emacsomancer has quit [Read error: Connection reset by peer]
alexteves has quit [Remote host closed the connection]
hakujin1 has joined #nixos
Fare has joined #nixos
hakujin1 has quit [Ping timeout: 265 seconds]
mayhewluke has quit [Ping timeout: 240 seconds]
mayhewluke has joined #nixos
Anton-Latukha has quit [Quit: Leaving.]
thc202 has quit [Ping timeout: 240 seconds]
hakujin1 has joined #nixos
emacsomancer has joined #nixos
trylist has joined #nixos
silver has quit [Read error: Connection reset by peer]
Ariakenom_ has quit [Quit: Leaving]
hakujin1 has quit [Ping timeout: 256 seconds]
hakujin1 has joined #nixos
barryfm has joined #nixos
barryfm has quit [Quit: Leaving]
emacsomancer has quit [Remote host closed the connection]
trylist has quit [Ping timeout: 248 seconds]
hakujin1 has quit [Ping timeout: 240 seconds]
baconicsynergy has joined #nixos
ris has quit [Ping timeout: 256 seconds]
iyzsong has joined #nixos
vodurden has joined #nixos
lassulus_ has joined #nixos
<vodurden>
/topic
<vodurden>
Supersonic has quit [Ping timeout: 240 seconds]
<vodurden>
Hello all! I'm new to IRC so apologies if I accidentally post a command. I wanted to ask about submitting a PR to nixpkgs. Recently I needed to improve the thermals of my new laptop running NixOS
lassulus has quit [Ping timeout: 272 seconds]
lassulus_ is now known as lassulus
<vodurden>
To accomplish that I modified the thermald package to add manual configuration, and I added a package to help with undervolting newer Intel CPUs to my fork of nixpkgs
<samueldr>
hi vodurden; no worries, everyone makes mistakes, you'll see we're generally friendly (I think) here
<vodurden>
I was wondering if those sorts of changes are useful to the wider group, and if so should I just submit a PR or is there a process I need to follow beforehand?
<vodurden>
It made a huge difference to the temperatures of my laptop configuring it manually, but I'm not sure if I should have just added a config option that accepts a file/string with the correct XML
baconicsynergy has quit [Ping timeout: 260 seconds]
<samueldr>
generally it's preferred to make it work with nix types
<samueldr>
either manually when dealing with a limited set, or like you apparently did
<samueldr>
I'm not 100% sure the way you did it is the entirely right approach, but you're at least on the right track
<samueldr>
(I'm not an expert on nixos module and I would defer that judement call to one)
hakujin1 has quit [Ping timeout: 256 seconds]
<vodurden>
Oh! One last question while I go sort out those PRs: To get thermald RAPL support working I needed to enable CONFIG_POWERCAP and CONFIG_INTEL_RAPL in the kernel. I noticed they're not set in the current nix kernal settings, is there anywhere I can look to figure out why they aren't set?
<samueldr>
one note for your undervolt commit, it would be **really** appreciated (required even) to split it into a commit adding the package, and one adding the module
<vodurden>
I saw that modules can declared `requiredKernelConfig` but that doesn't seem to be used for that manner so I ended up just manually enabling them in my config
<vodurden>
No worries, I can definitely do that. I'm assuming we'd still want one PR though?
<samueldr>
yes!
imalsogreg has joined #nixos
<samueldr>
requiredKernelConfig only *checks* the kernel's features AFAICT
<vodurden>
I guess those settings aren't technically required for thermald to work either. It just enables RAPL support which on my system was much more effective for cooling
<samueldr>
this'll create assertions based on the values in there, assertions will stop the build of the system if they don't pass
<samueldr>
sometimes it's as simple as "nobody did it" for options
<vodurden>
Would that be something that it would be appropriate to raise a Github issue about? Or just a PR?
<samueldr>
normally an issue should suffice, but I know for a fact they are often overlooked, Cunningham's Law will probably make a PR more effective
<samueldr>
alternatively, you could ask in the discourse
jperras has joined #nixos
<vodurden>
Fair enough. I might just make a PR if I have time later today and see how receptive people are :)
<samueldr>
though, make sure you explain the whole rationale :)
<{^_^}>
[nixpkgs] @Vodurden opened pull request #44919 → undervolt: init at 0.2.7 → https://git.io/fN5S7
<vodurden>
Ok, I've created a PR for undervolt ^^^, please let me know if I missed any crucial steps and I'll try to sort them out. And thanks for helping me work through the process :)
<vodurden>
I'll add another PR for thermald later today
<clever>
vodurden: ah, that reminds me, i was having thermal shutdown troubles on my amd fx-8350
<samueldr>
"later today", you must be living the future :) (EST timezone here)
<vodurden>
Haha I'm in Australia
<clever>
vodurden: i have since discovered that windows uses far less power at max load, but linux power usage matches the specs on the cpu, so windows is technically the broken one
<vodurden>
clever: Yeah I was hitting 95C temps on the CPU when running any games
<clever>
vodurden: mine would thermal shutdown around 85c
<vodurden>
Did a bunch of digging, it turns out the frequency scaling and power scaling wasn't working
Drakonis has quit [Remote host closed the connection]
<samueldr>
hmm, might check into that undervolt thing next heat wave if it reduces the head of my workstation
<vodurden>
Hence adding undervolt and thermald config to nixpkgs. It was also idling at 60C!
<clever>
vodurden: i also dont have any control over freq scaling, but it does seem to be doing some on its own internally
<vodurden>
Now it idles at 40C and I've configured it to stay under 80C
<samueldr>
ooh
<vodurden>
Hmm, you should have some control I _think_. Mainly via setting the p-states if it's a newer intel CPU
<clever>
windows on the same hardware definitely seems broken, it cant get over 60c at max load
<vodurden>
Which is what thermald seems to do
<clever>
amd cpu
<vodurden>
Ahhhh
<clever>
fx-8350
<vodurden>
Sorry I missed that :)
<samueldr>
vodurden: minor minor nitpick, but the module itself doesn't have a version, so simply "nixos/undervolt: adding undervolt module" would be better :)
<vodurden>
Oh of course.
<clever>
[root@amd-nixos:~]# cpufreq-info
<samueldr>
but as for the following the process, you're doing fine
<clever>
no or unknown cpufreq driver is active on this CPU
<clever>
vodurden: but despite it claiming to not have a cpufreq driver, the freqs in `grep MHz /proc/cpuinfo` change on their own
<clever>
cpu MHz : 3844.199
<clever>
cpu MHz : 3412.972
<vodurden>
Interesting
<clever>
and i can even see a 400mhz difference between 2 cores
<clever>
cpu MHz : 2684.090
<clever>
cpu MHz : 2331.485
<vodurden>
Though mine doesn't seem to use cpufreq either, it seems to set something under `/sys` to achieve the desired p-states. Not 100% how it works though
<vodurden>
Maybe amd has something equivalent? I know the newer intel CPUs don't work with the old frequency methods and require a bunch of new nonsense through the intel_pstate driver
hakujin1 has joined #nixos
<vodurden>
Anyway I've gotta run. I'll be back later to sort out the rest of those PRs. Thanks again for all the help :)
Drakonis has quit [Read error: error:1408F10B:SSL routines:ssl3_get_record:wrong version number]
Drakonis has joined #nixos
Drakonis has quit [Remote host closed the connection]
Fare has quit [Ping timeout: 240 seconds]
<vodurden>
I'm trying to test a PR I'm working on for thermald and when I run `nix-shell -p nox --run "nix-review wip"` I get `nox-review: command not found`. Is there a different command I should be running or have I missed a step?
<clever>
vodurden: you need to `nix run nixpkgs.nox` first
<clever>
ah, but nix-shell -p nox should do the same..
jb55 has quit [Quit: WeeChat 2.1]
<vodurden>
It seems to work after I ran `nix run nixpkgs.nox`, hmm
<clever>
it may be changes with nix-shell, but id call that a bug
<vodurden>
It's possible I ran it in the wrong context or with a partially rebased nixpkgs.
Mr_Keyser_Soze has quit [Ping timeout: 272 seconds]
<clever>
vodurden: nix-shell uses <nixpkgs> from $NIX_PATH
<vodurden>
Yeah my <nixpkgs> is a local copy that I _might_ have been in the middle of rebasing. I wasn't paying a huge amount of attention so my memory is a bit fuzzy :)
<clever>
ah
<clever>
though `nix run nixpkgs.nox` will also use <nixpkgs>
<vodurden>
Hmm, now I'm getting "Malformed database file header" when I run `nix-shell -p nox --run "nox-review pr 44923"`
<vodurden>
The PR number is my thermald PR and I'm running it in `/nix/nixpkgs` which is my <nixpkgs>
<clever>
not sure on that one
<vodurden>
I have `/nix/nixpkgs` checked out to the PR so maybe I need to run the command with nixpkgs checked out to the correct base, hmm.
sir_guy_carleton has quit [Quit: WeeChat 2.0]
<vodurden>
Nope same issue. Guess I'll poke around the documentation and see if I can figure out what's going on
<vodurden>
Thanks for the help regardless :)
Maxdaman1us is now known as Maxdamantus
orivej has quit [Ping timeout: 256 seconds]
Mr_Keyser_Soze has joined #nixos
kyren has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
endformationage has quit [Ping timeout: 240 seconds]
Mr_Keyser_Soze has quit [Ping timeout: 260 seconds]
FireFly has quit [Quit: Goodbye]
<dmj`>
clever: ping
FireFly has joined #nixos
<clever>
dmj`: pong
vmandela has joined #nixos
<dmj`>
clever: along with github_authorization, does <githubstatus> need to be specified as well in the hydra config for PRs
kyren has joined #nixos
<clever>
dmj`: yeah
kyren has quit [Client Quit]
Dedalo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
hakujin2 has joined #nixos
johnw has joined #nixos
vmandela has quit [Quit: Leaving]
johnw has quit [Read error: Connection reset by peer]
<maerwald>
can I provide an environment for vim maybe
<leotaku>
clever: Just a small question: Is it expected that qemu "cross-compilation?" for arm is very slow? I kinda have been compiling the linux kernel for 10+ hours now...
<clever>
leotaku: yeah, it will be pretty slow
<clever>
leotaku: proper cross-compile will be faster, but is more complex to setup, and doesnt work for all packages
init_6 has joined #nixos
jD91mZM2 has joined #nixos
rauno has joined #nixos
simukis has joined #nixos
init_6 has quit [Ping timeout: 240 seconds]
<leotaku>
clever: Hmmm, that's unfortunate... But if I managed to build something with a cross-compiling snippet nixops would pick it up from the local cache right?
<clever>
leotaku: proper cross-compile and qemu "cross" compile are seperate compilers as far as nix is concerned, and it will never mix the 2
<leotaku>
Maybe building on the actual arm machine + cachix could also work.
<leotaku>
clever,
<leotaku>
ok
alex`` has joined #nixos
alex`` has quit [Quit: WeeChat 2.2]
<maerwald>
seems youcompleteme plugin definition is broken in nix
<clever>
maerwald: broken how?, it works for me
<maerwald>
you cannot build
<maerwald>
ERROR: Unable to find executable 'cmake'. CMake is required to build ycmd
<maerwald>
so there is no actual completer
<maerwald>
just buffer
<clever>
how are you trying to build it?
<maerwald>
in a nix-shell
<clever>
there is already a nix package for it
<maerwald>
yes, I'm using that for the nix-shell
<clever>
ah
<clever>
what command did you run in nix-shell to build it?
<maerwald>
the one that you are supposed to
sb0_ has joined #nixos
<clever>
which is?
<sb0_>
hello. does anyone have an example of using swap.randomEncryption in configuration.nix?
<sb0_>
if I simply add "swap.randomEncryption.enable = true;" then I get "error: The option `swap' defined in `/etc/nixos/configuration.nix' does not exist."
<dhess>
LnL: you changed the order of hashes reported by nix download. It's only a problem for me because I was inadvertently using nixUnstable since back when that was the only way to get 2.0
<mpickering>
You know I think it would be easier if the format was `Expected: ....\nActual: ... `
<mpickering>
Having the hashes inline makes it annoying when they wrap and harder to copy
<dhess>
It was because you could just go to the end of the line and grab the last hash
<dhess>
in emacs, anyway
poapoa has quit [Ping timeout: 272 seconds]
<joepie91>
mpickering++
<{^_^}>
mpickering's karma got increased to 1
<joepie91>
(color-coded for bonus points)
<dhess>
Anyway it's not LnL's fault, he was just matching what fetchurl does
<dhess>
mpickering: even better would be "Expected:\n<hash>\nActual:\n<hash>\n" because then you could just easily double-click copy
<mpickering>
tbh I can already double click copy
<mpickering>
but it's not as easy as it could be to hunt for the right hash when wrapping happens
<LnL>
yeah, I noticed at some point that the order of builtins vs nixpkgs was different
<dhess>
well ok, if you're in emacs you can easily go to the next line and copy line without taking your hands off the keyboard (vi too?)
<LnL>
my terminal can find hashes now
<dhess>
oh that's nice
<LnL>
just have to press a number for the one I want
<dhess>
johnw wrote something for emacs that can find them in some cases but I don't think it handles the "pin nixpkgs and fetch it" case
<mpickering>
I think a by default clearer easier way is desirable as most people will not have their terminal set up to do this
<LnL>
sure, the current message isn't ideal
<Myrl-saki>
It seems like it'd be nice if this works.
dhess has quit [Remote host closed the connection]
thblt has joined #nixos
<thblt>
I'm using an HiDPI laptop (4K XPS 9650). With boot.earlyVconsoleSetup = true; I've been able to get a readable font for LUKS passphrase prompt. Is there a way to bring this to systemd-boot too (NixOS generations menu)?
<mpickering>
I am trying to use `nix-store -q --graph` but the generated graph has 8000 edges, as I'm just interested in the relationship between two nodes, is there a way to filter the edges somehow?
aanderse has joined #nixos
* thblt
looks
<thblt>
srhb: what's @this@ syntax? Does nix preprocess the Python file before running it?
<Myrl-saki>
Seems like memmove is okay though, it's just autotools.
hakujin2 has quit [Ping timeout: 240 seconds]
<jD91mZM2>
Myrl-saki: Ah yes, autotools does that, a lot
<jD91mZM2>
Been struggling it trying to get it to detect functions in Redox' relibc.
<Myrl-saki>
jD91mZM2: Well, crap.
<jD91mZM2>
thIS FuNCtIon DoEs Not ExISt
<Myrl-saki>
LMAO
<joepie91>
lol
<jD91mZM2>
comments out check in configure
<jD91mZM2>
works
<jtojnar>
is it possible to stop python application (meson) from propagating python?
<Myrl-saki>
jD91mZM2: On the script itself?
<jD91mZM2>
Not sure, lemme check
<jD91mZM2>
In my case, compiling bash, it was aclocal.m4
<jD91mZM2>
and configure.ac
<Myrl-saki>
Oh, hm.
<Myrl-saki>
Aaaa
<Myrl-saki>
Maybe I should try against master.
<Myrl-saki>
I literally have no system libraries lol
orivej has quit [Ping timeout: 272 seconds]
<Myrl-saki>
Ah, ftruncate is in unistd.h
haitlah has joined #nixos
<haitlah>
hey :)
<haitlah>
may I ask you guys if there is a way to move the /nix/store after having installed nixos ?
<haitlah>
I would like to put it in a ssd, tried to copy the actual store and setting a new mount in the configuration
<haitlah>
failed miserably
orivej has joined #nixos
<Myrl-saki>
I hate this already. ;~ ;
<srhb>
haitlah: Honestly, the absolute most resistant-to-small-errors way would seem to be reinstalling with the new mount in place. You can keep the rest of your system in place, and it should be fine :)
<Myrl-saki>
I think I know what's happening.
<Myrl-saki>
It's not detecting the ncurses library(no shit lmao)
<haitlah>
srhb: :'(. Any chance I can add a mount point and bind mount it ?
<Myrl-saki>
nativeBuildInputs run on host, right?
<srhb>
haitlah: I'm sure it's possible, I just don't feel like I know exactly what steps to take to recommend anything. What you already did would have been my first attempt. :)
<haitlah>
srhb: addind a /mnt/store neededForboot=true and then filesystem.'/nix/store' -> /mnt/store
<haitlah>
will try that :)
<haitlah>
thanks!
<Myrl-saki>
Oh, wow. Amazing. NCurses isn't being built either.
<{^_^}>
[nixpkgs] @matklad closed pull request #44897 → Make it possible to override VSCode version → https://git.io/fN54B
<Myrl-saki>
Wait, wtf, what's host and what's target?
tertle||eltret has quit [Quit: Connection closed for inactivity]
<Myrl-saki>
Ah, I know now. That's not on 18.03
<Myrl-saki>
Heck, the commit that added that is only 23 days old.
<srhb>
I guess it's best never to think about target if you can avoid it
<srhb>
(And instead think only of build and host)
<Myrl-saki>
Manually added the flags and it still doesn't work.
<Myrl-saki>
; ~ ;
hakujin2 has joined #nixos
Mateon1 has quit [Quit: Mateon1]
hakujin2 has quit [Ping timeout: 240 seconds]
<Myrl-saki>
I'm starting to hate this more and more. :/
<timokau[m]>
Why does it regularly happen that `nixos-unstable` is newer than `nixpkgs-unstable`? If `nixos-unstable` is ready to advance, shouldn't `nixpkgs-unstable` be rady as well?
orivej has quit [Ping timeout: 256 seconds]
<srhb>
timokau[m]: They look like completely different test sets to me.
<timokau[m]>
Does that make sense? I thought `pkgs-unstable` is supposed to be a subset of `unstable`. The hydra interface is still a bit of a mystery to me.
<Myrl-saki>
Well
<srhb>
timokau[m]: I think for that to be true every job in pkgs-unstable should _also_ be in nixos-unstable. There's no dependency from nixos-unstable to nixpkgs-unstable.
<srhb>
(... that I know of)
<timokau[m]>
srhb: Yes I thought that already was the case. Doesn't `nixos-unstable` also include all the packages?
<srhb>
timokau[m]: No, we don't block the channel based on arbitrary package failure.
<srhb>
timokau[m]: It does _include_ all the packages, in the sense that it's all based on the same nixpkgs expression.
<srhb>
Which becomes the channel iff the tests in trunk-combined pass.
<timokau[m]>
srhb: but nixpkgs-unstable isn't blocked on arbitrary package failures either, otherwise it could never advance right?
<srhb>
(and the eval is complete...)
<Myrl-saki>
How do I access arguments on overrideAttrs?
<srhb>
timokau[m]: RIght, only the tests you see in the above link :)
<srhb>
Did you mean override instead of overrideAttrs?
<Myrl-saki>
srhb: Like, the arguments, not the derivation.
orivej has joined #nixos
<timokau[m]>
srhb: I feel like those should also be in nixos-unstable if they are in pkgs-unstable. Otherwise I don't really see the point of pkgs-unstable.
<srhb>
Myrl-saki: It sounds like you mean override. :)
<srhb>
timokau[m]: I sympathize with that sentiment. :)
hakujin2 has joined #nixos
<Myrl-saki>
srhb: I kinda need both. I'm using `override` to give the correct argument, and `overrideAttrs` which gets the `override` argument.
<timokau[m]>
srhb: glad we agree :D
<srhb>
Myrl-saki: Then use both :)
<srhb>
Myrl-saki: compositionality and stuff.
<Myrl-saki>
srhb: I am using both, just that I can't get the override without using something like, `let/in override { inherit foo; }`
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/fN5jp
<jluttine>
is there something similar for haskell as there is for python: i can launch a python repl with some deps as `nix-shell -p pythonPackages.numpy --run python`
<sphaugh>
I'm trying to include the libwidevinecdm plugins with qutebrowser but I get "cannot open shared object file: Operation not permitted", anyone familiar with this?
<imalsogreg>
o/ Hello! I'm trying to do a fresh install of nixos with encrypted zfs. I'm following the nixos wiki here, but I'm running into an error when I try to use zpool to create a zroot. I'm working from a nixos-18.03 ISO on a usb stick... and I wonder if that is the wrong approach?
<imalsogreg>
Ummm, no error. It took me a while just now to delete the ext4 partition I had created and get back to last night's state when I originally saw the error.
<srhb>
Ah. :P
<imalsogreg>
Thanks! :P
<srhb>
By the way, if you choose to encrypt the entire zpool, you will not be able to add an unencrypted dataset afterwards
<srhb>
Which may not matter to you, but I find the latter option more flexible :)
ng0 has quit [Quit: Alexa, when is the end of world?]
<imalsogreg>
srhb: That layout looks good - does it rule out calling this Full Disk Encryption?
<leotaku>
Is it possible to run a "nixops build" that doesn't deploy the newly built stuff?
<srhb>
imalsogreg: Full Disk Encryption usually implies "everything but what is necessary to start the boot process is encrypted"
<srhb>
imalsogreg: I think this definitely covers it, except you can, if need be, add an unencrypted volume for a short while if for some reason you need to mount it without encryption powers.
<imalsogreg>
srhb: Oh nice, I was worried an unencrypted /boot would also invalidate us here.
<srhb>
(In short, the layout as pasted is FDE, but _COULD_ have a zpool/unencrypted dataset added at some later point)
<imalsogreg>
:thumbsup:
<srhb>
imalsogreg: No, definitely not. I have my first partition unencrypted VFAT.
<srhb>
Technically I think it's possible to have grub boot a disk that's entirely one encrypted zpool, but security-wise I see no difference (boot stuff is in efi/mbr vs boot stuff is in /boot)
<srhb>
And the latter is a simpler setup :)
<imalsogreg>
srhb: Great - I'll try to copy the layout in your gist.
<srhb>
imalsogreg: One thing, I've never gotten swap to work well with zfs, so if you need it I suggest setting aside space for it on a separate partition
<srhb>
OK, that's all the notes I can think of at this point :D
<imalsogreg>
srhb: Ok. I'm using a machine previously partitioned for normal nixos EFI install (separate partitions for /boot, /, and swap), so here I just deleted the / partition and created a new one.
<srhb>
imalsogreg: Makes perfect sense.
<srhb>
(For FDE you will have to set up encryption for your swap partition, but that can be done afterwards with no trouble)
<srhb>
(See. swapDevices.*.randomEncryption)
<imalsogreg>
srhb: Good point. Thanks for the notes! I am hopefully unstuck for a while.
<srhb>
Great, enjoy :)
<srhb>
leotaku: --build-only
hakujin2 has joined #nixos
hakujin2 has quit [Ping timeout: 256 seconds]
Mr_Keyser_Soze has joined #nixos
Izorkin has quit [Ping timeout: 240 seconds]
<infinisil>
leotaku: nixops deploy --build-only
<leotaku>
srhb, infinisil: thx
<infinisil>
Oh, I didn't see srhb's reply :')
jb55 has joined #nixos
rhomboidcactus has joined #nixos
ryanartecona has joined #nixos
nuncanada has quit [Ping timeout: 240 seconds]
Izorkin has joined #nixos
chpatrick has joined #nixos
<chpatrick>
I've got a weird problem with nixops
<chpatrick>
after deploying to an existing nixos system with targetHost, the deployment succeeds successfully but after that I can't ssh into the machine
<chpatrick>
there isn't even an sshd.service on it
Mr_Keyser_Soze has quit [Remote host closed the connection]
<chpatrick>
even if I do services.openssh.enable = true, which I don't think should be necessary
<chpatrick>
do you know what's going on?
leotaku has quit [Ping timeout: 256 seconds]
<infinisil>
chpatrick: Can you `nixops ssh` into it?
<Thra11>
I have a failed build of qtwebengine (from `nix build nixpkgs.qt5Full`, with keep-failed = true). How do I start a nix-shell with the correct build environment for that package?
<chpatrick>
infinisil: no, connection refused
<chpatrick>
but if I check it with a mouse and keyboard the ssh service doesn't seem to be installed
<{^_^}>
[nixpkgs] @viric opened pull request #44937 → coreutils: use openssl for faster *sum tools → https://git.io/fNdtP
viric has quit [Remote host closed the connection]
<teoma>
SuperCollider (scsynth) is crashing immediately for me on NixOS 18.03, but it works when I build it with genericBuild in a nix-shell and run it from the build directory. Any tips for debugging or preparing a bug report?
<srhb>
chpatrick: Are you sure services.openssh.enable isn't necessary on a none-backend-type? Not that that's the current problem, but I thought it was.
<srhb>
teoma: Sounds like you should get a good strance from the crash.
hakujin2 has joined #nixos
<chpatrick>
srhb: hmm I don't know if it's necessary but I'm pretty certain I tried it with it on. doesn't nixops always turn it on because it needs it to work?
<srhb>
chpatrick: I _think_ it's a product of the deployment.targetEnv and not just automatically set for all targetEnvs
<teoma>
srhb: So far I've gathered from strace that the crash happens while loading the plugin directory, which is /run/current-system/sw/lib/SuperCollider/plugins in both cases.
hakujin3 has joined #nixos
<srhb>
teoma: That sounds like a good start for a bug report. :)
hakujin2 has quit [Ping timeout: 256 seconds]
orivej has joined #nixos
hakujin4 has joined #nixos
<jD91mZM2>
Oh noes, nix filled up my root
<jD91mZM2>
Time to garbage collect I guess
<srhb>
jD91mZM2: Remember to post the amount freed for the weekly leaderboard.
trylist has joined #nixos
<teoma>
srhb: OK, thanks!
<teoma>
One thing I wanted to make sure of was whether the installed program and the one I built in nix-shell come from the same derivation. How would I check that?
<teoma>
symphorien: It's only installed with configuration.nix.
<symphorien>
mmh, then it's not a version mismatch between the executable and the plugins...
<teoma>
No, the executable and plugins are from the same directory under /nix/store. If there's any mismatch it would be with the nix-shell built version, which uses the installed plugins...
<teoma>
I have to go but I'll try nix-instantiate soon -- thank you!
hakujin has joined #nixos
<srhb>
symphorien: Any idea whether guarding mknod in the build-vm-with-bootloader failure is safe? if [ ! -e ... ?
<srhb>
I'm not sure why it started existing on its own :-)
hakujin4 has quit [Ping timeout: 276 seconds]
<symphorien>
srhb: honnestly I have no idea how all this works
ambro718 has quit [Quit: Konversation terminated!]
<symphorien>
well I let you know that it does not boot here :þ
<srhb>
Riiiight...
hakujin2 has quit [Ping timeout: 256 seconds]
<srhb>
Are we both attempting EFI, because I would understand how that can fail?
<symphorien>
hum there is no efi system partition in the partitionning iirc
<srhb>
Right. Need to try an mbr setup...
<srhb>
Or, well, non-efi.
hakujin has joined #nixos
<Dezgeg>
"# Create a /boot EFI partition with 40M and arbitrary but fixed GUIDs for reproducibility"
<Dezgeg>
is there not?
ryanartecona has joined #nixos
<srhb>
Truly..
hakujin3 has quit [Ping timeout: 240 seconds]
<symphorien>
hum. sorry
<srhb>
Doesn't appear to work either though. Hum.
<Myrl-saki>
ghc.mk:209: *** Can not build User's Guide when CrossCompiling or Stage1Only. Set BUILD_SPHINX_HTML=NO, BUILD_SPHINX_PDF=NO in your mk/build.mk file. See Note [No stage2 packages when CrossComp
<{^_^}>
[nixpkgs] @xeji pushed commit from @r-ryantm to master « redis: 4.0.10 -> 4.0.11 (#44684) »: https://git.io/fNdYG
<MichaelRaskin>
jD91mZM2: you need to shrink the filesystem; then you can use LVM2 to shrink the space allocated to one partition and give it to another one quite quickly
<MichaelRaskin>
I prefer to have some unallocated space in the LVM2 pool
haitlah has joined #nixos
hakujin has joined #nixos
rhomboidcactus has quit [Quit: Leaving]
haitlah has quit [Remote host closed the connection]
<MichaelRaskin>
jD91mZM2: LVM2 is based on device mapper
<MichaelRaskin>
The partitions are virtual
<LnL>
samueldr: just tested your installer thing, the boot menu looks good and if I select one of the hidpi options I get nicely readable text
<MichaelRaskin>
I.e. the actual partition is combined from a few slices of actual disk space
<jD91mZM2>
That's amazing
<LnL>
samueldr: anything else I should be looking at?
<chris|>
quick question: when building multi-output packages, how do I reference any non-binary output from another derivation? If I am reading the manual right, I assume this is not supported?
<MichaelRaskin>
And main partition table just sees a pool of space for LVM2
hakujin has quit [Ping timeout: 260 seconds]
<jD91mZM2>
Sounds like a great alternative to ZFS if I don't want to use that for whatever reason
<{^_^}>
[nixpkgs] @etu opened pull request #44940 → gitea: 1.4.3 -> 1.5.0 [backport to stable] → https://git.io/fNdYM
<jD91mZM2>
Can you install LVM without reformatting your entire disk?
<MichaelRaskin>
So you unmap a slice in the end of / and map it at the end of /home
<MichaelRaskin>
Or vice versa
hakujin has joined #nixos
<MichaelRaskin>
Well, you can feed it a partition.
<MichaelRaskin>
Then move some data
<MichaelRaskin>
Then feed it one more partition, which is now free
<samueldr>
LnL: I don't think so, that's 95% of the implementation, you could verify whether rEFInd and Firmware Setup do anything on your hardware
<samueldr>
I'm kinda expecting Firmware Setup will not work, and rEFInd is a wild card
<symphorien>
chris|: not sure I really understand, but to reference the dev output of foo, write foo.dev
reinzelmann has joined #nixos
sigmundv has quit [Ping timeout: 268 seconds]
<samueldr>
LnL: though, extremely happy that the hidpi options work (just to verify you booted to a readable linux VT, right?)
ng0 has joined #nixos
<thblt>
I just encountered a strange behavior in nixos-rebuild switch. Impurity, actually.
<MichaelRaskin>
Would be cool to just move it under the existing layout, but I am not sure a) there eare no header-in-beginning problems b) there is any tool that can be relied upon for doing this right
<chris|>
symphorien: just like that, I could have thought of that... thanks :)
<thblt>
I was playing with displayManagers, and tried services.displayManager.slim.enable = true; so far so good
<thblt>
BUT s/slim/xpra , nixos-rebuild switch complains that:
<jD91mZM2>
MichaelRaskin: So everything has to be copied over, am I reading that correctly?
<thblt>
The unique option `services.xserver.displayManager.job.execCmd' is defined multiple times, in `/nix/var/nix/prof
<thblt>
iles/per-user/root/channels/nixos/nixos/modules/services/x11/display-managers/xpra.nix' and `/nix/var/nix/profiles/per-
<MichaelRaskin>
jD91mZM2: unless you want to do some true magic by hand
<thblt>
After that, I need to explicitely services.displayManager.slim.enable = false; for nixos-rebuild switch to work
<jD91mZM2>
MichaelRaskin: Cool! Thanks for introducing me to this! Looks awesome :D
<thblt>
Anyone here has an idea what might be happening before i file a bug?
<MichaelRaskin>
Of course, in any case you are experimenting with partitions, and in any case at all you need to have a backup of your data, right?
<symphorien>
thblt: what if you disable explicitely slim ?
<LnL>
samueldr: yes, the bootlog itself is still tiny but once you get the promt it's pretty big with 720p
<thblt>
symphorien: that's what I do, and yes it works. But it's not exactly expected behavior.
<symphorien>
ah I had missed this line
<samueldr>
LnL: great, works like my tests did
<symphorien>
I guess the default is slim.enabled=true;
<symphorien>
which explains the behavior
<thblt>
symphorien: also notice that you only need slim explicitely disabled for nixos-rebuild to work once, the line can be safely removed after
<samueldr>
(it's AFAIUI and implified, whenever modesetting would start, that the resolution changes)
hakujin1 has joined #nixos
mayhewluke has quit [Ping timeout: 240 seconds]
<samueldr>
and simplified*
<symphorien>
hum strange
<thblt>
symphorien: that's not the case, the default is false
<jD91mZM2>
MichaelRaskin: I do have backups, but I don't want to imagine how long they'd take to restore
mayhewluke has joined #nixos
hakujin2 has joined #nixos
hakujin has quit [Ping timeout: 240 seconds]
<MichaelRaskin>
With minimal amount of care that should be on the safer side of data migration risk scale
<jD91mZM2>
Oh, and everything I care about is open source anyway :D
<jD91mZM2>
Well, except for the very secret name generator that generates a name in the style of jD91mZM2
<LnL>
samueldr: rEFInd also seems to work, firmware does not
hakujin1 has quit [Ping timeout: 256 seconds]
<samueldr>
good to confirm
<samueldr>
thanks!
<samueldr>
LnL++
<{^_^}>
LnL's karma got increased to 4
<Dedalo>
clever Why on the https://nixos.wiki/wiki/NixOS_on_ZFS page rpool/root is mounted with the option mountpoint=none, when the other are legacy instead?
hakujin3 has joined #nixos
ongy has joined #nixos
<ongy>
is there anything I need to look out for (nixos specific) when I want to have it on btrfs?
<srhb>
Dedalo: Because it's never meant to be mounted,
<ongy>
I have an existing setup, and just setting it into a btrfs mountpoint makes it easy to test-migrate without destroying the old system
hakujin2 has quit [Ping timeout: 256 seconds]
hakujin4 has joined #nixos
<srhb>
ongy: Not really.
<srhb>
Works out of the box insofar as you believe btrfs works. :)
<MichaelRaskin>
Don't let /nix/store completely fill BtrFS
<ongy>
didn't fail on me yet
semilattice has joined #nixos
<Dedalo>
srhb I still don't know if does make any sense to use l4z compression (for the whole disk or just the home directory). I mean, I'm on a laptop and my disk has plenty of space, but I read that compression may reduce the disk access but it uses more CPU (of course). So I don't know...
hakujin has joined #nixos
<srhb>
Dedalo: I have limited disk space on my laptop and I use it
<Dedalo>
srhb I guess I will go without compression. I have a 512 GB SSD with just NixOS, and when I download movies I use an external disk as storage.
jtojnar has joined #nixos
<Dedalo>
It's true that I have many docker containers, but 512 GB is a lot of space...
hakujin1 has joined #nixos
<ongy>
does nix use that much more than other systems? I have (had, I'm cleaning now) 3 full systems on my 256GB SSD
<srhb>
Dedalo: Again, you can always enable it later if you want to. or disable it.
<srhb>
ongy: Yes.
<Dedalo>
yep
<symphorien>
ongy: one of the features is "boot older generations of the system". you have to pay for that with disk space.
hakujin2 has joined #nixos
<MichaelRaskin>
A glibc update is basically «install a second copy of your current system, reboot to it, then you may choose to GC the old system»
hakujin has quit [Ping timeout: 240 seconds]
<srhb>
ongy: And even if you never kept more than one system around (that's not really possible...) you'd need around twice the space for every time gcc is bumped, or something else far down the dependency chain. This is not necessary in other distros.
<ongy>
ahh, so the store just grows over time, it's not cleaned from time to time?
<srhb>
ongy: Yes.
<MichaelRaskin>
You can enable periodical GC
<ongy>
can I clean it, e.g. specify I want only the current and one earlier (+ dependencies) of my packages?
<ongy>
cool
<srhb>
Yeah
<MichaelRaskin>
Anyway, double-space-use-during-upgrade is hard to work around
<symphorien>
MichaelRaskin │ Don't let /nix/store completely fill BtrFS << is there a smart way to prevent this ?
<srhb>
nix-collect-garbage --delete-older-than 14d is useful
<ongy>
yea, that's fine. Reminds me of GHC GC :D
<ongy>
can I not specify something closer to generation? days feels dirty
hakujin3 has joined #nixos
<MichaelRaskin>
symphorien: not sure; I use LVM2, so I just feed it a few extra GiBs when I forget to clean up in advance
<srhb>
ongy: Mind, keeping around a few generations is generally a really good idea. It's one of the immense superpowers of NixOS.
<symphorien>
I see
<srhb>
ongy: Yes, but I think you'd have to do it manually.
<srhb>
Essentially removing the symlink to the generations you want gone, then collecting garbage.
<symphorien>
and is btrfs that problematic when out of space ?
<srhb>
It shouldn't be hard to add that functionality to --delete-old though...
hakujin1 has quit [Ping timeout: 276 seconds]
<MichaelRaskin>
ongy: the basic idea of Nix is «convert sanity damage in pointlessly spent gigabytes of storage»
<ongy>
btrfs can be a bit... like a young brat
<MichaelRaskin>
symphorien: completely filled BtrFS might return ENOSPC on the unlink syscall
<MichaelRaskin>
Annoying
<ongy>
it generally works well for me, but I ran into stuff like "running out of space" while I still have ~1/2 of my disk free, because some allocation went haywire
<symphorien>
I have heard this but does it still happen ?
<MichaelRaskin>
(usually work-arounds exist een without LVM tricks, like rebalancing works almost always)
<Myrl-saki>
Um
hakujin4 has joined #nixos
<Myrl-saki>
How do I do something like `localSystem = "x86_64-linux"`
<MichaelRaskin>
Not sure — nowadays I usually hit the ENOSPC during a build, then add some space, so no problems with GC afterwards
<srhb>
symphorien: I never managed to have any idea how much space I really had on btrfs, but that's about a year ago, not sure if anything changed.
hakujin2 has quit [Ping timeout: 265 seconds]
<Myrl-saki>
Basically, I want to nix-instantiate something on my ARM machine, but set the cross builder to x86_64-linux.
<symphorien>
srhb: what do you mean ?
inquisitiv3 has joined #nixos
<Myrl-saki>
So I copy the drv to my x86 machine to build.
<MichaelRaskin>
BtrFS disk space usage reporting is intersting
<srhb>
symphorien: I mean, with all the tools to report disk usage at hand, I'd still arbitrarily run out and fail to allocate while all the tools reported I had space left.
<srhb>
Only remedy I found was to delete stuff.
<MichaelRaskin>
btrfs filesystem df is usuallly OK for me
<srhb>
It's definitely better than df which is straight up useless :P
<ongy>
srhb: the btrfs command has its own df. Anything else is wrong. That one is just rather specific
<srhb>
ongy: I know. :)
hakujin has joined #nixos
hakujin3 has quit [Ping timeout: 256 seconds]
<Myrl-saki>
Is `localSystem.system = "x86_64-linux";` enough?
<Myrl-saki>
symphorien: I'll consider that, but I don't know how to set the `builder`, on 18.03
hakujin4 has quit [Ping timeout: 276 seconds]
hakujin1 has joined #nixos
<inquisitiv3>
gchristensen: Thanks! I didn't know that the changelog was written during the development.
<gchristensen>
there are >10k commits per release, it'd be too hard to go back to find out :P
<inquisitiv3>
Haha, true
hakujin2 has joined #nixos
<inquisitiv3>
gchristensen: I remember your questions you asked me after finishing my first install of NixOS in a VM. You asked some questions how I thought of the UX installing the OS.
<gchristensen>
yea!
hakujin has quit [Ping timeout: 276 seconds]
<inquisitiv3>
I have given it some thought.
<gchristensen>
oh good!
hakujin1 has quit [Ping timeout: 256 seconds]
<inquisitiv3>
I think that making the automatically generated `configuration.nix` file preconfigured defaults that match the most common use cases for desktop would make the hardest part much easier
<inquisitiv3>
Why desktop you ask? Because the people wanting to install NixOS and needs most help is probably those that try to install it on a PC for desktop use.
hakujin3 has joined #nixos
<inquisitiv3>
People wanting to use it for servers need less hand helding.
<srhb>
I've been thinking that we should collect a few sample (say, gnome.nix, kde.nix) configurations and just throw them in /mnt/etc/nixos/examples and have a few commented-out lines in imports in /mnt/etc/nixos/configuration.nix
<srhb>
(But don't let me stop you)
<gchristensen>
both of these sound cool
<symphorien>
srhb: I'd say that for a newcomer, an import is really magic, and it's not easy to figure it's the same as copying the content of the imported file in configuration.nix
<srhb>
symphorien: Fair point. But I don't think it's much worse than having them all inlined in a big soup :)
<symphorien>
I remember being completely overwhelmed by nix syntax on the fist install
<inquisitiv3>
But having several different templates accessible at generation is also OK. E.g. getting a desktop template with the option `--desktop` and so on.
hakujin4 has joined #nixos
* srhb
nods
<symphorien>
so the big soup is fine, because it is somewhat understandable
<srhb>
symphorien: OK :) Good to know.
<samueldr>
^ switches for templates could make it easier indeed
hakujin2 has quit [Ping timeout: 265 seconds]
<symphorien>
true
<srhb>
I think switches are inferior to interactive questions, if we go down that route
<samueldr>
but nothing is interactive (yet!)
<srhb>
(choose 1 for kde, 2 for gnome, ...)
<srhb>
Right right!
<srhb>
But since we're spitballing :D
<samueldr>
just making sure, the important part here is to not mix-and-match the user interfaces :)
* srhb
nods
<inquisitiv3>
The optimal would have a GUI, but I'm thinking of simple changes that would have a big impact.
<gchristensen>
interactive is cool, b/c as long as the no stdin case emits what it does now, we could easily make it nicely interactive
hakujin3 has quit [Ping timeout: 240 seconds]
<dhess>
Does `imports` merge attrsets?
hakujin has joined #nixos
<gchristensen>
yeah they merge nicely
<dhess>
cooool thanks
<srhb>
dhess: It merges types based on the types' merging rule :)
<dhess>
srhb: right I guess that's what I meant more specifically
<dhess>
thank you
hakujin4 has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @flokli opened pull request #44943 → jameica: init at 2.8.1 → https://git.io/fNd3X
hakujin1 has joined #nixos
<inquisitiv3>
I'm of the same mind as symphorien. The hardest part of installing NixOS is `configuration.nix` because of Nix expressions. It's the most unique part of Nix, and I think it's hard on new users to expose them directly (even before they have a working installation).
<gchristensen>
the risk of an interactive thing is keeping it small and simple for newbies, and we still have the "leading the user to water" problem
reinzelmann has quit [Quit: Leaving]
<thblt>
Are we trying to drown users or am I missing the meaning of the phrase?
<srhb>
I'm still not quite sure whether that means "we should have a check early that ensures that people consent to working with configuration.nix before they regret everything" :P
<srhb>
Or "we should ensure there's a way to get the know-how to not regret"
<srhb>
(I'm guessing the latter!)
hakujin3 has joined #nixos
<MichaelRaskin>
«OK, I don't have much to GC»
<MichaelRaskin>
6183 store paths deleted, 31415.26 MiB freed
<gchristensen>
haha
hakujin1 has quit [Ping timeout: 265 seconds]
<srhb>
jD91mZM2: You have been beaten.
<gchristensen>
srhb: are you asking w.r.t. leading to water, or some other part?
<srhb>
gchristensen: Leading to water, yes :)
<srhb>
Or more correctly "force to drink"
<gchristensen>
ah
<symphorien>
I think we should ensure we have a base of valid examples to copy paste from. Like the manual, the wiki, templates or whatever.
hakujin4 has joined #nixos
<gchristensen>
I'm not saying we should get consent about a configuration.nix, but provide them obvious tools and path-ways to the docs they need to do what they want
<srhb>
gchristensen: Thanks, that makes sense :)
<MichaelRaskin>
Frankly, if initial generation of configuration.nix is one-time service, maybe the installation should include at least some editing of configuration.nix
<gchristensen>
right now the best you get is:
<MichaelRaskin>
Informed consent via revealed preferences…
<gchristensen>
# Edit this configuration file to define what should be installed on
<gchristensen>
# your system. Help is available in the configuration.nix(5) man page
hakujin2 has quit [Ping timeout: 256 seconds]
<Dedalo>
srhb I'm following this guide: https://nixos.wiki/wiki/NixOS_on_ZFS. When I try to mount using `mount -t zfs rpool/root/nixos` I get the following error: filesystem 'rpool/root/nixos' cannot be mounted, unable to open the dataset. Any idea?
<srhb>
Dedalo: What does zfs list say?
<srhb>
Dedalo: (Did you give it a target to mount on, by the way? eg. /mnt at the end?)
<gchristensen>
like you can't even copy-paste a `man` command, you have to know how to interpret "configuration.nix(5) man page" which tbh I don't know how to do, I just type 'man configuration.nix' and hope it goes to the right place
<inquisitiv3>
MichaelRaskin: Users will have to edit the file sooner or later (installing software globally, enabling services, etc). I don't think news users should be forced to edit the file before getting a working system.
<Dedalo>
It says no database available. But I have created them using `zfs create`...
<jD91mZM2>
srhb: We must bow before our new leader
hakujin3 has quit [Ping timeout: 256 seconds]
<Dedalo>
I have even tried to export them with zfs export
<srhb>
jD91mZM2: :D
<srhb>
Dedalo: Wait, if you exported them, that's probably why
<jD91mZM2>
srhb: Oh and almost all space was filled back up when I ran nixos-rebuild :'(
<srhb>
Or exported the zpool rather
<MichaelRaskin>
inquisitiv3: if they will be forced eventually, should we try to hide that fact?
<srhb>
jD91mZM2: hah
<Dedalo>
oh...
<Dedalo>
srhb ops, should I create them again?
leotaku_ has joined #nixos
<inquisitiv3>
MichaelRaskin: At least I feel much more comfortable reading docs when I got a working system and not when I'm trying to install it.
<jD91mZM2>
srhb: I have no option but to install ZFS or LVM or SOMETHING soon. Oh boy. Erasing all my data, even if I keep backups, will be painful
<jD91mZM2>
At least I won't have to worry about reinstalling the actual OS. Thank god for Nix
<srhb>
Dedalo: I think you can just zpool import
<gchristensen>
I think having the "wizard" take you to the point of "here's <kde> with your user and password setup..." is nice
hakujin has joined #nixos
<Dedalo>
well I have overwritten the zpool and I have recreated the datasets
<symphorien>
An interactive prompt to ask you neutrally if you want pulseaudio :p
<gchristensen>
probably not
<Dedalo>
srhb and now the mount works, thanks!
<MichaelRaskin>
inquisitiv3: installation medium comes with a system that should be close enough for reading docs…
<MichaelRaskin>
In our browser-centric world, why even bother installing something anywya
haitlah has quit [Remote host closed the connection]
hakujin4 has quit [Ping timeout: 265 seconds]
<inquisitiv3>
MichaelRaskin: But can I take a break for documentation reading and do some work to come back to the docs? You don't learn Nix in one go, you do it gradually with experience over time.
<jtojnar>
gah, why does git repo get horribly corrupted on power failure
<MichaelRaskin>
Dunno, read the three manuals in one go back in the day
<MichaelRaskin>
(they were indeed smaller back then, but not so much smaller)
hakujin1 has joined #nixos
<srhb>
Speaking of audio, has anyone experienced on recent unstable that they're unstable to connect to pulseaudio unless they kill the currently running server and restart it?
<{^_^}>
#44876 (by eadwu, 2 days ago, open): display-managers: rely on socket to activate pulseaudio
hakujin has quit [Ping timeout: 256 seconds]
<srhb>
That fix seems weird...
<inquisitiv3>
MichaelRaskin: Not all people have time to do that. They got other things that take time and doesn't have "infinite" time to dedicate to NixOS alone.
hakujin2 has joined #nixos
<MichaelRaskin>
Not sure installing and configuring NixOS can ever take less than half a day
<inquisitiv3>
Ideally editing `configuration.nix` should be like using Ansible or Salt. Everyone doesn't need to do that, but if you do it's a very powerfull tool to have.
<jtojnar>
srhb: why?
<MichaelRaskin>
That might be nice, but we are not even close to that.
<inquisitiv3>
MichaelRaskin: Half a day is much less than it took me. And it's still to much time to dedicate to install an OS for regular users.
<inquisitiv3>
Mich
hakujin1 has quit [Ping timeout: 240 seconds]
<srhb>
jtojnar: I'd have thought that the socket activation would do the exact same thing that the removed version does?
<srhb>
jtojnar: Well, I guess I'll just test.
hakujin3 has joined #nixos
<MichaelRaskin>
Right now regular users would still be overwhelmed with Nix
aanderse has quit [Ping timeout: 240 seconds]
<inquisitiv3>
MichaelRaskin: Yes, it's a long way to go. But if you want to force every new NixOS to read several hundred pages of technical documentation before being able to install NixOS we'll never come there.
jD91mZM2 has quit [Quit: WeeChat 2.0]
hakujin2 has quit [Ping timeout: 260 seconds]
chpatrick has quit [Quit: Connection closed for inactivity]
<inquisitiv3>
But even if we changed the templates people would have to edit the file manually. Options like `hostname` would have to be set manually in the file.
<MichaelRaskin>
Ah. I thought the plan was to ask that interactively
hakujin4 has joined #nixos
freeman42x]NixOS has joined #nixos
<inquisitiv3>
MichaelRaskin: My suggestion was for a simple change that is easily done now.
<inquisitiv3>
But has a big impact for new users coming to NixOS.
hakujin3 has quit [Ping timeout: 240 seconds]
garbas has joined #nixos
<srhb>
Guh, is there a fast way to git commit --amend on a PR without going through adding the other user's remote?
Dedalo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
hakujin has joined #nixos
<MichaelRaskin>
I have a feeling that we need to generate _two_ configuration.nix templates — last attempt to include a lot of easy-to-fill pieces in the template got objections about overbloating the generated config
<dhess>
so if x = pkgs.lib.mkMerge [ a b c ]; how do I get the "resolved" x, i.e., the result of the merge ?
<dhess>
x seems to be the thunk
<sphalerite>
srhb: also `git fetch origin refs/pulls/$number/head && git checkout -b pr-name FETCH_HEAD` to checkout the PR is nice. Or do you have a better way to share? ;)
<srhb>
sphalerite: I think that's what I have hidden in some alias somewhere that I forget how works :D
<srhb>
alanz: I am able to build it on both nixos-unstable and nixos-18.03
shabius has quit [Quit: Leaving]
<alanz>
ok, its not urgent, I have nix installed on my debian box, was just doing an update
<alanz>
and was surprised
<infinisil>
dhess: This depends on the type of the option you're assigning it to, and therefore can't be evaluated just like that
<srhb>
alanz: Looks likely that nixpkgs-unstable has a broken version of it :)
<dhess>
inif
<dhess>
infinisil: ok thanks
Fare has quit [Ping timeout: 240 seconds]
blahdodo has joined #nixos
<srhb>
alanz: It will be alright once 94815133604a0ff3b8831a6322c3b0138c2f0d96 lands in nixpkgs-unstable: cabal2nix: fix build by overriding hpack to 0.29.6
<alanz>
srhb, I will try again tomorrow or the next day
<alanz>
ok
shabius has joined #nixos
<bendlas[m]>
hey, can somebody tell me, why we have `libchop`, but it doesn't have `chop-backup` in its `bin` folder? cc rnhmjoj viric
sigmundv has joined #nixos
<{^_^}>
[nixpkgs] @srhb merged pull request #44876 → display-managers: rely on socket to activate pulseaudio → https://git.io/fN7rw
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
hakujin3 has quit [Ping timeout: 256 seconds]
noam has joined #nixos
<dhess>
domenkozar: you around?
KennosiS has joined #nixos
simukis has quit [Quit: simukis]
gchristensen has joined #nixos
leotaku has joined #nixos
<leotaku>
Is there any reason why a nixops deployment would try to write multiple kernels to /boot?
<Dedalo>
clever in the guide to setup ZFS (which I did), there is a step to modify the `hardware-configuration.nix`. When I have generated this file, I have seen there is not the boot file system entry. In the example it should be EXT4, my the boot partition that I have created is FAT32 because of UEFI. It's a EFI System. How can I configure the file system in the configuration file?
ng0 has joined #nixos
<clever>
Dedalo: /boot will be missing if you ran nixos-generate-config before mounting it
<clever>
Dedalo: mount /boot and re-run generate config
<clever>
the tool just scans what is currently mounted, and creates config to keep those mounted
<dhess>
I was trying to accomplish something similar with that mkMerge stuff but quickly realized that I really needed to call NixOps's eval-machine-info.nix with the right args, and that wasn't going to work.
<clever>
dhess: this branch of nixops has a new sub-command, dump-debug-expression
<clever>
dhess: that generates a nix file that calls eval-machine-info.nix with mostly the right args
<dhess>
clever: ohhhh haha I was considering writing the same thing!
<Dedalo>
clever I have mounted with `mount /dev/disk/by-label/boot /mnt/boot`, but now I get an error when I run the generator: ERROR: not a subvolume: /mnt/boot
<clever>
Dedalo: what does "mount | grep boot" say?
<Dedalo>
I have twice: /dev/nvme0n1p1 on /mnt/boot type vfat (etc.)
<clever>
Dedalo: umount one of the duplicates, then try nixos-generate-config again
<Dedalo>
clever here it is!
<Dedalo>
clever strange things are happening. I'm sure I had mounted the boot before, in fact it was there twice
<Dedalo>
dunno why it didn't pickup the first time
<dhess>
clever: I need something like dump-debug-expression and/or infinisil's little hack to build my config in Hydra. I started with your stuff in your nixos-config repo, but my NixOps expressions are more complicated than yours -- logical machine definitions, physical machine definitions, deployment directives, network defaults, etc.
<Dedalo>
clever I can see that now I have: /boot, /, /home, the swap device which is mounted using the uuid
<dhess>
so then I thought I'd just call eval-machine-info.nix, and then needed to figure out the right args.
<clever>
dhess: yeah, you should be able to tweak the nix that sub-command outputs, and then load it from hydra
<dhess>
cool
<leotaku>
Anyone got an idea? It only seems to happen if I use clevers quemu.nix for aarch64 compilation.
<clever>
leotaku: nixops will write one kernel to /boot for every kernel in /nix/var/nix/profiles/system*
silver has quit [Ping timeout: 244 seconds]
__monty__ has quit [Quit: leaving]
<Dedalo>
clever how can I enable WiFi when the system starts?
<clever>
Dedalo: did you set networking.wireless.enable = true; ?
<leotaku>
Dedalo: If I were you I'd just enable networkmanager.
<Dedalo>
clever I did, right now. Is there something else that I need to do? Or I can just launch nixos install?
<clever>
Dedalo: is the install done or not?
<Dedalo>
clever well I have done all the step in the guide, I'm at point 13 of the guide, "Do the installation"
<clever>
Dedalo: ah, then yeah, nixos-install
<clever>
Dedalo: and also youll want to copy /etc/wpa_supplicant.conf over
<Dedalo>
he wants networking.hostId to be set
<Dedalo>
that's my hostname, right?
<clever>
no
<clever>
read the hostid section in the man page of configuration.nix
mayhewluke has quit [Ping timeout: 272 seconds]
mayhewluke has joined #nixos
leotaku has quit [Read error: Connection reset by peer]
<Dedalo>
clever ok, I have generated it. Do I put it in the configuration.nix?
Lisanna has quit [Remote host closed the connection]
<Dedalo>
clever the install failed to build the keymap driver. I have changed the i18n section of the config file like follows: `consoleKeyMap = "en"` and `defaultLocale = "en_GB.UTF8"`.
<clever>
Dedalo: you could just comment those out, and get them to work later
<Dedalo>
ok
<Dedalo>
I did, let's see if it works
bennofs has joined #nixos
<Dedalo>
clever now it says the file system /boot is not FAT EFI, I'm going to check that.
<{^_^}>
[nixpkgs] @xeji pushed commit from @etu to release-18.03 « gitea: 1.4.3 -> 1.5.0 (#44940) »: https://git.io/fNd8v
slack1256 has quit [Ping timeout: 272 seconds]
nonfreeblob has joined #nixos
<Dedalo>
clever the volume for some reason that I can't understand was not mounted, but you know that I have mounted it just before. I have done a `mount /dev/nvme0n1p1 /mnt/boot`, then nixos-install worked
<clever>
Dedalo: oh, maybe you mounted /mnt/boot, then /mnt, which shadowed it
<clever>
check the order of things in `mount`
<Dedalo>
clever before I have mounted using `/dev/disk/by-label/boot`, that's the problem I think
<clever>
Dedalo: can you pastebin all of `mount`?
<Dedalo>
clever I'm in, I have done the reboot and it works. Thank you very much!
<clever>
ah, nvm then, should be all good
<Dedalo>
sure I can
<Dedalo>
wifi is not working
<clever>
iwconfig
<{^_^}>
[nixpkgs] @markuskowa opened pull request #44954 → qradiolink: init at 0.5.0 → https://git.io/fNd8g
<Dedalo>
I can see the wlp2s0
<Dedalo>
in the configuration.nix wireless is enabled
<clever>
wpa_cli -i wlp2s0
<clever>
status
<Dedalo>
I guess I didn't copy the wpa_supplicant, that's why
<Dedalo>
I forgot
<clever>
then youll need to run wpa_passphrase again to create a new one
<clever>
and restart the service
<nonfreeblob>
Guys, I'm planning to recreate my github account (with new user name). If you don't mind, I'll create PR to remove myself from the list of maintainers. And will create PR from a new account to add my new information.
<clever>
nonfreeblob: you should be able to just rename your account in the github settings
<Dedalo>
clever do I need to add `ctrl_interface=DIR=/var/run/wpa_supplicant GROUP=wheel` this as well, right?
<clever>
yeah, that is needed to allow wpa_cli
<clever>
and it lets members of the wheel group connect, so you dont need sudo for that
<clever>
wpa_gui also uses it
<nonfreeblob>
Yes I know. But I do it for reasons of privacy.
<Dedalo>
clever ok it works but I reboot to see if it still works
<Dedalo>
yes it does
<clever>
Dedalo: let me grab you some other useful things
<Dedalo>
clever thank you
<clever>
Dedalo: you can set priority=10 inside a network block, so when 2 are available, it will prefer one
<nonfreeblob>
clever: I cannot overwrite commits in other repos that have merged my PR's.
<clever>
Dedalo: for non-encrypted networks, its just a network block with only an ssid="foo" and key_mgmt=NONE
<clever>
Dedalo: and there is something else you can probably just google, to just allow it to auto-connect to any open wifi
<clever>
wpa_cli and wpa_gui can be used to control it more
<Dedalo>
I guess wpi_gui requires X11
<clever>
yeah
<Dedalo>
I need to install Wayladn and Sway
<Dedalo>
Wayland
<Dedalo>
clever I have some more generic question for you: I need to configure the locale because I have an English keyboard. If I fuck up, am I able to rollback and use the previous conf? I guess that's the aim of this distribution, right? And, second question, what I have to do to create a system that I can install, eventually on another laptop, let's say in 2 years because I change this one? I have made a readme with all the step
<Dedalo>
till here, but the goal is to install on another machine in a few minutes. I read that I can eventually export the zpool or reimport elsewere.
<clever>
Dedalo: as long as the filesystems are not corrupt, you can always rollback from the bootloader, and just select an olver version there
<clever>
Dedalo: and if the keyboard is still in a semi-usable state (or ssh works), you can use `nixos-rebuild --rollback` to undo the last change
<clever>
you can also just undo the changes and switch again
<Dedalo>
great
goibhniu has quit [Ping timeout: 260 seconds]
Lisanna has joined #nixos
Ariakenom has quit [Quit: Leaving]
jperras has joined #nixos
hakujin3 has joined #nixos
<Lisanna>
Hey guys, today I'm going to be working on an extension to hydra, and need a basic VPS setup or something so I can test my patch. Can't use my normal server resources since I want to be able to upstream the patch. Any recommendations on a cheap VPS solution I can use to quickly spin up a small NixOS + hydra instance?
<{^_^}>
[nixpkgs] @nonfreeblob opened pull request #44955 → Delete outdated info → https://git.io/fNd4a
<gchristensen>
AWS is pretty cheap for small stuff, I understand nixos can be had on digitalocean
hakujin3 has quit [Ping timeout: 265 seconds]
<Lisanna>
gchristensen AWS = EC2?
<gchristensen>
ya
<Lisanna>
okay, I'll give that a go
<Dedalo>
To install sshd should I just type: `nix-env -i sshd`?
<Lisanna>
oh, and EC2 has a lot of nixops support, so that's great
<{^_^}>
[nixpkgs] @puffnfresh opened pull request #44956 → blastem: init at 0.5.1 → https://git.io/fNd4o
<Dedalo>
Lisanna Linode is another alternative
* emily
briefly interpreted "Lisanna Linode" as a name
<elvishjerricco>
Is there any way to use `boot.initrd.network` without dropping NetworkManager?
<vodurden>
Howdy Nix friends, I just had a quick question about the PR process: Yesterday I submitted a few PRs to nixpkgs and got some feedback on one of them which I've resolved. Overall the process has been pretty smooth but I'm wondering if I need to proactivly reach out to people to keep the feedback coming, or is the process to wait until someone takes a look at them?
NegaNexus has joined #nixos
<samueldr>
vodurden: welcome back o/
<elvishjerricco>
Alternatively: What can I use instead of NetworkManager? :P I'm extremely un-knowledgeable with linux networking, so I prefer something like nmcli that just does all the complicated stuff for me.
<samueldr>
vodurden: ideally, the process would be first in first out, but it isn't exactly, so I can't just say "there are 777 other PRs", but there are certainly a good deal of PRs that are, too, ready to be checked
<samueldr>
(I basically checked yours yesterday only because of opportunistic reasons; you asked, I checked)
<samueldr>
I mean, you asked about the PR process*
<Dedalo>
clever the i18n.consoleKeyMap wasn't working because it's not `en`, neither `gb`, but instead is `uk`.
<nonfreeblob>
samuelrd: I've updated my PR
<samueldr>
vodurden: I think a kind of rule of thumb is that if there were no reviews, no comments, nothing for a week, it's probably good to ask
<samueldr>
especially for non-trivial ones
<nonfreeblob>
samueldr: sorry for spelling :)
<samueldr>
(in your case, vodurden, they're non-trivial ones :))
<samueldr>
nonfreeblob: no worries
<samueldr>
nonfreeblob: once the checkmar shows up I'll close (that one is a trivial one)
<nonfreeblob>
samueldr: thanks!
<samueldr>
(there's no meaning in waiting until someone else sees it :/)
<vodurden>
samueldr: No worries, I'm happy to wait I just wanted to make sure there wasn't extra input needed from me :)
<samueldr>
vodurden: you did everything right, good PR description and everything
<samueldr>
and I hope you continue doing fine contributions :)
<vodurden>
samueldr: Thanks! Yeah I'm happy to keep upstreaming changes I make if they're useful, there's a few other things I want to fix since I'm using NixOS the OS for my main computer. I appreciate all the help getting this sorted out
<samueldr>
do ask :)
<vodurden>
I have a feeling I'll be bugging people in this channel about how the fhs chroot stuff works since I want to get playonlinux downloaded wine binaries working.
<samueldr>
did you take a look at the steam stuff?
<samueldr>
and IIRC there were other people looking at PoL, or was it lutris
<vodurden>
Yeah I did, I have a wip branch trying to package playonlinux in a similar manner. But I haven't got it working yet
<vodurden>
I saw a bunch of work on lutris. Nothing on PoL last I checked
<vodurden>
lutris would be nice too but I'm using PoL right now so I figured it'd be a good place to start
<samueldr>
yeah, it must have been Lutris, no POL in the IRC logs