladis>
jD91mZM2: You like to read exactly what you wrote to the disk? ;)
<adisbladis>
jD91mZM2: You want efficient incremental backups
<adisbladis>
jD91mZM2: You don't want to wait for the fscking fsck to finish when your remote server needs a reboot
Arcaelyx has quit [Read error: Connection reset by peer]
<jD91mZM2>
adisbladis: Is ext4 more likely to get corrupted than zfs?
<adisbladis>
jD91mZM2: Yes
<adisbladis>
jD91mZM2: And you wouldn't even know
<adisbladis>
With zfs at the very least you _know_ that the data you are getting is faulty
<adisbladis>
I have actually noticed loose cables thanks to zfs checksumming
<adisbladis>
While ext4 would happily chug along
spear2 has joined #nixos
<jD91mZM2>
Ok so in other words zfs isn't just a nice extra, it's a must have?
<adisbladis>
I've even heard stories of people uncovering bugs in storage appliances (I think it was a netapp) thanks to zfs
<adisbladis>
jD91mZM2: Well.. Theoretically any CoW with checksumming would do. Currently zfs is the only mature one.
<adisbladis>
Btrfs is just not mature enough that I feel confident using it for production workloads
<adisbladis>
I've had some pretty nasty issues with it
<adisbladis>
Bcachefs seems promising but it's still some time before that is in the mainline kernel
<adisbladis>
The upstreaming process has started
<adisbladis>
jD91mZM2: So yes. I'd say it's a must for any critical systems.
<jD91mZM2>
Ah crap, I'd have to format my whole drive to install zfs (right?)
<Lears>
In principle you can convert in place, but I gather it's virtually untested.
<ckauhaus>
hi, I wonder that the current 18.03 channel is based on ef74cafd3e5 which is nearly a month old
<adisbladis>
ckauhaus: Looks like it was backported a few days ago
<adisbladis>
srhb: Me too :)
<ckauhaus>
ic
<LnL>
ckauhaus: are you sure? it was updated 2 days ago
<jD91mZM2>
adisbladis: Wait so zfs handles its own partitions sort of using pools. But would I still keep my EFI partition in a normal GPT table?
<ckauhaus>
the Hydra eval is only 2 days old
<adisbladis>
LnL: Looks like it was a commit that took a while to get backported
<ckauhaus>
a bit confusing
<kai_w>
my two pain points with btrfs is the lack of built-in caching and the fact that quotas don't mix well with many snapshots. on the plus side, I'm told it doesn't need as much RAM as zfs.
<srhb>
jD91mZM2: That's the easiest way at least.
<adisbladis>
ckauhaus: Yeah I've reacted the same before :)
<LnL>
ah the commit date, yes that's normal
<srhb>
jD91mZM2: fwiw I have two identical disks that are partitioned like <bootmirror|zfspool>
<adisbladis>
jD91mZM2: Yeah zfs doesnt really care about partitions
<etu>
srhb: Me as well. Been following that for a few years. And with the news about upstreaming it's very exciting!
<jD91mZM2>
srhb: So it'd look something like this: [ efi, sun reserved(?), zfs[ /, /home ] ]
<srhb>
etu: Yeah, Kent is doing good work :)
<ckauhaus>
I think the interesting point is the merge of staging-18.03
* ckauhaus
is shutting up
<srhb>
jD91mZM2: What on earth is sun reserved? :-)
<srhb>
jD91mZM2: But yes, something like that.
<jD91mZM2>
srhb: That reddit comment that's probably from 200 years ago lied to me!1!!111
<srhb>
jD91mZM2: I have special zfs sets for nix and nix store because I want different parameters for those re. compression and stuff
<jD91mZM2>
srhb: I thought I read somewhere that it needed a reserved partition. Turns out that's wrong then :^)
Ariakenom has joined #nixos
<srhb>
Definitely not. :)
<adisbladis>
jD91mZM2: I make an EFI partition and one huge ZFS partition for the rest
<adisbladis>
ZFS has subvolumes and such
<jD91mZM2>
adisbladis: Are those the "pools" I keep hearing about?
<srhb>
A pool can span many partitions
<srhb>
They are purely logicla
<Myrl-saki>
Is there a way to set nixops' deployment ordering?
<adisbladis>
jD91mZM2: Let me explain :)
<srhb>
Storytiem!
<srhb>
:-)
<adisbladis>
jD91mZM2: A zpool contains of one or multiple vdevs, vdevs can be either a single disk, raidz etc
<adisbladis>
jD91mZM2: Inside a zpool you can have one or multiple file systems (zfs)
* jD91mZM2
googles raidz
<adisbladis>
jD91mZM2: Raidz is basically fancy-speak for raid-levels in the context of zfs
unkn has joined #nixos
<jD91mZM2>
Yeah I don't know what a raid is either :^)
<avn>
jD91mZM2: depend of what you need from raid. If doubt -- use mirror ;)
yastero has quit [Ping timeout: 256 seconds]
<adisbladis>
jD91mZM2: raid is about either duplicating or spreading data over multiple drives and achieve the semantics of a single coherent drive
<jD91mZM2>
Ah
smallville7123 has quit [Quit: rosa]
<adisbladis>
jD91mZM2: Mirroring writes the same data to two or more drives, striping evenly distributes data over the drives
<adisbladis>
For example
yastero has joined #nixos
amir_ is now known as amir
<sphalerit>
Where the benefits of doing so are performance (for both spreading and duplicating) and reliability (for duplicating)
<adisbladis>
:)
<adisbladis>
Then we can get into more fancy stuff like raid5 where you have the redundancy of one disk regardless of how many drives you add to the array
<adisbladis>
jD91mZM2: And pretty much no one is using raid-4 or raid-3
<sphalerit>
But that's weird and has confusing performance characteristics. So I stick to mirroring only myself (haven't ever managed a system with enough drives for striping as well as mirroring to make sense)
<jD91mZM2>
Awesome, thanks for explaining :)
<adisbladis>
sphalerit: I'm using raidz1 for my "vacation pictures" at home
<jD91mZM2>
Does it make sense to make a separate /home "partition"/dataset in zfs?
<adisbladis>
jD91mZM2: It might?
<adisbladis>
jD91mZM2: Personally I only snapshot my /home
<sphalerit>
Yes it does
<sphalerit>
You might even want to go more granular than that
<adisbladis>
Everything else is managed by nix anyway so it can be rebuilt easily
<adisbladis>
jD91mZM2: Subvolumes are pretty much "free" anyway
<adisbladis>
No extra overhead, so just add subvolumes to your heart's content
<sphalerit>
adisbladis: /var is important too!
<jD91mZM2>
wait why
<jD91mZM2>
Isn't /var just in memory?
<jD91mZM2>
Or is that /run?
<adisbladis>
sphalerit: Oh yes! Not on my laptop though :)
<adisbladis>
jD91mZM2: /run is tmpfs
<adisbladis>
jD91mZM2: /var/db is the default for most databases
<sphalerit>
jD91mZM2: /var/run is temporary stuff, usually symlinked to /run and only exists for historical reasons
<jD91mZM2>
ah
<jD91mZM2>
Wait so one dataset in zfs can hold multiple "partitions" and can be reverted to a snapshot?
<sphalerit>
But most of /var is persistent data that you typically don't want to lose
<sphalerit>
One pool can hold multiple datasets
<jD91mZM2>
So I most likely do not want / and /home in the same dataset?
<adisbladis>
jD91mZM2: Yeah, make /home separate
johann__ has quit [Quit: Leaving.]
<sphalerit>
Each dataset can have snapshots made of it
<sphalerit>
And rolled back to them
jtojnar has joined #nixos
Freneticks has joined #nixos
<adisbladis>
And each dataset inherits the property of the parent dataset
<sphalerit>
Yep. Most of the time the deciding factor for whether to put things in separate datasets is "will I want to snapshot them separately?"
<jD91mZM2>
wow this is really confusing
<adisbladis>
jD91mZM2: Yeah it is a bit of stuff to wrap your head around
<adisbladis>
But I don't ever want to to ye olde partition management ever again
<jD91mZM2>
Okay so each zfs gpt-partiton has a bunch of pools. Each pool has a bunch of datasets. Each dataset can be rolled back. Each dataset can contain another dataset(?). Each dataset can contain a mountpoint, or "partition" or "container" or whatever it's called?
logzet has joined #nixos
<avn>
jD91mZM2: actually you need only one pool, spanning on one or more drives (or partitions)
<jD91mZM2>
You only need. But can you have multiple?
<adisbladis>
Yes you can
<avn>
I am actually have 1 EFI (1g) + 1 zfs (1t - 1g)
jensens has joined #nixos
<jD91mZM2>
So in the end my layout would look something like: GPT [ EFI, ZFS [ Pool1 [ dataset [ / ], dataset [ /home ] ] ] ]
<avn>
jD91mZM2: actually you need more than one pool if you experiment/debug something in zfs, so one is work/system, and one is experimental
<avn>
yep
<avn>
I have two disks with layout I mentioned above, and both action like halves of mirror
<adisbladis>
jD91mZM2: Yeah :)
<jD91mZM2>
Great, thanks for the help! :D
ryantm has joined #nixos
<avn>
if you use systemd-boot btw, you possible need more than 1g there
<jD91mZM2>
Would it make sense to create a new zfs partition (in GPT) before my current partitions, and then copy everything over and delete the original partitions?
griff_ has joined #nixos
<avn>
I used grub, and have only grub-x64.efi on this partition
<jD91mZM2>
avn: Currently I'm usng 69% of /boot, and it's 256M
<jD91mZM2>
I guess I garbage collect too often :P
<jD91mZM2>
On second thoughs, it would be easier to backup my entire drive and then just redo the whole thing
<srhb>
jD91mZM2: I've started using 1G for boot on NixOS. Lots and lots of kernels lying around.
<jD91mZM2>
I wish I could just resize the partition without moving everything which takes 500 years. I guess ZFS would also solve this kind of since each container doesn't have a fixed size?
<avn>
jD91mZM2: if you have two disks of same size -- just create efi+zfs on second, then you can attach old one as mirror if you need extra safety
<adisbladis>
Actually you can set quotas and such on individual subvolumes
Guest67305 has joined #nixos
<jD91mZM2>
adisbladis: In this case I'd want / and /home in separate dataset I assume?
<adisbladis>
jD91mZM2: Yeah I think so
<adisbladis>
Maybe /var separate too
<jD91mZM2>
why
<adisbladis>
jD91mZM2: You might want to snapshot it separately
<adisbladis>
But I dont know. It's up to you
<jD91mZM2>
Currently the only reason I'm using a separate /home is so I can reinstall OS again... and again... and again
<jD91mZM2>
I can't ever find peace
<adisbladis>
jD91mZM2: That should not be required with nixos :)
<jD91mZM2>
It probably isn't, no - I should look into merging them
<jD91mZM2>
Or maybe it is. nixos-rebuild can restore missing needed files, but it doesn't delete uneeded ones
<adisbladis>
jD91mZM2: It does delete unused config files.
<adisbladis>
And packages, and ....
<jD91mZM2>
Configs from nix packages, yes. But if I somehow end up with a bunch of useless files manually, nix doesn't find them
<adisbladis>
Yeah, of course.
<jD91mZM2>
Like a 5GB /etc/some-useless-file-i-managed-to-get-because-i-suck file filled with 0-bytes
<adisbladis>
But that would mostly be in /home anyway
yastero has quit [Ping timeout: 256 seconds]
<jD91mZM2>
true
<adisbladis>
I use filelight to find the biggest culprits
<etu>
ncdu is nice for finding big files as well :)
<jD91mZM2>
^ second that
<adisbladis>
etu: Yeah. Though I think this is one use case where a gui really is useful
<adisbladis>
And I say this as someone who lives in emacs 90% of the day
<jD91mZM2>
Anyway, thanks for teaching me about ZFS :D. Sadly I don't want to risk getting it right now since I want my system to still be working when Redox Summer of Code starts, but I'll try to remember to try it some time after
maingo has quit [Quit: Connection closed for inactivity]
xcmw has joined #nixos
<adisbladis>
jD91mZM2: Ohh! Redox!
<adisbladis>
I remember seeing nix mentioned in their bug tracker as a possible future option
<jD91mZM2>
adisbladis: Sadly there hasn't been much focus on a Nix-like system yet AFAIK, but it'd definitely be good with an OS that did this right from the start
<adisbladis>
jD91mZM2: Yeah.. I'm hoping that the Redox ppl wont fall for the "worse is better" that is most package managers :/
Arcaelyx has quit [Read error: Connection reset by peer]
<jD91mZM2>
adisbladis: Hopefully Redox will become more like GNU/Linux in the sense that it has more distributions and stuff in the future. Currently it's more of a complete OS, which seems sad.
<adisbladis>
jD91mZM2: I dont know. There are both strenghts and weaknesses in that
<adisbladis>
The level of coherency that something like FreeBSD provides is quite impressive
<adisbladis>
Cohesion is probably the right word
<jD91mZM2>
adisbladis: Having multiple distributions would mean there could be an unofficial flavor of Redox with Nix-type package management even if they end up with a normal package manager.
<jD91mZM2>
Also that could allow for more window managers :>
<FruitieX>
Hi! I'm trying to get some binaries to run inside an FHS environment without doing any patching to them. I'm getting this error: "bash: ./aapt: No such file or directory"
ericsagnes has quit [Ping timeout: 265 seconds]
<jD91mZM2>
FruitieX: That's usually because it expects the linker to exist in a specific place where it doesn't
<FruitieX>
I suspect it has to do with the binary wanting to use /lib/ld-linux.so.2 (per readelf -l) which does not exist
<jD91mZM2>
FruitieX: You can patch it with `patchelf --set-interpreter <linker location> <file>`
<FruitieX>
Right, is there a way around this that does not involve patching?
<jD91mZM2>
I mean... not really
<FruitieX>
Can I do something to my shell.nix to have the linker appear in the correct place
<jD91mZM2>
You could symlink the linker but that's a very bad idea
Arcaelyx has joined #nixos
<jD91mZM2>
So the answer is no I think
<adisbladis>
FruitieX: It looks to me like bash is trying to call the "aapt" tool with a fixed location where it does not exist.
<adisbladis>
I dont think you are having a dynamic linking problem.
<FruitieX>
adisbladis: I'm calling it myself, it does exist
<adisbladis>
Oh wait, aapt is loading something that doesnt exist
<adisbladis>
Gah. Im too tired to help debugging stuff
<adisbladis>
FruitieX: What's the exact output of ldd?
<FruitieX>
hmm what's a pastebin that doesn't suck
<sphalerite>
I have a server where lm_sensors only reports temperatures for the RAM. What might be missing for me to get hold of the CPU temperatures?
<FruitieX>
adisbladis: sorry for ping :) symlinking /lib/ld-linux.so.2 to /nix/store/8229.../lib/32/ld-linux.so.2 makes the binary work for me... wtf
<bkchr[m]>
Hmm, my "~/.nix-profile" points to an invalid path 🤔
<clever>
FruitieX: you need to run patchelf on the binary to fix the ld.so path
<FruitieX>
so even though ldd shows the reloc correctly, this is not happening when I run the 32-bit binary, but for whatever reason it works for the 64-bit binary
<clever>
bkchr[m]: thats normal for a new install, running nix-env -i will fix it
<FruitieX>
clever: I have some 32-bit binaries which I cannot patch, such as Steam games
<clever>
FruitieX: steam runs under an FHS chroot, that provides files at the right paths
<clever>
FruitieX: and there is also the steamrun package, which lets you run things in that chroot
<FruitieX>
For some reason Steam doesn't launch for me with the same error
<bkchr[m]>
clever: Ahh! Then this is should not be the root cause of my zsh problem :(
orivej has quit [Ping timeout: 265 seconds]
<FruitieX>
I'm trying to get this one simple binary to run, maybe I'll understand more how to debug the Steam problem then :)
<clever>
FruitieX: steamrun should help, try debugging under that
<adisbladis>
Man android tooling is _weird_
<FruitieX>
yep!
<FruitieX>
Seems like neither of the binaries work under steam-run
<adisbladis>
FruitieX: What are you trying to accomplish? Building android apps with nix?
<clever>
FruitieX: what does file say on them, and do those paths exist if you do `steamrun bash` ?
<FruitieX>
Mysterious, as steam-run ldd <binary> looks all good
<bkchr[m]>
Does someone know why in my `/run/current-system/sw/share/terminfo` path I have folders with `\~nix\~case\~hack\~1`?
<clever>
bkchr[m]: are you on a mac?
<bkchr[m]>
clever: No
acarrico has quit [Ping timeout: 264 seconds]
<clever>
bkchr[m]: was something copied from a mac with rsync?
<bkchr[m]>
clever: I just have the following patch applied: https://pastebin.com/UniUeX69 With this patch I'm able to boot my raspberry pi again with glibc 2.27
asuryawanshi has quit [Remote host closed the connection]
asuryawanshi has joined #nixos
maccampus has left #nixos [#nixos]
<FruitieX>
Is there any way to supply a custom derivation in configuration.nix, so that it is automatically built on nixos-rebuild?
<sphalerite>
fogbugz: it sounds like it needs a native library to work correctly. The R package set is automatically generated and any additional dependencies are layered on top manually, which presumably hasn't been done for Seurat
<sphalerite>
FruitieX: of course. Depending on what sort of derivation it is, you may want to put it in environment.systemPackages or something
<FruitieX>
Oh, it goes in systemPackages?
acarrico has joined #nixos
<sphalerite>
FruitieX: it depends on what sort of derivation it is and how you want it available
<FruitieX>
It's a firmware blob
<FruitieX>
I have a default.nix which simply copies a file to $out/lib/firmware
<sphalerite>
then you probably want to put it in hardware.firmware
<sphalerite>
hardware.firmware = [(pkgs.callPackage ./foo-firmware.nix {})]; # or something like that
Twey has quit [Ping timeout: 276 seconds]
<{^_^}>
[nixpkgs] @Zimmi48 opened pull request #40552 → Add support for SSL in ocamlPackage.conduit. → https://git.io/vp7Fy
Twey has joined #nixos
asuryawanshi has quit [Remote host closed the connection]
<FruitieX>
However I get "error: attempt to call something which is not a function but a set, at /nix/store/rdd9nqgc688bcbwahpw44v283bgci621-nixos-18.03.132336.ef74cafd3e5/nixos/lib/customisation.nix:74:12"
iyzsong has joined #nixos
<FruitieX>
when doing a nixos-rebuild
<FruitieX>
Probably something wrong with my derivation I'm guessing
<FruitieX>
it does work when just running nix-build
<sphalerite>
change `with import <nixpkgs> {};` to { stdenv }:
<sphalerite>
it's not inherently wrong, just less suitable for composing with other stuff :)
<FruitieX>
oh yep, that did the trick I think, thanks!
spear2 has quit [Ping timeout: 256 seconds]
asuryawanshi has joined #nixos
disasm has quit [Ping timeout: 260 seconds]
disasm has joined #nixos
athan has quit [Ping timeout: 250 seconds]
<angerman>
can I have nix-shell drop me into a temporary directory?
<angerman>
hmm... alright, I don't have a special shell derivation.
<sphalerite>
then make one, if you want that sort of complex behaviour it's not something you'd want to invoke from your shell history each time anyway :D
<sphalerite>
mkShell is nice for keeping shell derivations fairly succinct
asuryawanshi has quit [Ping timeout: 256 seconds]
<angerman>
sphalerite: ha, I'm working on some derivations, I'd rather not be working on, but ohh well ;-)
xcmw has joined #nixos
<yorick>
Synthetica: I've been discovered
<Synthetica>
yorick: 😈
<yorick>
your emoji doesn't support my glorious terminal
<sphalerite>
what is your glorious terminal?
<srhb>
"SMILING FACE WITH HORNS"
balsoft has joined #nixos
<yorick>
urxvt, of course
<Synthetica>
>not using irccloud
<Synthetica>
we're living in 2018, damnit
<srhb>
Synthetica: Speak for yourself!
<srhb>
I'm happily somewhere in the mid-to-late-nineties.
<joko>
yorick: it works here, you could use Symbola :D
ThatDocsLady_ has joined #nixos
<yorick>
I'm still looking for a better terminal emulator. all I need is at-runtime DPI setting (like all gtk3 apps everywhere), transparent background, not an electron monstrosity
ThatDocsLady has quit [Ping timeout: 240 seconds]
<Synthetica>
QTerminal is pretty decent
<Synthetica>
I'd use Konsole, but it's bugged for me
xcmw has quit [Ping timeout: 256 seconds]
simukis has quit [Ping timeout: 256 seconds]
asuryawanshi has joined #nixos
applePrincess has joined #nixos
pxc has joined #nixos
<angerman>
Sonarpulse:
<angerman>
Sonarpulse: I might need your help in a few days, with some gcc-ng integration.
<pxc>
ok, so with the new change in how libGL is setup on NixOS, a (precompiled) package is broken. I have succeeded in working around the issue using LD_PRELOAD, but I want to fix the package. I presume adding a dependency on Mesa is the wrong way to go
asuryawanshi has quit [Remote host closed the connection]
endformationage has joined #nixos
<pxc>
does anyone know the right way to handle this?
asuryawanshi has joined #nixos
MercurialAlchemi has quit [Ping timeout: 268 seconds]
<johanot>
srhb: cat mid-to-late-nineties | grep -v Aqua
Twey has quit [Ping timeout: 255 seconds]
Twey has joined #nixos
xcmw has joined #nixos
<srhb>
:-)
balsoft has quit [Quit: balsoft]
erasmas has joined #nixos
spacefrogg has quit [Remote host closed the connection]
aw has quit [Remote host closed the connection]
nuncanada has joined #nixos
applePrincess has quit [Ping timeout: 240 seconds]
<srhb>
pxc: The right way to do it, if possible, is to inject the gl neutral vendor library instead of mesa directly
<srhb>
vendor neutral even.
<pxc>
srhb: How do I do that? I don't know anything about libglvnd
<srhb>
pxc: In case you just need libgl.so or whatever, you should just be able to depend on libGl
<pxc>
srhb: oh nvm. the libglvnd package has a libGL.so.1 in it. Just use that one instead of mesa_noglu's or whatever?
<srhb>
I think so, yes.
<pxc>
ok, let's give it a try :-)
<sphalerite>
pxc: I think you should depend on the libGL name in nixpkgs
kreisys has joined #nixos
<adisbladis>
yorick: Konsole is pretty decent
<adisbladis>
Not the most minimal but has all the features I want
<adisbladis>
I'm finding it very hard to switch to another one
<sphalerite>
pxc: since it just gives you the right thing no matter which platform you're on
<adisbladis>
Most terminals dont really handle reflowing on zoom and stuff like that so well
<adisbladis>
And doesn't have search in scrollback buffer
<adisbladis>
Which is the killer-feature of any terminal for me
<sphalerite>
adisbladis: tmux has both of those :p
<adisbladis>
sphalerite: It's way more clunky imho
<pxc>
sphalerite: tmux's reflow gets all messed up if your terminal has a right prompt
<sphalerite>
right prompt? eww
<adisbladis>
And I ran into issues with my right prompt
<adisbladis>
But it doesnt work well in emacs terminal emulators
<pxc>
adisbladis: me, too. I love fish. I think some day soon, Elvish is going to be even better
<pxc>
but Fish is great
<Synthetica>
... Seriously, Elvish?
<fogbugz>
sphalerite: yes, I've tracked down the issue to rPackages.SDMTools itself, but it's not obvious (even by grepping the source) which dependency is missing.
griff_ has joined #nixos
<pxc>
Synthetica: yeah, have you seen it?
<Synthetica>
I just nix-shelled it
<Synthetica>
typed `git [tab]`
<Synthetica>
No completions
<Synthetica>
~literal trash~
<pxc>
hahahaha
<pxc>
the version that comes in Nixpkgs isn't latest, and some good work on completions has just started in the (tiny) community
<Synthetica>
Ah, cool
<Synthetica>
I'll keep an eye on it, looks promising
<yorick>
elvish is really fast
<pxc>
the maintainer is super nice, and a former fish contributor
<yorick>
but I want emacs shell because the gui does what I want :P
<pxc>
I think he's drawing inspiration from all the right places
<Synthetica>
I'll try again once `thefuck` supports it
winem_ has quit [Read error: Connection reset by peer]
<Synthetica>
Can't live without that anymore :(
<yorick>
Synthetica: just use emacs for everything
<Synthetica>
I already use Electron-emacs
<pxc>
yorick: I haven't gotten into Emacs shell yet, but I feel like I'll have to at some point. Do any Emacs shells support truecolor?
<pxc>
Synthetica: wait, what's Electron-emacs?
<Synthetica>
Atom
<bkchr[m]>
Is `/run/current-system/sw/share/terminfo` rebuilded when I use a new nix?
winem_ has joined #nixos
<yorick>
pxc: it's emacs. there are patches and it's probably in it nowadays
<sphalerite>
bkchr[m]: it may be updated whenever you run nixos-rebuild switch
<sphalerite>
bkchr[m]: (or nixos-rebuild test, or nixos-rebuild boot and reboot, etc)
<yorick>
git push origin && fuck -> C-g P p RET
<pxc>
what's the point of electron-emacs? does it have some nice web content rendering features or something?
<yorick>
"I personally enjoy Atom's 250+ ms keyboard lag. That's something emacs can never do for me unless I'm using it over a slow SSH connection."
<bkchr[m]>
sphalerite: hmm okay :(
<sphalerite>
bkchr[m]: why?
<pxc>
yorick: hahahahaha! I'm inclined to think similarly. Surely there must be some advantage, though, or Electron-emacs wouldn't exist
<sphalerite>
what's the actual problem you're having?
<adisbladis>
yorick: Do you know of a decent emacs shell?
<yorick>
adisbladis: install emacs, type M-x shell
<bkchr[m]>
sphalerite: I updated nix to circument the following folder naming "x~nix~case~hack~1"
<adisbladis>
yorick: That shell sucks though
<bkchr[m]>
I use linux
<Synthetica>
pxc: it's just the editor I picked up when I started my study, haven't really found a reason to switch yet ¯\_(ツ)_/¯
<Synthetica>
I tried Vim for a bit, didn't expecially like it
<yorick>
adisbladis: escape codes are in M-x term
<yorick>
adisbladis: also try eshell
<Synthetica>
Intend to switch to Xi when it becomes usable
silver has joined #nixos
<pxc>
Synthetica: why not try GNU emacs?
<adisbladis>
yorick: Neither of them can render my prompt properly
<sphalerite>
bkchr[m]: oh right, you want to have those paths fixed? I think doing `sudo nix-store --repair-path /run/current-system` should hopefully fix it
<yorick>
adisbladis: it's not a real terminal, and it's not really meant to be
<Synthetica>
pxc: not sure tbh, mostly the learning curve
<pxc>
adisbladis: !! if that works well I might not use a normal terminal emulator at all anymore. All Emacs! :-D
<yorick>
it's annoying that emacs doesn't have true background transparency, but at least it handles fonts and dpi switching well
<Synthetica>
Can't you just `compton` the transparancy?
<yorick>
Synthetica: I can make the window transparent, this makes the text and images transparent
<Synthetica>
Oh, right
Neo-- has quit [Ping timeout: 268 seconds]
coot has joined #nixos
<pxc>
sphalerite, srhb: ok, this is weird. When I run zoom-us with `env LD_PRELOAD=...` and add libGL.so.1 and libEGL.so.1 from libGL manually, it works
smallville7123 has joined #nixos
<pxc>
but when I modify the package to prepend those to the LD_PRELOAD it already uses in its makeWrapper, zoom-us starts (the process stays open) but the window never appears
<pxc>
it doesn't output anything to the terminal, either, but I get terminal output when I launch with the manual LD_PRELOAD stuff
<sphalerite>
that is odd
<pxc>
I'm using Bumblebee on this laptop, with a config that worked on 18.03 and that I didn't modify
aw has joined #nixos
spacefrogg has joined #nixos
<{^_^}>
[nixpkgs] @Anton-Latukha opened pull request #40553 → handbrake: platforms: linux -> unix; rm cmake dep → https://git.io/vp5J1
iyzsong has quit [Ping timeout: 255 seconds]
<sphalerite>
oh yeah, bumblebee hasn't been working for a while for me
* Synthetica
is really glad they got a intel-only laptop
<pxc>
sphalerite: ok the thing that works is
hyper_ch2 has quit [Quit: Page closed]
<pxc>
hm nvm I just realized something
xcmw has quit [Ping timeout: 264 seconds]
<pxc>
my local nixpkgs checkout I was modifying the Zoom package in was still 18.03 -_-
<sphalerite>
Synthetica: I muse mine as an intel-only laptop usually ;)
<pxc>
isn't the AMD hybrid graphics situation better?
<sphalerite>
oh, my bumblebee issues are just because of a kernel/userspace driver version mismatch apparently
<sphalerite>
I haven't rebooted in a while, so I guess htat's the issue
<{^_^}>
[nixpkgs] @matthewbauer merged pull request #40519 → gcc,gcc8: {revert gcc to gcc7, mark gcc8 as broken} on Darwin → https://git.io/vpQB1
<{^_^}>
→ 836404df by @Synthetica9: gcc: reset default to gcc7 on darwin
<{^_^}>
→ 87fd200b by @Synthetica9: gcc8: mark as broken on darwin
<{^_^}>
→ 7feb7623 by @Synthetica9: gcc8: s/hostPlatform/stdenv/
<sphalerite>
but yeah screw nvidia for making our lives complicated.
<pxc>
I didn't have a lot of choices with this laptop; I wanted to grab the last okay-ish thinkpad before they ruined the keyboard forever
<srhb>
pxc: fwiw I already fixed that package in master.
<pxc>
but for everything else in my life I'm all AMD now
<pxc>
srhb: aaah thanks. I still have to figure out what I was doing wrong completely; I think I'm close
smallville7123 has quit [Quit: rosa]
<pxc>
I'll treat your patch like the back of the book, for when I'm done :-P
<Synthetica>
pxc: t470 keyboard is fine to me, but I'm not used to the old Thinkpad keyboards :P
<srhb>
It was just a quick hack, so that's probably not recommended :P
<{^_^}>
[nixpkgs] @dezgeg closed pull request #40539 → treewide: Remove uses of crossConfig and add strictDeps → https://git.io/vp7et
<sphalerit>
It's been 30 days :o
<sphalerite>
yep now bumblebee works again
<{^_^}>
[nixpkgs] @Ma27 opened pull request #40554 → nixos/statsd: refactor test → https://git.io/vp5IN
<__monty__>
So I'm trying to install nix on a mac. Ran the curl ... | sh and installation finished telling me to restart the shell but the nix commands aren't available, .nix-profile doesn't exist, etc. What step am I missing?
<tilpner>
Hey sphalerite, do you have a declarative weechat config I could steal?
<sphalerite>
__monty__: at risk of stating the obvious, did you restart the shell?
<pxc>
srhb, sphalerite: ok yeah, I fixed Zoom.us using the same fix that wasn't working before because I was doing it on nixos-18.03 nixpkgs instead of nixos-unstable, which my system is running. whoooooooops
<sphalerite>
tilpner: nope, sorry. Weechat is horrific.
<__monty__>
I quit terminal.app multiple times.
<sphalerite>
tilpner: no separation of config and state whatsoever.
<tilpner>
Yeah :/
<sphalerite>
pxc: \o/
<pxc>
I need to run nixos-unstable more so that I actually at least attempt to contribute to nixpkgs
<pxc>
I just never do anything unless my system is broken lol
<pxc>
srhb: the difference in our fixes is that I (gratuitously, it seems) used LD_PRELOAD instead of just adding libGL to the libPath. I think yours is better
<boxofrox>
sphalerite: i'm curious which irc client you prefer.
<pxc>
sphalerite: I'm using weechat now but I don't love it. Recommendation?
<pxc>
sphalerite: programs like that really drive home the difference between CLI and TUI applications
<boxofrox>
lol, fifo based irc. I might have to try that sometime
<gchristensen>
__monty__: for me because irssi takes more config than weechat, which mostly is just fine out of the box
<sphalerite>
__monty__: I'm not sure either actually. I switched at one point but I don't remember
<sphalerite>
but yeah the out-of-the-box setup is much nicer with weechat
<tilpner>
irssi has a weird config, that looks kind of like Nix data, but not really
<boxofrox>
__monty__: same reason as gchristensen
winem_ has quit [Ping timeout: 256 seconds]
<__monty__>
I just slowly customize so it doesn't matter that much to me. I get the convenience but I like having zero unnecessary crap and the easy (as in perl) customizability.
<adisbladis>
Imho erc has by far the best user interface ;)
<tilpner>
I've tried to like emacs a few times :/
<__monty__>
If I used emacs I'd definitely use erc or circe.
xAFFE has left #nixos ["Error from remote client"]
<adisbladis>
Though erc has some issues if you leave it running for a long time
<adisbladis>
Memory usage grows and scrollback buffers start getting sluggish
<__monty__>
Sounds like everything emacs : p
<adisbladis>
__monty__: I'm talking about runnig it for weeks now
<adisbladis>
My emacs sessions typically live only for a few days
<sphalerite>
well yeah that's what you do with IRC clients :D
<__monty__>
Even though I don't, I'd like to have the option : )
griff_ has quit [Quit: griff_]
MercurialAlchemi has joined #nixos
iyzsong has joined #nixos
jperras has quit [Ping timeout: 240 seconds]
iqubic has joined #nixos
<iqubic>
samueldr: The thing about the printer drivers you linked me too yesterday (the ones I was going to make an overlay for) is that they have hardecoded the printer model into the install script. While it makes sense for their usecase, I'll basically need to modify the install script along with the download location.
<iqubic>
Of course all of this assumes that this will even work in the first place, which it very well might not.
griff_ has joined #nixos
<iqubic>
Getting printing to work on NixOS is the largest pain that I have ever been through in my life.
<iqubic>
I'm trying to get the drivers for the Brother HL-L6300DW to work on NixOS
<__monty__>
Ok, found it. Apparently the install ignored the presence of zsh. Is nix install and co ready for use or should I stick to nix-env for now.
adisbladis has quit [Remote host closed the connection]
<iqubic>
No. But that's for the 6400, and not the 6300. Do you think that will work for me?
<goibhniu>
in the comments it mentions it works on various similar printers
<iqubic>
I have the 6300.
<{^_^}>
[nixpkgs] @yorickvP opened pull request #40555 → module-list: add youtrack module to list → https://git.io/vp5YK
<iqubic>
And is it possible to get that working for NixOS. I'm not at home right now, so I can't actually test and of this right now.
<iqubic>
I'm at school right now. I'll be home at about 14:45
<goibhniu>
you can try adding it manually via the cups web interface, that has worked for me
<samueldr>
iqubic: obviously, the suggestion was to use those as a base since they probably used similar structure for those packages
<iqubic>
Oh, I see.
<iqubic>
I'm going to try the openprinting drivers first, and then if that fails write the overlay for the printer drivers you recommended.
<iqubic>
samueldr: Once I write the overlays, how do I tell Nix that I want to use them as printer drivers?
<samueldr>
everything in an overlay works just like it was a normal package, so follow the normal nixos instructions for printing
<{^_^}>
[nixpkgs] @yegortimoshenko merged pull request #40555 → module-list: add youtrack module to list → https://git.io/vp5YK
<{^_^}>
[nixpkgs] @yegortimoshenko pushed 2 commits to master: https://git.io/vp5Yb
<{^_^}>
→ 261d9935 by @yorickvP: module-list: add youtrack module to list
<{^_^}>
→ 14d7d39d by @yegortimoshenko: Merge pull request #40555 from yorickvP/patch-3
<{^_^}>
[nixpkgs] @vbgl merged pull request #40552 → Add support for SSL in ocamlPackage.conduit. → https://git.io/vp7Fy
<{^_^}>
[nixpkgs] @vbgl pushed commit from @Zimmi48 to master « ocamlPackages.conduit-lwt-unix: add SSL support »: https://git.io/vp5YA
<iqubic>
I guess that makes sense. But what package name do I give to the list of printer drivers? Just use the overlay's file name?
lonokhov has quit [Quit: WeeChat 2.0]
Arcaelyx has quit [Read error: Connection reset by peer]
<iqubic>
But these are good things for me to try when I get home from school.
smallville7123 has joined #nixos
<samueldr>
the overlay file returns an attribute set, it would be the attribute name in that attribute set
dbmikus_ has quit [Quit: WeeChat 2.1]
<samueldr>
so if you used something like gary = self.callPackage ./gary.nix; (bogus example) you would use pkgs.gary
dbmikus has joined #nixos
<{^_^}>
[nixpkgs] @matthewbauer pushed to master « telnet: use inetutils »: https://git.io/vp5OP
<iqubic>
I see. So basically, because an overlay modifies a package, anywhere I want to use the overlay, I just use the name of the modified package?
<samueldr>
or the added package (you don't have to replace an existing one, and in your case, you should probably add one)
<iqubic>
Oh, I see. How do you add a package? Is there a guide for that anywhere.
unkn has quit [Quit: a]
<iqubic>
Several people here helped me understand overlays last night. I still don't have full intuition for the actually syntax and function calls needed.
<infinisil>
There's a presentation from nixcon 2017 by pierron about overlays which i think was pretty good
<iqubic>
inifinisil: I watched that last night actually.
<samueldr>
(I know I personally did do nix stuff without knowing stuff from chapter 14, and it was confusing)
<iqubic>
samueldr: That's an odd choice.
<sphalerite>
So the sound chip in my chromebook has an insane amount of controls, some of which are reportedly very much capable of frying the speakers. How do I deal with this manageably?
<boxofrox>
iqubic: chapters 12, 13, & 14 of nixpkgs manual also discusses overlays or developing packages for submission. https://nixos.org/nixpkgs/manual/
johann__ has quit [Quit: Leaving.]
Arcaelyx has joined #nixos
<iqubic>
I saw that last night. It's basically a text summary of pierron's 2017 NixCon talk, which I also saw.
<jtojnar>
cool, pulseaudio 12.0 RC1
<infinisil>
jtojnar: Any highlights?
<jtojnar>
infinisil: GConf dependency can be finally removed
<iqubic>
boxofrox: Chapter 12 is all about sharing packages between multiple machines.
<bkchr[m]>
sphalerite: thanks for your help with the nix-store repair, it seems to have fixed my problem :)
<iqubic>
Solution: Never use libraries. Copy and Paste relevent code.
Freneticks has quit [Quit: Leaving]
<iqubic>
^^^ the above is not a serious suggestion.
<Synthetica>
iqubic: I was gonna say, slow there Satan
<infinisil>
I'd like to explore the possibility of not using libraries (collection of functions), but individual functions instead, every one being "versioned" (but not really)
iyzsong has quit [Ping timeout: 240 seconds]
__Sander__ has quit [Quit: Konversation terminated!]
alex`` has quit [Ping timeout: 260 seconds]
<iqubic>
infinisil: So like, git/github for individual functions?
jperras has quit [Ping timeout: 276 seconds]
applePrincess has joined #nixos
<infinisil>
Hmm hard to say
<iqubic>
Why does a single simple package have to about 6 or 7 moving parts?
olynch has joined #nixos
<infinisil>
I could elaborate on my idea in #nixos-chat
<iqubic>
Is there a list of all the options that needs to be passed to mkDerivation that I can just read?
<das_j>
Hey everyone, is there an option to automatically fix the rpath for all binaries in a package instead of doing patchelf --set-rpath for each file?
<das_j>
Something that detects 'this requires libatk and there is a packge with libatk.so in the input path so I'll patch the rpath'
<iqubic>
Don't we already have a bot with that name that spews out github info for pull requests and things on the nixpkgs repo?
<infinisil>
iqubic: Yeah, {^_^} is a single frontend for unlimited backends, the github notifications being one of them, the one I coded being another one
srl295 has joined #nixos
<samueldr>
the channel updates notifications another one
Tucky has quit [Remote host closed the connection]
<pierron>
iqubic: buildFHSUserEnv is a different function, which is likely expressed in terms of mkDerivation, and will create a derivation which can be used as a shell which runs a program with a few extra mount-points, emulating an ordinary Linux distribution.
coot has joined #nixos
<iqubic>
I see. I understand how that works.
<pierron>
This is quite useful when dealing with unpatchable software.
<dtz>
das_j: idk if you're not on NixOS I've found it tricky to setup-- but definitely worth the trouble! err if you're referring to shell tab-completion anyway
<iqubic>
Is the vidyo overlay violating the rule where overlays are only supposed to depend on super and self?
<pierron>
iqubic: but this should be avoided in Nixpkgs.
<das_j>
dtz: I am, but I am currently trying to package Citrix software, which is a real pain
<iqubic>
Because it runs callPackage, which depends on a whole lot of things.
<das_j>
Nothing is even close to FHS
<pierron>
iqubic: it depends on super.callPackage.
<iqubic>
And it's fine that the package it is calling, with super.callPackage, depends on a whole lot of things? I thought one of the rules of overlays was to only depend on self and super.
<pierron>
iqubic: `super` provides the `callPackage` function, and the `callPackage` function is extracting from `self` the arguments needed to call the function from the Vidyo expression.
<nyanloutre[m]>
{^_^}: nice tool !
<infinisil>
It's a bit confusing
jperras has joined #nixos
<pierron>
iqubic: you should not depends on another imported version of Nixpkgs, yes. It is fine to have a custom library for your packages, or have separated files for the Nix expressions for individual packages, such that they might be integrated in Nixpkgs later on.
<neonfuz>
Anyone know a decent sheet music editor that's in nixpkgs?
<pierron>
infinisil: I hope you cap the memory / cpu time / running time / output size of these nix expressions ;)
<{^_^}>
[nixos-hardware] @yegortimoshenko pushed 3 commits to master: https://git.io/vp58X
<{^_^}>
→ b586e105 by @yorickvP: add XPS 13 (9360)
<{^_^}>
→ 0794564c by @yegortimoshenko: dell/xps/13-9360: review
<{^_^}>
→ fbf167ed by @yegortimoshenko: Merge pull request #58 from yorickvP/master
<infinisil>
,whomademe = The PR linking on #<number>, the "," prefix commands and the "> " nix eval have been implemented by infinisil, my source code can be found at https://github.com/infinisil/nixbot, the other functions of me have been implemented by other people
<{^_^}>
whomademe defined
<das_j>
So, I'm pretty desperate now. auto-patchelf.sh should automatically fixup my dependencies to "runtimeDependencies", but it doesn't. Not even the log message from the function is printed
<das_j>
Does anybody know why that would happen?
asuryawanshi has joined #nixos
simpson has joined #nixos
<goibhniu>
hi das_j, can you post your expression?
simpson has left #nixos ["WeeChat 1.0.1"]
elasticdog has quit [Ping timeout: 248 seconds]
<neonfuz>
wtf
<neonfuz>
so I wanted to write a nix package, so I copied default.nix from the hello package, and modified the source to point to the new url to start out
<neonfuz>
but when I do nix-shell and genericBuild, or nix-build it builds hello still...
<neonfuz>
I don't understand, I edit the nix file and it doesn't seem to know that I did...
Arcaelyx has quit [Read error: Connection reset by peer]
<jD91mZM2>
boomshroom: Did you mean: Good evening
<jD91mZM2>
:^)
<boomshroom>
jD91mZM2: Well it's morning for me. Good evening for you!
Arcaelyx_ has joined #nixos
ajs124 has joined #nixos
<boomshroom>
jD91mZM2: Hejsan!
pxc has joined #nixos
<jD91mZM2>
boomshroom: Where are you from btw?
<boomshroom>
jD91mZM2: Canada. My mom told me how to spell that. :P
<Notkea>
hello, can Hydra be installed in a nixos container? I see an issue saying that it didn't interacted well with the host nix store, but this was from two years ago
<jD91mZM2>
boomshroom: I would make a joke and write to you in your native language.... buuuuut
<boomshroom>
jD91mZM2: Which native language? My first language is English, and I don't know any German, Italian, Swedish, or Finish.
<jD91mZM2>
boomshroom: thatsthejoke.jpg
fendor has joined #nixos
<{^_^}>
[nixpkgs] @yegortimoshenko merged pull request #39534 → tracker-miners: switch to meson, some tweaks → https://git.io/vpChw
<{^_^}>
[nixpkgs] @yegortimoshenko pushed to master « tracker-miners: switch to meson, some tweaks (#39534) »: https://git.io/vp5zy
<{^_^}>
[nixpkgs] @thoughtpolice pushed commit from R. RyanTM to master « metabase: 0.29.0 -> 0.29.2 »: https://git.io/vp5ag
nur0n0 has joined #nixos
<{^_^}>
[nixpkgs] @LnL7 pushed commit from @oxij to release-18.03 « nixos: systemd.services.*.environment: fix type »: https://git.io/vp5Vf
asuryawanshi has quit [Ping timeout: 240 seconds]
<nur0n0>
is there a canonical way to tell you are already in a nix-shell? maybe through env? I have a program that sets up a nix-shell before doing its core job, but setting up the nix-shell introduces a significant delay. Some sort cache would be the best solution but AFAIK that is not implemented (yet?)
iqubic has quit [Ping timeout: 260 seconds]
<boomshroom>
nur0n0: I'm pretty sure nix-shell sets an environment variable.
<symphorien>
nur0n0: there is the env var IN_NIX_SHELL
tmaekawa has joined #nixos
szicari has quit [Quit: szicari]
<nur0n0>
symphorien: thanks! that's exactly what I was looking for
<nur0n0>
is there any plans for some sort of cache? maybe there is a good reason why we don't have one, or maybe no one has implemented it yet. the delay is acceptable for long running processes, but my tool is meant to be an interactive cli tool.
<samueldr>
depending on what the shell uses as source, it may be possible that it picks file changes if e.g. it uses anything from the CWD as build inputs
<samueldr>
e.g. ./. and there is a log file there, or even the result symlink of nix-build
<johnw>
pierron: although, I thought we were supposed to use super for "the package we're updating"
<johnw>
pierron: then you must mean just that my dependencies for that package should be from self?
<nur0n0>
jD91mZM2: sure. /nix/store does store the packages. practically, the job of nix-shell is to make a certain set of binaries available in your environment. but when you point nix-shell at a default.nix file it goes through the trouble of re-evaluating a set of nix files (not sure what that is about, I just see 'evaluating file' when I run `nix-shell
<nur0n0>
-v`). from an user's perspective, I just want default.nix -> set of binaries. given a default.nix file, the set of binaries remains static. so why is it redoing work when the output is the same?
<nur0n0>
from my limited perspective it seems that the only necessary work should be to set the PATH accordingly, not to re-evaluate nix files
lord| has quit [Quit: WeeChat 2.1]
<nur0n0>
maybe there is a good reason why it is not done this way?
kreisys has quit [Ping timeout: 255 seconds]
<gchristensen>
how can you know what the binaries are without evaluating the nix?
<boomshroom>
nur0n0: It has to evaluate the Nix expressions to determine what should go in the path.
<nur0n0>
yes, but if the default.nix file does not change, why does it need to re-evaluate?
<gchristensen>
what if any of the things default.nix depends on?
szicari has joined #nixos
<boomshroom>
nur0n0: If you know the .drv file that it resulted in, you probably could create a shell just from that.
<{^_^}>
→ 36a4e543 by @Synthetica9: atom: add myself as maintainer
<{^_^}>
→ 87b0afb0 by @Synthetica9: atom: 1.26.1 -> 1.27.0
<{^_^}>
→ 80fd6cff by @xeji: Merge pull request #40563 from Synthetica9/atom-1-27-0
szicari has joined #nixos
<nur0n0>
boomshroom: so I should be pointing nix-shell at the .drv file, not at the default.nix file? that makes more sense, I can't seem to find such a command in the nix-shell man pages, but I'll look online
<fogbugz>
Hi, how can I alter the compilation flags used by GCC, using a shell variable? I'm trying to debug a problem with some R packages nix compiles.
<boomshroom>
nur0n0: nix-shelll by default takes a filepath argument that gets substituted with default.nix or shell.nix if it's omitted. `nix-shell /nix/store/...drv` should be enough provided the path exists.
<nur0n0>
boomshroom: ok, that would be the ideal solution. conceptually, it is cleaner to just re-evaluate everything from scratch. in practice, I think deteriorates the usability (especially on a weak machine like mine!)
<jD91mZM2>
How does one bypass string interpolation in nix?
<sphalerite>
"bypass" it?
<jD91mZM2>
I'mthinking something like \${test} but that doesn't work
<{^_^}>
→ decaeb1b by @dtzWill: nano: 2.9.4 -> 2.9.6
<jD91mZM2>
infinisil: Yeah well how else did you intend on solving the 2038 problem? /s
<{^_^}>
→ 0f30db42 by @dtzWill: nano: 2.9.6 -> 2.9.7
<{^_^}>
→ 4046c49f by @dtzWill: Merge pull request #40562 from dtzWill/update/nano-2.9.7
<sphalerite>
I think you probably could reset it by doing rm /nix/var/nix/profiles/system-* then rebuilding. But YMMV, this may break your system so don't do it if you don't want to spend an hour fixing it afterwards :p
<tilpner>
D:
<infinisil>
jD91mZM2: True dat!
MercurialAlchemi has quit [Ping timeout: 248 seconds]
<infinisil>
> builtins = { currentTime = 0; }
<{^_^}>
builtins defined
<samueldr>
wasn't there an OS that used 2000 as its epoch?
<infinisil>
> builtins.currentTime
<{^_^}>
0
<infinisil>
There we go
dbmikus has joined #nixos
<etu>
sphalerite: well, in theory you could just boot a live-image and rename the system-* symlink and change a boot-entry? :p
<infinisil>
Btw don't use the :d command a lot, it can break my nixbot
<boomshroom>
And everything was right with the world again.
<infinisil>
I think
robstr has quit [Quit: WeeChat 1.9.1]
dbmikus has joined #nixos
sigmundv_ has joined #nixos
hotfuzz_ is now known as hotfuzz
balsoft has joined #nixos
reinzelmann has quit [Quit: Leaving]
sigmundv_ has quit [Ping timeout: 276 seconds]
<etu>
Well, that was easy... :)
<iqubic>
What does :d do?
dbmikus has quit [Quit: WeeChat 2.1]
<infinisil>
iqubic: Undefine variables
<iqubic>
I see.
<infinisil>
I should either fix the semantics of my bot or just remove it
<infinisil>
the :d command I mean
knupfer has joined #nixos
<{^_^}>
[nixpkgs] @etu opened pull request #40565 → nixos/gitea: Enable configuring of derivation and a systemd timer for dumping of configs → https://git.io/vp5Pi
<etu>
Well, that was thatn
<iqubic>
Is there a good reason to keep :d around?
xy2_ has quit [Ping timeout: 240 seconds]
hhomar has joined #nixos
<infinisil>
Well it was useful just now, but ideally you shouldn't be able to redefine builtins
<infinisil>
Other than that I can't see a reason to keep it
<{^_^}>
[nixpkgs] @matthewbauer merged pull request #40553 → handbrake: platforms: linux -> unix; rm cmake dep → https://git.io/vp5J1
<{^_^}>
[nixpkgs] @matthewbauer pushed 3 commits to master: https://git.io/vp5XT
<{^_^}>
→ 56149c24 by @Anton-Latukha: handbrake: rm dependency on cmake
<{^_^}>
→ b54f76e8 by @matthewbauer: Merge pull request #40553 from Anton-Latukha/handbrake-platforms-cmake
<{^_^}>
→ 1eea58eb by @Anton-Latukha: handbrake: platforms: linux -> unix
jnoah has quit [Ping timeout: 268 seconds]
<{^_^}>
[nixpkgs] @matthewbauer merged pull request #40234 → Do not kill udev during boot 17.09 → https://git.io/vpXey
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to release-17.09: https://git.io/vp5Xs
<{^_^}>
→ 41307a7b by @ngortheone: Fixes #39867
<{^_^}>
→ 73d231c4 by @matthewbauer: Merge pull request #40234 from ngortheone/release-17.09
asuryawanshi has quit [Remote host closed the connection]
MichaelRaskin has joined #nixos
<{^_^}>
[nixpkgs] @matthewbauer merged pull request #39978 → Make plymouth work with gdm wayland → https://git.io/vpV8G
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/vp5X8
<{^_^}>
→ 5777272b by @hedning: nixos/plymouth: multi-user.target wants plymouth-quit-wait.service
<{^_^}>
→ 2a3399b3 by @matthewbauer: Merge pull request #39978 from hedning/plymouth-quit-wait
asuryawanshi has joined #nixos
ckauhaus has quit [Quit: Leaving.]
carlmare has joined #nixos
infinisil has quit [Quit: Configuring ZNC, sorry for the join/quits!]
infinisil has joined #nixos
<carlmare>
Hi, I'm trying to build the FeedReader 2.2 and I'm getting this error "UnicodeDecodeError: 'ascii' codec can't decode byte 0xc2 in position 922: ordinal not in range(128)" Here's a full paste: https://paste.sh/A6HOWieR#H51R8wZles71ynyj8a1g5y6M
<boomshroom>
Has anyone managed to build shofel (Nintendo Switch Linux launcher) on NixOS?
xy2_ has joined #nixos
<carlmare>
gchristensen: Thank you so much. Now I just need to find where :)
<boomshroom>
When using `(import <nixpkgs> { crossSystem = lib.systems.examples.armv7l-hf-multiplatform; }).stdenv.cc` as the compiler, I get `common.h:60:1: sorry, unimplemented: Thumb-1 hard-float VFP ABI`
humanoyd has quit [Quit: WeeChat 2.1]
drakonis has quit [Remote host closed the connection]
BigMassive has quit [Ping timeout: 255 seconds]
orivej has quit [Ping timeout: 265 seconds]
sary has quit [Remote host closed the connection]
<boomshroom>
shofel2 recommends using the arm toolchains provided by Linaro, but I'm trying to use Nix toolchains.
pxc has quit [Ping timeout: 256 seconds]
infinisil has quit [Quit: Configuring ZNC, sorry for the join/quits!]
balsoft has quit [Quit: balsoft]
infinisil has joined #nixos
<alexteves>
can I use requireFile on a directory? I did `nix-store --add-fixed sha256 --recursive myDir` and got the sha256 in base32, but it's not taking
<alexteves>
can only get it to work by taring myDir
iqubic has quit [Quit: Page closed]
<alexteves>
and doing nix-store --add-fixed on that
<alexteves>
er, I mean that if I do both `nix-store --add` the .tar then `nix-store --dump`to get the hash, requireFile works
<alexteves>
but if I do the same things for the dir, requireFile doesn't seem happy
<alexteves>
I'm the one writing (well, trying to) requireFile
saati_ has joined #nixos
xcmw has joined #nixos
<infinisil>
Ah
<{^_^}>
[nixpkgs] @Ericson2314 opened pull request #40566 → cmake: Fix build and setup hook → https://git.io/vp5D3
<infinisil>
What's the error you get?
<infinisil>
Wrong hash or does the directory add not even work?
civodul has joined #nixos
<alexteves>
the add works, I can browse its contents inside the /nix/store, and the hash from nix-store --dump is the same as running nix-hash on the original directory
<samueldr>
boomshroom: you forgot I did?
<alexteves>
but requireFile just spits its generic "can't find file" message
<pxc>
jtojnar: hey!! Thanks for all your work on Flatpak support! Obviously, Nixpkgs is the One True Way™, but Flatpak support is going to make a huge difference in making NixOS an even more viable desktop distro for many people
<jtojnar>
✌
<bebarker>
Is it expected that after doing `nix-env --set /nix/store/.../something.drv, a subsequent call like `nix-env -i tree` will make it so the only binary in one's environment is `tree`? previously this was mutable in the sense that `tree` would temporarily be added to the existing environment specified by `nix-env -if foo_env.nix`
<clever>
bebarker: --set isnt compatible with -i, you cant mix the 2
<clever>
bebarker: -i expects to find a manifest.nix in the profile, that details the contents
pxc1 has joined #nixos
<bebarker>
clever, ok - I'm not sure I tried it, do you know off hand if I can do something like nix-env --if /nix/store/.../foo.drv? will try as soon as I rollback
<clever>
i know you can do nix-env -i /nix/store/hash-name
<clever>
ive not tried it on a .drv though
pxc has quit [Ping timeout: 256 seconds]
das_j has quit [Quit: Page closed]
<boomshroom>
samueldr: It looks like your overlay requires cbfs.bin and other stuff, but doesn't build them.
<samueldr>
ah, that's the bit I couldn't get to buil
<samueldr>
though it's all documented what's needed
<samueldr>
(there's support for all the tethered host stuff, none for the target device)
<samueldr>
the kernel is missing two things IIRC, but nothing too hard, just haven't got around to implementing them
hhomar has quit [Quit: leaving]
<samueldr>
the MTC thing for ram timings (iirc)
szicari has quit [Quit: szicari]
<boomshroom>
clever: Is your QEMU Nix working?
<samueldr>
I was stumped at coreboot and stopped since I'd need to have serial to test it anyway
<samueldr>
but coreboot may need serious work to build
<clever>
boomshroom: last i checked its working fine
<bebarker>
clever, it appears not as far as I know; I'll just use nix-env -if from the start, and continue to run nix-env --set before that just to check that the API doesn't change, in case I need it
<{^_^}>
[nixpkgs] @alexfmpe opened pull request #40568 → Allow recursive hash for requireFile so it can support directories → https://git.io/vp5SP
<{^_^}>
[nixpkgs] @matthewbauer pushed to master « pspp: provide default icons & settings »: https://git.io/vp5Q6
balsoft has quit [Quit: balsoft]
fendor has quit [Quit: Leaving]
<boomshroom>
The magic of clever's QEMU patch: spoofing the local system and having it work anyways.
<boomshroom>
I currently have 3 nix's running plus a nix-shell. That's 4 things being built. One if GCC, another is Linux.
balsoft has joined #nixos
<joko>
clever: I noticed on your nixos config repo that your netboot server supports now partially uefi
<joko>
Why partially, is anything missing?
<clever>
joko: i havent been able to get any uefi system to actually boot the uefi files that the dhcp serves
<{^_^}>
[nixpkgs] @Mic92 pushed 3 commits to release-18.03: https://git.io/vp57A
<{^_^}>
→ bbb46b14 by @Mic92: nix-review: 0.1.0 -> 0.1.2
<{^_^}>
→ 303cb2e5 by @Mic92: nix-review: init at 0.1.0
<{^_^}>
→ 380d7d09 by @Mic92: nix-review: 0.1.2 -> 0.2.0
<joko>
clever: ok, I think the issue is with dhcpcd, I have a similar system with dnsmasq
<joko>
which works
<joko>
But couldn't make it with dhcpcd
<clever>
joko: what differs in how they reply? can you capture both with tcpdump -w and then compare them in wireshark?
<joko>
I will do so tomorrow, it's in my work and I don't have remote access
<clever>
ah
<joko>
The other thing missing was initrd=initrd as kernel boot argument
<joko>
I think gchristensen mentioned it here sometime ago
<joko>
Without it I was getting kernel panics
infinisil has quit [Quit: Configuring ZNC, sorry for the join/quits!]
broccoli_ has quit [Ping timeout: 264 seconds]
infinisil has joined #nixos
lord| has joined #nixos
<infinisil>
Hmm.. I'm not sure what to do with my nixops state
<clever>
infinisil: ?
<infinisil>
For now I committed it to my private git repo, which I can sync between machines
balsoft has quit [Read error: Connection reset by peer]
<clever>
ah
<clever>
i just only ever deploy from 1 machine
<infinisil>
But it updates on every deploy..
balsoft has joined #nixos
<sphalerite>
boomshroom: the nix patch is upstream now btw :)
<boomshroom>
sphalerite: Cool!
<sphalerite>
not in a release yet, and there's some talk of removing it again, but it's there for now :p
<infinisil>
Maybe it would still work if I just don't commit the new state everytime, I think it just updates the /nix/store/<hash>-nixops-machines part
<sphalerite>
nixops state seems like a potential use case for something like syncthing
<infinisil>
I'll do everything to stay away from syncthing, not sure why
<infinisil>
But I heard it has a GUI and I don't like those
<infinisil>
Especially for seemingly simple stuff such as syncing
<sphalerite>
Because it's go and only provides a web interface and JSON API?
<joko>
infinisil: you just set it once and then you don't have to use it
<joko>
But what about the state? Since it's a private repo, you could include the state, too
<infinisil>
What I should really do is finally implement my zfs sync tool, which would be exactly what I need, and be faster than syncthing
<infinisil>
joko: Yeah but it updates on every deploy
<joko>
Even use git-crypt for added security
<{^_^}>
[nixpkgs] @Mic92 pushed commit from @jbgi to release-18.03 « Add bash to jira PATH: required by health checks. »: https://git.io/vp5dO
broccoli_ has joined #nixos
<infinisil>
I don't want to commit the new state every time I deploy
<{^_^}>
[nixpkgs] @vbgl opened pull request #40571 → coqPackages.stdpp: init at 1.1 → https://git.io/vp5dZ
<sphalerite>
What's nice about syncthing is that it's opportunistic — it will use a LAN connection when possible, will sync between a pool of devices even if they're offline half the time
lord| has quit [Quit: WeeChat 2.0]
<sphalerite>
only providing a web interface is the only disadvantage it has really IMHO
<infinisil>
I can tell syncthing that it should try ip 192.168.1.25 first and then try another one?
<sphalerite>
yes. Or it can work that out for itself.
<boomshroom>
Now I just need to build cbfs.bin
<infinisil>
And I don't like merge conflicts
<infinisil>
Syncthing has those afaik
<sphalerite>
There's no way to avoid those, except never modifying the file in two places "at once"
<infinisil>
Yeah, that's exactly what my tool would do
<sphalerite>
how do you achieve that?
<sphalerite>
read-only files that are only modified through your tool?
<infinisil>
Nah, zfs can set a dataset as readonly
<infinisil>
Then it acts as if it's a ro mount
<sphalerite>
yes so the files are read-only.
<infinisil>
Yup
<sphalerite>
that seems like a pain.
<manveru>
is anyone working on the ff 60 update?
<infinisil>
I can explain further in #nixos-chat
balsoft has quit [Remote host closed the connection]
<manveru>
syncthing never really worked for me... i moved all my stuff to seafile instead :P
<manveru>
syncthing had a crazy amount of conflicts, plus every other version is incompatible with the previous ones, so you have to keep everyone updated
asuryawanshi has quit [Ping timeout: 276 seconds]
rihards has quit [Quit: rihards]
karlguy has joined #nixos
<sphalerite>
keeping stuff up to date is a good idea :p
<infinisil>
manveru: Whew, I almost thought of using syncthing
broccoli_ has quit [Ping timeout: 260 seconds]
<infinisil>
Yeah conflicts are a no-go for me
<sphalerite>
I don't know what you're using it for to get conflicts
<manveru>
sphalerite: i couldn't keep my friends up to date, with whom i shared some folders
balsoft has joined #nixos
<sphalerite>
I use syncthing for syncing music and photos between my laptop and phone, and have never encountered a conflict with that
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<samueldr>
boomshroom: if you make headway in getting coreboot built, pleas tell me :)
<manveru>
i'm not sure, it just had conflicts even for files i never even touched
<infinisil>
Should I tackle rewriting the znc module for nixos?
<infinisil>
It's definitely in need of one
<gchristensen>
joko: indeed, seems mandatory for uefi
sigmundv_ has quit [Ping timeout: 268 seconds]
sigmundv_ has joined #nixos
asuryawanshi has joined #nixos
<infinisil>
Hey does anybody have a good emacs config for nix indentation?
<infinisil>
The default nix-mode always messes it up for some reason
<{^_^}>
[nixpkgs] @xeji pushed 1000 commits to cpan-update: https://git.io/vp5F7
<{^_^}>
→ d21ff871 by @Synthetica9: gcc8: Removed java as a target language
<{^_^}>
→ ae4fe67c by @jtojnar: open-pdf-presenter: remove
<{^_^}>
→ 3ae3c0c0 by @jtojnar: python2.pkgs.popler—qt4: remove
<{^_^}>
→ 203d8dfb by @volth: [cpan2nix] perlPackages.Appcpanminus: 1.7043 -> 1.7044
<{^_^}>
→ 9fe92199 by @volth: [cpan2nix] perlPackages.JSON: cleanup
<{^_^}>
→ 5133812f by @volth: [cpan2nix] perlPackages.CPANPerlReleases: 3.50 -> 3.56
<nh2[m]>
dtz: hey, I saw your work on musl stuff. I'm a bit unclear on how I can start playing with it though. I see you have a binary cache, but how does one start writing derivations that use it?
<{^_^}>
[nixpkgs] @the-kenny pushed to master « pythonPackages.podcats: Init at 0.5.0 »: https://git.io/vp5Nf
balsoft has quit [Quit: balsoft]
<{^_^}>
[nixpkgs] @bkchr opened pull request #40572 → gcc7: Enable arm erratum 843419 on aarch64 → https://git.io/vp5Ng
<boomshroom>
Who could have guessed that Linux takes forever to cross-compile?
<symphorien>
the default nixos kernel config already takes hours to compile normally
<sphalerite>
boomshroom: it shouldn't take that long…
<sphalerite>
the default nixos kernel takes about 40min to compile on my 2011 laptop
Have-Quick has quit [Quit: Have-Quick]
<samueldr>
boomshroom: cross compiling using cross compilers or compiling using the qemu trick?&
<symphorien>
oh it takes about two hours on mine
<boomshroom>
samueldr: Cross-compiler. I'm going to wait for a reboot before testing the qemu-kludge.
<samueldr>
:)
<samueldr>
qemu-kludge is going to be slower
<samueldr>
much slower, pretty sure
<samueldr>
never got around to check the state of cross compiling on nixos, any good pointers?
<samueldr>
I was under the impression that the resulting hashes would differ from native compilations? how much an inconvenience it is if I want to mix cross-compilation with native builds on one system?
<clever>
samueldr: you can also cheat some by running the native compiler under qemu (either system or user), but then you loose performance
<samueldr>
yes, I did it in the past already (worked fine!) though I don't know what issues I may face if I mix native and cross compilation on one target system
<clever>
samueldr: in general, it would probably have 2 copies of all libs, and have to re-build them natively
<samueldr>
that's a thing I don't even know what to search for :/
<{^_^}>
[nixpkgs] @xeji pushed to release-18.03 « nixos/tests/kafka: fix and refactor tests »: https://git.io/vp5xf
<clever>
it will behave as if they had come from different nixpkgs revisions
<sphalerite>
I'd really like to have mixing working nicely
Neo-- has joined #nixos
<samueldr>
ah, so pretty much as I thought :/ at least I got that right
romildo has joined #nixos
<sphalerite>
like have an overlay of sorts that composes native packages with cross packages
c0ffee152 has joined #nixos
<sphalerite>
which can then be built on an ARM machine that has an x86_64 remote builder configured
hamishmack has joined #nixos
<samueldr>
so, when using cross, I assume it's best to treat it as if it was some kind of buildroot-built system? how would I handle a "nixos-rebuild ... --upgrade" like scenario with a cross-built system?
<sphalerite>
I don't think that's really possible currently
<samueldr>
ah, so probably the easiest would be to have it as a stateless built root system, and I replace it all at once
<samueldr>
with mounts for state where needed
tertle||eltret has joined #nixos
hiratara has quit [Ping timeout: 276 seconds]
<sphalerite>
for audio to work properly on my chromebook, I'd need to put an updated ALSA UCM file in. However, I'd like to avoid modifying alsaLib (and causing a bunch of rebuilds) itself. Any good way to do this?
<samueldr>
nothing :) simply showing that in a way, information can be queried
<samueldr>
(like clever did)
<iqubic>
I see. I was actually asking for a way to search and see if a package exists or not.
<iqubic>
Or see packages relating to "Printing" or "Emacs" or something.
tertle||eltret has joined #nixos
<clever>
if the attribute name is predictable, then you can try to just eval it
inabsentia has joined #nixos
<inabsentia>
Hi, I'm trying to install NixOs over Arch using the lustrate install method (https://nixos.org/nixos/manual/index.html#sec-installing-from-other-distro) and am stuck on step 13 running `switch-to-configuration boot`. I get no output except for a perl warning about locales. Anyone able to help?
<clever>
inabsentia: has boot.loader.grub.device been set in the configuration.nix?
<inabsentia>
yes
<inabsentia>
and there is now a nix directory in the boot partition
commander has quit [Remote host closed the connection]
<inabsentia>
but my system doesn't recognise any linux distro or windows anymore
<iqubic>
After you did step 13, did you reboot?
<inabsentia>
yes
<iqubic>
And what was the result?
<inabsentia>
got Dell SupportAssist
<inabsentia>
scanning for errors then shutting down
<iqubic>
I see. Do you have a back-up linux usb/cd you can boot from?
thblt has quit [Ping timeout: 260 seconds]
<inabsentia>
worried `switch-to-configuration boot` didn't work properly as I was expecting a success message instead of just perl warnings
<iqubic>
You were right. It didn't work correctly.
<inabsentia>
yes, I have booted with my live arch usb
halfbit has quit [Ping timeout: 240 seconds]
rain1 has joined #nixos
<iqubic>
clever: You seem more capable of helpping. Would you like to take over?
<inabsentia>
I tried to solve the perl locales errors in case it helped the command run, but haven't been successful
<inabsentia>
*warnings
jperras has quit [Ping timeout: 240 seconds]
c0ffee152 has quit [Ping timeout: 264 seconds]
<sphalerite>
inabsentia: I'm guessing this is an EFI system. This may just be a matter of adding the appropriate EFI boot entry
<inabsentia>
clever: yes just wanted to update that I have boot.loader.systemd-boot.enable = true
<inabsentia>
not boot.loader.grub.device like you asked
lord| has joined #nixos
<iqubic>
inabsentia: It should work either way.
<sphalerite>
inabsentia: try setting boot.loader.efi.canTouchEfiVariables = true; if you haven't already
<iqubic>
And if that line is missing, then go ahead and add it.
Ariakenom has quit [Quit: Leaving]
<inabsentia>
I have that line and also boot.loader.grub.useOsProber = true since I'm dual booting windows 10
<iqubic>
So aside from one being a gui app and the other being cli app, are there any differences between pkgs.keepass and pkgs.kpcli?
c0ffee152 has joined #nixos
<iqubic>
I assume both are fine ways to interface with keepass databases, right?
<infinisil>
That's something i want nixbot to do too eventually
<infinisil>
iqubic: xclip
<infinisil>
(duh)
logzet has quit [Remote host closed the connection]
<iqubic>
I wasn't sure.
<johnw>
can I override inNixShell and force it to be false, even when I'm in a nix-shell?
<johnw>
I want to import in a default.nix that uses developPackage, which returns a different result if inNixShell = true, and I want to access the result it returns for inNixShell = false
<iqubic>
Well this stinks...
erasmas has quit [Quit: leaving]
<iqubic>
Actually, I was able to fix my issue.
sanscoeur has quit [Ping timeout: 248 seconds]
c0ffee152 has joined #nixos
stumble has joined #nixos
<infinisil>
johnw: Not sure if this works for you, but with my recently merged #38522 you should be able to change lib functions, and inNixShell comes from lib
<infinisil>
I'm not entirely sure how that lib gets into pkgs
<johnw>
infinisil: oh right, it's being loaded later
<infinisil>
Yeah something like that
<infinisil>
Um, now here is something super odd with the module systems typing system
<johnw>
i don't have a setEnv either
<johnw>
is there any way, within the Nix expression being evaluated by '-p', to tell it that I don't want to be considered "in a shell" during that evaluation?
<infinisil>
> typecheck (lib.types.attrsOf lib.types.str) { x = null; }
<{^_^}>
value is a set while a list was expected, at /var/lib/nixbot/state/nixpkgs/lib/types.nix:269:32
<infinisil>
But this error
<infinisil>
Is just plain misleading and wrong
pako_ has joined #nixos
<infinisil>
It seems that for all nested types the error is the same (value is a set while a list was expected) on any error
<infinisil>
Oh
<infinisil>
Never mind
ThatPako has quit [Ping timeout: 248 seconds]
<infinisil>
The first argument to type.merge is the problem.. It's supposed to be a list
<infinisil>
How could I miss this..
daveo has quit [Remote host closed the connection]
joehh has joined #nixos
jperras has joined #nixos
<Dezgeg>
gchristensen: huh yeah, sounds possible
coot has quit [Quit: coot]
joehh1 has joined #nixos
jperras has quit [Ping timeout: 248 seconds]
joehh2 has joined #nixos
joehh has quit [Ping timeout: 256 seconds]
joehh3 has joined #nixos
__monty__ has quit [Quit: leaving]
joehh1 has quit [Ping timeout: 256 seconds]
joehh4 has joined #nixos
joehh2 has quit [Ping timeout: 255 seconds]
joehh has joined #nixos
joehh3 has quit [Ping timeout: 256 seconds]
<palo>
nix-shell -p haskellPackages.haskell98 is not compiling I need it for some old packages. I'm just supprised it's failing with that exception. But it's not critical