<srhb>
wilornel: If you followed the instructions you ought to be on some 2.0 version (possibly pre, if on 17.09) if your channel is up-to-date
kelleyNif has joined #nixos
<ottidmes>
wilornel: nix.package = (import <nixpkgs-unstable> { config = { allowUnfree = true; }; }).nixUnstable; should give you an even more recent version (the <nixpkgs-unstable> looks for nixpkgs-unstable on the NIX_PATH, not sure if you have that though, you can also just inline a path to a checkout that uses channels/nixos-unstable)
<clever>
ottidmes: shouldnt need allowUnfree
<ottidmes>
The config = { allowUnfree = true; }; bit can be ignored
<ottidmes>
clever: Yeah, was just writing that, (I just copied parts of my config, should have removed that bit)
<clever>
ah
vandenoever has quit [Ping timeout: 276 seconds]
<ottidmes>
I have unstable = import <nixpkgs-unstable> { config = { allowUnfree = true; }; }; in my nixpkgs overlay and just refer to unstable.<pkg-name> all over where I happen to need a more bleeding edge version, some of which might be unfree, hence the config
kelleyNif has quit [Client Quit]
kelleyNif has joined #nixos
<wilornel>
srhb: how do I update my channel? so 1.12pre IS 2.0? `nix-env -qaP | grep bundler` still gives me version 1.14.6 and not 1.16
lopsided98 has joined #nixos
mekeor has joined #nixos
<wilornel>
I'm confused, why should I do your method ottidmes if nix1.12pre is 2.0?
<wilornel>
error: file ‘nixpkgs-unstable’ was not found in the Nix search path (add it using $NIX_PATH or -I), at /etc/nixos/configuration.nix:29:23
<wilornel>
(use ‘--show-trace’ to show detailed location information)
BlessJah has quit [Ping timeout: 245 seconds]
lopsided98 has quit [Ping timeout: 245 seconds]
<clever>
wilornel: you need to add a channel called nixpkgs-unstable with nix-channel, as root
Neo-- has quit [Ping timeout: 265 seconds]
<ottidmes>
I really feel like I should know this by now, but I am getting "Host key verification failed.", which is because it is missing from known_hosts, but it is not prompting to add it, is just exits with that error
<bsima>
How can I check if a library is installed on NixOS and available in my current env?
<johnw>
bsima: that's a really great question, actually
<bsima>
ldconfig doesn't seem to work
<clever>
ottidmes: which user is causing that error?
<clever>
bsima: ldconfig isnt supposed to work on nixos
<MichaelRaskin>
Libraries are expected not to be available by installing
<MichaelRaskin>
Inside nix-shell — that's another story
<bsima>
MichaelRaskin: yeah, i'm inside nix-shell
<ottidmes>
clever: root
<clever>
ottidmes: when you just run ssh normally?
<ottidmes>
clever: yeah, ssh root@mymachine fails with the host key error
<clever>
ottidmes: can you gist the full error?
<ottidmes>
clever: Normally I would get a prompt asking me to add it to the known hosts
<bsima>
I'm actually trying to reproduce a bug which seems to be cauesd by a missing lib
<MichaelRaskin>
bsima: we-e-e-ell
Arcaelyx has quit [Client Quit]
<MichaelRaskin>
I mean, store enumeration, propagated build inputs, LD_LIBRARY_PATH from external shell…
<ottidmes>
clever: Thanks, that helped me realize what likely the problem is, I have "BatchMode yes", which I read does basically the same as some other options I had combined, but I think it has other side effects, including this one, i.e. becoming non-interactive
<bsima>
MichaelRaskin: how about nix-shell --pure?
BlessJah has joined #nixos
<MichaelRaskin>
That helps with inherited stuff
<MichaelRaskin>
And propagated build inputs should work the same for the Nix build
<bsima>
ok
<MichaelRaskin>
(and store enumeration is just something I hope nobody ever tries to do automatically)
erasmas has quit [Quit: leaving]
<clever>
ottidmes: ah, that feels like it would do it
kelleyNif has quit [Quit: Leaving.]
<manveru>
is there a channel for home-manager?
<MichaelRaskin>
Why would it need one?
<ottidmes>
clever: It actually is appropriate for my use case, but I would have to generate my own known_hosts file, or only enable the option after populating the file with the necessary entries
<manveru>
just wondering :)
mekeor has quit [Ping timeout: 256 seconds]
<clever>
ottidmes: i think you can also use -O to set the flag on the CLI
kelleyNif has joined #nixos
<clever>
ottidmes: so scripts get it, but manual ssh doesnt
<ottidmes>
clever: ssh -o BatchMode=yes probably
<clever>
yeah
<ottidmes>
I of course had it in my programs.ssh.extraConfig... cannot recommend :P
<NixOS_GitHub>
[nixpkgs] Ma27 opened pull request #36234: uutils-coreutils: disable general unix build for now (master...disable-uutils-coreutils-on-darwin) https://git.io/vASFq
NixOS_GitHub has left #nixos [#nixos]
ottidmes has quit [Ping timeout: 256 seconds]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master ae7efee aszlig: systemd: Reident the expression file...
<NixOS_GitHub>
[nixpkgs] aszlig pushed 1 new commit to master: https://git.io/vASFZ
<abathur>
to hack on a local copy of Doxygen I had before moving to nixos--since Doxygen happens to already be in nixpkgs--I guess I should be writing a shell/default.nix in the directory that overrides the doxygen derivation in nixpkgs? probably to override the src attribute and add an env?
<halfbit>
so is there a way to make like the equivalent of a PKGBUILD for archlinux or.spec file for rpm or something
<halfbit>
I still can't understand how I can package my own stuff, it seems like the general guide says git clone all of nixos and go from there
<halfbit>
that seems insane to me, the git repo is huge
<clever>
halfbit: you can just make your own default.nix
<clever>
halfbit: and import <nixpkgs> {} loads the nixpkgs you already have
<wilornel>
error: file ‘nixpkgs-unstable’ was not found in the Nix search path (add it using $NIX_PATH or -I), at /etc/nixos/configuration.nix:29:23
<wilornel>
(use ‘--show-trace’ to show detailed location information)
<halfbit>
so like, mkdir mypackage, cd mypackage, vim default.nix, nix-env default.nix?
<halfbit>
seems like this would be a really common thing, or is it more common to keep everything in a fork of nixos?
<simpson>
halfbit: If you intend to contribute to nixpkgs, then you'll likely find it much easier at some point to establish a local git checkout. It's not necessary to get started, and not necessary if you just need stuff imported from nixpkgs.
<clever>
halfbit: i always put default.nix's in the root of my projects
<clever>
halfbit: and it builds without having to clone all of nixpkgs
<halfbit>
yeah I mean, I just want to be able to package company software for install
freeman42x]NixOS has quit [Ping timeout: 256 seconds]
<halfbit>
clever: thats what I'm looking for
<halfbit>
there's some other stuff I might contribute if I can get far enough, like a scylladb package
<halfbit>
I feel like I'm in a fog of confusion still though
<adelbertc>
i also have logs for `strace` of the lookup on another Pod that wasnt built by Nix
<adelbertc>
`/etc/resolv/conf` and `/etc/hosts` are more orless identical on both pods
<adelbertc>
i thought installing `bind` on the Nix-built one would help, but no go
<adelbertc>
`spark-test` and `spark-test-original` are the pod/container names - K8s adds this as an entry in `/etc/hosts` automagically
<adelbertc>
so the expectation is `nslookup` works
<abathur>
hmm
lopsided98 has joined #nixos
<wilornel>
clever: I did that
<pxc2>
halfbit: working directly in Nixpkgs is sometimes easier
thc202 has quit [Ping timeout: 276 seconds]
ma27 has quit [Ping timeout: 252 seconds]
<pxc2>
if you're packaging things for work, you might want a local clone of Nixpkgs for adding dependencies to Nixpkgs, since that will make your life easier. But it should be easy to spare end-users all that stuff
<MichaelRaskin>
Well, it is easy to import Nixpkgs and use their callPackage on an out-of-tree file
<MichaelRaskin>
Or write an out-of-tree file that imports things from Nixpkgs
<MichaelRaskin>
In any case you are going to want to pin the Nixpkgs version because of compatibility testing, so you can as well just develop it in a local Nixpkgs branch…
pxc2 has quit [Ping timeout: 252 seconds]
<wilornel>
warning: in configuration file '/etc/nix/nix.conf': unknown setting 'signed-binary-caches'
<clever>
wilornel: usually safe to ignore that one
<lejonet>
What is wrong with my machine or iso if I get "error: substituter 'download-from-binary-cache.pl' died unexpectedly" when trying to do nixos-install
<clever>
lejonet: did it say anything before that?
<wilornel>
Woot I got nix-env 2.0!
<lejonet>
clever: nope, that is the first error at all
<wilornel>
but bundler it still version 1.14.6!
<wilornel>
I don't get it
<clever>
lejonet: strange, does the machine have internet access?
<clever>
wilornel: updating nix wont update other packages
<lejonet>
clever: it should have, it can ping google for example
<manveru>
wilornel: sorry, i had bundler updated in my nixpkgs, forgot it's not there yet :(
<lejonet>
clever: I think I know what the problem could be -.- that domain is still running emulated, haven't switched to kvm, thought it was a bit too slow still
<lejonet>
clever: my guess is that it timed out and stuff, due to being very slow due to emulation
itscaleb has left #nixos ["ERC (IRC client for Emacs 25.3.1)"]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] jtojnar opened pull request #36236: tree-wide: Add some updateScripts (master...updatescripts) https://git.io/vASxf
NixOS_GitHub has left #nixos [#nixos]
kelleyNif has quit [Quit: Leaving.]
<wilornel>
I just tried to install nix-env --version 2.0 on my vagrant instance and it did work but ultimately did not install 2.0
<wilornel>
I should be able to get it with this version right?
<wilornel>
[vagrant@nixbox:~]$ nixos-version
<wilornel>
17.09.2683.360089b3521 (Hummingbird)
<lejonet>
clever: switching it to kvm made everything work ^^ sorry for noise :P
<wilornel>
for some reason, nix-env --version 1.12pre I think acts as 2.0 on that vagrant machine
<wilornel>
I'm still confused about how someone said that 1.12pre == 2.0
drakonis_ has joined #nixos
amfl_ has joined #nixos
amfl_ has quit [Client Quit]
amfl has quit [Quit: WeeChat 1.6]
amfl has joined #nixos
<manveru>
well, they're the same basically
<manveru>
1.12 was released as 2.0
acarrico has quit [Ping timeout: 268 seconds]
Lisanna has joined #nixos
acertain has quit [Ping timeout: 268 seconds]
acarrico has joined #nixos
Supersonic has quit [Ping timeout: 240 seconds]
Lisanna has quit [Quit: Lisanna]
Lisanna has joined #nixos
Supersonic112 has joined #nixos
Supersonic112 is now known as Supersonic
phdoerfler_ has quit [Quit: Page closed]
d4g_ has joined #nixos
d4g has quit [Ping timeout: 260 seconds]
pxc has quit [Ping timeout: 256 seconds]
ryantm_ has joined #nixos
phdoerfler has quit [Quit: Leaving.]
ryantm_ has quit [Ping timeout: 260 seconds]
<clever>
adelbertc: i dont see the child proc, -f is missing
<clever>
this says how old it all is, and links to the tests
<aardito>
oh ok it is 8 days old
<aardito>
I'm seeing nixos-18.03pre129076.831ef4756e3 from 2/22
<samueldr>
it's not unusual to be days behind, but it's not necessarily the norm
<aardito>
I wouldn't even have noticed but I went to install spotify
<clever>
ah, but the 'last updated' on howoldis is not how old the commit is
<aardito>
only to find out it was broken, but working in master
<clever>
youll notice, 17.03 and 17.03-small have different ages, yet are on the exact same commit
<clever>
17.03-small updated first, because thats the whole design of the -small jobs
<clever>
and then the full build finished a month later, causing 17.03 to update to the rev
<aardito>
looking at that though it seems like nixpkgs should be very recent, with nixos a bit older
<aardito>
just wondering if I have my channels set up right
<aardito>
I'm on both currently
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] womfoo opened pull request #36240: conkeror: use firefox-esr, 1.0.3 -> 1.0.4 (master...fix-bump/conkeror-1.0.4) https://git.io/vA9e9
NixOS_GitHub has left #nixos [#nixos]
* clever
heads off to bed
<wilornel>
good night clever
Myrl-saki has quit [Ping timeout: 256 seconds]
acarrico has quit [Ping timeout: 240 seconds]
blankhar1 has joined #nixos
raynold has quit [Ping timeout: 265 seconds]
dgonyeo has quit [Ping timeout: 256 seconds]
dgonyeo has joined #nixos
raynold has joined #nixos
schoppenhauer has quit [Ping timeout: 240 seconds]
Guest50636 has quit [Ping timeout: 256 seconds]
schoppenhauer has joined #nixos
aardito has quit [Quit: Page closed]
Guest50636 has joined #nixos
hellrazor has joined #nixos
marusich has quit [Ping timeout: 256 seconds]
Arcaelyx has joined #nixos
chocopuff has joined #nixos
marusich has joined #nixos
pps_ has joined #nixos
r5d has quit [Ping timeout: 240 seconds]
<mfiano>
I can't seem to install NixOS on my netbook
r5d has joined #nixos
<mfiano>
It uses an MMC storage device, but despite including the MMC kernel module that nixos-generate-config detected, it times out looking for the root partition, and then times out looking for the swap partition
<mfiano>
Odd thing is the livecd works. This is very annoying, because the livecd doesn't store the closure on the disk, so nixos-install takes about a half hour everytime
<samueldr>
err, every time I had to nixos-install, over an existing partition, it never had to re-download
<samueldr>
are you re-partitionning every time?
<mfiano>
It doesn't redownload
marusich has quit [Ping timeout: 245 seconds]
<mfiano>
But its very slow CPU and disk, and takes forever
<samueldr>
ah, I see
<mfiano>
Think raspberry pi in the shape of a laptop :(
<mfiano>
The strange thing is I can't get the initrd to know about my storage device, even though its what was detected and booted from with the livecd
neonfuz has joined #nixos
<samueldr>
I'm curious, though it's not to help, what kind of laptop is it?
<neonfuz>
major issues after latest update :|
<mfiano>
samueldr: Acer Cloudbook
marusich has joined #nixos
<neonfuz>
before reboot I couldn't open dmenu or urxvt windows, said no display, after update (and GC because I was an idiot) my wifi is broken and X reboots after a few minutes
<samueldr>
ah, Intel® Celeron® N3050
<samueldr>
1.6 GHz; Dual-core
<mfiano>
Yep
<neonfuz>
and on boot I get some error about TSC_DEADLINE, maybe this is to do with the meltdown update?
<neonfuz>
this only happened after doing a nixos-rebuild switch --upgrade to the latest unstable
<neonfuz>
any way to use not the latest unstable, but a slightly older one?
<samueldr>
neonfuz: if you know a commit ID, you could either git clone / git checkout that commit, or change the channel to use that commit
blankhar1 has quit [Quit: WeeChat 1.9.1]
<neonfuz>
how can I set the channel to use a certain commit?
<samueldr>
neonfuz: do you know around when was the last channel update you did that worked?
<neonfuz>
no :/
<samueldr>
:/ well, you'll have to try some
<neonfuz>
oh god
<neonfuz>
now when I try to run urxvt I get Invalid MIT-MAGIC-COOKIE-1 keyurxvt: can't open display :0, aborting.
<samueldr>
I don't know how "normal" it is, and you'll have to fix it later on, but you can `sudo nix-channel --add <full url>` to what the channel URL (https://nixos.org/channels/nixos-unstable) previously resolved too
<neonfuz>
this sounds like something to do with meltdown / spectre fix
<neonfuz>
alright
<samueldr>
but uh, I don't think that's anything meltdown / spectre related
<samueldr>
MIT-MAGIC-COOKIE is the way x11 programs can tell the server "look, I have the rights to talk to you"
<neonfuz>
hmm, any idea what the mit-magic-cookie thing is? I
<samueldr>
non-LTS releases may be removed, this would have to be actually verified
<leocp1>
hi, first time installing. Assuming I've created a /boot partition and have a preexisting ESP from windows 10 mounted to /boot/efi, how do I tell configuration.nix to install grub with the command ` grub-install --efi-directory=/boot/efi `? Should I use `boot.loader.grub.device = "nodev"` then install grub myself manually?
<mfiano>
Did you see boot.loader.grub.efiSupport and friends?
<samueldr>
didn't try it myself, but AFAIUI this should be how nixos-install knows how to find the ESP
<mfiano>
samueldr: nixos-install is actually downloading everytime I run it if the machine was rebooted. It keeps everything in tmpfs at /nix/.rw-store
<samueldr>
and looking through nixpkgs, the grub modules effectively uses that option
<mfiano>
So that adds to the long time it takes
<samueldr>
mfiano: ah, I must have been wrong then, I am lucky and never really had to go back through the installer
<mfiano>
I've ran it about 10 times already, or about 5 hours
<mfiano>
I don't see an end in sight yet :)
<samueldr>
all-hardware.nix looked promising to me
<mfiano>
Yep, that's next to try. Had a build running for a while I don't want to interrupt
<neonfuz>
sweet yeah I totally had kernel patches being set to only apply to 4_9 when my kernel updated to 4_14... oops lol
<leocp1>
mfiano,samueldr: thanks. what do I set boot.loader.grub.devices to then? I was worried about nix trying to run `grub-install /dev/sda` for a MBR style install. is the variable just thrown out?
<samueldr>
leocp1: "nodev" for EFI
<samueldr>
that's AFAIUI only used for legacy bios boot
<samueldr>
when installed as EFI it will add a .efi program to the ESP, and configure it using efibootmgr
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nix] shlevy opened pull request #1942: installables: Support unprefixed NIX_PATH lookups. (master...installables-unnamed-nix-path) https://git.io/vA9fb
NixOS_GitHub has left #nixos [#nixos]
<mfiano>
samueldr: Is there a way to find out what modules the initrd has after nixos-install before I reboot? I'm not actually seeing the eMMC module being loaded at boot, so wondering if nixos is not regenerating the initrd or something
hamishmack has quit [Remote host closed the connection]
Tobba has quit [Read error: Connection reset by peer]
hamishmack has joined #nixos
Tobba has joined #nixos
dkao has quit [Quit: WeeChat 2.0]
pps_ has quit [Quit: Leaving]
Guest50636 has quit [Ping timeout: 245 seconds]
<mfiano>
including all-hardware.nix didn't even work
<mfiano>
I don't think nixos is including the module it needs
Rusty1_ has quit [Quit: Konversation terminated!]
<leocp1>
samueldr: thanks for your help! it seems to have installed
yegortim1 has quit [Ping timeout: 255 seconds]
yegortim1 has joined #nixos
yegortim1 has quit [Remote host closed the connection]
oahong has quit [Ping timeout: 240 seconds]
Guest50636 has joined #nixos
yegortim1 has joined #nixos
oahong has joined #nixos
Guanin has quit [Ping timeout: 260 seconds]
adisbladis has quit [Ping timeout: 240 seconds]
pxc2 has joined #nixos
Guanin has joined #nixos
pxc2 has quit [Ping timeout: 256 seconds]
rogue_koder has quit [Quit: Konversation terminated!]
jtojnar has quit [Ping timeout: 240 seconds]
<abathur>
hrm
Fare has joined #nixos
drakonis_ has quit [Read error: Connection reset by peer]
marusich has quit [Ping timeout: 240 seconds]
marusich has joined #nixos
<hyper_ch>
hwat is NixOS:nixos-v237 ?
marusich has quit [Ping timeout: 240 seconds]
marusich has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] womfoo opened pull request #36241: facter: 3.9.3 -> 3.10.0 and deps (master...bump/facter-3.10.0-and-deps) https://git.io/vA9TI
NixOS_GitHub has left #nixos [#nixos]
freeman42x]NixOS has joined #nixos
leocp1 has quit [Quit: Lost terminal]
justanotheruser has joined #nixos
justan0theruser has quit [Ping timeout: 240 seconds]
knupfer has joined #nixos
aarvar has quit [Ping timeout: 276 seconds]
afics has quit [Quit: afics]
<rycee>
manveru: No, there is no specific channel for hm. Typically we have borrowed this channel when discussing :-)
<neonfuz>
now nixos-rebuild switch is getting
<neonfuz>
inconsistent kernel version 4.14.20
<neonfuz>
after setting kernel to 4_9
afics has joined #nixos
drdaeman has quit []
NightTrain has quit [Ping timeout: 256 seconds]
robstr has joined #nixos
<etu>
neonfuz: Which channel?
<Fare>
has 18.03 been cut yet?
<MichaelRaskin>
I haven't seen any announcements like that.
<NixOS_GitHub>
[nixpkgs] Mic92 closed pull request #36234: uutils-coreutils: disable general unix build for now (master...disable-uutils-coreutils-on-darwin) https://git.io/vASFq
NixOS_GitHub has left #nixos [#nixos]
vandenoever has joined #nixos
vandenoever has quit [Changing host]
vandenoever has joined #nixos
<Fare>
what is the branch-cutting process?
<MichaelRaskin>
I think it is vcunat as the release manager who will do the git branch command
tA63 has joined #nixos
<tA63>
Hey, I'm trying to update one of the nixpkgs, havent written an expression before so learning a bunch. When I run nix-build -A, I get a bunch of "Wrong ELF Type" errors, where can I find some info on this?
<robstr>
Is there a relaxed way of specifying a copy file instruction from configuration.nix ? I want to copy few dot files but i dont want to use the git / home-manaer solution.
<srhb>
tA63: That's usually just the result of one of the patching tools iterating over all files, even the ones it has nothing to do with. Usually harmless.
<srhb>
robstr: can you elaborate? Copy files from where to where?
<tA63>
@srhb Ahh ok, awesome. So I'll be good to continue on then?
<srhb>
tA63: Most likely yeah :)
Guanin has quit [Remote host closed the connection]
<robstr>
I have one directory containing my nixos configurations, i created only one import from `/etc/nixos/configuration.nix` to my seperated configuration file. I want to have everything inside this directory. I got vimrc already done with the `vim_configurable`.. But i want to move my xmonad.hs file from this directory to /home/whatever/.xmonad/
<robstr>
hope this makes sense :P
<robstr>
srhb: ^
vandenoever has quit [Ping timeout: 240 seconds]
<neonfuz>
etu: unstable
<neonfuz>
sry for late response i had to drive somewhere
<etu>
neonfuz: nixos i presume?
<neonfuz>
yep
<etu>
Do you know which commit you're on?
<neonfuz>
i set the kernel packages option to linuxPackages-4_9
<neonfuz>
no
<etu>
what does nix-info say?
<neonfuz>
a lot of stuff but id have to hand type it back to you as im on a diffwrent machine
chrisburr has joined #nixos
<neonfuz>
18.03-pre129076.831ef4756e3
<neonfuz>
the package that failed was kernel-modules.drv
<etu>
I think you want: linuxPackages_4_9 and not linuxPackages-4_9
<neonfuz>
this is after setting boot.kernelPackages to pkgs.linuxPackages_4_9
<etu>
If that wasn't a typo before?
<neonfuz>
was a typo i didnt have my machine on at the time lol
ottidmes has joined #nixos
<neonfuz>
im typing on my phone as im currently trying to fix my wifi lol
<srhb>
robstr: No, there's really no NixOS'y way of doing that, short of home manager and company.
<neonfuz>
the kernel compiled fine
<neonfuz>
(compiled because i do sone changes to it)
<neonfuz>
oh wait
<robstr>
srhb: sad, we should add it :P but thank you
<neonfuz>
i think its udev related?
<neonfuz>
unit.systemd-udev.service
<tA63>
Do I not understand how nix-env -f works? If I do nix-env -i <package> -f /path/to/nixpkg/folder/ will that use the default.nix from that folder rather than the usual one?
<MichaelRaskin>
robstr: activationCommands or something like that
<MichaelRaskin>
Look in the configuration.nix manual
timon37 has joined #nixos
<srhb>
It's system.activationScripts and I don't recommend it.
<robstr>
srhb: ok, whats the reason against it ?
<tA63>
home-manager seems good, maybe give that a try. I'm trying to get it set up now looks promising
<robstr>
tA63: maybe you can report your experience soon ? :P
<MichaelRaskin>
I would actually have a separate script — not usually executed to re-create _symlinks_ to the store that point into current-system
<srhb>
robstr: It's vital for boot and activation, so you need to get it Very Right
<robstr>
srhb: I see
<srhb>
robstr: For the record, I too use home-manager and it's very good :)
<neonfuz>
well i have to sleep now, im gonna fix this in thr morning...
<srhb>
MichaelRaskin: That's a nice solution :)
<tA63>
robstr I haven't gotten it going well haha, at the moment just using it to hardcode some dotfiles and have a list of installed packages same way as the systemPackages
<tA63>
srhb Will be able to give a much better argument haha
<MichaelRaskin>
My current system (not exactly NixOS) actually makes sure that everything is switched atomically via the current-system symlink, even /etc
<srhb>
robstr: Mine isn't very advanced either, but I do some nice things generating xsession commands based on the machine I'm on. :)
<srhb>
robstr: essentially I used it to join my machinex/machiney/machinez dotfiles repo into one parameterized specification
<mfiano>
I've seen the last 8 hours trying to get this laptop to boot the NixOS installation. I think I give up
<tA63>
mfiano Don't give up, you can do it
<robstr>
srhb: I only want to copy two files, it feels a bit overkill -.- ^^
<MichaelRaskin>
Symlinks it is?
Myrl-saki has joined #nixos
<robstr>
srhb: the workflow would be soooo great if we could specify files to copy inside the `user.extraUser` stuff
<mfiano>
tA63: Nah, eMMC module is not available during boot, depsite using the hardware-configuration.nix generated by the install medium AND the all-hardware.nix that the livecd itself uses
<mfiano>
So it times out during Stage 1 looking for partitions
<MichaelRaskin>
robstr: I dunno, having a writeable file that gets auto-overwritten sounds like a bad idea
<srhb>
robstr: I agree with MichaelRaskins symlink proposal :)
<srhb>
robstr: Have a nix expression that generates the file you're interested in, and have a script that symlinks that into place.
simukis has joined #nixos
<robstr>
srhb: could you create small gist with it ?
<sphalerite>
mfiano: what kind of laptop is it?
Fare has quit [Ping timeout: 276 seconds]
<mfiano>
sphalerite: Acer Cloudbook. There are a few different variants. This is AO1-431-C8G8
<mfiano>
It uses internal eMMC storage
<mfiano>
the eMMC is so slow that it takes about 30 minutes to run nixos-install everytime I try to fix it...and it has to redownload everything again because the livecd keeps everything in tmpfs at /nix/.rw-store
<mfiano>
So until the system can be booted, it takes forever to try again
<sphalerite>
and the installer boots on it fine?
<mfiano>
Correct, which leads me to believe it is not generating the initrd correctly but have no way to inspect that
<srhb>
robstr: Which part? The script is just ln -s $(nix-build yourNixfile.nix) destination
<mfiano>
During Stage 1, it times out looking for the root partition, and then times out looking for the swap partition
<sphalerite>
mfiano: if you pass debug1devices on the kernel command line you can get a shell to inspect what's going on
<robstr>
srhb: I see, thank you.
<sphalerite>
you should be able to put that in by pressing e on the grub or systemd-boot entry, not sure about other loaders
<mfiano>
sphalerite: Let me try that
<mfiano>
sphalerite: debug1devices?
<sphalerite>
yes
<mfiano>
sphalerite: It still times out
<mfiano>
I'll take a picture
<sphalerite>
oops sorry boot.debug1devices
<mfiano>
sphalerite: I'm in a shell
fxr has joined #nixos
<mfiano>
and of course, I don't have a /dev/mmcblk0 block device
<sphalerite>
anything at all in /sys/block?
<mfiano>
but the kernel modules are all loaded, even the rtsx_usb_sdmmc that the livecd used
<mfiano>
sec
<mfiano>
ram{0,15} and sda (which is the usb installer)
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] andersk opened pull request #36242: Give working advice for running nix-prefetch-url (master...nix-prefetch-url) https://git.io/vA9qz
NixOS_GitHub has left #nixos [#nixos]
<sphalerite>
how strange
<mfiano>
Yeah. it works fine on Arch Linux though
<mfiano>
and the nixos livecd ha
<sphalerite>
maybe it's missing a USB driver
<sphalerite>
anything in /sys/bus/usb?
<sphalerite>
/devices
<mfiano>
It's got a few loaded, that were added to hardware-configuration.nix, so I assume the same as the livecd had
<mfiano>
Yeah quite a few
<mfiano>
/sys/bus/usb/drivers/ has hub, rtsx_usb, usb, usb-storage, and usbfs
stphrolland has joined #nixos
<mfiano>
ad ./devices/ has 14
<sphalerite>
anything in dmesg about the rtsx thing?
<sphalerite>
mfiano: maybe you can manually bind the rtsx driver to the appropriate USB device
<mfiano>
no idea how to do that
<sphalerite>
find the eMMC's address on the USB bus (check /sys/bus/usb/devices/*/id{Vendor,Product})
<sphalerite>
then echo that into /sys/modules/rtsx_usb_sdmmc/drivers/*/bind
<mfiano>
and if there is no id*?
the_real_plumps has quit [Quit: No Ping reply in 180 seconds.]
<sphalerite>
I then not what you're looking for, I think
<sphalerite>
clever: you here? This seems like the kind of thing you'd be able to help with as well
<sphalerite>
mfiano: some of them should have those ids though
<mfiano>
I'm not sure how to find the correct vendor product, when I don't know the vendor or product of the emmc hardware, and all i get is 2 bytes of hex
Arcaelyx has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<sphalerite>
ah actually try looking at /sys/bus/usb/devices/**/product, those should be more human-readable
<tnks>
man, the "nix subcommand --help" really suck.
<tnks>
the examples of "nix copy --help" show a "-r" switch, but have no mention of it as a flag.
<tnks>
someone here care to explain it?
<sphalerite>
tnks: my best guess is that it used to enable recursive behaviour, until that was made into the default without the examples being updated
<tnks>
sphalerite: I was hoping it was a default.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] aszlig pushed 1 new commit to master: https://git.io/vA9mI
<NixOS_GitHub>
nixpkgs/master 88530e0 aszlig: systemd: Update to latest NixOS branch...
NixOS_GitHub has left #nixos [#nixos]
<tnks>
copying a nix expression without closures seems wrong, certainly not a good default.
<sphalerite>
rizary: I think so, although nixops also has azure support iirc and it might have a more efficient way of setting nixos up initially. I don't really know anything about azure though so YMMV
the_real_plumps has joined #nixos
<mfiano>
sphalerite: I see my flash disk, webcam, USB2.0-CRW (I don't know what that means), and a device that doesn't have a product node at all.
<sphalerite>
tnks: actually Mic92 already fixed the examples
<sphalerite>
mfiano: USB2.0-Card Reader/Writer? Maybe try it with that one
<sphalerite>
that's just a guess though
<sphalerite>
if that doesn't work I'd probably have a look at the journal in the installer system (journalctl) and see what's there that might be related to the eMMC
<mfiano>
/sys/bus/usb/devices/1-3:1.0/ has rtsx_usb_sdmmc.0.auto and rtsx_usb_ms.1.auto directories but not sure what to access in there
<sphalerite>
or what are their contents, if they're not symlinks?
<mfiano>
they got a bunch of stuff in there, most of which seem to be some type of links and recurse infinitely. i rebooted into the installer and i see mmc stuff in journalctl
ma27 has joined #nixos
<sphalerite>
is it early on?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master 0615366 Roman Volosatovs: Revert "spotify: 1.0.69.336.g7edcc575-39 -> 1.0.70.399.g5ffabd56-26"...
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 2 new commits to master: https://git.io/vA9mM
<mfiano>
It's even more confusing because this laptop has sdcard drive and internal emmc storage, so not sure which is which
<mfiano>
s/drive/port/
<sphalerite>
aaaah
<sphalerite>
I suspect what you actually need in that case is the sdhci driver
<mfiano>
looking at the linux kernel source, HS200 device is a eMMC
ma27 has joined #nixos
<mfiano>
sphalerite: the hardware-configuration.nix that it generated for me includes xhci_pci usb_storage sd_mod sdhci_acpi and rtsx_usb_sdmmc
<sphalerite>
hm ok
<mfiano>
and i am loading that
pxc2 has joined #nixos
Guest50636 has quit [Ping timeout: 240 seconds]
thc202 has joined #nixos
<mfiano>
dmesg of the early boot shell of debug1devices does not have any mention of mmc
ma27 has quit [Ping timeout: 252 seconds]
<mfiano>
I have an idea
pxc2 has quit [Ping timeout: 240 seconds]
ma27 has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to staging: https://git.io/vA9Ya
<NixOS_GitHub>
nixpkgs/staging 6a9ab8d Vladimír Čunát: mesa: 17.3.5 -> 17.3.6 (bugfix)...
NixOS_GitHub has left #nixos [#nixos]
<sphalerite>
yay ideas!
<mfiano>
didn't help. i read that this can be caused by the module loading too late, and rootdelay or rootwait parameters may help, but nixos stage1 happens anyway without waiting
stphrolland has quit [Quit: leaving]
wavewave has quit [Read error: Connection reset by peer]
wavewave has joined #nixos
<mfiano>
I added a bunch more kernel modules to the initrd. attempting a rebuild
<schoppenhauer>
hi. where can I find the python package "gi"?
<mfiano>
added a bunch of mmc related modules to the initrd, and now i get a huge stream of "/dev/mmcblk0rpmb: read failed after 0 of 4096 at 4096: Input/output error" lines. googling suggests this is the read-protected memory block used by the manufacturer, and it's encrypted.
mirlur[m]1 has joined #nixos
<schoppenhauer>
sphalerite: ah. for the beta?
alphor_ has joined #nixos
<mfiano>
and then it times out again lol
goibhniu1 has joined #nixos
<schoppenhauer>
sphalerite: could you send it to me?
dash1 has joined #nixos
scott11 has joined #nixos
<sphalerite>
schoppenhauer: I was running off the gtk3 mercurial branch, apparently last touched it in May last year. I don't think the beta existed at that point
<sphalerite>
mfiano: that sounds like progress to me!
<sphalerite>
if you use boot.debug1devices again, do you find any new stuff in /sys/block?
grenade has quit [Ping timeout: 256 seconds]
grenade_ is now known as grenade
dottedmag_ has joined #nixos
[0__0] has quit [Read error: Connection reset by peer]
goibhniu has quit [Ping timeout: 256 seconds]
mirlur[m] has quit [Ping timeout: 256 seconds]
alphor has quit [Ping timeout: 256 seconds]
dash has quit [Ping timeout: 256 seconds]
simpson has quit [Ping timeout: 256 seconds]
scott1 has quit [Ping timeout: 256 seconds]
justanotheruser has quit [Ping timeout: 256 seconds]
dottedmag has quit [Ping timeout: 256 seconds]
dottedmag_ is now known as dottedmag
<mfiano>
that's what i'm trying now
<mfiano>
sphalerite: huh. i have a /dev/mmcblk0 and partitions now
[0__0] has joined #nixos
<mfiano>
but it's not finding them on boot
the_real_plumps has quit [Quit: No Ping reply in 180 seconds.]
mbrgm has quit [Ping timeout: 256 seconds]
akl has quit [Ping timeout: 256 seconds]
efx[m] has quit [Ping timeout: 256 seconds]
smbarte2[m] has quit [Ping timeout: 256 seconds]
Obscurity[m] has quit [Ping timeout: 256 seconds]
bdimcheff has quit [Ping timeout: 256 seconds]
ycy[m] has quit [Ping timeout: 256 seconds]
Pneumaticat has quit [Ping timeout: 256 seconds]
WilliamHamilton has quit [Ping timeout: 256 seconds]
puffnfresh has quit [Ping timeout: 256 seconds]
aspiwack[m] has quit [Ping timeout: 256 seconds]
offlinehacker[m] has quit [Ping timeout: 256 seconds]
Ralith has quit [Ping timeout: 256 seconds]
skrzyp has quit [Ping timeout: 256 seconds]
<schoppenhauer>
sphalerite: hm. what's in ./python-nbxmpp?
<sphalerite>
oh boy, matrix is falling over again.
Pneumaticat has joined #nixos
smbarte2[m] has joined #nixos
efx[m] has joined #nixos
offlinehacker[m] has joined #nixos
<schoppenhauer>
(the packages version of python3-nbxmpp is too old)
WilliamHamilton has joined #nixos
Ralith has joined #nixos
jophish has quit [Ping timeout: 256 seconds]
Obscurity[m] has joined #nixos
aspiwack[m] has joined #nixos
ycy[m] has joined #nixos
puffnfresh has joined #nixos
jophish has joined #nixos
bdimcheff has joined #nixos
mbrgm has joined #nixos
<sphalerite>
schoppenhauer: oops :D that would be python-nbxmpp, just the git repo from gajim.org
skrzyp has joined #nixos
akl has joined #nixos
skrzyp is now known as Guest64734
akl has joined #nixos
akl has quit [Changing host]
akl has joined #nixos
the_real_plumps has joined #nixos
<mfiano>
sphalerite: I totally have the mmc device and dmesg has all the expected devices detected now...but Stage 1 has all those errors trying to read the r/o encrypted storage device instead of the real one
<mfiano>
i think the encrypted one is the actual uefi and secure boot keys
<sphalerite>
that's odd
hotfuzz has joined #nixos
<sphalerite>
Can you mount the root filesystem yourself?
<mfiano>
yup
justanotheruser has joined #nixos
ma27 has quit [Quit: WeeChat 2.0]
<sphalerite>
if you started the shell as PID1 you should be able to actually boot the system by doing `exec switch_root /mnt /nix/var/nix/profiles/system/init`
<sphalerite>
assuming you mounted it on /mnt
hotfuzz_ has quit [Ping timeout: 240 seconds]
<sphalerite>
if that works you'll at least have a system to iterate on faster than with half-hour rebuilds :D
ma27 has joined #nixos
<mfiano>
sphalerite: that resulted in a Kernel panic
<sphalerite>
the attempting to kill init one?
<mfiano>
started up with boot.debug1devices, hit `f` for PID1 shell, then `exec switch_root /mnt /nix/var/nix/profiles/system/init` = instant panic
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peti pushed 1 new commit to release-17.09: https://git.io/vA9O0
<NixOS_GitHub>
nixpkgs/release-17.09 700a717 Peter Simons: stack: fix build with GHC 8.0.x
NixOS_GitHub has left #nixos [#nixos]
<mfiano>
well mounted before that
<sphalerite>
what did the panic say though? Attempting to kill init?
<mfiano>
Yes
<mfiano>
Kernel panic - not syncing: Attempted to kill init! exit code=0x00000100
<sphalerite>
that's good, means it's just a userspace issue
telent[m] has joined #nixos
<mfiano>
heh
<sphalerite>
you'll get the same thing if you type "exit" at the f shell
<mfiano>
yes i did
<mfiano>
i accidentally hit ^D earlier
<sphalerite>
:D
ma27 has quit [Client Quit]
<mfiano>
so any other ideas? :) feels like I'm almost there
<sphalerite>
does /mnt/nix/var/nix/profiles/system/init exit?
<sphalerite>
exist*
<ottidmes>
I read on an old mailing list thread that ssh-substituter-hosts only supports one host, is this still the case? I tried looking up the implementation, but searching 'ssh-substituter-hosts' in the code base of Nix just gives me documentation
<mfiano>
oh i know the problem
<mfiano>
sigh
bitnotri has quit [Quit: Connection closed for inactivity]
<sphalerite>
ottidmes: yeah that sounds outdated
<mfiano>
well jeez. I'm in
<sphalerite>
ottidmes: just substituters = https://cache.nixos.org/ ssh://user@host1/ ssh://user@host2/ should do it
<sphalerite>
mfiano: wooo! What was it?
<mfiano>
oh stupid me was mounting the partition instead of the btrfs sub volume on that partition, so /mnt/nix... was really /mnt/subvolname/nix...
<mfiano>
so now I have to figure out which module is causing those massive errors and blacklist it/remove it
<sphalerite>
:D
<ottidmes>
sphalerite: Is that a new option in nix 2.0? Or is it also already available in 1.11.x? I have used ssh-substituter-hosts with a single host successfully under 1.11.x
<sphalerite>
oh yeah that's a 2.0 thing, I think in 1.11 it's called binary-caches rather than substituters but the ssh URLs should still work I think
<sphalerite>
wait no
<sphalerite>
I'm not sure.
<ottidmes>
sphalerite: Well, why make things complicated for myself, I will just start using 2.0 then, thanks for the tip :)
<mfiano>
hopefully i don't need no network to further debug this, because i built a very minimal config without wireless tools
periklis has joined #nixos
<sphalerite>
mfiano: depending on how adventurous you're feeling you could copy wpa_supplicant from the installer USB :p
<mfiano>
note to self. actually run bixos-rebuild before rebooting and having to do that long boot process needlessly again
<sphalerite>
so you know how to fix it already?
<mfiano>
no but i'm being dumb this late at night. i edited the config to remove a module, and just rebooted without rebuilding
<sphalerite>
oh haha
<mfiano>
really...
<mfiano>
i need network connection to issue nixos-rebuild even though i'm only changing 1 module in boot.loader.initrd?
<mfiano>
this is going to waste so much time
pie_ has joined #nixos
<mfiano>
sphalerite: Do you know which package I need to add to systemPackages to have the nmtui tool from the installer usb?
NixOS_GitHub has joined #nixos
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/staging ce34a9a Frederik Rietdijk: python: botocore: 1.9.0 -> 1.9.3
<NixOS_GitHub>
[nixpkgs] FRidh pushed 14 new commits to staging: https://git.io/vA93k
<NixOS_GitHub>
nixpkgs/staging 7ee5833 Frederik Rietdijk: python: zc.buildout: 2.11.0 -> 2.11.1
<NixOS_GitHub>
nixpkgs/staging 0e5e184 Frederik Rietdijk: python: cffi: 1.11.4 -> 1.11.5
<sphalerite>
enable network-manager
<mfiano>
that's it? cool
<sphalerite>
that'll bring in a bunch of graphical deps though
<mfiano>
at this point. ask me if i care
<manveru>
rycee: k, just wanted to say thanks, i was up until 5am just fiddling with colors and sizes :)
<mfiano>
sphalerite: do you think it's a bug that the installer is not adding all of the needed modules to hardware-configuration.nix? When I get to the bottom of this I'll know exactly the module(s) that are needed and are not added by the installer
Lisanna has quit [Quit: Lisanna]
polytux has joined #nixos
<mfiano>
heh, well out of about 10 modules that i added, i figured out the one that's needed first try
the_real_plumps has quit [Quit: No Ping reply in 180 seconds.]
<mfiano>
We've been debugging kernel modules for hours. This is all testing in order to get a rebuild that doesn't take a half hour each
<romildo>
I am waiting for it to happen in order to add more deepin DE packages.
<ij>
Yes, it's because of programs.zsh.enable = true; That functionality is weird.
<polytux>
mfiano: which chipset is that? I'm pretty sure I have the same chipset and no issues with it
chisui has joined #nixos
<polytux>
mfiano: networking.networkmanager.enable = true; is everything I needed in my configuration.nix
<mfiano>
intel 3160. dhcp times out. but works fine on the livecd
<polytux>
oh, mine is ac 8260
<mfiano>
it's not a network-manager issue or chipset
<mfiano>
botrh work on the livecd
<polytux>
does it work when assigning a static IP?
<mfiano>
I'll have to figure out how to do that with just ifconfig, because rebuilding takes a half hour on the livecd since it has to redownload everything to tmpfs
<polytux>
do you have any firewall running?
<mfiano>
no
romildo has quit [Quit: Leaving]
<polytux>
hm to tmpfs? I thought it was smart enough to download to your filesystem instead
<mfiano>
not the livecd. it downloads to /nix/.rw-store which is tmpfs
<polytux>
aah
robstr has quit [Remote host closed the connection]
periklis has quit [Ping timeout: 252 seconds]
Myrl-saki has quit [Ping timeout: 276 seconds]
robstr has joined #nixos
<Orbstheorem[m]>
Hello, how can I install gnumake “man” output using nix-env -i ? I tryed this: https://paste.gnugen.ch/raw/seQX but doesn't seem to be working
Myrl-saki has joined #nixos
ebzzry_ has quit [Ping timeout: 240 seconds]
pkill9 has joined #nixos
robstr has quit [Ping timeout: 248 seconds]
<polytux>
mfiano: did you enable unfree stuff as well?
<mfiano>
no. this is a bare minimal config that only serves to debug initrd modules at this point.
<polytux>
ah. not even sure it is related, but in case your wifi card needs a proprietary blob I guess it is necessary?
periklis has joined #nixos
<mfiano>
maybe. I'll have to figure out how to add that
knupfer has quit [Ping timeout: 268 seconds]
GiGa has joined #nixos
GiGa has quit [Changing host]
GiGa has joined #nixos
<GiGa>
Hello all
<GiGa>
Anyone else finding gedit _really_ unstable lately, even with quite small files?
<GiGa>
I'm just trying to scroll a file of about 280 lines and it's just painful
<rycee>
manveru: Hehe, glad you like it :-)
<mfiano>
sphalerite: still around?
jensens has joined #nixos
cfricke has joined #nixos
cfricke has quit [Client Quit]
<hyper_ch>
hmmm, nixos-unstasble small hasn't been updated for 3 days :(
<sphalerite>
mfiano: back now!
blankhar1 has joined #nixos
<sphalerite>
sorry about the unannounced disappearance :p
blankhar1 has quit [Quit: WeeChat 1.9.1]
<mfiano>
sphalerite: further debugging issues
<mfiano>
sphalerite: The real problem is that the livecd has the module "sdhci-pci" loaded, but does not add this to hardware-configuration.nix for some reason. Adding that is what makes booting possible with your switch_root hack.
<sphalerite>
right
<mfiano>
the livecd not adding all the proper modules is a NixOS bug. The IO errors trying to read the mfg secure boot partitions for no reason is a systemd bug that was supposed fixed in a later version
<mfiano>
234.11-8 to be precise. journalctl only shows me that i'm running 234. Do you know how to find the minor version?
<sphalerite>
huh ok. Weird, I didn't think systemd would be running that early on
<sphalerite>
I don't think systemd has minor versions at all
blonkhart has joined #nixos
<sphalerite>
that's probably a downstream version number
<mfiano>
ah then i think i need 235 or later
<mfiano>
can you verify for me what version unstable has?
<sphalerite>
234 as well
<mfiano>
damn
<sphalerite>
although I'm on a firly old unstable, hang on
<mfiano>
236 is latest i think
<sphalerite>
unstable has 237
<mfiano>
ok. so if i update that i have a better chance at success
<sphalerite>
wait no that might be master
<sphalerite>
but yes unstable might install just fine
<mfiano>
and i cant run nixos-rebuild without another kernel boot parameter :)
<sphalerite>
what's that?
ThatOtherGuy has quit [Read error: Connection reset by peer]
<mfiano>
systemd.gpt_auto=0, otherwise systemd errors because of the io error :)
<sphalerite>
ah right
<mfiano>
this is mildly amusing, all these issues i'm running into
<hyper_ch>
hmmm, stuff is blocking unstable small :(
isHavvy has joined #nixos
<mfiano>
hyper_ch: at least its not unstable...like 10 days stalled. there was a database outage
<hyper_ch>
unstable hasn't been updated for 8 days
<sphalerite>
you need the per-user setting if you want to use nix-env
<fearlessKim[m]>
is there a bounty procedure ? Like being able to bet some money on an issue to get it solved ?
<sphalerite>
mfiano: unless it's in udev, but udev runs *before* debug1devices
timon37 has quit []
<sphalerite>
mfiano: but yeah if you need it running urgently probably going back to arch is best (for now! ;) )
<sphalerite>
fearlessKim[m]: not a formally approved one, I think people have done stuff on bountysource in the past though
<mfiano>
Well I have until tonight to figure it out
<ottidmes>
I am confused, I have extra-substituters = ssh://root@server in my nix.extraOptions, but I get error: cannot connect to 'root@server', while I when logged in as root I can run: ssh root@server, without any problems. Any ideas? Does it look for ssh keys somewhere else than /root/.ssh in this case?
<MichaelRaskin>
Well, Arch + Nix-on-Arch is a way safer bet to learn Nixpkgs in details…
<MichaelRaskin>
(If you know Arch well already)
<sphalerite>
ottidmes:might be a known_hosts issue
<ottidmes>
sphalerite: Yeah, but shouldn't that have come up with I ssh'ed as root into root@server?
<sphalerite>
ottidmes: yeah I think so…
<mfiano>
sphalerite: The problem is whatever is probing block devices to find the corrrect UUID/label/whatever. When it even attempts to read a mmcblk*rpmb device, it is supposed to I/O error, as the RP stands for Read Protected, and only readable through specific firmware api's
<steve-chavez>
My bad didn't read the part about `NIXPKGS_ALLOW_UNFREE=1`, thanks a lot sphalerite.
<sphalerite>
mfiano: aaah so maybe you can, as a workaround, not specify your filesystems by UUID but by device path
<sphalerite>
mfiano: or if you don't like that you could do boot.initrd.postDeviceCommands = "rm /dev/mmcblk0rpmb"; :D
<sphalerite>
(not sure if that will work though)
<mfiano>
ha, will try
<Dezgeg>
can't one just ignore the warnings?
<sphalerite>
apparently not since it stops the root fs from getting mounted
<mfiano>
sphalerite: the former worked anyway!
pie___ has joined #nixos
<sphalerite>
wooo!
<mfiano>
i still see an IO error but it skips over it with device names
<Dezgeg>
huh, it has always worked for me on my arm devices to ignore them
<mfiano>
sphalerite: Thanks a lot! That was pretty crazy debugging that for like 20 hours
<sphalerite>
xD
<sphalerite>
now let's hope you don't get any more kernel panics
<sphalerite>
if the rpmb device is supposed to give I/O errors, why is it even exposed?
<mfiano>
i believe that was due to not having the systemd.gpt_auto=0 boot flag that boot, and doing a switch causes systemd to restart services
<mfiano>
sphalerite: good question
<Dezgeg>
you need a specific sequence to access them, not just normal reads
<sphalerite>
oh right so the appropriate ioctls will let you read it?
<Dezgeg>
read requests need to be signed with a private key with a nonce from the device or something to counter replay protection
pie__ has quit [Ping timeout: 268 seconds]
<Dezgeg>
but yeah, I suppose udev should really skip over them
* mfiano
remembers a simpler time before udev with devfs
<mfiano>
i just wouldn't make that node :P
robstr has joined #nixos
nuncanada has joined #nixos
<sphalerite>
yeah that seems like a sensible behaviour really
<Dezgeg>
well somebody needs to code that logic to udev
<mfiano>
Dezgeg: They're not warnings. When looking for a block device that matches the UUID or label you specify in fileSystems.*, it just errors infinitely until it times out and halts the system.
<gchristensen>
lol $100 bounty on a massive todo list
<Dezgeg>
you mean it never creates the /dev/disk/by-label/ symlink then?
<MichaelRaskin>
gchristensen: that might be a failed attempt at crowd-bountying
<mfiano>
I'm not sure. I just know that if you specify either a uuid or a label rather than a device path, it tries to touch the read protected device while resolving the uuid/label and gets stuck there due to i/o errors until it times out
<gchristensen>
manveru: spotify is on dbus?
<manveru>
jep
<Dezgeg>
well, I have a mmcblk0rpmb as well and specifying /dev/disk/by-label/NIXOS as the device works for me
* gchristensen
just got a weekend project
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Ma27 opened pull request #36246: idea: add `libsecret` to the library path (master...solve-keychain-issues-for-intellij) https://git.io/vA9W3
NixOS_GitHub has left #nixos [#nixos]
<mfiano>
Dezgeg: actually we worked around some other issues and I never properly tested a label yet with the other issues out of the way...only a UUID. let me try right now
<manveru>
gchristensen: that's how the emacs spotify client works too, pretty neat
<Dezgeg>
is your UUID actually correct then? x)
<fearlessKim[m]>
sphalerite: I see thanks
<mfiano>
Dezgeg: Another good question. It's what nixos-generate-config provided :)
<mfiano>
Let me run some more tests
<gchristensen>
mfiano: thank you for being a good sport about it
<mfiano>
gchristensen: what did i do? :)
<gchristensen>
mfiano: you've remained a positive participant in our community through frustrating installation problems :)
<sphalerite>
didn't blame nixos, say we're useless, etc
oahong has quit [Ping timeout: 240 seconds]
<gchristensen>
manveru: it seems maybe I should learn more about this here dbus business
* gchristensen
can hear MichaelRaskin writhing at the idea of _embracing_ dbus
<MichaelRaskin>
I mean, I am OK with it, but I prefer to have it neatly isolated per every task that it is needed for
<mfiano>
ok interesting. a label works. uuid generated with generate-nixos-config does not. so I found 2 bugs with generate-nixos-config today that i need to somehow document for a proper bug report
<manveru>
tested it with a few medium size projects, and saves me the pain of go2nix
<mfiano>
if that's even possible
<MichaelRaskin>
I don't like package conflicts, neither do I like interaction between dbus requirements of different packages
<mfiano>
sphalerite: ^
<manveru>
MichaelRaskin: ++
<mfiano>
gchristensen: no, nixos is fantastic and the community just as much so. sphalerite and clever especially have helped me get through some tough times
<gchristensen>
manveru: can polybar keep your program running in the background? to avoid startup costs every 5 seconds
<gchristensen>
mfiano: clever and sphalerite are both gems
<sphalerite>
mfiano: the uuid thing is a good feature IMHO, labels can collide while it's extremely unlikely for UUIDs
<MichaelRaskin>
Actually, I cannot exclude that at some point I will run some service by running it in a container with a systemd PID1 — as long as it doesn't get a chance to manage my entire system, only the pieces ascribed to it.
<sphalerite>
mfiano: and for detecting the kernel module, that is strange since nixos-generate-config.pl does handle MMCs specifically
<mfiano>
sphalerite: the other bug is that the sdhci-pci module is not being added to the hardware file
<mfiano>
it adds a few other mmc modules, but not sdhci-pci
<manveru>
gchristensen: i think you could use the polybar ipc for it
<gchristensen>
(I don't use polybar, but I remember the i3bar manpage telling me to be careful about segments where a program has to start every refresh)
<mfiano>
it adds sdhci-usb, sdhci, etc...but missing the pci one that is actually loaded by the livecd, just not copied to the config for some reason
<angerman>
How do I modify all haskellPackages? Eg iterate over all packages and apply overrideAttrs?
<sphalerite>
MichaelRaskin: sounds unlikely :D
<manveru>
yeah... most bars work this way
<fearlessKim[m]>
I feel like a headless running chicken trying to package gpgme python bindings, like I generate a wheel through occult wizardry as requested per nix python, but my wheel is empty :s https://github.com/NixOS/nixpkgs/pull/30429#discussion_r152653134
<manveru>
that's why i didn't write that tool in ruby, so startup is pretty fast
<sphalerite>
MichaelRaskin: iirc the only way to get systemd running in docker (lol) is to give it --privileged, which prevents cap dropping and seccomp filtering, or at least not dropping CAP_SYS_ADMIN
<gchristensen>
sphalerite: MichaelRaskin's pid1 is a magical lisp program, equivalent to the UniverseBrain meme guy
<mfiano>
sphalerite: Anyway, thanks again...you rock. I can now work on my config
<sphalerite>
gchristensen: :D
<gchristensen>
I trust it can start systemd
<MichaelRaskin>
gchristensen: wrong
<gchristensen>
awww
<MichaelRaskin>
My PID1 is sinit
<MichaelRaskin>
Because I want to be able to load code into my system management daemon without caring if it crashes
<MichaelRaskin>
(And crashed PID1 is a bit annoying)
<gchristensen>
indeed
<gchristensen>
todo: erlang pid1 for that sweet sweet infinity-9's goodness? ;)
<MichaelRaskin>
sphalerite: I would expect systemd to work in a user+PID namespace as PID1-UID0?
GiGa has quit [Remote host closed the connection]
<sphalerite>
MichaelRaskin: no, it wants more >_> can't remember the exact details
<sphalerite>
it will run in a systemd-nspawn container, but that requires the host to be running systemd as well \o/
<MichaelRaskin>
(I mean _without_ seccomp etc)
<clever>
sphalerite: ive worked on a bash script before that can boot a nixos container
<MichaelRaskin>
Yeah, systemd breaks too many things to be allowed on the top level
<clever>
sphalerite: the idea, was to half a self-contained bash script that can launch a nixos container without relying on the host systemd
<sphalerite>
it doesn't work without root-namespace CAP_SYS_ADMIN though, right?
steve-chavez has quit [Remote host closed the connection]
<clever>
sphalerite: cant remember, but it probably needs root
<MichaelRaskin>
I am OK with using CAP_SYS_ADMIN for initial setup, as long as by the time systemd is exec-ed the global capability is dropped.
<clever>
sphalerite: i also wrote the container-generator before reading the source of nix-user-chroot, which may help things a lot if i was to rewrite it
oahong has quit [Ping timeout: 248 seconds]
humanoyd has joined #nixos
<ottidmes>
It is annoying that when you rebuild the old values of /etc/nix/nix.conf are still in place, I am getting: download-via-ssh: error: secret key is corrupt, but haven't got a clue why, the signing keys in /etc/nix exists and have the right perms, the ssh keys in root exist and also have valid perms.
<sphalerite>
haven't seen that one before…
oahong has joined #nixos
<sphalerite>
as a workaround you can mv /etc/nix/nix.conf{,~} && cat /etc/nix/nix.conf~ >/etc/nix/nix.conf , then edit it to your liking
pxc2 has joined #nixos
<ottidmes>
sphalerite: I tried that, I removed ssh-substituter-hosts, but still get the same error, so it seems as if if is ignoring that files contents
<sphalerite>
you'll also need to restat the daemon for the change to take effect
<ottidmes>
sphalerite: Ah, right, thanks
yegortim1 has quit [Write error: Broken pipe]
Rusty1_ has joined #nixos
<ottidmes>
sphalerite: yep, that worked, thanks, still have to figure out how to get substitutors properly working under 2.0, I keep running into problems
<sphalerite>
:/
pkill9 has quit [Ping timeout: 260 seconds]
pxc2 has quit [Ping timeout: 240 seconds]
dan_b has joined #nixos
<ottidmes>
sphalerite: I think I get it now... I have 2 machines, but one is on Nix 1.x and the other Nix 2.x, and apparently they are not backwards compatible with everything
<sphalerite>
aaaaaah
<sphalerite>
we do need better errors for that though >.<
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peterhoeg opened pull request #36248: mycroft: init at 0.9.19 [WIP] (master...p/mycroft) https://git.io/vA98m
<ottidmes>
sphalerite: Now I was getting: error: cannot open connection to remote store 'daemon': secret key is corrupt, as well, which made me believe this to be the cause
<obadz>
so ermmm… when are we forking 18.03?
<gchristensen>
obadz: maybe ping fplet-z and vcuna-t in #nixos-dev?
<obadz>
gchristensen: done! thx
Guest64734 is now known as skrzyp
pkill9 has joined #nixos
drakonis has quit [Ping timeout: 256 seconds]
<manveru>
rycee: if i want to add a module/service to my home-manager, is there an easy way to do that?
<sphalerite>
manveru: You should be able to use imports like with nixos config
<sphalerite>
AFAIU
<sphalerite>
(I don't actually use HM though :P )
leo60228 has joined #nixos
Mateon1 has quit [Ping timeout: 255 seconds]
Mateon1 has joined #nixos
<leo60228>
how would I go about packaging a free windows game as a nix package for personal use via wine?
<leo60228>
*derivation
<clever>
leo60228: theres not much to really do for wine things, just install wine and run it against the installer.exe
<leo60228>
thanks
<manveru>
sphalerite: seems to work, thanks :)
<clever>
wine has so much state that you cant really manage it with nix
<leo60228>
will it show up in the kde application launcher?
<clever>
leo60228: i believe wine will generate .desktop files when the installer tries to generate windows shortcuts
<rycee>
manveru: Yeah, like sphalerite said. If you have a locally made module then you can use the `imports` option just like in NixOS.
<manveru>
yeah, now i just have to get the QT_PLUGIN_PATH into my service somehow
<rycee>
If you want to include as an "official" module in HM then it's just a matter of making a PR.
<leo60228>
why are my speakers not making noise
<leo60228>
they're on max volume
<leo60228>
kde shows them
<leo60228>
but no noise
<manveru>
leo60228: did you check pavucontrol?
<manveru>
rycee: trying to make a service for the seafile-applet
<rycee>
manveru: If this is for a systemd service then you may have to add `systemctl --user import-environment` to `xsession.profileExtra`.
<rycee>
`systemctl --user import-environment QT_PLUGIN_PATH` that is.
<leo60228>
manveru: i don't see anything weird there
<manveru>
i'll try that
yegortim1 has joined #nixos
<rycee>
leo60228: Maybe the audio is sent to the wrong device?
<leo60228>
only one device plugged in, and it's selected
<leo60228>
it's worked before
<manveru>
leo60228: well, it's a bit hard to say what's the issue... did you plug the speakers in right? does your soundcard show up under pavucontrol configuration, is your sound application under playback and has the right sink, is the port in output devices correct?
<clever>
leo60228: also try toggling each of the mute flags in `alsamixer -c 0` they are sometimes on mute by default
<leo60228>
it's a laptop, kde shows it and lets me change to volume
<rycee>
leo60228: I have that problem sometimes. I change the output device and then forget to change it back and later wonder why I can't hear anything.
<leo60228>
how do i toggle a mute flag
<clever>
leo60228: the m key
<clever>
and if it says MM, its on mute, an OO lit up is enabled
<leo60228>
nop
<leo60228>
e
<clever>
leo60228: do you see the VU meter moving in pavucontrol?
<leo60228>
what's that?
<clever>
there should be a moving line in pavucontrol, that moves with the level of the noise the program is making
<leo60228>
it's moving
<clever>
and what output is it set to?
<leo60228>
speakers
<leo60228>
the only other one is headphones (unplugged)
<clever>
what is the sound card set to on the configuration tab?
<leo60228>
analog stereo duplex
<clever>
try changing that configuration
<leo60228>
i changed it to analog stereo output, which didn't fix it, all the other ones say unplugged
<sphalerite>
duplex should be the right one
<sphalerite>
I've found that sometimes plugging something in and unplugging it helps
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] sengaya opened pull request #36250: ansible-lint: init at 3.4.20 (master...add-ansible-lint) https://git.io/vA94z
NixOS_GitHub has left #nixos [#nixos]
<leo60228>
it's hardware though
alex`` has quit [Ping timeout: 240 seconds]
<clever>
ive found that the internal/external switching is driven by software
<clever>
on one laptop, it could only drive one at a time, but the system beep demanded to go from the internal speaker
<leo60228>
i don't have headphones plugged in, i could try plugging them in though
<clever>
so it would temporarily switch from headphones to internal, beep (and leak some music), then switch back
<clever>
so other people in the room would randomly hear a beep mixed with music, lol
<sphalerite>
leo60228: yeah try plugging them in and unplugging them
<leo60228>
didn't switch automatically, manually switched to headphones and it worked, switched back and it didn't
<clever>
leo60228: the speakers might be damaged
<mfiano>
which manpage describes valid values for `period` as mentioned in `--delete-older-than period` in the nix-collect-garbage manpage?
<clever>
yeah
<manveru>
mfiano: afaict, the only valid option is the 'd' for days one
<manveru>
at least from reading the nix source, and no other ones are documented either
<mfiano>
manveru: Ok thanks
leo60228 has quit [Ping timeout: 260 seconds]
<manveru>
rycee: any idea why `home-manager generations | cut -d' ' -f5 | xargs -n1 echo home-manager remove-generations` doesn't work but manually passing the id does?
<manveru>
- the echo
<manveru>
maybe the coloring...
<manveru>
ah, yeah, a hexdump shows there's a control code before the space :)
<clever>
most programs will do istty() on stdout, and turn colors off automatically
<mfiano>
Where is the documentation for lib.mkDefault? Can't seem to find it :)
qewzee has joined #nixos
<manveru>
well, `home-manager generations | sed 's/.*id \([1-9][0-9]*\).*/\1/' | xargs home-manager remove-generations` works for now
<mfiano>
LnL: Ah hmm. Can you birefly describe what it is doing? I see this a lot in expressions I read
Fare has joined #nixos
<fyuuri>
I am new to nixos. Is the general workflow for installing packages using configuration.nix or nix-env? I am the only user of the system and I want to share the config among multiple PCs.
<LnL>
manveru: I usually use awk '{print $5}' instead of cut, it handles whitespace and maybe things like ansi escapes better
<LnL>
mfiano: it's similar to mkForce but with a low priority instead
<manveru>
LnL: no, that also includes the escape code
<mfiano>
fyuuri: nix-env installed packages are not upgraded or managed when you do a rebuild. You can use it to install local packages for your user, etc that won't be added to your config.
<manveru>
awk splits only at spaces too
<mfiano>
LnL: Thanks
periklis has quit [Remote host closed the connection]
<fyuuri>
mfiano: what do you normally use? Sry. I am completely new to systems like nixos. (I used arch for a long time)
ryanartecona has joined #nixos
<mfiano>
fyuuri: I am completely new this week myself, but I use the config for everything except packages I only need temporarily or for a specific user, to be honest.
<fyuuri>
Do the nix-env and configuration.nix methods interfere with each other? I gues the nix-env takes precedence when both methods were used to install the same package?
<mfiano>
fyuuri: I also used Arch for almost 2 decades before this week
<fyuuri>
:))
<srhb>
fyuuri: They do not interfere with each other. The sharable way is to use configuration.nix or something similar. nix-env is very ad-hoc and imperative (install this package now, with no record of it)
<LnL>
manveru: not just spaces: echo "foo$(echo 'foo\t bar' | awk '{print $2}')foo"
<manveru>
LnL: :)
<srhb>
fyuuri: With configuration.nix you can easily have a list of packages that you share between machines (imports = [ ... ./mypackages.nix ];)
<LnL>
but yeah control character are probably not included
<fyuuri>
Will the usage of nix-env "trash" my system? Or can I cleanly manage this packets and remove them aswell? Will these packets survice a nixos-rebuild switch?
<srhb>
fyuuri: nix-env installed packages are in your user environment. configuration.nix ones are in your system environment. Both can coexist peacefully.
<fyuuri>
so nixos-rebuild does not affect the home?
<srhb>
Indeed.
<fyuuri>
are the packages installed to .local/ ?
<srhb>
No, they're still in /nix
<srhb>
fyuuri: Take a look at /nix/var/nix/profiles
<fyuuri>
ok thx!
<srhb>
fyuuri: You'll have some system links and some per-user links
<fyuuri>
thx!
<fyuuri>
that clarifies a lot :)
<srhb>
fyuuri: ~/.nix-profile will be a symlink to /nix/var/nix/profiles/per-user/you/profile
<srhb>
That in turn will be a link to your current user env generation
<srhb>
(For purposes of rollbacks and such)
<fyuuri>
thats nice
<srhb>
Yes it is :)
<srhb>
It's all a bunch of symlinks in the end. One for your user profile, one for your system profile.
<fyuuri>
My main reason for switching is to be able to do frictionless updates with rollback capabilities even if I did not update for a while
<srhb>
fyuuri: That's a good reason.
<srhb>
Being able to roll back is nigh-impossible on other systems.
<fyuuri>
And, of course, being able to set up a new system fast :)
<mfiano>
fyuuri: If that's the case you will soon forget Arch. It only took me a couple days until I was sold for all of my devices.
<srhb>
fyuuri: I avoid nix-env because of that last argument.
<fyuuri>
yes. That is what I thought.
<srhb>
fyuuri: Being able to carry a configuration.nix from one system to another with only minor corrections (or even parameterization) is a blessing.
<srhb>
fyuuri: In the end my main configuration.nix is essentially a small wrapper with system-dependent setup (hostname, ip, ...) which then includes other more generic files.
<srhb>
Eh, it was a long learning experience for me :-)
<fyuuri>
srhb: I will try to do the same.
<manveru>
something like this is my configuration.nix :)
<srhb>
(I came from Arch too, approximately... two years ago?)
<fyuuri>
thats nice :)
<fyuuri>
I will manage mine in git aswell :)
<ottidmes>
I am getting this warning: substituter 'ssh://root@server' does not have a valid signature for path, I have set secret-key-files and trusted-public-keys to what I have generated with nix-store --generate-binary-cache-key, is there something else that I need to do to make it a valid substitutor?
<srhb>
ottidmes: Did you upgrade to Nix 2.0 recently?
<ottidmes>
srhb: Jep, today
<srhb>
ottidmes: I think you'll essentially have to sign them all with nix sign-paths, but I haven't done it myself yet.
<LnL>
did you restart the daemon after adding the key?
<srhb>
Or listen to LnL :-)
<ottidmes>
LnL: Yeah, I have restarted it, and NixOS did when I updated the nix.conf, but can try to do so again
<fyuuri>
Do you have to manually install "nix"?
<ottidmes>
Nope, no change (as expected, it should already have been restarted)
<ottidmes>
fyuuri: On any other distribution than NixOS you will have to do so yeah, but in NixOS it is given, part of the core tools, but you can specify which package of nix you want to use via nix.package in your configuration.nix file, which I did to update to a pre version of Nix 2.0
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 1 new commit to master: https://git.io/vA9R9
NixOS_GitHub has left #nixos [#nixos]
<clever>
confusingly, they didnt change the default when doing the release
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] abbradar pushed 1 new commit to master: https://git.io/vA9RH
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master d5e82d6 Nikolay Amiantov: ibus-engines.uniemoji: add pygobject3 as a dependency...
<ottidmes>
fyuuri: nix-store --version will for example show the version of Nix you are running, in Nix 1.x there are a bunch of nix-* binaries, in Nix 2.x they are still there (at least for now) but you also have a more unified nix binary
<fyuuri>
ok. thx!
knupfer has quit [Ping timeout: 256 seconds]
<fyuuri>
1.11.16
<fyuuri>
Is the nix repl only available in 2.0?
<ottidmes>
clever: do you know anything about Nix 2.0 substitutors and how to make them be signed correctly?
<fearlessKim[m]>
since I switched to nix 2.0, my nixos-rebuild are stuck :/
<clever>
ottidmes: not yet
<ottidmes>
fyuuri: nix-repl is available in 1.x, might be some differences between the two though (that goes for all, nix-build is not identical to nix build)
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master 88dae1a Matthieu Coudron: protocol: module to application...
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 2 new commits to master: https://git.io/vA9RF
<fearlessKim[m]>
well I guess it's just the network being very slow... and in absence of feedback :s
badi has joined #nixos
<sphalerite>
fearlessKim[m]: you can do nix build -f '<nixpkgs/nixos>' config.system.build.toplevel to build it before running nixos-rebuild, that way you get fancy progress info
<fearlessKim[m]>
I would like to see the packets download advance, I often have connectivity issues and it's annoying to just see "copying paths"
<sphalerite>
nix build should show progress in MB
<fearlessKim[m]>
I checked with wireshark, it does seem to download but nix started downloading several big packages in parallel (kernel & co ) so it seemed stuck
<fearlessKim[m]>
sphalerite: your command is much better !
<gchristensen>
is there a way with nixos' modules to set the default gateway for two different IPv6 IPs?
<gchristensen>
derp, IPv4*
<sphalerite>
gchristensen: you mean setting non-default routes, or setting multiple default routes?
<gchristensen>
I have three IPs, a private IPv4, public IPv4 and an IPv6... each has its own gateway, and I'm not sure how to handle that
ryanartecona has quit [Quit: ryanartecona]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 closed pull request #31019: [RFC/RDY] make l2tp work with Strongswan (master...strongswan_rebased) https://git.io/vFqxI
NixOS_GitHub has left #nixos [#nixos]
<sphalerite>
gchristensen: you should only need one default gateway per protocol
<sphalerite>
IPv6 should be fine since you only have one case to cover
<sphalerite>
For IPv4, I'm not quite sure what you mean. The default gateway is the next hop that your machine will send packets to when their destination isn't a local address
<mkaito>
the nixops docs say to build with `nix-build release.nix -A build.x86_64-linux`, which results in a `result` symlink in current folder. How would I go about installing it to my profile?
<Mic92>
Are there any pull requests that you feel are ready for quite a while, but no maintainer has looked at them lately?
<sphalerite>
mkaito: nix-env -i ./result
<gchristensen>
sphalerite: I guess they want me to route it to a different IP depending on if it is a private IP or public IP
<mkaito>
sphalerite: cheers
<sphalerite>
mkaito: or you can just run it from there — ./result/bin/foo
<clever>
mkaito: ive also done nix-env -f release.nix -iA build.x86_64-linux
<sphalerite>
gchristensen: the public one is the default gateway, for the private one it's not quite clear, usually private addresses are local ones
<clever>
gchristensen: i think for the private ip, you need a route for its private LAN, with the gateway set
<clever>
gchristensen: since you cant have 2 default gateways on v4
<sphalerite>
gchristensen: a lot of home networks are on the 192.168.1.0/24 subnet and have the default gateway 192.168.1.1. That means that all addresses in 192.168.1.0/24 (192.168.1.(0-254)) can be reached directly on the LAN, and anything else goes through 192.168.1.1
<clever>
ip route add 10.0.0.0/8 dev wan2 via 10.1.2.3; for example
<sphalerite>
you shouldn't need a gateway for private networks usually
<clever>
the command i typed above, forced a certain subnet to go via a certain gateway, and the rest goes to the default
badi has quit [Ping timeout: 260 seconds]
<clever>
in my case, the ISP has a 10.x.y.z network for the tv service, complete with multicast, and undocumented routing tables
<gchristensen>
maybe I don't need one, we'll find out :)
<sphalerite>
samueldr: oooooh I never saw that PR! Pretty!
<mkaito>
clever: that's even easier!
<xnaveira[m]>
Iäm trying to package this python module https://github.com/Juniper/py-junos-eznc and i am having trouble figuring out the checkPhase (spoiler alert i am no python guru)
<samueldr>
let me verify everything is in the repo
<samueldr>
I woke up ~25 minutes ago
<sphalerite>
xnaveira[m]: note that long or multiline messages don't get relayed to IRC properly
<xnaveira[m]>
sorry i ment checkPhase = ''py.test ''; @spha
<xnaveira[m]>
sphalerite:
<samueldr>
Mic92: eveything should be there
<samueldr>
(on the nixos-artwork repo)
<sphalerite>
xnaveira[m]: the standard way is not specifying a checkPhase at all AFAIK, and making sure doCheck is set to true (I think it defaults to it on python)
<sphalerite>
xnaveira[m]: whether that actually works depends on upstream though
<neonfuz>
After updating from linux 4.9 to 4.14 I get some errors at boot, can anyone help me figure out some things?
<neonfuz>
I get some errors related to TSC_DEADLINE, and says I need to update my microcode, how do I do that on nixos?
<infinisil>
ottidmes: i don't know of such a thing
<sphalerite>
xnaveira[m]: well by the looks of it (looking at travis.yml) they use nose to run their tests. If just setting doCheck = true; doesn't run them try just copying the nosetests line from the travis config
<mt_caret_>
I'm currently trying to update a library that lot's of packages depend on (of which a few break when updated). If I retain the old library under a different attribute name, is it possible to get a broken package (along with its dependencies, recursively) to build using the old library?
<mt_caret_>
s/lot's/lots/g
teto2 is now known as saber
<sphalerite>
mt_caret_: change the callPackage in all-packages.nix — see docker for an example
saber is now known as Guest46892
<rycee>
manveru: Good point. I'll try fixing that when I get back in a few hours, it should only use colors when running in an interactive shell.
Guest46892 is now known as teto2
<rycee>
Sorry, when the output is a terminal with color support…
<xnaveira[m]>
The nosetests line did the trick, thx sphalerite !
<neonfuz>
any idea on the "error parsing PCC subspaces from PCCT" error?
<neonfuz>
some people say this happens on dual boot systems, though I don't dual boot
<neonfuz>
something to do with bios / uefi? I'm pretty sure I'm booting through uefi but not 100% sure, any way to check on a booted system if it's been booted from uefi?
<samueldr>
it could come from a bios(uefi) setting used for that feature
<Mic92>
ryantm: now I hate myself for doing bikeshedding. I only feel that there were some bad naming descicions in the past like nativeBuildInputs
<mt_caret_>
sphalerite: The problem is that other dependencies for the broken package also depend on the library I'm trying to bump, I need to use the old library for them too. Does docker do this?
<ottidmes>
infinisil: just went through all options on https://nixos.org/nixos/options.html and cannot find it either, might have been confused, thought I remember seeing such an option being mentioned once, but I can just (mis)use boot.loader.grub.extraPrepareConfig for now since I am using grub, and I am actually using it for something related
badi has quit [Read error: Connection reset by peer]
<sphalerite>
mt_caret_: I'm not really sure. Is there no more up-to-date version of the broken one which works with the updated library?
<infinisil>
ottidmes: what are you using it for?
<neonfuz>
samueldr: if you're talking to me, I'm not sure exactly what PCCT and PCC are even
<samueldr>
I don't know either, but from a quick googling of your error message, it looks like "fast boot" is a recurring theme for that message, which is sometimes an option in the UEFI settings
<mt_caret_>
sphalerite: unfortunately not. To name the package, it's bap (which is written in OCaml). It depends on v0.9.0 of some widely-used libraries, but it's been a few months since v0.10.0 has been out, and the two are not really compatible.
<ottidmes>
infinisil: I have an hardware encrypted drive, but it allows me to write an small image to boot into, in the initrd of the image I unlock the drive and then want to boot into the most recent kernel/initrd pair on the unlocked disk, so I want to make this info available in a file on /boot
jensens has quit [Ping timeout: 268 seconds]
<ottidmes>
infinisil: I had it working by just parsing /boot/grub/grub.cfg, but apparently it was not resistant enough, it stopped working because probably one of my greps got different results than I assumed, so I want to use something less hacky instead
<clever>
ottidmes: have you seen the extlinux.conf stuff?
<mt_caret_>
sphalerite: the callPackage { a = use 0.9.0 } trick doesn't work since the other libraries it depends on depends on the new one, and causes breakage.
<ottidmes>
clever: Yeah, I am using that to build the image
<neonfuz>
samueldr: hmm... well I hope I don't have errors down the road but things seem to be working alright now
<neonfuz>
I'm just paranoid because last night I accidentally GC'd after nuking my wifi driver,and the only way to get ethernet is with an adapter that is frayed and might stop working at any moment *_*
<ottidmes>
clever: Unless I can make it to only generate the extlinux config file, I will just generate something myself
<neonfuz>
damn lenovo and their proprietary $30 ethernet "adapter"
<clever>
ottidmes: you can also look at how bootloaders work in nixos, and just define a custom one,one sec
<mt_caret_>
I mean, there must be a way for multiple versions of libraries to coexist relatively seamlessly, or it'll mean that every backwards-incompatible version bump of a library will permanently break some app that isn't maintained anymore.
<clever>
ottidmes: when enabled, a bootloader must set system.build.installBootLoader to a string that properly configures /boot, dealing with gathering all the generations, generating the config, and installing things into MBR's
<ottidmes>
clever: Yeah, that is what I am doing now, inject my code in boot.loader.grub.extraPrepareConfig that will generate 3 lines, the kernel, the initrd, and the kernel params, and read this file in the unlock image
<clever>
ottidmes: you could also just disable grub, and make your own bootloader, that directly sets system.build.installBootLoader
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] earldouglas opened pull request #36255: rtl_433: init at 51d275c (master...rtl_433) https://git.io/vA9ut
NixOS_GitHub has left #nixos [#nixos]
<clever>
ottidmes: oh, but if grub is gone, it probably wont be able to reboot, lack of an MBR on the unlocked disk
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] PierreR opened pull request #36256: Update salt-pepper from 0.5.0 to 0.5.5 (master...master) https://git.io/vA9uq
NixOS_GitHub has left #nixos [#nixos]
<clever>
yeah, keep the mutated grub your already doing
badi has joined #nixos
<ottidmes>
clever: the file I am after is only used in the unlock image, the whole unlock image is hidden when the drive is unlocked, so then it should just be using grub (when rebooting e.g.)
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master f4d1d26 MERCIER Michael: pass: Make tomb plugin optional (disabled by default)
<ottidmes>
clever: But the line you linked to system.build, I know it from doing nix-build -A config.system.build.<...> but can I just add a builder and it will be build when I am doing a nixos-rebuild?
simpson1 is now known as simpson
the_real_plumps has quit [Quit: No Ping reply in 180 seconds.]
<clever>
ottidmes: should be able to
<clever>
just refer to ${config.system.build.whatever} inside a string in any nixos module
<ottidmes>
clever: But would I be able to use it to run anything outside of Nix, write a file to /boot
<xnaveira[m]>
ok so now i have this other python package where they have "test.py" in the root, i do checkPhase = '' SCPPY_PORT=10022 ${python.interpreter} test.py ''; and it gives me /nix/store/7b2z0vfbs9539ga4pxx5gmli47rz5y3n-python-2.7.14/bin/python2.7: can't open file 'test.py': [Errno 2] No such file or directory
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master 9b29c06 Pierre Radermecker: Update salt-pepper from 0.5.0 to 0.5.5
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 2 new commits to master: https://git.io/vA9ug
<xnaveira[m]>
ok, nevermind, it turns out that the pip package doesn't include the test
<clever>
ottidmes: and you should be able to run it whenever you want, provided you give that $out path still
nuncanada has quit [Ping timeout: 268 seconds]
<ottidmes>
clever: Yeah so the bootloaders are special cases in that they are allowed to affect the world outside nix, so I have to write my own, or inject my code in an existing
<simpson>
xnaveira[m]: Yeah, some Python folks have a bad habit of shipping broken tests like this. It happens.
<clever>
ottidmes: yep
<teto2>
xnaveira[m]: actually I had two similar cases recently where tests files were missing
<xnaveira[m]>
ok, as I said I\m no python guru, i just need to get a bunch of dependencies working for an ansible playbook :P
<xnaveira[m]>
and I thouht i might as well contribute them and learn some nixos expressions stuff
<xnaveira[m]>
I think i am done now, going to submit to gitub, thx guys
<simpson>
xnaveira[m]: There is zero shame in `doCheck = false;` on Python packages.
<gchristensen>
yeah I was worried that might be it :)
<LnL>
I'm using networking.localCommands for something similar :/
<clever>
gchristensen: i cant think of a good spot off the top of my head, and i have found the network layer to be a bit statefull, and doesnt always update itself right
<gchristensen>
indeed
<sphalerite>
gchristensen: networking.interfaces.<name>.ipv4.routes on unstable/18.03
<clever>
sphalerite: oh, that sounds perfect
<gchristensen>
wow! I might have to upgrade early :)
<clever>
sphalerite: also, nixos-unstable cant build hydra, so my router is stuck on 17.09
badi has quit [Quit: WeeChat 1.9.1]
<attila`>
Hi, I am creating my first derivation on NixOS for qtcam, one of its dependencies is libusb, so I added it to nativeBuildInputs. However nix-build can not find libusb.h.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to master: https://git.io/vA9zB
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master ac3b537 Vladimír Čunát: 18.03 release notes: highlight linux and gcc updates
<sphalerite>
attila`: you may need to add pkgconfig to nativeBuildInputs; libusb itself should be a buildInput, not nativeBuildInput
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/nix-2.0 b70c93f Vladimír Čunát: Merge branch 'master' into nix-2.0
<NixOS_GitHub>
nixpkgs/nix-2.0 729d72f Vladimír Čunát: 18.03 release notes: nix-2.0 by default
<NixOS_GitHub>
[nixpkgs] vcunat pushed 2 new commits to nix-2.0: https://git.io/vA9z0
NixOS_GitHub has left #nixos [#nixos]
<sphalerite>
attila`: looks like it's built with qmake and just expects the headers to be in /usr. You may need to set INCLUDEPATH manually using qmakeFlags to get it to find them correctly
iyzsong has quit [Ping timeout: 240 seconds]
<clever>
sphalerite: is the qmake file checking for the existance, or is it the compiler that fails to find it?
<attila`>
sphalerite`: Thank you. I'll try, then report back
badi has joined #nixos
<attila`>
it is a compiler error
nixer101 has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 1 new commit to master: https://git.io/vA9zy
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master c94ad64 Erik Arvstedt: rsbep: init at 0.1.0...
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to master: https://git.io/vA9zF
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master d586f5d Vladimír Čunát: Merge branch 'nix-2.0'
<nixer101>
Hello everyone! Can you comment the situation with nixos-unstable and nixos-17.09 split (further on "stable") ? I noticed that the version in 17.09 are all at least 3-6 months old, even if there is constant activity. The nixos-unstable may be days old, but has recent packages and sometimes encounters build errors with expressions, but not with packages themselves. Is there a channel with cutting edge packages that actually pass build?
<nixer101>
Specifically, the palemoon package in stable is half year old and segfaults, where version from unstable is not unstable at all and has no segfaults.
<gchristensen>
nixer101: stable backports are best-effort, and typically only for security updates
<sphalerite>
nixer101: the idea is that versions stay stable in stable and only major bug fixes get backported
<gchristensen>
(palemoon updates should all qualify for backport, though)
<nixer101>
I mean, what is the best option for the channel with cutting edge software that has no build-time and expression errors. I am fine with package-related errors.
<nixer101>
gchristensen, sphalerite: please pardon my goofiness, but where can I find nixos stable backports?
<sphalerite>
nixer101: the master branch is the danger zone, the various unstable channels are based on master but only advance when certain test sets pass on them
<nixer101>
gchristensen: unfortunately, the kde packages are all 6 months old and palemoon specifically segfaults. I would like to find a channel that is cutting edge, so that the expression always builds, but the packages themselves may have own errors (and help fix them).
<gchristensen>
yes, unstable is unstable. there is no perfectly stable unstable.
<sphalerite>
nixer101: yes. You're looking for nixos-unstable
Itkovian has joined #nixos
<gchristensen>
but we can't give you exactly what you want
<sphalerite>
gchristensen: "I would like to find a channel that is cuttingedge, so that the expression always builds, but the packages themselves may have own errors (and help fix them)." sounds like unstable to me :)
<LnL>
you can use 2 channels an only use what's broken from unstable
k2s has joined #nixos
<gchristensen>
sphalerite: they don't always build
badi has joined #nixos
<mfiano>
to be fair, stable doesn't always build either
<mfiano>
zathura for example, fails, or it did 2 days ago anyway
<nixer101>
sphalerite: the problem with nixpkgs-unstable is that expressions themselves can fail. I would appreciate the channel, that is build-tested, but the bugs of the package itself are omitted.
<LnL>
those are mostly edgecases
drakonis has joined #nixos
<MichaelRaskin>
Making everything build is a huge undertaking which is only rarely _completely_ succesful
<MichaelRaskin>
_Checking_ that everything builds takes multiple days.
<nixer101>
sphalerite: there is limitation with stable, the 6 months changes freeze, that prevents pulling in new versions. These new versions may actually be bugfix releases. This is the problem that makes stable "unstable" in practice.
<sphalerite>
yes so use unstable
<simpson>
nixer101: To spell it all out, fresher packages are inherently more unstable because they have had less time to be in contact with other packages. The feature that makes our unstable channel fresh is what makes it unstable.
<nixer101>
LnL: I have added unstable injects yesterday. From about 20 injected unstable packages, 5 have failed to build and about 5 had same versions as in stable.
<simpson>
You can't get one without the other.
Itkovian has quit [Ping timeout: 256 seconds]
<nixer101>
LnL: and they all added about 5 gigabyte of binaries.
mkoenig_ has joined #nixos
<simpson>
(You *can* have a custom nixpkgs, based on applying your own pile of commits, but it doesn't sound like you're at that stage yet.)
mkoenig has quit [Ping timeout: 240 seconds]
<nixer101>
simpson: no, I am not, this is right. But I pulled a browser from stable on stable system, and it segfaulted. Luckily the Konversation did not segfault, so our members have given me directions how to add unstable. After I injected unstable palemoon into stable, it was rock solid.
<sphalerite>
or you could just use a browser that many people use and that gets tested better
<sphalerite>
using a niche browser on a(n unfortunately still) niche distro is just naturally more likely to be problematic
dash1 is now known as dash
<nixer101>
sphalerite: the palemoon version in the stable is half a year old. The version in unstable is fresh and is stable by upstream. Between these there were several bugfix releases.
<simpson>
sphalerite: We should not paper over bugs when we could fix them.
<gchristensen>
sphalerite's comment is totally correct that the size of both the communities together means it is less likely to work and be upto date
<sphalerite>
I don't mean to paper over it, but fixing everything takes a lot of effort and that effort isn't always available. And other browsers do work
Arcaelyx has joined #nixos
<sphalerite>
nixer101: it's normal for the version in stable to be half a year old, since that's how old stable is
<LnL>
nixer101: what you're running into is something that should get fixed on stable, however there's a _lot_ of software in nixpkgs so in practice this won't always happen for things that are not used by many people
<nixer101>
sphalerite: yes, the stable version is unstable. Thats the problem I would like to mention. Unstable because its old and was long patched.
<nixer101>
sphalerite: is this okay to request for version bump for the stable in this case? Is such version bump possible per policy?
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 4 new commits to master: https://git.io/vA920
<NixOS_GitHub>
nixpkgs/master 22c5082 Xavier Naveira: pythonPackages.ncclient: init a 0.5.3
<NixOS_GitHub>
nixpkgs/master 284c6a5 Xavier Naveira: pythonPackages.scp: init at 0.10.2
<NixOS_GitHub>
nixpkgs/master 1689c3c Xavier Naveira: pythonPackages.junos-eznc: init at 2.1.7...
NixOS_GitHub has left #nixos [#nixos]
<LnL>
nixer101: yes definitively, there's no point in keeping something on an old version if it doesn't work
<nixer101>
simpson: There are three different categories of bugs. 1) bugs in build expression, where package does not build. 2) bugs in how package interacts with rest of ecosystem, - runtime bugs. 3) finally application specific bugs that are to be reported upstream. I am looking for a channel that filters 1+2, but allows 3.
<nixer101>
simpson: Three categories as I see it.*
<MichaelRaskin>
2 requires manual testing
<nixer101>
LnL: thank you
<MichaelRaskin>
Or many many more tests
<sphalerite>
nixer101: we don't have that
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 2 new commits to master: https://git.io/vA929
<NixOS_GitHub>
nixpkgs/master bd27fbb bernie: update dashd to 12.2.3
<MichaelRaskin>
1 is a goal that is sometimes susccesfully approximated, but not always achieved in Nixpkgs (stable or unstable)
<MichaelRaskin>
Just because even checking it takes _a lot_ of time.
<simpson>
nixer101: (1) on its own is something we care about and generally something we try to fix even on unstable. (1+2+3) doesn't exist in any distro I've ever used or seen, but it can be aspired to.
<LnL>
nixer101: for something with a subtle bug that have a lot of dependencies we might not want to do it to avoid breaking more (or more important things) then it would fix
<simpson>
(1+2) is something that can only be achieved if software in general makes massive improvements in quality in many different ways; I doubt it'll ever happen.
<nixer101>
simpson: well, Gentoo. Although its specific to package manager, as it explicitly allows choice of versions.
<MichaelRaskin>
Nix technically supports it, and Nixpkgs doesn't do it more mostly because of combinatorial explosion
<simpson>
nixer101: Choice of version isn't the only thing going on here; we might be talking past each other here.
<MichaelRaskin>
We try to have _some_ binary cache
<nixer101>
simpson: I don't want to bring Gentoo to discussion at all though. Its completely different design, with problems in package dependencies and requirement for manual configuration management.
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 2 new commits to master: https://git.io/vA92h
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master 023d584 Väinö Järvelä: wireshark: Add Darwin support
<simpson>
nixer101: Sorry, I'm out of polite words. Your desired channel would not be feasible. Feel free to start your own channel if you disagree.
<MichaelRaskin>
Well, there are multiple questions. What actually happens in Nixpkgs, what is a priority for improvement in Nixpkgs, what is >70% likely to get accepted as a PR if someone does the work, what doesn't have any conceptual issues but is not currently considered a good idea.
<nixer101>
simpson: No, I understand you perfectly sir. I am trying to find the way in current situation that just suits my needs better, perhaps motivate to improvements, if there is place for them.
<simpson>
nixer101: Learn to write Nix expressions and submit PRs. Be the change you wish to see in-tree.
<MichaelRaskin>
Nixpkgs does have a long tail. We accept more packages than we can polish or test for each commit or whatever.
<nixer101>
simpson: Thats definitely on my horizon of actions, sor.
<nixer101>
sir*
<MichaelRaskin>
There is a binary cache that targets building a snapshot of master (after testing that becomes the unstable channel) and the stable.
<simpson>
nixer101: Rude~
simpson has left #nixos ["WeeChat 1.0.1"]
<MichaelRaskin>
This disincentivizes accepting too many versions of a package without a compelling reason.
<MichaelRaskin>
If a package is broken on the segfault or build failure level, any fix will be accepted into stable (either upgrade or patch), which doesn't mean someone actually cares enough to have already applied it.
<LnL>
something like requiring an active maintainer for a package to be in nixpkgs would be one way of improving the general quality of things
<nixer101>
MichaelRaskin: Is there guarantee that nixos-unstable is free from 1. category of bugs (evaluation, build-time problems)? The question is because I have experimented with nixpgs-stable channel and found it problematic to be used as main channel.
ElGoreLoco has joined #nixos
coot has joined #nixos
ElGoreLoco is now known as jorge-jbs
<nixer101>
LnL: understandable, thanks!
<MichaelRaskin>
LnL: it would also require cutting down the number of packages in three or so
<sphalerite>
nixer101: if nixos-17.09 and nixos-unstable are both problematic then nixos is unfortunately not the distro for oyu
<MichaelRaskin>
nixer101: search bug tracker for ZHF
<LnL>
yes, that obviously has some implications :)
<MichaelRaskin>
LnL: I would probably abandon that version and switch to a fork that doesn't care and keeps coverage.
<nixer101>
sphalerite: no, I would like to betone specifically, that nixos is for me and I find it the by far best distribution. I am just trying to figure out the way.
<MichaelRaskin>
nixer101: these are the attempts to achieve full build success.
<nixer101>
sphalerite: please don't see my questions as some kind of demands for changes or something.
<MichaelRaskin>
nixer101: I think you have a wrong impression of some of the scaling issues; reading a few latest ZHF sprints should help.
<sphalerite>
nixer101: that's nice to hear! :D but yeah mixing channels is probably the easiest way to go
sigmundv__ has quit [Quit: Leaving]
aarvar has quit [Ping timeout: 260 seconds]
<MichaelRaskin>
And also ZHF threads show where Nixpkgs usually falls behind on fixing things
k2s has quit [Remote host closed the connection]
<nixer101>
MichaelRaskin: Could I ask you what ZHF means? Sorry if I sound stupid.
<LnL>
MichaelRaskin: yeah, I don't think it would be a great solution. there's a ton of stuff in nixpkgs that might not have been touched in a year but still works perfectly fine
<sphalerite>
zero hydra failures
<MichaelRaskin>
They _are_ named ZHF in the issue tracker, though.
<LnL>
that's the effort we do to go from unstable to zero failures for a new release channel
<mfiano>
sphalerite: Is it worth filing an issue about sdhci-pci not propagating from the installer environment to the generated harware-configuration.nix?
<sphalerite>
definitely
<Dezgeg>
it's definitely a hard problem to solve, do you know if other distros manage to automatically detect that?
<mfiano>
Ok, I've seen so much weirdness with this awful hardware that it's hard to tell what should be reported
<mfiano>
Dezgeg: I'm not sure, but I've installed Arch and Ubuntu on this hardware multiple times without issue
<Dezgeg>
okay
<mfiano>
Well different hardware, same make/model (wife's laptop is the same)
halfbit has quit [Ping timeout: 255 seconds]
<Dezgeg>
I think it's enough to have the script look at parent devices of all block devices (and recursive all the way up the tree) but would be useful to use something that other distros use as well
<sphalerite>
couldn't you also rmmod it and it would complain if it's the eMMC?
<mfiano>
Yeah I removed them all just now
<mfiano>
Ok good. So I only really need the generic sdhci-pci module for the internal storage
acertain has quit [Remote host closed the connection]
<clever>
Mar 03 14:40:21 amd-nixos display-manager[9218]: (EE) Screen 0 deleted because of no matching config section.
<mfiano>
Yeah, I think a better strategy is needed for the .nix config detection. it added usb-storage because i had the livecd usb plugged in, even though its not required to boot the root filesystem. it's adding sdcard, and failing to add the only module it actually needs to boot the rootfs
<clever>
Mar 03 14:40:21 amd-nixos display-manager[9218]: (II) UnloadModule: "amdgpu"
<clever>
power went out for a split second, and now i discover that my gpu drivers are toast
<sphalerite>
clever: as long as it's the drivers and not the GPU I guess..?
<mfiano>
but i admit it's a hard problem to solve. no real way to detect what device the root fs will be installed on
<clever>
355 and 354 wont start...
<Dezgeg>
during nixos-generate-config it does know that
<sphalerite>
^
<clever>
352 wont start
<mfiano>
how?
michas_ has quit [Ping timeout: 265 seconds]
<sphalerite>
mfiano: well it's mounted at that point
<clever>
sphalerite: ok, thats strange, i rolled back to a generation older then the last boot, and X still cant start
<sphalerite>
clever: uh oh.
<mfiano>
sphalerite: ah very true. I'm not sure why it's adding a bunch of storage modules then, even for the device the livecd is mounted on
<mfiano>
i think this really needs to be fixed proper
<clever>
mfiano: i found an impurity, where it will add usb storage drivers if any usb storage is connected
<sphalerite>
it just goes through all the block and mmc devices in the system
<mfiano>
I'll document all this into an issue and begin writing a hardware-specific wiki page tomorrow when i'm more awake
<sphalerite>
and works out what they're using
<sphalerite>
as opposed to basing it on what's mounted
<sphalerite>
clever: does it really count as an impurity? The whole of nixos-generate-config is impure :p
acertain has joined #nixos
<clever>
sphalerite: its a variable that doesnt depend on what you mounted to /mnt/
<sphalerite>
for purity to make sense here we'd have to have a well-defined set of inputs
<sphalerite>
which we don't really
<mfiano>
Well at the very least, it fails wrt sdhci-pci then. I guess it doesn't hurt the initrd to have a few extra modules that aren't needed, but it hurts it when the only 1 that is needed, isn't included
nixer101 has quit [Quit: Konversation terminated!]
<sphalerite>
yeah
<clever>
sphalerite: i suspect its a kernel side issue, switching to an old generation doesnt fix it, but booting into an old one does
chocopuff298 has joined #nixos
<sphalerite>
oh so the failure was when starting back up after the power failure
<Dezgeg>
are the drivers the same version in the generations? I wouldn't expect that to work with proprietary drivers
<sphalerite>
I misunderstood, I thought the power failure was short enough for the computer to stay on but X crashed
acertain has quit [Ping timeout: 245 seconds]
<sphalerite>
amdgpu isn't proprietary is it?
<clever>
sphalerite: i think thats the open-source one
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dtzWill opened pull request #36259: clementine: apply upstream patch to fix build w/gcc7 (master...fix/clementine-gcc7) https://git.io/vA9rz
<NixOS_GitHub>
nixpkgs/master b2acbe5 Jörg Thalheim: nixos/openldap: add test
Myrl-saki has quit [Ping timeout: 256 seconds]
<mfiano>
Sleep. Thanks for all the help today
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 closed pull request #36163: deepin: add meta package for the Deepin Desktop Environment (DDE) (master...new.deepin) https://git.io/vAMVE
NixOS_GitHub has left #nixos [#nixos]
<clever>
sphalerite, lejonet: i should probably experiment with the options before i open everything back up
<lejonet>
clever: opencl is interesting, I'm waiting for 4.16 when ROCm hits mainline, so that you don't have to use the proprietary driver to do opencl
<attila`>
sphalerite: You were right. I added `sed -i 's|/usr/include/libusb-1.0|${libusb1.dev}/include/libusb-1.0|' src/qtcam.pro` into the preconfigure phase, and now it can find libusb.h . Thank you.
reinzelmann has joined #nixos
jorge-jbs has quit [Quit: jorge-jbs]
<clever>
attila`: ah, so its trying to be too smart and looking for the file, rather then letting gcc find it within the search path nix setup
ElGoreLoco has joined #nixos
ElGoreLoco is now known as jorge-jbs
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] kragniz opened pull request #36261: nixos/networking: add hostname to /etc/hosts by default (master...hostname-in-etc-hosts) https://git.io/vA9oU
NixOS_GitHub has left #nixos [#nixos]
acertain has quit [Ping timeout: 240 seconds]
nick_l has joined #nixos
acertain has joined #nixos
<attila`>
clever: Yep, the target.path was also hardcoded.
michas_ has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master 99d9c35 Will Dietz: clementine: apply upstream patch to fix build w/gcc7...
<NixOS_GitHub>
nixpkgs/master 1fb765d Will Dietz: Merge pull request #36259 from dtzWill/fix/clementine-gcc7...
<NixOS_GitHub>
[nixpkgs] dtzWill pushed 2 new commits to master: https://git.io/vA9oB
NixOS_GitHub has left #nixos [#nixos]
saksham has joined #nixos
<saksham>
Hi all, a Gentoo to Nix convert here. Been trying to get a particular haskell package to build locally, but so far pkgconfig refuses to acknowledge that gtk3 is available in the shell environment.
<clever>
saksham: it helps if you put a fetchFromGitHub into the nix file, then it doesnt need a local clone
<saksham>
Right. Will do that in the future. Meanwhile, this is the my local .config/nixpkgs/config.nix => https://pastebin.com/w3T5ZFL0
<clever>
*edits*
k2s has joined #nixos
<saksham>
Funnily enough, building gtk-traymanager had the same issue. I made it build by *removing* gtk3 from the pkgconfigDepends in its default.nix (the one I pasted above).
<clever>
!tofu
<{^_^}>
To get a sha256 hash of a new source, you can use the Trust On First Use model: use probably-wrong hash (for example: 0000000000000000000000000000000000000000000000000000) then replace it with the correct hash Nix expected.
<nick_l>
Why does nixops create a type "none" entry when I create a deployment in nixops info output?
<clever>
nick_l: did you set the targetEnv to something?
<clever>
saksham: what us the url for taffybar's src?
<nick_l>
clever: yes, to "ec2".
<nick_l>
clever: I am deploying a bunch of other machines using the exact same logic.
<clever>
nick_l: can you gist the deployment file and the output of nixops info?
<nick_l>
clever: those display "ec2" in the type column.
<NixOS_GitHub>
[nix] dtzWill opened pull request #1943: release.nix: don't try to use nix-2.0 branch, no longer exists (master...fix/nixpkgs-branch) https://git.io/vA96k
NixOS_GitHub has left #nixos [#nixos]
<nick_l>
clever: I had to remove the subnetId declaration from all network descriptions to get nixops info in a sane state.
<coconnor>
sphalerite: not intentionally. That build is done by my CI to integrate nixpkgs changes
<coconnor>
sphalerite: I would presume the nixos release tests would succeed with the nix 2.0 merge
<coconnor>
sphalerite: but I'm not terribly sure what the relation is between branches and stability in nixpkgs repo X-D
<saksham>
clever: Right, that's nice. Will take care from next time. Meanwhile, as you can see, the nix-shell doesn't load up since gtk-traymanager doesn't build.
<coconnor>
If anyone already has a nix 2.0 env up and can try that test then that would narrow it down to my specific setup
<clever>
saksham: got more edits coming up
<nick_l>
clever: how does adding a single declaration modify the type in nixops info output? I thought nixops info didn't contact the network.
<clever>
saksham: refresh the gist link
<clever>
nick_l: info doesnt contact aws, but it does re-eval the deployment files
<nick_l>
clever: and what determines whether something is of type 'ec2' then?
Fare has quit [Ping timeout: 255 seconds]
<clever>
nick_l: can you gist the output of it before and after the change?, i suspect something else is being missed
<nick_l>
clever: there are no other changes and the exit code is 0 in both cases.
<clever>
nick_l: what is the status in both cases?
<Judson>
There was an issue on nixpkgs (I think) about "too many wiki pages about language packaging" (or something like) Does that strike anyone's recollection?
<nick_l>
| the_querya | Missing / New | ec2 [eu-west-1] | | | vs | the_querya | Missing / New | none | | |
<nick_l>
clever: that's all.
<coconnor>
I don't recall there being a specific issue, but I know documentation of packaging for each language is lacking ;)
<coconnor>
except, maybe, haskell
<clever>
nick_l: i cant see anything obvious, id need more of the nix expression to know whats going on
<Judson>
That's certainly true. It's tautological given a particular audience, even.
acarrico has joined #nixos
<Judson>
I'm partilcularly interested in overall strategies for language packaging in Nix.
<Judson>
*Nixpkgs
<coconnor>
haha aye
k2s has quit [Remote host closed the connection]
<nick_l>
clever: I have a function computing an attribute set common to all machines, then I apply that to the standard arguments provided and do // {deployment.ec2.subnetId = "subnet-fdfsdf";}
<coconnor>
I think a year ago there wouldn't of been much to say about overall strategies. Now, tho, there is similarity between a number of languages
<coconnor>
can't think of any documentation addressing that tho
<clever>
nick_l: and theres your problem
<nick_l>
clever: it is my understanding that {<anything>} // {<else>} is equal to merging right into left with right taking precedence.
<saksham>
clever: Still trying to make it work. There's allowBroken requirement coming up for Agda, then allowUnfree for StrategyLib, and then an insecure issue coming up for erlang. Is this expected? (I have allowUnfree and allowBroken set to true in my user config, yet...)
k2s has quit [Client Quit]
<nick_l>
clever: OMFG
<nick_l>
clever: OK, so how do I merge recursively then?
<clever>
nick_l: first, if you have a 2nd deployment file (and you can use requires to chain them), nixops will merge things for you
<clever>
but thats not aware of types, and cant merge things like functions
Itkovian has joined #nixos
<clever>
in the require example, i set report-server twice, with 2 different functions, and each sets a number of nixos options for the report-server machine, which all get merged up nicely
<nick_l>
clever: where is "require" documented?
<clever>
nick_l: i think its in the nixops manual
<clever>
nick_l: it mostly works the same as imports does within nixos
<nick_l>
clever: It didn't seem like it was documented (I just looked)
<clever>
saksham: your nix-build is trying to build all of haskell packages
<nick_l>
clever: what does it mean to merge two functions?
<clever>
saksham: you want nix-build taffybar.nix --fallback -A gtk-traymanager
<clever>
nick_l: in my case, nixos will run both functions (line 6 of the file i linked, plus 6 of the file in require)
<clever>
nick_l: then nixos will merge the nixos options they both set
<blonkhart>
for a haskell build, is there a way to ensure that nix passes a flag to cabal to pull up the right version of a library (ghc versus ghcjs)?
<Judson>
In fact, I'm re-reading "Nix For Lang Packaging" and came across this passage: "Extending stdenv.mkDerivation - It's tricky to extend the stdenv.mkDerivation to make it open for extension. TODO: add more on the subject"
<Judson>
I'm not clear on what it takes to make a derevation admit extension, honestly.
<Judson>
Is there docs for that somewhere?
<nick_l>
clever: what do you do when you find a missing resource in nixops? E.g. vpc are in only since Jan.
<clever>
nick_l: if you run `nix-env -f release.nix -iA build.x86_64-linux` in the root of a clone of nixops, it will build that exact version and install it
sigmundv_ has joined #nixos
<clever>
nick_l: so you can just skip right to master if you want to
<nick_l>
clever: yes, I was running master already via that method (which works well).
<nick_l>
clever: that code looks like gibberish to me.
<clever>
nick_l: it took me an hour to figure out how the nixPath can be set within nixops, lol
<nick_l>
clever: there is no mention of what "f" is, although it is probably a file.
<clever>
nick_l: line 382, f is an argument to the function called getRequires
raynold has joined #nixos
<nick_l>
clever: yes, but what is it?
<clever>
nick_l: and line 391 recursively calls itself on every file in requires
<nick_l>
clever: i.e., its type.
<nick_l>
clever: right, so it's supposed to be some kind of of file.
<clever>
nick_l: it can be either a function or a path
<clever>
nick_l: the -I path you set gets saved into the state file, and can be seen in nixops info
<nick_l>
clever: oh, I am already doing that :)
<blonkhart>
oh, i see, yeah i didn't give you enough to do that. i haven't used github this way before
<nick_l>
clever: I was just wondering whether I missed something.
<clever>
blonkhart: gist also doesnt support directories, they dont want you abusing it as free private repos
<rizary>
hi all, i look into https://nixos.org/nixos/download.html in its azure section. It says that the version is still in 16.03. But now the stable version is 17.09. How to contact the maintainer who handle the azure blob? is it possible to use the 17.09 one?
<nick_l>
clever: is there a way to do require semantics without extra files?
<clever>
nick_l: i dont think so
<nick_l>
clever: ok, that sounds hacky, but thanks a lot.
<saksham>
clever: Sorry for the delay. So, was that gist supposed to allow building taffybar/gtk-traymanager? It didn't for me. I ran "nix-build taffybar.nix --fallback -A gtk-traymanager". Even tried "nix-shell taffybar.nix" after modifying the last line to nixpkgs.haskellPackages.taffybar
<clever>
nick_l: i find it more powerful, the report-server-bucket-storage.nix example i linked, can just be removed from require, and the deployment still works perfectly fine
<clever>
more modular
<clever>
saksham: yeah, it fails with a gtk error within traymanager for me, i havent figured that out yet
<clever>
saksham: but its at least all in one file, so its easyer to pass around
<nick_l>
clever: How do I port this to your approach then? All free variables here are in a surrounded let binding now. machine = { resources, pkgs, lib, nodes, ... }@args: let common = (common_machine args); in common // { deployment.ec2.securityGroupIds = common.deployment.resources.ec2.securityGroupIds
PMS has joined #nixos
<clever>
nick_l: you may need a 3rd file, that contains the stuff from the let block, and then import that, or only set it once in a single deployment file
<tnks>
hmm... the fancy new "nix build" is terrible for capturing output in a script.
<saksham>
clever: Right, thanks! Any idea why I could build gtk-traymanager with this: https://pastebin.com/w3T5ZFL0 ? (
<clever>
saksham: you have 2 super's in there, and its not entirely clear which it will use, and i may have also picked the wrong revision from github when i switched it over
obadz has quit [Quit: WeeChat 2.0]
<clever>
blonkhart: have you tried the nix files already in the miso repo?
<blonkhart>
i am learning nix and was trying to go by analogy to something in the miso repo
<blonkhart>
putting together a simplified gist to illustrate
obadz has joined #nixos
<clever>
blonkhart: what is the exact error you get when building this?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master 9041eff Tuomas Tynkkynen: diffoscope: Update tools list
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to master: https://git.io/vA9ip
<muzzy>
Hey clever, I saw that you use system76 in your github repo. What's your opinion on them bc I opted for a thinkpad
<blonkhart>
the error is that when building the client, the "Effect" type is out of scope
<clever>
muzzy: the laptop has worked great so far, only a minor issue with suspend sometimes locking up
<nick_l>
clever: so, I need to convert let foo=1;bar=2; in <blah> to {foo = 1;bar = 2;} in a file old_let.nix and then with import ./old_let.nix; <blah>?
DavidEGrayson has joined #nixos
<clever>
nick_l: yeah
<muzzy>
clever: which model do you have?
<clever>
muzzy: kudu
<blonkhart>
the module is only exposed on the ghcjs side
<nick_l>
clever: my mind is heavily protesting the idea that there are semantics which don't work as a function, though.
kp__ has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master 6326053 Tuomas Tynkkynen: diffoscope: Add script to list required tools
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to master: https://git.io/vA9PI
NixOS_GitHub has left #nixos [#nixos]
jorge-jbs has quit [Quit: jorge-jbs]
<saksham>
clever: I kinda got it to work, and it is hilarious. I had spent so much time on this that I forgot I removed 'pkgconfig-depends: gtk+3.0' from the actual cabal file of gtk-traymanager (in my local sources) to make it build. Somehow, pkg-config does not find gtk+3.0 if it is listed in the cabal file, otherwise it is good to go.
<clever>
saksham: there is a bug with pkgconfig in nix, pkgconfig will refuse to give the flags for gtk, if it cant find the .pc files for gtk's dependencies
<clever>
saksham: and due to how nix works, those dependencies wont be in the search path for traymanager
<clever>
and pkgconfig just silently doesnt tell you so
abathur has quit [Ping timeout: 252 seconds]
<saksham>
clever: I gotta say "that's not clever" :) Anyhow, this was painful. So, any sane way to build such packages without losing my mind / modifying source files?
<clever>
saksham: read the .pc file for gtk, and add its dependencies to gtk-traymanager
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
NixOS_GitHub has joined #nixos
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/staging 1e10744 Tuomas Tynkkynen: Merge commit '618ac29687a650d854c8bea7efd4490387589ce2' into staging
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to staging: https://git.io/vA9PW
<saksham>
Oh. Okay. So, can I generalize this to: "all gtk3-based haskell packages need to list deps of gtk3 in their cabal files" ?
<saksham>
clever: ^
<clever>
saksham: the proper fix, is to put the gtk3 deps into the propagatedBuildInputs of gtk3
coot has quit [Quit: coot]
<clever>
saksham: then what you said happens entirely automatically, by only putting gtk3 into the deps
<nick_l>
clever: in the terraform folder of your repository it says that you need to run nix-shell in the README.md, but there is no default.nix.
<clever>
nick_l: is there a shell.nix?
<nick_l>
clever: no
<saksham>
clever: Aah I see. Will try that. Thanks a lot !
<clever>
nick_l: you run the nix-shell command one directory up from terraform
coot has joined #nixos
<clever>
nick_l: shell.nix is present there
<nick_l>
clever: in that case the documentation needs to be fixed.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] launerbie opened pull request #36263: dcrd: init at version 1.1.2 (master...decred) https://git.io/vA9Pg
NixOS_GitHub has left #nixos [#nixos]
<clever>
nick_l: it does say to "cd terraform" after you nix-shell
<clever>
which implies you have to be one dir up
<nick_l>
clever: sure, but documentation is read in a current directory.
<clever>
the linebreaks in the readme.md got eaten by github, it looks better if you open it in raw mode
<gchristensen>
man, I think I just did a super cool thing
<nick_l>
clever: I found a potential security issue in your repo, btw.
<clever>
nick_l: where?
<nick_l>
clever: I sent it via a private message.
<ryantm>
I think nix-repl is failing to build for me on nixpkgs/master but I might be just be using nix 2.0 wrong. I'm doing `nix build -f . nix-repl`
<clever>
ryantm: 2.0 also has `nix repl` with a space, but its tab-completion is a bit too different for me
<ryantm>
clever: oh right, I forgot about that, and that works.
<blonkhart>
clever: this is impressive, it is like watching a chess master play 8 games at once
nuncanada has joined #nixos
robstr has quit [Remote host closed the connection]
<clever>
blonkhart: i think you killed them all, lol
coot has quit [Quit: coot]
MP2E has joined #nixos
reinzelmann has quit [Quit: Leaving]
<MichaelRaskin>
gchristensen: ?
<MichaelRaskin>
A Borg feature?
<blonkhart>
put some time on your clock anyway
<blonkhart>
were you able to reproduce from the gist? if not no worries i need to sit down with the nix docs and think it through from first principles
<blonkhart>
the main ghcjs frp frameworks express a strong preference for nix
<rycee>
manveru: If you update HM now then your pipeline with `home-manager generations` should work better.
<blonkhart>
which suggests it is worth the investment as i head in that direction
BenceF has joined #nixos
<BenceF>
Hi is anyone using sway (the tiling wm for wayland)? I can't seem to run X11 apps
<BenceF>
also GTK3 apps want to start as X11 apps even though GDB_BACKEND is set to wayland
<Baughn>
Turns out my workstation has bad ram. Which explains a few glitches.
<Baughn>
Is there any way to regenerate the entire Nix store?
<MichaelRaskin>
Wipe /nix, reinstall with the same configuration.nix?
<clever>
blonkhart: it fails because i dont have the client or server directory
<Baughn>
Without doing that, preferably. :P
<MichaelRaskin>
Well, if you want to regenerate the entire store, what would be the difference?
<blonkhart>
oh sorry the gist just indicated the directories with a hyphen rather than a slash
johnw has quit [Ping timeout: 265 seconds]
<clever>
blonkhart: yeah, gist doesnt support directories, simpler to use a full git repo at that point
<blonkhart>
so client-Main.hs and client-default.nix are in a directory and similarly for server-
<blonkhart>
everything else is at the project root
lord| has quit [Quit: WeeChat 2.0.1]
<Baughn>
MichaelRaskin: Whether or not I need to bootstrap from scratch or can do it in the background.
<symphorien>
wilornel: did you use nix-shell ? if yes how ? you may add pkgconfig to the build-inputs
<sphalerite>
clever: brk(NULL)
<symphorien>
sometimes build systems need it
pxc has joined #nixos
i-am-the-slime has quit [Remote host closed the connection]
winem_ has joined #nixos
<wilornel>
symphorien: yes, I made a mistake. I was outside of a nix-shell. I put sqlite in my temp/default.nix then ran `nix-shell default.nix` on it and then `cd ~/human`. `bundix -m` shows "/nix/store/gd5f1i038kh8wcglqjlbyx640r1jnsxx-bundler-1.14.6/lib/ruby/gems/2.3.0/gems/bundler-1.14.6/lib/bundler/runtime.rb:40:in `block in setup': You have already activated bundler 1.16.1, but your Gemfile requires bundler 1.14.6. Prepending
<wilornel>
`bundle exec` to your command may solve this. (Gem::LoadError)
<wilornel>
"
<manveru>
never use bundle exec
<clever>
sphalerite: thats a pretty standard syscall...
<sphalerite>
clever: indeed
<manveru>
wilornel: what's your version of bundix? did you try `bundix -l`?
<clever>
sphalerite: what if you use strace -f
<clever>
sphalerite: maybe a different thread used something bad at the same time
<sphalerite>
clever: I used strace -ff
<wilornel>
2.3.1
<wilornel>
manveru: Hey!
<wilornel>
I did `bundix -l` but then I get issues when trying to install webpacker
<clever>
sphalerite: and you confirmed which thread was the failing one, or the exit code from the syscall?
<sphalerite>
clever: all 4 of the processes killed by SIGSYS were sh, but this happened very early on, just after exec, and before any of the actual loading happened
<clever>
sphalerite: can you gist the logs?
<wilornel>
`bundix -l` works
PMS has quit [Ping timeout: 252 seconds]
<manveru>
wilornel: if you need sqlite as a dependency for one of your gems, you can put it into gemConfig