pkx3 has quit [Read error: Connection reset by peer]
m0rphism has joined #nixos
<porrifolius>
So, here's a weird one... every second urxvtc I run doesn't work properly. Anyone have any ideas?
<porrifolius>
I have two lots of keybindings set up, one to change font size (using font-size), and one to change colours (using color-themes). Every second terminal I run the colour changing keybindings work, but they report that they can't find any themes to switch between. Xmonad's terminal keybinding calls urxvtc, but the same thing happens if I run urxtvc from either a working or non-working terminal. It seems that urxvt (no 'c') works
<porrifolius>
fine all the time, so some sort of client/daemon problem.
mr_noobs has joined #nixos
<porrifolius>
If I do a ls in the non-working terminals there are the themes still in .config/themes/urxvt of course.
<porrifolius>
I've tried killing off all daemons and clients for the user, logging out, logging in... a new daemon comes up but still the same problem.
o1lo01ol1o has quit [Remote host closed the connection]
<porrifolius>
Ah, the non working ones say "No themes found in " [no directory]. In the working terminals, if there are no themes, it will say "No themes found in /home/user/.config/themes/urxvt" so it seems to be _some_ of the settings being overwritten.
<porrifolius>
Or not initialised.
<jackdk>
is there a way to ask for specific metadata revisions when doing a haskellPackages.callHackage?
<{^_^}>
[nix-pills] @Ma27 opened pull request #90 → Update documentation after Nix 2.0 switch → https://git.io/fp1hA
<{^_^}>
[nixpkgs] @bhpdt opened pull request #51597 → heaptrack: wrap GUI for QT plugins → https://git.io/fp1jO
<{^_^}>
[nixpkgs] @AndersonTorres opened pull request #51598 → tecoc: add BSD and Darwin/OSX support → https://git.io/fp1jG
mmlb has quit [Read error: Connection reset by peer]
mmlb has joined #nixos
DigitalKiwi has joined #nixos
cyounkin_ has quit [Remote host closed the connection]
cyounkins has joined #nixos
cyounkins has quit [Ping timeout: 246 seconds]
<DigitalKiwi>
I keep getting errors about a nar xz that won’t let me nixos-rebuild and found to do this http://ix.io/1vlM how do I fix? Sorry in advance I’m on mobile so can’t type well
<DigitalKiwi>
An irc client is one of the things I can’t get to install :|
<meizikyn>
sometimes this happens to me with shaky network connections
<meizikyn>
or bad binary-cache :/
<DigitalKiwi>
Oh yeah one of the things that was like that on the wiki was wrong I think
<DigitalKiwi>
Actually I probably have a ton of bug reports after this is all done to see if they’re new and I can remember them mhr
<DigitalKiwi>
And yes my internet is being exceptionally bad on this I hope there’s a way to make it better at some point
<DigitalKiwi>
It’s a late 2013 MacBook Pro using the b43 and sometimes it doesn’t even load the module when I boot ...
illegalprime has quit [Ping timeout: 246 seconds]
<meizikyn>
is the b43 a wireless nic, eiher way, does "~$ dmesg" give you anything about the module?
<meizikyn>
oh and do you mean the command arguments were different on the wiki, or wrong in that they didn't allow you to rebuild?
<DigitalKiwi>
Both. the one I saw when I did it said invalid option or slt
<meizikyn>
did the one I proposed work? if not, do you have an error trace? I've never for the death of me known what nix uses the nar files for, but usually are only necessary when pulling in new sources not already available in the store
<DigitalKiwi>
Yes that one is wireless I’m not finding anything in dmesg tho it is loaded currently it’s the one networking.enableB43Firmware = true and the one you said hasn’t not failed yet
<DigitalKiwi>
It could take a while I have a feeling it’s going to build ghc but so far it hasn’t stopped
Lears has joined #nixos
<DigitalKiwi>
On the manual on 36.4 and 36.5 I have tried the first three the third one is the one that complained
[Leary] has quit [Ping timeout: 246 seconds]
<meizikyn>
I've seen bug reports already complain about the first one in 36.5, had to do some digging in git to find the substitute option.
<meizikyn>
The online manual is terribly out-dated, I wouldn't recommend it for troubleshooting in general. unfortunately git complaints are more frequently well documented
<meizikyn>
I also don't know why substitute would be the name of the option, seems unintuitive to set it to false to not use default binary cache
<meizikyn>
but I also haven't found official documentation on it either, so theres that
<DigitalKiwi>
So anyway other problems I had were 2 of the most recent unstable nixos live usbs wouldn’t work it said dangling symlinks when I tried to work with them . And then the one before that zfs doesn’t work . Allegedly you can switch to the zfsUnstable but when you try to make a pool it won’t work because the kernel module is still 7.12 and can’t be replaced or at least i don’t know how
<DigitalKiwi>
So I had to make my own that actually had 8rc2
<DigitalKiwi>
Which was entirely not easy for me to figure out :(
jasongrossman has joined #nixos
Anton-Latukha has quit [Quit: Leaving.]
boxscapeR has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
<meizikyn>
hmm. Haven't used unstable. haven't used ZFS since the last 18.03, does the zfs module work with 18.09 stable?
<DigitalKiwi>
Not for me no
o1lo01ol1o has quit [Ping timeout: 272 seconds]
<DigitalKiwi>
I ended up eventually figuring out how to add a line to lib/eval-config.nix that got me a working version of 19.03 with the right version of zfs
<DigitalKiwi>
Because even the working 19.03 had the couldn’t change module to the right module which results in fun errors about invalid options that lead to git issues from 2010-14
sir_guy_carleton has quit [Quit: WeeChat 2.2]
<ivan>
I have a hydra and it continues building things for previous evaluations when I don't really care about older evaluations, is there a way to configure it to give up on them?
<ivan>
I don't want to manually abort and restart jobs
<DigitalKiwi>
I had to make a few of those too because some of them didn’t work and I made from a live disk optimistically... ooh and then since it wasn’t obvious to me how to make it do what I needed I did have one that I just swapped the function names which did actually work...I eventually did it right
<DigitalKiwi>
it kept only wanting to build zfsStable so I tricked it
<ldlework>
I'm really interested in having an encrypted USB key that is mounted/unmounted when I interact with my Yubikey. Anyone know anything at all about that? I don't really know anything about disk encryption.
aristid has joined #nixos
<ivan>
you're probably looking for cryptsetup/LUKS
<ivan>
googling finds some cryptsetup yubikey integrations
<DigitalKiwi>
If you search for yubikey encrypts there’s a git issue you might like
<DigitalKiwi>
It’s on yubico/yubico-pam
<ivan>
I would recommend having a secondary keyslot with a proper password that you can type in without the yubikey
<DigitalKiwi>
Also depending on your needs zfs native encryption might be an option though it is new so be careful is all
jperras has quit [Ping timeout: 250 seconds]
<DigitalKiwi>
meizikyn: It’s still building stuff
hamishmack has quit [Remote host closed the connection]
<meizikyn>
great scott, is this a system upgrade?
<DigitalKiwi>
It’s a few Haskell things
<meizikyn>
do you mean building as in unpacking and configuring or is it compiling from source
<DigitalKiwi>
Like I might be better off trying to get more things installed individually first...
<meizikyn>
I don't use Haskell, but I figure its a prety light weight language, shouldn't be building this long
<DigitalKiwi>
It’s building the world
<meizikyn>
the world?
<DigitalKiwi>
It might be compiling gcc right now I’m not sure
jperras has joined #nixos
<DigitalKiwi>
Binutils
joehh has joined #nixos
<joehh>
anyone having issues with cache.nixos.org at the moment?
<joehh>
getting lots of timed out requests
<DigitalKiwi>
Or gdb maybe I don’t know it all looks the same but my battery is actually going down even though it’s plugged in LOL
<joehh>
is there a way to debug cache issues?
<DigitalKiwi>
That’s what I’m here for
<meizikyn>
nixos-rebuild switch --option substitute false, give it a try if you're not downloading new packages
<meizikyn>
it may however attempt to build from source, as in DigitalKiwi's case
DigitalKiwi is now known as CodeKiwi
<CodeKiwi>
I’m going at this a different way I don’t have the patience anymore to build that much
jperras has quit [Ping timeout: 250 seconds]
<CodeKiwi>
I may just go plug into the modem to help... I don’t think this is all on my end tho
<CodeKiwi>
While I’m waiting does anyone have suggestions for switching between intel/nvidia on a MacBook Pro ... it’s not Optimus so as far as I know though I’d love to be wrong the options are stick with one or use macOS
<meizikyn>
It might not be, Forums and error traces may resolve larger-scale issues like this
jperras has joined #nixos
<meizikyn>
if your on linux, I'd look on nvidia forums for how to switch using the driver
<jackdk>
ldlework was talking about dual-gpu things in nixos-chat earlier today
<meizikyn>
I think you need both drivers active for them to switch properly on your display
<meizikyn>
so swapping modules is out
<CodeKiwi>
I’d settle for an easy way to switch by rebooting
<CodeKiwi>
So maybe like two marginally different config and rebuild switch ...
mmlb has quit [Read error: Connection reset by peer]
mmlb has joined #nixos
jonaswouters has quit [Quit: Connection closed for inactivity]
vidbina has joined #nixos
vk3wtf has quit [Ping timeout: 252 seconds]
jperras has quit [Ping timeout: 250 seconds]
epicmetal has quit [Quit: epicmetal]
jperras has joined #nixos
porrifolius has quit [Remote host closed the connection]
porrifolius has joined #nixos
slack1256 has quit [Remote host closed the connection]
agander has joined #nixos
jperras has quit [Ping timeout: 250 seconds]
porrifolius has quit [Remote host closed the connection]
palo1 has joined #nixos
<Myrl-saki>
What if we add parallel execution to Nix?
jperras has joined #nixos
<Yaniel>
how about making downloads not take forever first :(
<Myrl-saki>
Yaniel: :'(
<Myrl-saki>
Yaniel: Why do you say it's slow?
<Yaniel>
are those even done in parallel?
<Myrl-saki>
Yes, they are, by the amount of cores you have.
<Yaniel>
I'm not sure what the issue was
palo has quit [Ping timeout: 268 seconds]
<Yaniel>
I remember seeing some actual discussion about it too
<Myrl-saki>
Yaniel: I'm guessing 'by the amount of cores you have'
<Yaniel>
but I've been away from nixos for almost a year now so things mayy have changed
<Myrl-saki>
Yaniel: A single-core device will only download one derivation at a time.
<Yaniel>
I don't think I've even tried on anything with less than 4 cores
<Myrl-saki>
Yaniel: Non-resumable nlaods?
<Myrl-saki>
Non-resumable downlaods?
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
jackdk has quit [Ping timeout: 244 seconds]
<Yaniel>
maybe something about connection reuse?
<Yaniel>
idk I didn't pay much attention at the time
<Yaniel>
but did note that even a rather bare bones installation takes its sweet time
jperras has quit [Ping timeout: 272 seconds]
jperras has joined #nixos
Ariakenom has joined #nixos
crmlt has joined #nixos
crmlt has quit [Ping timeout: 252 seconds]
hamishmack has joined #nixos
rprije has quit [Ping timeout: 240 seconds]
vk3wtf has joined #nixos
jasongrossman has quit [Ping timeout: 250 seconds]
jonaswouters has joined #nixos
jperras has quit [Ping timeout: 268 seconds]
jperras has joined #nixos
Ariakenom has quit [Quit: Leaving]
o1lo01ol1o has joined #nixos
<makefu>
Myrl-saki: isn't there "nix.maxJobs" which can scale up execution?
o1lo01ol1o has quit [Ping timeout: 268 seconds]
m0rphism has quit [Quit: WeeChat 2.2]
rprije has joined #nixos
jperras has quit [Ping timeout: 246 seconds]
ckauhaus has quit [Quit: WeeChat 2.2]
reinhardt has joined #nixos
jperras has joined #nixos
[Leary] has joined #nixos
goibhniu has joined #nixos
Lears has quit [Ping timeout: 250 seconds]
ckauhaus has joined #nixos
rprije has quit [Ping timeout: 240 seconds]
freeman42x has quit [Ping timeout: 252 seconds]
irdr has quit [Ping timeout: 268 seconds]
Radivarig has joined #nixos
jperras has quit [Ping timeout: 246 seconds]
porrifolius has joined #nixos
jperras has joined #nixos
<porrifolius>
I've got a special key on my laptop, alongside all the +/- volume etc, which is hooked into suspend-to-disk... this immediately hard locks the computer. How can I disable or re-map this key?
<porrifolius>
This key did not trigger the suspend to disk when I was running ubuntu. I've tried to get the keycode using showkey and xev, no luck. I tried to run xinit xterm from another VTY so as to try and rule out xmonad and such. But I can't get xinit to work.
<porrifolius>
Any ideas?
hamishmack has quit [Ping timeout: 246 seconds]
<symphorien>
I think logind itself handles it
<v0|d>
nice, I am able to obtain a arm pi3 sdImage on amd64.
monokrome has quit [Ping timeout: 244 seconds]
<Ralith>
porrifolius: yeah, definitely a logind thing
<{^_^}>
[nixpkgs] @dotlambda merged pull request #49741 → Python package set major updates → https://git.io/fxhSs
<{^_^}>
[nixpkgs] @dotlambda pushed 544 commits to staging: https://git.io/fpMGe
<porrifolius>
symphorien Ralith: Thanks, I'm having a look at man logind.conf now. Looks like there is the nixos services.logind.extraConfig option... here's hoping that does the trick.
<Ralith>
sounds like the right path
klntsky has joined #nixos
jperras has joined #nixos
Ariakenom has joined #nixos
Ariakenom has quit [Client Quit]
Ariakenom has joined #nixos
lnikkila has joined #nixos
Ariakenom has quit [Client Quit]
Ariakenom has joined #nixos
<Myrl-saki>
makefu: It can scale up building, but not execution.
<porrifolius>
Hmmm... I'm being sill I'm sure, but... what is the syntax for options 'strings concatenated with "\n"'?
<Ralith>
lines
<porrifolius>
Not = [ "blah=foo" "otheroption=zob" ]; it seems
<Ralith>
oh, the values? just pass a string
<porrifolius>
So I've just set s.l.extraConfig = "HandleHibernateKey=ignore HandleSuspendKey=ignore";
<porrifolius>
Is that right?
<porrifolius>
Thought they'd need to be in []
ThatDocsLady has joined #nixos
<Ralith>
normallly you use double 's for multiline strings, but yes
<Ralith>
(and logind will indeed likely require that those options be on separate lines)
<porrifolius>
Ralith: ok, the first one I showed you, with ", was a single line, no \n between the two options. It n-r switch'd ok. I've just tried with '' as you suggested and this time the second option/value is on another line. It switch'd ok.
<porrifolius>
I guess they're different in the logind will see 1 or 2 arguments respectively?
<{^_^}>
[nixpkgs] @vbgl opened pull request #51611 → ocamlPackages.lablgtk3: init at 3.0.beta1 → https://git.io/fpMZx
* Ralith
shrug
<Ralith>
I wouldn't like to guess at what logind might do with unusual config parameters
civodul has joined #nixos
<porrifolius>
It seems that either way the suspend to ram is still happening. Is there some way to reload logind or do I need to logout, or restart?
jperras has quit [Ping timeout: 250 seconds]
jtcs has quit [Ping timeout: 250 seconds]
__monty__ has joined #nixos
shreyansh_k has quit [Ping timeout: 246 seconds]
jperras has joined #nixos
porrifolius has quit [Remote host closed the connection]
mmlb has quit [Read error: Connection reset by peer]
mmlb has joined #nixos
jtcs has joined #nixos
porrifolius has joined #nixos
<porrifolius>
Ralith: well, I'm feeling confident. A reboot and the suspend key is no longer putting to to sleep. Now to chance the hibernate key!
<Ralith>
\o/
<Ralith>
restarting display-manager might be sufficient, but I'm not certain
<porrifolius>
Sweet relief! It's now surviving the key-combo of death. Didn't try with my dodgy " form of parameter though. Don't think I bother.
<porrifolius>
Thanks for the help!
jperras has quit [Ping timeout: 250 seconds]
jperras has joined #nixos
thc202 has joined #nixos
porrifolius has quit [Remote host closed the connection]
porrifolius_ has joined #nixos
<porrifolius_>
Ralith: Do you know how I can find what is capturing a given key press? Got some more special keys, no visible effect on pressing them, I can get their codes using showkey in a VTY but they don't show up with xev.
<Ralith>
evtest might give you more to go on
cyphase has quit [Read error: Connection reset by peer]
jperras has quit [Ping timeout: 250 seconds]
<porrifolius_>
Ralith: gave it a go, they don't show up. Perhaps they're capturend by the bios, they're mousepad and screen related. A battle for another day I think. Thanks for the pointer though.
<Ralith>
that seems plausible, the linux stack doesn't usually mask things off entirely
jperras has joined #nixos
<teto>
does `nix search -u stackage2nix` return any result for you ? none here :/
<ldlework>
My laptop has bluetooth, I have hardware.bluetooth.enable = true; but when I try to use bluetoothctrl it says there's no controllers
elgoosy has joined #nixos
jperras has quit [Ping timeout: 250 seconds]
agander has quit [Ping timeout: 246 seconds]
<ldlework>
hm rfkill unblock worked
drk_lrd_877_ has joined #nixos
jperras has joined #nixos
<sphalerite>
porrifolius_: you mean they don't show with showkey either?
<symphorien>
I think it goes through acpi, not as a regular keyboard event
cyphase has joined #nixos
<porrifolius_>
sphalerite symphorien: correct, they don't show with showkey within a VTY. I have found out that one of them messes with my keys though, sort of like numlock (which is a special 'Fn' key on my laptop as well but it shows in xev/showkey/evtest).
sigmundv has joined #nixos
<dhess>
Anyone around who has good knowledge of the NixOS test harness?
<dhess>
and/or QEMU?
<sphalerite>
porrifolius_: ooooh boy yeah I had an issue like that a long time ago, was extremely confused.
<sphalerite>
dhess: probably. Maybe you should try asking what you want to ask those people ;)
Jackneill has quit [Ping timeout: 250 seconds]
rprije has joined #nixos
elgoosy has quit [Remote host closed the connection]
elgoosy has joined #nixos
elgoosy has quit [Client Quit]
elgoosy has joined #nixos
<sphalerite>
teto: none here either
trevthedev has joined #nixos
<sphalerite>
teto: I think it's because haskellPackages is excluded from searches
<trevthedev>
Hi, I'm trying to package DaggerfallUnity(A game) to NixPkgs, the download is a zip file, does Nix allow that?
<elgoosy>
hi, after setting nixos channel to 18.09 and `sudo nixos-rebuild switch --upgrade` my nixos-version outputs 18.03.133381.a18112a7e8f (Impala)
<teto>
sphalerite I tried nix search -u haskellPackages.stackage2nix too with the same result
<sphalerite>
teto: yes, because haskellPackages is excluded from searches.
<teto>
sphalerite: is there any way to bypass this , it's kinda annoying :s
<sphalerite>
teto: I have an overlay that includes it in searches, I can share it with you in about 20min
<sphalerite>
teto: in the meantime, you can use nix-env -f '<nixpkgs>' -qaA haskellPackages <search term>
<{^_^}>
[nixpkgs] @Izorkin opened pull request #51612 → unit: init at v1.6 → https://git.io/fpMBQ
<teto>
sphalerite: no hurry/pb. I usually grep nixpkgs. Just this time tried nix search...
<sphalerite>
yeah, I should probably put this trick somewhere easier to find. Did you look anywhere?
<avn>
sphalerite: `substitute false` force it to download bootstrap and rebuild everything
vidbina has quit [Ping timeout: 268 seconds]
<sphalerite>
avn: oh yeah of course, because then it doesn't use the ISO store, silly me
<sphalerite>
avn: try --option susbtituters ''
<sphalerite>
I think it uses extra-substituters
elgoosy has quit [Remote host closed the connection]
worldofpeace has joined #nixos
<jasongrossman>
symphorien: It surely can't be guaranteed to work on all hardware, can it?
elgoosy has joined #nixos
<symphorien>
sure, there are quite a lot of tests, though
<elgoosy>
why after upgrading to 18.09 is the system still in 18.03? what am i doing wrong?
<symphorien>
how did you upgrade ?
<symphorien>
ah you said it earlier, never mind
<symphorien>
did you set root's channel to 18.09 or your user's ?
<elgoosy>
ohh..user's
<elgoosy>
symphorien, thank you
<elgoosy>
just one more question: is it better to fresh install 18.09 or upgrade from 18.03 to 18.09?
jrolfs_ has joined #nixos
<symphorien>
an upgrade allows you to rollback :)
<elgoosy>
symphorien, when i change root's nix channel tp 18.09 i must also change system.stateVersion = "18.03"; in configuration.nix before upgrading, right?
<symphorien>
no
<symphorien>
,stateVersion
<{^_^}>
Setting stateVersion to the latest release doesn't update anything and can only break your setup at best. If you want to regardless, Ctrl-F for "stateVersion" in https://nixos.org/nixos/manual/release-notes.html to see things that need to be manually upgraded with the new value
<symphorien>
don't change stateVersion
<goibhniu>
AFAIK stateVersion is only used for a few system services, which you may not even be using, e.g. postgres
<kipari>
hi. I want to install the QuickChick module for Coq, but I can't quite get how to install modules for coq using nix. anyone able to help with this?
<{^_^}>
elgoosy: If a Nix file foo.nix starts with something like `{ stdenv, cmake }:`, you can build it with `nix-build -E '(import <nixpkgs> {}).callPackage ./foo.nix {}'`
<elgoosy>
i've tried this nix-build -E 'with import <nixpkgs> {}; callPackage ./default.nix {} but then it complains about missing argument qtbase
<symphorien>
otherwise, nix-build /path/to/nixpkgs -A digitalbox
<avn>
symphorien: -A pkgs.digitalbox, isn't it? ;)
<avn>
I always prefer second, when test my stuff before PR
<dramforever>
It's the same thing
<trevthedev>
i havent even gotten to the packaging part, so i cant use autopatchelfhook yet
<trevthedev>
error still throws regardless
<symphorien>
avn: not when you provide the path to nix-build
<trevthedev>
goes*
jperras has joined #nixos
<elgoosy>
there's that package online but it is very old, so i thought i build for first time a package locally just updating the version of digitalbitbox source code
<symphorien>
avn: but it also works because pkgs contains a pkgs argument which is very similar to itself
<avn>
symphorien: well, you right. I just have NIX_PATH always pointing to my worktree
<symphorien>
ah I see :)
wacke has joined #nixos
<dramforever>
This one by symphorien then: nix-build /path/to/nixpkgs -A digitalbox
<wacke>
hi all, after upgrade to 18.09 i get following error message, what can i do? nix-env -qc
<wacke>
error: cannot import '/nix/store/7vl1nnj0w5565g8hy16yzi9sh9y1l1kg-references.nix', since path '/nix/store/njm3nb198zzpln0wi4s2n0jzhl88f7aa-references.nix.drv' is not valid, at /nix/store/43vrvm0isa550h06c43ckx935hzqd6mp-nixos-18.09.1228.a4c4cbb613c/nixos/pkgs/build-support/replace-dependency.nix:26:16
<wacke>
(use '--show-trace' to show detailed location information)
<wacke>
tried sudo nix-store --verify --check-contents --repair, didn't help
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
<wacke>
ok, nixos-rebuild switch did help
adamt has joined #nixos
<elgoosy>
nix-build -E '(import <nixpkgs> {}).callPackage ./foo.nix {} does not work either. I get `default.nix:1:1 called without required argument 'qtbase',....nixos/lib/customization.nix`
mmlb has quit [Read error: Connection reset by peer]
mmlb has joined #nixos
jperras has quit [Ping timeout: 250 seconds]
<adamt>
Hiya. I got some Mellanox nics I'm trying to get up and running. Does anybody know whether the required software is accesible in an easier way than manually packaging the driver they provide on their website? lspci reports the NIC's as "Mellanox Technologies MT27710 Family [ConnectX-4 Lx]"
vk3wtf has quit [Ping timeout: 252 seconds]
<adamt>
(for some reason the host is loading the kernel mod "mlx5_ib" and friends, which sounds wrong, and loading "mlx4_ib" doesn't make the NIC's show up with ip a on its own
<srhb>
adamt: "
<srhb>
Woops
<srhb>
"mlx5 is the low level driver implementation for the ConnectX®-4 adapters designed by Mellanox Technologies. ConnectX®-4 operates as a VPI adapter. The mlx5 driver is comprised of the following kernel modules: [...]"
<srhb>
From the user manual.
jperras has joined #nixos
pie__ has joined #nixos
<srhb>
adamt: Is mlx5_core also loaded?
<srhb>
(I assume the infiniband part depends on it, just checking)
<adamt>
srhb: Which manual did you find? I didn't find one with that phrase in it
<adamt>
srhb: Yeah, mlx5_core is also loaded -- i reversed the mod dependency order in my head by mistake. :P
<srhb>
adamt: 1.1.2 Software Components
<srhb>
adamt: Peculiar. And you're looking for ethernet, not infiniband, right?
<adamt>
srhb: I'm certainly expecting it to be ethernet, yes
elgoosy has quit [Remote host closed the connection]
<srhb>
adamt: OK. The first thing I would try if you haven't already is grab a way newer kernel (because it's a cheap test)
<srhb>
adamt: Second, dmesg would be useful for debugging.
<adamt>
srhb: Been there, done that, now running 4.19.5. But the docs confuse me, because sections 1.1.5 says "mlx4 is the low level driver implementation for the ConnectX" ... (without specifying what version of ConnectX)
irdr has joined #nixos
<srhb>
adamt: I'm pretty sure mlx5 is correct.
elgoosy has joined #nixos
<srhb>
adamt: Third: Firmware updates all around.
<adamt>
dmesg only reports two instances of "mlx5_ib: Mellanox Connect-IB Infiniband driver v5.0-0" when searching for mlx or Mellanox
<srhb>
adamt: Can you pastebin the entire thing?
<adamt>
srhb: Firmware updates requires the mellanox firmware tools, i think
<adamt>
(no sure anybody packaged them, didn't find it on the usual pkg search page)
<cyounkins>
What's the difference between nixos-unstable and nixpkgs-unstable?
jtojnar has joined #nixos
<gchristensen>
nixos-unstable is tested to ensure it will produce a bootable system, whereas nixpkgs-unstable could potentially render your system unusable
<gchristensen>
("but the bootloader rollbacks" doesn't count here, because the bootloader is rewritten by the untested version, and could break it)
<{^_^}>
[nixpkgs] @dasJ opened pull request #51624 → nixos/openldap: Support configuring the log level → https://git.io/fpM5o
smolboye has quit [Ping timeout: 252 seconds]
<dtz>
nh2: yep! :D :D
<nh2>
dtz: then, after the whole cross/musl/static stuff is done, let us go and hunt down integer overflows in NixOS and Haskell, making a hardened overlay where overflow crashes!
<pie_>
nh2: ooh!
<pie_>
nh2: also where can i read about this cross/musl/static stuff
<pie_>
i was looking at running on musl with a friend the other day, and i dug into the code a little. x)
<pie_>
what is the buddha nature? 5 rounds of bootstrapping xD
<gchristensen>
dtz: awesome!
<hodapp>
A monk asked Joshu, the Zen master: "Master, how do I bootstrap the system?" Joshu answered, "By first bootstrapping the system," and was silent. At that moment, the monk was enlightened.
<dtz>
;D
<nh2>
dtz: if 10 Linux distributions and OpenBSD can't get this done, I think it's time that somebody tackles it with better tooling, like the power of a programmable package ecosystem and an easy way for multiple hands to help and fix the fallout
<hodapp>
pie_: you could call it "The Stateless State" as a play on "The Gateless Gate".
<pie_>
:D
layus has joined #nixos
ma27 has joined #nixos
reinhardt has quit [Quit: Leaving]
<pie_>
now that i think about it, being stateless is pretty buddhist isnt it?
<hodapp>
MU!
alex`` has joined #nixos
jabranham has joined #nixos
hotfuzz has joined #nixos
<dtz>
So I definitely started down that path, and you're right Nix is amazing for this! The situation is a bit tricky but it's the sort of thing I spend my time building and exploring regardless... Im very interested, we should do it!
mmlb has quit [Read error: Connection reset by peer]
mmlb has joined #nixos
zolk3ri has quit [Remote host closed the connection]
dbmikus has joined #nixos
<nh2>
dtz: my plan is to start with C and make overflow panic. https://danluu.com/integer-overflow says that the runtime cost for that shouldn't be very large. Put that into an overlay and iterate, patching C libraries (submitting patches upstream) until the base system works. Then do the same with GHC and iterate there until it works. Bundle the whole thing into a "hardened" overlay advertised for people who want high security
<bigvalen>
For some reason, the getty@tty1 service is started by systemd quite late, and some dependent services are failing to start. Is it easy to move that service to start earlier, so I can login & diagnose ?
<thebloggu>
hello! I'm a python developer currently using macos and decided to use nix to manage my packages. however I still can't figure out how should I manage my python packages (every project I work on has a virtualenv and a requirements file). can someone provide some insight on how to build an environment from my requirements?
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/fpMx0
vidbina has quit [Ping timeout: 250 seconds]
erasmas has joined #nixos
Ariakenom has quit [Ping timeout: 244 seconds]
vidbina has joined #nixos
Lears has joined #nixos
[Leary] has quit [Ping timeout: 250 seconds]
vidbina has quit [Ping timeout: 240 seconds]
<andi->
Is there a `toHex` function for base 10 -> base16 conversion somewhere? I fail to find one and it would be really handy right now.. Thought I might ask before writing my own.
<Taneb>
Is it possible to define a derivation with no source, just build inputs?
crmlt has joined #nixos
<symphorien>
with runCommand or mkShell depending on your usecase ?
<Taneb>
symphorien: hmm, it didn't occur to me to use runCommand for this
<mdash>
thebloggu: the other tool to know about is pypi2nix which generates a nix expr from requirements.txt
ThatDocsLady has quit [Ping timeout: 250 seconds]
crmlt has quit [Ping timeout: 245 seconds]
<cyounkins>
How do I find the latest revision of staging for which Hydra has built binaries? I thought it would be https://hydra.nixos.org/jobset/nixpkgs/staging-next which has 13ee780 but that isn't in the log for staging
<thebloggu>
mdash: pypi2nix only allows building packages right? I can't take a requirements and start a shell with it with prebuilt packaged?
<thebloggu>
s/packaged/packages
<thebloggu>
symphorien: probably the reason why some install work. thank you
<elgoosy>
i've downloaded source code, i have to execute `make dockerinit`. nixos says i dont have make, i install it, do it again and get *docker build --pull --force-rm -t bitbox-wallet .make: /bin/bash: Command not found*
<pie_>
thats actually kind of a weird place for documentation, does that get pulled into the nixpkgs manual?
__Sander__ has quit [Quit: Konversation terminated!]
<WhittlesJr>
So why does doing nixops deploy require `nixos-config` in your NIX_PATH? Isn't that irrelevant to the actual deployment?
metastance has quit [Quit: WeeChat 1.4]
<ben>
How do I make `nix run nixpkgs.{pkgconfig,openssl} --command pkg-config --cflags --libs openssl` work?
<fendor>
huh, failed to find it on github... thanks!
<clever>
its in the /topic of #nix-tools
aleph- has joined #nixos
<fendor>
thanks!
emily has quit [Quit: down for repairs]
<{^_^}>
[nixpkgs] @orivej-nixos pushed commit from @orivej to master « libpqxx: fix license »: https://git.io/fpDJM
zolk3ri has joined #nixos
Ariakenom has joined #nixos
endformationage has joined #nixos
<WhittlesJr>
So a while back, samueldr told me that I could get the latest fully-built checkout of nixpkgs for aarch64 by looking at hydra. Well, even when tracking the "Latest Finished Evaluation," it still wants to build the kernel.
<tilpner>
WhittlesJr - Evaluation is just the process to go from a nix expression to a list of derivations
<clever>
WhittlesJr: the last eval to finish is the one with the worst caching, since it has just started building things in it
<clever>
you want the most recent eval to have zero pending jobs
<typetetris>
Hi. I would like nix to install clang on an old machine, but If I compile stuff with it, I can't call it, error message is : `libstdc++.so.6 => not found`. Could someone point me in the right direction, what is happening here?
<samueldr>
clever: know of a good job to use for aarch64 to mean "mostly finished and mostly successful?" :/ there doesn't seem to have an equivalent enough "tested" or "unstable"
<clever>
but also, that eval had 134 new failures
<clever>
samueldr: normally, channels take care of all of this for you
<samueldr>
clever: that's how nixos.org channes are updated
<WhittlesJr>
clever: Yeah I decided to go with the one prior, but I'd be interested in the answer to samueldr's question
<dtz>
nh2: in the journal version of that paper I looked at top 10k packages from Debian, as well as report on overheads and such. https://wdtz.org/pages/about.html lists them, direct PDF link: https://wdtz.org/files/tosem15.pdf and as you can see between that and the ICSE12 paper, finding and reporting such things is basically my favorite ;).
<WhittlesJr>
samueldr: Have you ever gotten "unable to download '...narinfo': Peer cetrificate cannot be authenticated" when deploying to aarch64? My x86_64 deploys go fine but I can't deploy to aarch64. This wasn't always the case...
rihardsk has joined #nixos
<tilpner>
But does it query individually for each path? That's what I'm trying to figure out
rihardsk has quit [Remote host closed the connection]
rihardsk has joined #nixos
rihardsk has quit [Remote host closed the connection]
<WhittlesJr>
the URLs of the .narinfo return 404
<gchristensen>
if WantMassQuery is 1, yes, it will be asked for every path
<{^_^}>
[nixpkgs] @ju1m opened pull request #51628 → stdenv: add armv7a-linux system → https://git.io/fpDIx
rihardsk has joined #nixos
<dtz>
If it's 0 will it ..ever ask?
<gchristensen>
I don't know, maybe if you explicitly say nix copy or something?
rawreraw has quit [Quit: WeeChat 2.3]
crmlt has joined #nixos
<dtz>
No further insights on that, sorry tilpner
Tucky has quit [Quit: WeeChat 2.2]
Havvy has quit [Ping timeout: 246 seconds]
rawreraw has joined #nixos
<gchristensen>
it isn't clear to me why you would have a binary cache with WantMassQuery set to 0 -- though it is interesting that Hydra set its binary cache WantMassQuery to 0
<tilpner>
Maybe they want people to use cache.nixos.org insetad
<gchristensen>
so, passing -s could result in 10s of thousands of queries in one shot, and maybe you don't want that since you don't actually host nixpkgs, so it would return almost entirely 404's
<tilpner>
Thanks for commenting on the issue!
<gchristensen>
Cunningham's law
<gchristensen>
dictates that if I posted the wrong answer, someone will replly with the right one
jperras_ has joined #nixos
crmlt has quit [Ping timeout: 268 seconds]
rawreraw has joined #nixos
fendor has quit [Remote host closed the connection]
jperras has quit [Ping timeout: 252 seconds]
<nh2>
dtz: good, good! Let me swiftly open an issue to track ideas there
rawreraw has quit [Client Quit]
rawreraw has joined #nixos
fendor has joined #nixos
dbmikus has quit [Ping timeout: 272 seconds]
worldofpeace has quit [Ping timeout: 250 seconds]
fendor has quit [Remote host closed the connection]
<aleph->
What's the best way to install a python package on pip but not in nixpkgs?
dbmikus has joined #nixos
<aleph->
Just use pip, or do we have a way to use nix as an intermediary?
emily has joined #nixos
pie_ has quit [Ping timeout: 256 seconds]
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
emily has quit [Ping timeout: 264 seconds]
alex`` has quit [Ping timeout: 250 seconds]
hotfuzz_ has joined #nixos
fendor has joined #nixos
hotfuzz has quit [Ping timeout: 250 seconds]
alex`` has joined #nixos
<meizikyn>
aleph: I think theres documentation for this with nix-shell on the wiki
<meizikyn>
but virtualenv should work just as well
emily has joined #nixos
worldofpeace has joined #nixos
jonaswouters has quit [Quit: Connection closed for inactivity]
Radivarig has quit [Ping timeout: 246 seconds]
<worldofpeace>
Anyone hove Jack2 and pulseaudio working?
<goibhniu>
worldofpeace: yep
<aleph->
meizikyn: Okay guess I'll look into it.
<worldofpeace>
goibhniu: Care to share relevant nixos config?
<goibhniu>
did you see the wiki page about it? I think it's all there
<worldofpeace>
Yeah but it's labeled as very outdated
<goibhniu>
I don't know, I have been using the same config for ages.
<worldofpeace>
the override shouldn't be need anymore I think so `package = pulseaudioFull;`
<goibhniu>
that would be nice
jrolfs_ has joined #nixos
sigmundv has quit [Ping timeout: 268 seconds]
mmlb has quit [Read error: Connection reset by peer]
mmlb has joined #nixos
<WhittlesJr>
Ah! For posterity, in case anyone else searches the chat logs: my problem was that my Raspberry Pis had the wrong date/time, thus the (current) SSL certs were in the future and thus were rejected.
<cyounkins>
doh!
<WhittlesJr>
haha
rihardsk has quit [Quit: rihardsk]
worldofpeace has quit [Remote host closed the connection]
worldofpeace has joined #nixos
smolboye has joined #nixos
jrolfs_ has quit [Ping timeout: 250 seconds]
seqizz has joined #nixos
seqizz has quit [Client Quit]
<develCuy>
o/
worldofpeace has quit [Remote host closed the connection]
<ik3123>
Hi, any chance someone can help me with an issue booting on zfs? It times out on importing the zpool but imports fine if I boot to the install media again.
<cyounkins>
ik3123: Do you have log output? Is it trying to import from cache or by name?
dsx has joined #nixos
<ik3123>
just name from what I can see in the boot log. I haven't been able to get to a prompt to check logs, it just goes to kernel panic
<cyounkins>
ik3123: What bootloader?
<ik3123>
grub2
<ik3123>
seems to be the way the wiki does it
<cyounkins>
If your system supports EFI, you could try systemd-boot. I don't know how grub2 loads drivers for disk controllers
<ik3123>
I thought maybe I'd try the zfs unstable, just running the nixos-install. Just thought maybe there was something obvious the wiki wasn't mentioning that I should know
<ik3123>
yah, I've gotten arch to go with that before, first attempt on nic
<ik3123>
nix
jperras_ has quit [Quit: WeeChat 2.2]
<cyounkins>
Wish I could help more, I'm not running ZFS on root.
<ik3123>
thx tho
vidbina has joined #nixos
betaboon has quit [Quit: WeeChat 2.2]
freeman42x has joined #nixos
<ottidmes>
develCuy: I am buliding it now
<develCuy>
ottidmes: you rock!
justanotheruser has quit [Ping timeout: 250 seconds]
aleph- has quit [Ping timeout: 268 seconds]
<tilpner>
Wasn't there some way to add out-of-tree "nix foo"-style commands? A quick skim of the source doesn't suggest so, but I haven't looked at the plugin API yet
<{^_^}>
#51630 (by cyounkins, 32 seconds ago, open): nixos: Decide and document scope of configuration
<ottidmes>
tilpner: I could only see something about adding builtins when searching for it, but you could always create a kind of wrapper yourself and point nix.package to it, I guess
<tilpner>
That's more ugly than the alternative of "nix-foo"
<tilpner>
I was asking because I vaguely remember someone talking how all the lang2nix "importers" could be pulled into the new CLI as "nix import-lang"
<tilpner>
So I suspected there might be some git-style extension mechanism, but I didn't manage to find it
<{^_^}>
[nixpkgs] @worldofpeace pushed commit from @prusnak to master « luppp: init at 1.2.0 »: https://git.io/fpDZ7
<ottidmes>
tilpner: if its just for you on the interactive shell, a simple shadowing shell function would work as well, but you might find that ugly too
<{^_^}>
[nixpkgs] @worldofpeace closed pull request #51473 → applications.audio.luppp: new package (at version 1.2.0) → https://git.io/fpiJH
rsully has joined #nixos
jrolfs_ has joined #nixos
worldofpeace has quit [Quit: worldofpeace]
<tilpner>
Yeah, nix-foo it is. Thank you! ottidmes++
<hakujin>
we have a hydra server that builds both github PRs and commits to `master`. is there a way we can coax hydra into re-using successful PR build products for the `master` build, post merge? would save a ton of time
<hakujin>
this question may be: "how do I get nix to build the same product twice so hydra can share?"
zolk3ri has joined #nixos
<symphorien>
why are the derivations different after merge ?
<hakujin>
yeah, great question
<symphorien>
you can use nix-diff to find out
<hakujin>
aha. will give that a shot. thanks!
crmlt has joined #nixos
<cyounkins>
ik3123: does your hardware-configuration have fsType = "zfs", configuration.nix have boot.supportedFilesystems = [ "zfs" ]; and boot.loader.grub.devices?
jrolfs_ has quit [Ping timeout: 250 seconds]
<ik3123>
I'll pastebin them, but yes, as far as I can see that's all correct
<cyounkins>
I actually did have root on ZFS working at one point in a VM using EFI
<cyounkins>
Since it can't find the pool I feel like it's related to the kernel module for the disk controller
<ik3123>
this is a vm, proxmox. just using basic disks though, and no efi boot
<ottidmes>
ik3123: I believe the liveCD now also support ZFS, so you could also try boot into that, and do the things manually to test if its all good that way. I have little ZFS experience, but I remember that having a different hostId caused my imports to fail, so also make sure that thats properly set, and you might need to add some extra import arguments to force an import initially after the host id change
<ik3123>
the live cd works fine, can import and manipulate. Initially tried to add that but it was already active
<ik3123>
hostid: how do you determine what that should be? I know you need it for the install compile, but I just used a 32bit value I got off another guys config.
<cyounkins>
It's a random value
crmlt has quit [Ping timeout: 246 seconds]
<cyounkins>
If you boot from the livecd do you need to force import it?
<ik3123>
yes
fendor has joined #nixos
Thra11 has joined #nixos
<ik3123>
which makes sense, I can't see that the livecd would autoimport would it?
<cyounkins>
then hostid may well be your problem
<ottidmes>
that could be it then, no? by force importing it, you need to then also force import it on the actual install hostId
<cyounkins>
Then make sure you export the pool before rebooting
<ik3123>
ok, so I've missed a step then. How do I determine the hostid I should use? And yes, I've exported the pool clean before rebooting a couple times, didn't help
<Laalf>
can i manage manage pulseaudio support via home-manager? mumble is installed as system at the moment and i dont really like that.
johnnyfive has joined #nixos
<ik3123>
so when I generate that 32bit id, it returns an 8 hex string. all well and good, I put thatinto the config. But if I cat the machine-id file, it returns a longer string starting with that 8 hex value. Is that OK?
kreisys has quit [Remote host closed the connection]
<ottidmes>
ik3123: so I guess you will have to enter the LiveCD, import the pools, run cat /your/pool/etc/hostid | od -A none -t x4, take the hex representation of the hostid and set the liveCD's hostid to that: sudo cp /etc/hostid{,~} && sudo mv /etc/hostid{~,} && sudo nano /etc/hostid, do the force import now with the correct hostid, and reboot and hope for the best?
<cyounkins>
ik3123: If you used `head -c 8 /etc/machine-id` to generate it, of course.
<cyounkins>
I don't think you even have to do the copy and force import so long as you export it before rebooting
<clever>
yeah, if you export the pool properly, it can be imported without a force
<clever>
and that import can come from a different hostid
<Ericson2314>
jophish: angerman angerman: yes the way we build GCC sucks, but should handle build != host != target in principle
<ottidmes>
ah, clever to the rescue, he knows!
<Ericson2314>
but you will need to invoke stage.nix yourself
<Ericson2314>
no combination of `buildPackages` and `targetPackages` crawling will give you a package set where all 3 are diffferent today
<Ericson2314>
that should also be fixed
<Ericson2314>
in addition to the monolithic GCC
<Ericson2314>
but neither of those block you per-se, just make your life more difficult
<cyounkins>
ik3123: looks ok from what I see
<ik3123>
and of course I forgot the refresh the install...
vidbina has quit [Ping timeout: 240 seconds]
<gchristensen>
,locate gmp.h
<{^_^}>
Found in packages: cgal, gmp5.dev
jperras has joined #nixos
_d0t has joined #nixos
<_d0t>
ohai! The latest unstable cannot be built. Is that alright?
<gchristensen>
what are you trying to build?
<_d0t>
nixos-rebuild switch
<_d0t>
I get 404 while downloading some packages and build errors while building others
<_d0t>
steam fails to update due to 404s in some dependencies
<hakujin>
symphorien: we use `pkgs.haskellPackages.callCabal2nix` to generate the input source so we don't have to manually update the .nix file if we add a dependency; my `nix-diff` investigation leads me to the `{HASH}-source` inputs to the PR/master derivations differing
<hakujin>
I looked at the actual files contained in the store paths for the supposedly diverging inputs and they're identical
<clever>
hakujin: if those paths are not fixed-output derivations, then it will depend on the hashes of its inputs, not the contents of its outputs
<hakujin>
ahh
<hakujin>
clever: happen to know a way to get `callCabal2nix` to generate deterministic outputs?
<tilpner>
Is there any easy-ish way to get the outputs of the runtime closure of the output of a derivation without having built that derivation?
<tilpner>
It doesn't seem like there is, but I'm asking anyway before I write my own thing to collect that closure
<clever>
hakujin: its using filterSource to do that already, but it doesnt work if the input src is already in the nix store
<clever>
tilpner: not really, nix only knows the runtime closure by grepping the output
<hakujin>
so, if I'm following: hydra builds the PR, registers the input source in the store, we merge, callCabal2nix already has input in the store, voila?
<clever>
hakujin: if any part of the source changes, the storepath for the .cabal file will change, and it has to at least re-run cabal2nix
<tilpner>
clever - So you're saying there's no hard way to do that either?
<clever>
hakujin: check to see if there is any other differences
<clever>
tilpner: yeah
<tilpner>
That makes sense, and complicates this a little. clever++
<cyounkins>
ik3123: does "no dice" mean it fails to import rpool?
<ik3123>
cyounkins: yes.
danielrf has joined #nixos
<cyris212_>
"if it’s (also) a runtime dependency, use propagatedBuildInputs" nvm, the manual answered my question :-)
<ottidmes>
ik3123: BTW, we prefer github gists, multiple files, syntax highlighting, allowing changes and seeing diffs what changed, all useful trying to figure out issues with a config
<ik3123>
I did run across this closed issue: https://github.com/NixOS/nixpkgs/issues/9904 but that seems to have been fixed a few years ago and legacy mountpoint seems the supported method
<{^_^}>
#9904 (by dasjoe, 3 years ago, closed): Booting off ZFS doesn't work as intended, as mount.zfs disallows legacy mounts
<ik3123>
ottidmes: will do
<Lears>
So I have some local Haskell projects that build properly with cabal new-build in nix-shell, but where nix-build fails on one of them with a cabal error saying it can't find the other, even though it's clearly included amongst the --extra-lib-dir configureFlags.
<ottidmes>
ik3123: hopefully someone with a ZFS setup can pitch in, because I dont see any glaring mistake in your configuration and although I run two servers with ZFS pools, they are otherwise managed outside of nixos (i.e. not using fileSystems), so I dont have ZFS as root on any, but I know there are plenty of people here that do
crmlt has joined #nixos
<sphalerite>
IIUC, services.nginx.virtualHosts.foo.locations."/bar".root = "/var/www/baz"; will mean that when someone accesses http://foo/bar/bletch, nginx will try to serve /var/www/baz/bar/bletch. Is there a way to get it to server /var/www/baz/bletch instead?
<ik3123>
ottidmes: fair enough, i really do appreciate your help, it did give me a better understanding of it.
EarlDeLaWarr has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<pie_>
root is supposed to be the "do what i want even if its stupid" user :(
<ik3123>
ottidmes: alrighty, I did a cursory attempt at that and it's telling me rpool doesn't exist, but can't get a print of it,or a console session. I have to head out for awhile now, I'll try again later and hopefully you or another person can help me make sense of it.
<ik3123>
ottidmes: again, ty for your time.
<pie_>
connection hiccup
<pie_>
[21:43] <pie_> yeah ok :c [21:43] <pie_> doesnt work with su either x) [21:43] <sphalerite> And because it's not configured systemwide (which you can do but you shouldn't do that either usually). [21:43] <pie_> ah su needs -l
<Mic92>
I have build for the other platforms already.
<infinisil>
mightybyte: because we need a hash to be able to download stuff at all in the sandbox
<gchristensen>
Mic92: sure
<mightybyte>
infinisil: Couldn't you just have the user specify a hash?
<infinisil>
mightybyte: yeah i guess, but why would you want to specify a hash when you can not do that too?
<infinisil>
It's much more convenient
<ik3123>
ottidmes: lol. So I start an interactive shell in the stage 1 failure. I try to manually zpool import rpool and it won't go. Come back to it and try again about 3 minutes later, it gives me the error the pool already exists. zpool status shows it mounted now. So apparently I have a timing issue?
fragamus has joined #nixos
<infinisil>
And you can still it with your own source if you want to (and provide the hash)
<infinisil>
do* it
<LnL>
Mic92: I have it queued, but hydra has been acting weird lately
mmlb has quit [Read error: Connection reset by peer]
jonaswouters has joined #nixos
mmlb has joined #nixos
<gchristensen>
Mic92: nix-build . -A fd?
<Mic92>
gchristensen: yes
<mightybyte>
infinisil: Because I want to be able to get the latest packages straight from hackage without updated all-cabal-hashes and triggering rebuilds of all kinds of things that depend on it.
<ottidmes>
ik3123: I remember this issue, now for the fix...
<develCuy>
ottidmes: perhaps it is entering the wrong folder, can you try to find the path of this path in your current broken build? "webide-prefs.js"
<develCuy>
s/path/file/
pxc has joined #nixos
<ik3123>
ottidmes: may not be timing, I just tried zpool import rpool, it gives me an i/o error, but if I check zpool status it shows it mounted.
<ik3123>
ottidmes: ok, now I really have to go, I will check back ina couple hours.
<ottidmes>
ik3123: things I could think of, just try and redo the ZFS, creating them anew, or else just dont use fileSystems to mount the stuff and do it yourself at e.g. preLVMCommands, but not sure if that works
<ottidmes>
ik3123: ok
jrolfs_ has joined #nixos
<Mic92>
gchristensen: if I could cover all plattforms, I would give this a try.
<ottidmes>
develCuy: how does one enter a failed build again
rhalff has joined #nixos
<develCuy>
afaik it is somewhere in the /nix directory or /tmp
<develCuy>
ottidmes: ls /nix | grep palemoon-src
freeman42x has quit [Read error: Connection reset by peer]
<ottidmes>
develCuy: its not there
freeman42x has joined #nixos
<{^_^}>
[nix] @jvassev opened pull request #2565 → Saner defaults for MANPATH → https://git.io/fpDzn
crmlt has quit [Ping timeout: 250 seconds]
<ottidmes>
this should be a really basic question, but its hard to google, anyone? how to inspect the failed build directory
<gchristensen>
Mic92: maybe I could set you up with access to my ofborg builder
<rhalff>
Hi, I'm trying to understand this code part: https://github.com/garbas/dotfiles/blob/master/config/default.nix#L35-L44 is it required to use 'inherit' when passing declared values to a function? Also not sure why all packages are hand picked and then passed into the function. Trying to determine whether it's coding style or something nix specific.
fendor has joined #nixos
<ivan>
chromium's multiple outputs seem to confuse hydra into showing that the build took 4 seconds instead of N hours
<ottidmes>
rhalff: thats not valid Nix code, so yeah
goibhniu has quit [Ping timeout: 250 seconds]
<aminechikhaoui>
ivan the build steps tab should show the actual build duration
<ottidmes>
> let foo = 5; in { foo }
<{^_^}>
error: syntax error, unexpected ')', expecting ':' or '@', at (string):211:1
<ivan>
aminechikhaoui: yep
<ottidmes>
> let foo = 5; in { inherit foo; }
<{^_^}>
{ foo = 5; }
<ottidmes>
> let foo = 5; in { foo; }
<{^_^}>
error: syntax error, unexpected ';', expecting '.' or '=', at (string):210:22
<rhalff>
ottidmes, ok thanks. I need babeljs for nix to add language some language features :-)
<ottidmes>
rhalff: haha, there actually are some projects that give other frontends for Nix, but I would not recommend them for regular use
fendor has quit [Remote host closed the connection]
<gchristensen>
yay, on to stage2
<ottidmes>
gchristensen: I really ought to know this, but if I build something, I see /build mentioned, any way for me too access that /build dir?
<symphorien>
with nix-build -K ?
<symphorien>
or breakPointHook ?
<gchristensen>
you can add --keep-failed to the nix-build command, and it'll keep the build dir after failure
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<ottidmes>
gchristensen: ah so its just gone, ok, makes sense then that I cannot find it...
fendor has joined #nixos
<gchristensen>
/build only exists inside the sandbox, never in your environment
<Mic92>
intentionally
pxc has quit [Ping timeout: 252 seconds]
vk3wtf has quit [Ping timeout: 252 seconds]
<ottidmes>
ok makes sense, I guess I will just use nix-shell, no point in waiting another 35 minutes just to test the install phase
<ottidmes>
got a weird bug, when using --pure, my overwritten command-not-found is being called, which in my case runs nix-locate /bin/$missing_bin, and apparently tput is missing, and it keeps trying to call tput unless I kill the shell
pxc has joined #nixos
<rizary>
a
rawreraw has quit [Ping timeout: 240 seconds]
drakonis has joined #nixos
dbmikus has quit [Quit: WeeChat 2.3]
porrifolius_ has quit [Remote host closed the connection]
porrifolius has joined #nixos
<{^_^}>
[nixpkgs] @c0bw3b merged pull request #51572 → qmltermwidget: add missing utmp lib on darwin → https://git.io/fp1qs
fendor has quit [Read error: Connection reset by peer]
fendor has joined #nixos
<fendor>
sorry, pc froze
<symphorien>
closure means inside nix-store -qR, ie even the smaller executable would pull glibcLocal from the internet because it would depend on it
<fendor>
symphorien, but if the locale is shared, shouldnt it only download if it is needed?
<symphorien>
sure, but people using nix for docker images for example would complain
<fendor>
docker is inherently large, I build my docker images with other docker images, do 150 mb have such a high impact?
<fendor>
what does impure mean in the sense of nixos actually? that the packages rely on it that they are already present?
Mr_Keyser_Soze99 has joined #nixos
alex`` has quit [Ping timeout: 246 seconds]
pxc has quit [Ping timeout: 252 seconds]
Mr_Keyser_Soze99 has quit [Quit: Leaving]
<ottidmes>
develCuy: if you are still here, the files its looking for are here: /tmp/palemoon/palemoon-src/obj-x86_64-pc-linux-gnu/dist/bin/browser/chrome.manifest
<develCuy>
ottidmes: great catch! Also, can you gist the change you did to the nix file?
<ottidmes>
develCuy: I will once I figure out how to fix the last two lines such that it builds succesfully
<develCuy>
k
<develCuy>
so the path is in temp... let's see the nix settings
<ottidmes>
no
<ottidmes>
thats only the case because I made it so with nix-shell
Mr_Keyser_Soze99 has joined #nixos
<ottidmes>
I wanted to be able to keep testing the installPhase without having to wait 35min each time :P so I just used nix-shell in a tmp dir and called the phases one-by-one
<develCuy>
clever
<develCuy>
you have this: cd $builddir; $src/mach install
<ottidmes>
I guess the `cd $builddir` is wrong
<develCuy>
what does that var has in it?
<ottidmes>
/tmp/palemoon/palemoon-src/build
<ottidmes>
which is basically the working directory of any build, just like /build is in the sandboxed nix-build
Mr_Keyser_Soze99 has quit [Remote host closed the connection]
Mr_Keyser_Soze has joined #nixos
<develCuy>
how about cd $builddir/dist
<ottidmes>
that does not exist, its being written to /tmp/palemoon/palemoon-src/obj-x86_64-pc-linux-gnu/dist instead
<develCuy>
oh, I meant that
pxc has joined #nixos
<ottidmes>
I guess I can try, but it seems wrong to hardcode: $builddir/../obj-x86_64-pc-linux-gnu/dist, I am more curious about $src/mach install
<develCuy>
let's use other var, like $src what do you have there?
pie___ has quit [Remote host closed the connection]
pie___ has joined #nixos
Ariakenom has quit [Quit: Leaving]
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<ottidmes>
develCuy: nothing useful, so I guess we have to patch /tmp/palemoon/palemoon-src/obj-x86_64-pc-linux-gnu/Makefile somehow
<ottidmes>
the installer folder has a Makefile too, so so far it makes sense, but that file loads a package-manifest file with files relative to target/dist rather than installer
<ottidmes>
develCuy: it points back to the target directory: /tmp/palemoon/palemoon-src/obj-x86_64-pc-linux-gnu, how is that bad?
<develCuy>
that one is calculated somehow, so we should fix autoconf perhaps?
ZeDestructor has quit [Quit: o.O]
epicmetal has quit [Quit: epicmetal]
<ottidmes>
this is not a tool chain I am that familiar with, I managed so far, but debugging this stuff is not really my cup of tea, Makefiles and such seem so unnecessary complicated
epicmetal has joined #nixos
ZeDestructor has joined #nixos
dslegends has quit [Quit: dslegends]
epicmetal has quit [Client Quit]
o1lo01ol1o has quit [Remote host closed the connection]
epicmetal has joined #nixos
ZeDestructor has quit [Client Quit]
<ottidmes>
develCuy: DIST = /tmp/palemoon/palemoon-src/obj-x86_64-pc-linux-gnu/dist in autoconf.mk seems fine though
<ottidmes>
develCuy: include $(DEPTH)/config/autoconf.mk is also correct
ZeDestructor has joined #nixos
pxc has quit [Ping timeout: 252 seconds]
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<develCuy>
ottidmes: can you try to build without that line?
<ottidmes>
develCuy: its the other way around, its been removed in the pull request
<ottidmes>
so we would need to add it again if anything
<develCuy>
k
<ottidmes>
develCuy: could you verify, I do not want to wait another 35min just to find out we had to add yet another of those lines, because quite a few have been removed
<develCuy>
ottidmes: I have replaced the full build step with the official palemoon instructions
<develCuy>
gtg
Mr_Keyser_Soze has quit [Ping timeout: 268 seconds]