<__monty__>
Ok, good to know. Probably more of a manual process because requirements.txt can be incomplete.
catch22 has joined #nixos
<dhess>
yeah I suspect that is one reason. And one reason why there isn't a clear-cut "Python 2 Nix" process.
<dhess>
but still... what a mess!
<dhess>
Unless, like I said, I'm missing something and there is some tool that makes this really easy and I just can't find it.
<dhess>
It's weird given how hugely popular Python is that their packaging is so broken.
<dhess>
after all these years.
<__monty__>
How would a tool find out the C libraries without specifying them manually though? Code analysis?
<dhess>
Even Cabal and Hackage are better, and that is saying something!
<dhess>
__monty__: I dunno. How does pip do it?
<dhess>
I guess pip just fails unless you've already installed the C bits? I can't remember anymore, it's been years since I did any non-trivial Python development.
Wizek_ has quit [(Ping timeout: 260 seconds)]
joehh has joined #nixos
Lisanna has quit [(Ping timeout: 260 seconds)]
<manveru>
there was a presentation at nixcon about pipy2nix and why packaging is so shitty :P
<dhess>
manveru: hehe. OK, thanks. Sounds like there's no magic bullet.
<manveru>
can't say much about it myself... i don't do python
<dhess>
wow, just a few days ago. Timely
<bgamari>
clever, I'll admit I'm still a bit puzzled by python
<dhess>
manveru: me neither, but so many useful tools use it, so I end up having to package them from time to time.
<bgamari>
clever, If I do run nix-shell, should I expect to see PYTHONPATH changed?
<dhess>
Just long enough between each that I apparently forget what a nightmare it is.
<__monty__>
Could be a killer feature for nix maybe, becoming the defacto python package manager.
<dhess>
__monty__: As far as I'm concerned, Nix's killer feature is that it's the defacto *everything* package manager. :)
<__monty__>
Yes but let's face it, I think most users come from haskell? So the userbase is part of a niche. If the same fraction of people working in python started using it the network effect would be insane.
alhariel has joined #nixos
<dhess>
you're right, it's not really "defacto." Just "best."
<MP2E>
sounds like a reasonable assumption, I wouldn't have found NixOS if not for Haskell
asuryawanshi has joined #nixos
primeos has quit [(Ping timeout: 240 seconds)]
<dhess>
I don't think I would have, either. In fact, I was quite surprised to find that pretty much none of the Nix* infrastructure is written in Haskell. :)
<__monty__>
Yes me too, outraged in fact.
<__monty__>
: )
<dhess>
It is a bit of a turnoff, yes :)
ylwghst has quit [(Remote host closed the connection)]
asuryawanshi has quit [(Ping timeout: 248 seconds)]
tempeh has joined #nixos
<tempeh>
so, i'm trying to build a C++ program with eigen. I'm using "nix-shell -p gcc eigen3_3", but I need the location of the eigen package. Is that provided somewhere? I see that it's in the buildInputs variable, is that what I should use?
johnw has quit [(Ping timeout: 246 seconds)]
johnw has joined #nixos
darlan has joined #nixos
<infinisil>
tempeh: In your nix-shell: `env | grep NIX`
<earldouglas>
fetchsvn is failing when it detects a URL redirect cycle. I need to fetch /foo/bar/ with a trailing slash, but it looks like nix is dropping that trailing slash. Any way to force it?
MP2E has quit [(Remote host closed the connection)]
erasmas has quit [(Quit: leaving)]
aristid has quit [(Ping timeout: 240 seconds)]
darlan has quit [(Client Quit)]
acarrico has quit [(Ping timeout: 252 seconds)]
<infinisil>
earldouglas: wild guess: If you didn't quote the url, try quoting it (so it's a string)
<earldouglas>
Yep, tried it as a string.
fragamus has joined #nixos
<infinisil>
then it's probably a limitation of the current fetchsvn implementation, I'd go check the source
<tempeh>
infinisil: doesn't seem to, but i don't know what i'm doing
Sonarpulse has joined #nixos
<tempeh>
man, I am having a lot of trouble with the basics. So I want to use a default.nix to call a Makefile to compile a C++ program that links against eigen. So I'm writing default.nix, and list gcc and eigen as inputs. I call that with nix-build '<nixpkgs>' in the directory?
mekeor has quit [(Ping timeout: 250 seconds)]
jb55 has quit [(Ping timeout: 248 seconds)]
<tempeh>
That's giving me a long winded error.. "error: Please be informed that this pseudo-package is not the only part of
<tempeh>
Nixpkgs that fails to evaluate.y
markus1199 has joined #nixos
ylwghst has joined #nixos
<ylwghst>
how should i edit /etc/systemd/system/systemd-suspend.service ?
markus1189 has quit [(Ping timeout: 260 seconds)]
iqubic has joined #nixos
<clever>
bgamari: if python is in the buildInputs, then the setup hook for python should be sourced
<ylwghst>
only systemd.services.<name>.unitConfig no serviceConfig :(
<iqubic>
Hello guys
<bgamari>
hmm
<clever>
bgamari: and i believe that will then add all other buildInputs to the PYTHONPATH
ryanartecona has joined #nixos
ryanartecona has quit [(Read error: Connection reset by peer)]
ryanarte_ has joined #nixos
<bgamari>
clever, thanks! that is a useful hint
fragamus has joined #nixos
<iqubic>
I just got my new laptop and I need to get nixos onto the new laptop. Is
<iqubic>
What filesystem type should I use.
Guest10 has quit [(Ping timeout: 268 seconds)]
<ylwghst>
iqubic: i recommend ext4 for root
<ylwghst>
iqubic: /boot should be vfat
<iqubic>
I am dual booting with windows10. /boot is formatted correctly.
MP2E has joined #nixos
<iqubic>
Why do you think that ext4 is the best for root?
<iqubic>
Isn't ZFS better?
ryanarte_ has quit [(Quit: ryanarte_)]
<ylwghst>
iqubic: depends how you will the machine use
<iqubic>
What uses are zfs better for?
primeos has joined #nixos
<__monty__>
Enterprise mostly.
<iqubic>
I don't enterprise.
<ylwghst>
iqubic: i will use it on machine which will handle huge amounts of data
<iqubic>
My machine is just for personal use.
<iqubic>
Nothing too taxing will be done on the machine.
<ylwghst>
i just prefer ext4 for my desktop workstation
dhess has quit [(Ping timeout: 258 seconds)]
<iqubic>
I think I'll do that then
fragamus has quit [(Ping timeout: 260 seconds)]
<__monty__>
iqubic: Ext4 is tried and true, xfs is newer and interesting but not revolutionary, btrfs is most like zfs but easier to run if you don't care as much about how finished the product is ; )
<ylwghst>
brb
ylwghst has quit [(Remote host closed the connection)]
<adisbladis>
I'd go with xfs over ext4
asuryawanshi has joined #nixos
<iqubic>
xfs, and not zfs?
<adisbladis>
Imho: ZFS > btrfs > xfs > ext4
<iqubic>
Do you recommend xfs for a daily driver?
ylwghst has joined #nixos
<adisbladis>
iqubic: If you want a more "traditional" file system xfs is the best of that bunch imho.
<iqubic>
Cool.
<adisbladis>
If you want things like volume management, full data checksumming etc you have 2 choices: zfs and btrfs
<adisbladis>
And out of those zfs is easily the best
<ylwghst>
i've xfs a long time too
<ylwghst>
used*
<__monty__>
I use it for a storage disk, no problems yet. I use btrfs daily on my laptop, no issues either.
<iqubic>
Which do you recommend __monty__?
<adisbladis>
iqubic: xfs is the current default in RHEL
<adisbladis>
You can never go wrong with zfs <3
srdqty has joined #nixos
<ylwghst>
xfs its bit old date range only to 2038, ext4 is kind of modern fs
<iqubic>
Too many options
<MP2E>
been using btrfs for 6 months on a laptop I use for work, had no issues ever since. I only picked btrfs because i was under the assumption that "in kernel == easier install" but on nixos that isn't necessarily the case
<MP2E>
when i reinstall at some point in the future i'll move over to zfs
<__monty__>
xfs is more modern than ext4 afaik.
asuryawanshi has quit [(Ping timeout: 248 seconds)]
<iqubic>
I am dual booting and have 100 GB for my FS + swap partitions
ryantrinkle has quit [(Ping timeout: 250 seconds)]
<iqubic>
Is that enough for running nixos?
<ylwghst>
xfs can handle up to 8EiB
<ylwghst>
ext4 only 1EiB
<__monty__>
I'd say play around with btrfs or zfs, they're just snazzier : )
<ylwghst>
but ext4 has date range up to 2446
<iqubic>
Which is better between zfs and btrfs?
srdqty has quit [(Client Quit)]
<__monty__>
100 GB isn't much to backup and reformat to xfs/ext4 at a later date.
<MP2E>
yeah that should be fine. I use a hybrid setup with / mounted on a 120gb SSD and my /home mounted on a 2TB HDD. I've *sometimes* maxed out my nix store when I haven't gc'd for a few months
srdqty has joined #nixos
<MP2E>
but it's rare and I rebuild from the latest git master a lot
<MP2E>
so it was bound to happen at some point heh
<__monty__>
iqubic: zfs has always seemed more complicated than necessary to me, btrfs still has a reputation for being somewhat unstable.
<iqubic>
I think I want to use zfs then
primeos has quit [(Ping timeout: 268 seconds)]
<iqubic>
In a while I will need help getting it running.
<ylwghst>
btw
tempeh has quit [(Ping timeout: 248 seconds)]
<iqubic>
Yes?
<__monty__>
ylwghst: Why exactly is the date range significant? I'm not well-versed enough in filesystems to judge?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] aszlig pushed 1 new commit to master: https://git.io/vFuEh
<bgamari>
Is it just me or is the line editing in the new `nix repl` a step backwards from the old `nix-repl`?
<bgamari>
tab completion seems to have lost the ability to list all completions
dejanr has joined #nixos
bts- has quit [(Ping timeout: 250 seconds)]
noobly has joined #nixos
bts- has joined #nixos
acarrico has joined #nixos
tempeh has joined #nixos
cwre has joined #nixos
orivej has joined #nixos
noobly has quit [(Quit: Page closed)]
cwre has quit [(Client Quit)]
Supersonic112 has quit [(Disconnected by services)]
Supersonic112_ has joined #nixos
cwre has joined #nixos
Supersonic112_ is now known as Supersonic112
radvendii has joined #nixos
w-spc-gir has joined #nixos
ylwghst_ has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] jerith666 opened pull request #31415: pythonPackages.titlecase: init at 0.12.0 (master...python-titlecase) https://git.io/vFu2e
NixOS_GitHub has left #nixos []
<clever>
bgamari: i have heard the same thing
ylwghst has quit [(Ping timeout: 240 seconds)]
asuryawanshi has joined #nixos
lambdamu has joined #nixos
asuryawanshi has quit [(Ping timeout: 268 seconds)]
lambdamu_ has quit [(Ping timeout: 268 seconds)]
jb55 has quit [(Ping timeout: 250 seconds)]
Wizek has quit [(Quit: Connection closed for inactivity)]
jb55 has joined #nixos
<dhess>
clever: I'm trying to override the "ghc" attribute of a ghc derivation as defined in pkgs/top-level/haskell-packages.nix. Any idea how to do that? I couldn't figure out the magic.
cifra[m] has joined #nixos
<dhess>
e.g., ghc802
cwre has quit [(Quit: Aliens ate me. I'm not good for them.)]
<dhess>
I have a special ghc binary derivation that I want to use for ARM hosts, because they take so long to bootstrap (and because armv7l doesn't have a GHC at all), and I want to shim it into the ghc802 derivation from my local overlays when running on those platforms.
<dhess>
geeze, you have a gist for everything, don't you? :)
<clever>
dhess: line 9 creates the haskellPackages set, with a given ghc, compilerconfig, and haskellLib
<clever>
dhess: in this case, i was using the ghc from an old nixpkgs, as the base for a new haskellPackages that had a modified generic-builder.nix
<clever>
and i wanted to avoid rebuilding the ghc for every single change
<dhess>
clever: ahh, callPackage within callPackage. Yeah I couldn't figure that bit out.
<clever>
but you can then customize it for anything
<dhess>
clever: ok cool. Let me try this
<dhess>
I also couldn't figure out the haskellLib bit
<dhess>
this is helpful
bts- has quit [(Quit: leaving)]
<clever>
that part changed within a month of me starting that, so it may be broken in master
<clever>
but it should fail in an obvious way
<dhess>
I also wanted to turn off tests for a few packages on armv7l as they fail in harmless ways, so the compilerConfig override is helpful for that as well.
<dhess>
for some reason, setting "dontCheck" on them in my haskell overrides didn't work.
ylwghst_ has quit [(Ping timeout: 260 seconds)]
<clever>
dhess: you need to override mkDerivation, and then // with { dontCheck = true; } i think
<NixOS_GitHub>
[nixpkgs] dhess opened pull request #31419: ProcFind: init at 0.04. (master...perl-procfind) https://git.io/vFuX2
NixOS_GitHub has left #nixos []
python476 has quit [(Ping timeout: 264 seconds)]
asuryawanshi has joined #nixos
asuryawanshi has quit [(Remote host closed the connection)]
ssmike has joined #nixos
Ivanych has joined #nixos
MoreTea2 is now known as MoreTea
<fearlessKim[m]>
isn't there a way to have $ configurePhase always be a function. I was wondering why it would run nothing, it appears this is a string so I should $exec $configurePhase instead. Having it stick to a function would reduce the cognitive load
<srhb>
eval "$configurePhase" I think.
<srhb>
There's an old issue on this, let me see if I can dig it out
<srhb>
fearlessKim[m]: Nope, can't find it. :/
kuznero_ has joined #nixos
kuznero has quit [(Ping timeout: 268 seconds)]
Isorkin_ has joined #nixos
<fearlessKim[m]>
srhb maybe https://github.com/NixOS/nixpkgs/issues/11069 ? if the variable $configurePhase exists, maybe nixos should save it, unset it and declare the configurePhase function that will execute the backuped value
fractal_ has quit [(Ping timeout: 240 seconds)]
<srhb>
fearlessKim[m]: Different one, but yeah, same idea.
ekleog_ has quit [(Ping timeout: 240 seconds)]
flyx has quit [(Ping timeout: 258 seconds)]
pierron has quit [(Ping timeout: 240 seconds)]
roblabla has quit [(Ping timeout: 258 seconds)]
Isorkin has quit [(Ping timeout: 268 seconds)]
ij has quit [(Ping timeout: 240 seconds)]
BlessJah has quit [(Ping timeout: 264 seconds)]
octe has quit [(Ping timeout: 240 seconds)]
Shados has quit [(Ping timeout: 240 seconds)]
S0rin has quit [(Ping timeout: 260 seconds)]
edef has quit [(Ping timeout: 250 seconds)]
sepi has quit [(Ping timeout: 255 seconds)]
layus has quit [(Ping timeout: 268 seconds)]
boegel has quit [(Ping timeout: 246 seconds)]
SuprDewd has quit [(Ping timeout: 258 seconds)]
CustosLimen has quit [(Ping timeout: 264 seconds)]
asuryawanshi has quit [(Ping timeout: 240 seconds)]
Guest27020 is now known as fgaz
MercurialAlchemi has joined #nixos
zzamboni has joined #nixos
ma27 has quit [(Ping timeout: 252 seconds)]
bennofs has quit [(Read error: Connection reset by peer)]
blahdodo has quit [(Ping timeout: 246 seconds)]
thc202 has joined #nixos
leat has joined #nixos
Kingsquee has joined #nixos
snikkers has joined #nixos
jensens has quit [(Ping timeout: 264 seconds)]
pie__ has quit [(Ping timeout: 268 seconds)]
blahdodo has joined #nixos
lars__ has joined #nixos
<fearlessKim[m]>
in a multiline string nix chokes on $(pwd) '' ln -sv ${configfile} \${buildRoot:=$(pwd)/../build}/.config'', I tried to escape with \$(pwd) like I did for \${buildRoot to no avail. is $() supposed to be the same as ${} ?
<makefu>
you do not need to escape $()
<makefu>
and of course $() has a different meaning than ${} , the first one runs a command (pwd in your case), the second one expands a variable
<makefu>
$() == `` ; ${var} == $var
<makefu>
and for nix context, ${} expands to the nix expression inside
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to master: https://git.io/vFuHF
<NixOS_GitHub>
nixpkgs/master d209612 Tuomas Tynkkynen: trinity: 1.6 -> 1.8
NixOS_GitHub has left #nixos []
<fearlessKim[m]>
makefu: that s what I though too but nix complains that pwd is a variable
zzamboni has joined #nixos
goibhniu has joined #nixos
<fearlessKim[m]>
https://pastebin.com/7YP5efGn (when hacking on the kernel I don't use the unpackPhase so buildRoot is not set, which is troublesome, so I want it to have a default value)
zzamboni has quit [(Client Quit)]
zzamboni has joined #nixos
<fearlessKim[m]>
in the preUnpakc of the same file there is a multilime string too `` export buildRoot="$(pwd)/build"`` but it does not trigger any warning
FRidh has quit [(Quit: Konversation terminated!)]
FRidh has joined #nixos
<Dezgeg>
maybe the \${buildRoot: part is misquoted
<Dezgeg>
maybe: ''${buildRoot:
<fearlessKim[m]>
good idea but nope no effect
kuznero_ has quit [(Ping timeout: 240 seconds)]
<fearlessKim[m]>
if I remove any of the \${ nix doesn t complain on the $(pwd)
proteus-guy has quit [(Ping timeout: 240 seconds)]
jensens has joined #nixos
ylwghst has joined #nixos
fractal_ has joined #nixos
<Enzime>
lluchs: thanks for the advice
<Enzime>
got it to flash
<Enzime>
now I just need to fix my keymap
Wizek has joined #nixos
<Enzime>
so that my keyboard doesn't just type the letter E...
justbeingglad has joined #nixos
justbeingglad has left #nixos []
snikkers has joined #nixos
quchen has joined #nixos
ylwghst has quit [(Remote host closed the connection)]
periklis- has quit [(Ping timeout: 250 seconds)]
takle has quit [(Remote host closed the connection)]
<quchen>
Hi! I’m using nix-the-package-manager (on Ubuntu), and I recently ran `nix-channel --update` to, well, update. However, I now get an error whenever I try to do something with nix-env (e.g. nix-env -u) – a syntax error in some default.nix of a package that I don’t recognize. nix-info: http://lpaste.net/359916
<quchen>
The specific error is »error: syntax error, unexpected $undefined […] at /nix/store/…hash…/nixos/pkgs/applications/misc/hubstaff/default.nix«
<quchen>
(Hubstaff is unknown to me, so I guess it’s some transitive dep)
<sphalerite>
quchen: nix 1.11.8 is outdated and has some security vulns. I suspect that updating nix may also fix the evaluation error, not sure about that bit though
takle has joined #nixos
BlessJah has joined #nixos
<sphalerite>
quchen: also you wrote that "löb and möb" article, right? I really enjoyed that :)
<quchen>
sphalerite: Hooray, it works again! Thanks. Now I’ll try to understand what that command did, actually.
<sphalerite>
quchen: just like installing anything else
<quchen>
The -f flag selects a release then?
<quchen>
And for some reason, that’s called <nixpkgs> and not nixpkgs?
<sphalerite>
I always use `nix-env -f '<nixpkgs>' -iA stuff` rather than `nix-env -iA nix{pkgs,os}.stuff` because it works both on nixos and non-nixos, and doesn't rely on nix-env's kind of stupid "nix-defexpr" logic
<sphalerite>
from a practical perspective it's usually equivalent
<sphalerite>
in fact, I have nix-env aliased to nix-env -f '<nixpkgs>'
<quchen>
I see.
<sphalerite>
-f specifies the file to install. '<nixpkgs>' means "nixpkgs from NIX_PATH"
<orivej>
Orbstheorem: there is nix-index
<sphalerite>
you can put a regular path in instead though.
iyzsong has joined #nixos
<avn>
Folks, how I can "mix" 2 or more pkgs.foo.env build environments in one nix-shell?
<NixOS_GitHub>
[nixpkgs] yegortimoshenko opened pull request #31436: meson: set install_rpath via patchelf (master...meson/patchelf) https://git.io/vFzUE
NixOS_GitHub has left #nixos []
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] jtojnar closed pull request #31228: [wip] meson: append new items to the original rpath (master...meson-append-rpath) https://git.io/vFCJB
<sphalerite>
kuznero: you mean enable/disable sandboxing on a per-build basis?
zzamboni has quit [(Ping timeout: 248 seconds)]
<hyper_ch>
sphalerite: I found something from Mic92 in an issue from 2016 and added this now to my config https://paste.simplylinux.ch/view/raw/7632c57a - in virt-manager I just select now the "br0" interface and the VMs and Host all can talk to one another :)
<kuznero>
sphalerite: yep
<hyper_ch>
one thing though, I do assign ip from router by mac address... and during initrd (for remote unlocking) it has now a different mac address..... I need to see how I can fix that
<yegortimoshenko>
kuznero: i don't think so, but nix-shell has it. open an issue if you want it.
zzamboni has joined #nixos
<yegortimoshenko>
then, performance degradation in sandbox is negligible, so if you want it, you might just as well enable it system-wide... i have it enabled for all my machines
<hyper_ch>
well, basically I want to run abby ocr 4 linux in there... so it should only startup when new file was sent from scanner to do ocr and then go back to sleep again... with openvz it was fired up very quickly....
<hyper_ch>
I'll have a look, thx
<larsvm>
aminechikhaoui: do you have and .zshrc file?
<larsvm>
*an
<hyper_ch>
larsvm: /me heard that everytime docker is mentions, god kills a cute kitten
<yegortimoshenko>
hyper_ch: why do you need a containers for that?
<yegortimoshenko>
sounds like buildFHSUserEnv would do
<larsvm>
hyper_ch: that's why i said _resort_
strobelight has joined #nixos
<yegortimoshenko>
s/containers/container/
<hyper_ch>
yegortimoshenko: for quick boot up and afterwards releasing the resource again instead of running permantently
<hyper_ch>
not even sure though if I can get abbyocr4linux run on nixos...
<yegortimoshenko>
hyper_ch: can't it just be killed after use?
<hyper_ch>
not sure what you mean by "just be killed after use"
zzamboni has joined #nixos
abcrawf_ has quit [(Remote host closed the connection)]
<yegortimoshenko>
`pkill abbyocr4linux`
<hyper_ch>
basically workflow was on debian: new file in scan folder -> incron notices -> fires up openvz, assignes 4 cores and 8gb ram to it, inside autostart of abby with parallel ocr -> shut down openvz
pie__ has joined #nixos
<hyper_ch>
yegortimoshenko: I don't want permanently assign 4 cores to the vm
<yegortimoshenko>
why does it have to be run in a vm?
<hyper_ch>
because the host doesn't do anything besides managing VMs
<yegortimoshenko>
oh, ok.
abcrawf has joined #nixos
joepie91 has quit [(Read error: Connection reset by peer)]
joepie91 has joined #nixos
joepie91 has quit [(Remote host closed the connection)]
<larsvm>
aminechikhaoui: how do you know there are those attributes to override?
joepie91___ has joined #nixos
joepie91___ has quit [(Changing host)]
joepie91___ has joined #nixos
<yegortimoshenko>
if it presumes FHS-compliance, you can port it to NixOS via `buildFHSUserEnv`.
<aminechikhaoui>
larsvm: I think the module should take care of the whole setup
<hyper_ch>
Nixos uses a whooping 1.88GB
<aminechikhaoui>
so no need for .zhrc afaik
<aminechikhaoui>
.zshrc *
<aminechikhaoui>
check nixos/modules/programs/zsh/oh-my-zsh.nix to see what's being done in the module
<larsvm>
speaking of rc files, how do people have their vim set up?
<larsvm>
i'd like to share the config with my work computer, so i'd prefer a .vimrc
<larsvm>
but then i run into problems with all the dependencies for the packages
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peterhoeg pushed 1 new commit to master: https://git.io/vFzZV
<NixOS_GitHub>
nixpkgs/master 1529295 Peter Hoeg: qtox: fix compilation
NixOS_GitHub has left #nixos []
scribbler has quit [(Ping timeout: 240 seconds)]
ThatDocsLady has quit [(Ping timeout: 240 seconds)]
roberth has joined #nixos
<kuznero>
yegortimoshenko: thanks, will enable that system-wide and will try to file an issue on nix-build
acarrico has joined #nixos
zzamboni has quit [(Quit: Leaving.)]
<betaboon>
hello #nixos. is the nix install on macos currently working? I'm running into the problem of install-darwin-multi-user.sh complaining about an existing /nix, which is the loopback devide mount .. any ideas ?
zzamboni has joined #nixos
<sphalerite>
as a workaround you could remove it and then move it onto wherever you want it to actually live and mount it there afterwards
<sphalerite>
betaboon: It just doesn't like a pre-existing /nix
<contrapumpkin>
why is /nix a loop mount?
<contrapumpkin>
that's not normal macOS setup
musicmatze has joined #nixos
ssmike has quit [(Remote host closed the connection)]
ssmike has joined #nixos
<betaboon>
contrapumpkin: i read that i have to create a case-sensitive loopback-mount because of the default-case-insensivitivy of HFS !?
roberth has quit [(Ping timeout: 248 seconds)]
scribbler has joined #nixos
<betaboon>
sphalerite: i got past that check with `ALLOW_PREEXISTING_INSTALLATION=true ./install....`
__monty__ has joined #nixos
c0bw3b has joined #nixos
<sphalerite>
betaboon: that's a neater solution!
<contrapumpkin>
betaboon: read where? you can do that, but almost nobody I know ever does that
<contrapumpkin>
most of the macOS nix maintainers do it on case-insensitive filesystems
<contrapumpkin>
if we decide to recommend that we should probably put it into the installer :)
roberth has joined #nixos
<betaboon>
contrapumpkin: a collegue read that on an extremly old article.
<__monty__>
Why is nix-env's memory pressure so high? Granted this would probably be a non issue on more recent hardware but 60% of 2G is a lot of memory. Is it because of xz? (I figure that's where the cpu load comes from.)
<betaboon>
nvm that problem we solved by RTFM ....
<betaboon>
stupid me
<contrapumpkin>
betaboon: it shouldn't hurt you to do that, but was it official on site or something? want to make sure our instructions are up to date and recommending stuff that we actually test :)
<contrapumpkin>
which article?
<betaboon>
contrapumpkin: the manual is 100% correct. by now following just the up-to-date manual, and not the ages-old-blog-article creates the correct installion
<contrapumpkin>
yeah, a lot has changed since then but maybe we should automate setup of a case-sensitive mount
joepie91___ is now known as joepie91
<contrapumpkin>
maybe not though, I don't want to make a separate partition/image
<contrapumpkin>
was hoping there was just a flag you could set at mount time rather than at FS create time
ssmike has quit [(Remote host closed the connection)]
ssmike has joined #nixos
roberth has quit [(Ping timeout: 248 seconds)]
yegortimoshenko has quit [(Ping timeout: 248 seconds)]
ThatDocsLady has joined #nixos
pxc has joined #nixos
strobelight has quit [(Quit: gtg)]
<fearlessKim[m]>
I tried fetchzip on a zip file (granted the extension is src.rock and not .zip) which triggers, "do not know how to unpack source archive /tmp/nix-build-lpeg-1.0.1-1.src.rock.drv-0/lpeg-1.0.1-1.src.rock" :/ an way to force unzipping ?
<sphalerite>
__monty__: What command are you running exactly?
michiel_l has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] orivej pushed 2 new commits to master: https://git.io/vFzlS
<__monty__>
sphalerite: Except for the one I'm running right now, of course : )
Guest21796 has quit [(Changing host)]
Guest21796 has joined #nixos
Guest21796 is now known as TweyII
pxc has quit [(Ping timeout: 240 seconds)]
<TweyII>
fearlessKim[m]: There's a ‘name’ argument to fetchurl that you can use to rename it to .zip locally before unpack
<sphalerite>
__monty__: are you using -i without -A?
<__monty__>
sphalerite: Yes.
<sphalerite>
__monty__: don't :) it evaluates all of nixpkgs
<__monty__>
Does that make all the difference? In the current one I did use -A so I could specify the nixos channel.
<__monty__>
Ok, noted.
<__monty__>
Another issue it looks like nix-env doesn't read .config/nix/nix.conf What's the proper location?
<betaboon>
installing nix on macos, then nixops and deploying to a nixos-virtualbox doesnt work, right ?
<betaboon>
due to missing cross-compile !?
<fearlessKim[m]>
TweyII: doesn't seem to make a differnece, it just changesthe output name. The error is triggeredby 'if ! runOneHook unpackCmd "$curSrc"; then'
<__monty__>
Are remote builds possible without a daemon on local nor remote?
<__monty__>
If no, does it require a daemon on both or just remote?
<Dezgeg>
neither side requires a daemon
jtojnar has joined #nixos
zzamboni1 has joined #nixos
zzamboni has quit [(Remote host closed the connection)]
<__monty__>
Dezgeg: This is what I get currently: http://ix.io/C69
<__monty__>
If I ssh toonn@nix on the shell it works without prompting for a passphrase.
<Dezgeg>
how about 'toonn@nix nix-store --version'
<Dezgeg>
err, 'ssh toonn@nix nix-store --version'
zzamboni1 has quit [(Client Quit)]
<__monty__>
Hrm, command not found. While running after ssh toonn@nix gives 1.11.15
zzamboni has joined #nixos
<Dezgeg>
yeah, sourcing the nix profile script needs to be somewhere that non-interactive shells execute it as well
scribbler has joined #nixos
zzamboni has quit [(Client Quit)]
zzamboni has joined #nixos
codygman has quit [(Quit: Page closed)]
<__monty__>
Dezgeg: I figured if I put it in .zshenv it would work for both (non-)interactive shells but it doesn't for interactive shells, does it need to be in both .zshenv and .zshrc?
<Dezgeg>
I don't know how those work for zsh
<manveru>
__monty__: man zsh :)
<Dezgeg>
but even for bash it's real pain to figure out what gets sourced where
<manveru>
__monty__: i actually configure my zsh via nixos config
<__monty__>
manveru: The manual says .zshenv is read for every shell invocation (non-)interactive and login shells. But the nix commands can no longer be found when ssh'ing to the host.
<__monty__>
I'm not on nixos.
<manveru>
ok
<manveru>
then make a .profile that sources it maybe
<manveru>
or .zprofile
<__monty__>
Mmm, I'd rather figure out the problem than sourcing files multiple times. NIX_PATH does seem to get set by the sourcing in .zshenv.
<__monty__>
Ok, it seems like PATH shouldn't be set in .zshenv so I'll just set that in zshrc.
<betaboon>
can soneone point me into the right direction on how to use fetchPrivateGit ?
<__monty__>
Nix.sh adds $NIX_LINK/sbin to the PATH but this directory doesn't exist so zsh complains when it's defined in .zshrc Should $NIX_LINK/sbin exist?
ssmike has quit [(Remote host closed the connection)]
patrl has quit [(Quit: WeeChat 1.9.1)]
patrl has joined #nixos
ssmike has joined #nixos
<__monty__>
Oh, figured it out, syntax error.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to master: https://git.io/vFzEE
<NixOS_GitHub>
nixpkgs/master b306298 Tuomas Tynkkynen: postfix: Fix build by adding libnsl...
NixOS_GitHub has left #nixos []
<__monty__>
Ok, so now the only thing still missing is a working nix.conf. How do I find out _sysconfdir_? The manual says the proper location is sysconfdir/nix/nix.conf, I figured sysconfdir'd be ~/.config by default but that assumption seems incorrect.
<Dezgeg>
it's /etc
ma27 has quit [(Ping timeout: 240 seconds)]
<bgamari>
steveeJ, are you using my branch?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] lsix pushed 2 new commits to master: https://git.io/vFzzk
<NixOS_GitHub>
nixpkgs/master e4c7cfc Marti Serra: pythonPackages.sybase: move to python-modules
<betaboon>
can anyone assist in getting fetchgitPrivate to work ?
<infinisil>
betaboon: Just ask your question, if somebody happens to be able to help you they'll respond
ryantrinkle has quit [(Ping timeout: 264 seconds)]
<hyper_ch>
hi infinisil
<infinisil>
hyper_ch: Good day!
pie__ has quit [(Ping timeout: 240 seconds)]
<hyper_ch>
you're in a good mood
<infinisil>
so so, could be better, i have an important exam tomorrow
FRidh has quit [(Ping timeout: 260 seconds)]
FRidh has joined #nixos
<averell>
how can i remap a single character into my layout? or get "ssharp" to work another way using standard dvorak?
acarrico has quit [(Ping timeout: 240 seconds)]
<TweyII>
averell: I use an xkboption, compose:caps
<TweyII>
averell: Then you can Compose s s to get ß
<TweyII>
You can also use xmodmap, but you're supposed to just write your own keyboard layout, these days — they can inherit from other layouts
<betaboon>
I'm trying to use fetchgitPrivate to build from a private Github repo. as explained in build-support/fetchgit/private.nix i set ssh-config-file=/home/betaboon/.ssh/config in NIX_PATH. I'm running into the following error: https://gist.github.com/betaboon/2841c07ef35655256f9e11cad2876513 . can anyone help?
<hyper_ch>
infinisil: oh.... what is it about?
iyzsong has quit [(Ping timeout: 240 seconds)]
<MoreTea>
betaboon: I think you need to give access to that file to the nixbld group
asuryawanshi has quit [(Ping timeout: 240 seconds)]
<MoreTea>
are you using a multi-user nix install?
<hyper_ch>
hi MoreTea
tgunb has quit [(Ping timeout: 240 seconds)]
<MoreTea>
hyper_ch: hey :)
<averell>
TweyII: thanks, i'll try compose first. caps is already full though.
<catern>
there should be a way to "ssh forward" a Nix store
<TweyII>
averell: Sure, you don't have to use caps :)
<catern>
so that I can install things on the remote host, using the store on my local host
joncfoo__ has joined #nixos
<__monty__>
catern: nix-copy-closure?
<catern>
it would be like nix-copy-closure, but seamless and automatic
<catern>
:)
<betaboon>
MoreTea: as far as i know I'm on a single-user system.
<catern>
one would just nix-env -i or nix-build or whatever
<hyper_ch>
MoreTea: you remember my rtorrent issue?
asuryawanshi has joined #nixos
<__monty__>
betaboon: Does the key actually exist?
<betaboon>
MoreTea: should changing the file with chmod :nixbld be sufficient ?
<catern>
(on the remote host)
<catern>
and it would fetch the artifacts from the "binary cache", namely the store on my local host
<infinisil>
hyper_ch: The test is about Computer Science Theory: finite state machines, regular/recursive languages, kolmogorov-complexity, etc.
<__monty__>
catern: ssh-substituter?
<__monty__>
catern: Or ssh-substiture-hosts to be more precise.
<betaboon>
__monty__: yes the IdentityFile referenced in ssh-config-file for the host github.com does indeed exist.
<__monty__>
-.- *substituter
<hyper_ch>
infinisil: I have no idea what you just said... but then, luckily I don't have to take that test
<betaboon>
infinisil: i remember that kind of stuff. i hated the exams in those classes. but by now I'm glad i have taken them :)
<__monty__>
infinisil: One of the best courses : )
<infinisil>
I failed last years exam in this course, so I'm retaking it now
simendsjo has joined #nixos
<infinisil>
Don't like that subject
<catern>
__monty__: it would be like ssh-substitute-hosts, but seamless and automatic
p4cman has quit [(Quit: Connection closed for inactivity)]
<catern>
__monty__: or more concretely, it would work without explicitly setting up a connection backwards
<betaboon>
infinisil: happened to me too
acarrico has joined #nixos
<betaboon>
i think the "fun"-aspect heavily depends on the tutor
<catern>
__monty__: there would just be some ssh invocation to forward a Unix socket
<__monty__>
catern: Backwards? Host to client?
<betaboon>
MoreTea: it might be relevant, that I'm using nixops ?!
simendsjo has quit [(Remote host closed the connection)]
<__monty__>
catern: I don't see what's less seamless about ssh substituters.
pie__ has joined #nixos
<catern>
__monty__: yes, host to client
erasmas has joined #nixos
<__monty__>
betaboon: Not sure about nixops. Still trying to understand the permission issue. Are you maybe running a single user on a multi-user nix install? That's different from a single user setup.
<catern>
__monty__: when I ssh from client C to host H, I have to configure an ssh-substituter on H, to C. it should be automatic instead
<catern>
well, automatic after some generic setup :)
<__monty__>
It'd be akin to X forwarding. Would be cool for sure.
<catern>
except nix-copy-closure is a "push" model of deployment
<catern>
I want a "pull" model of deployment
scribbler has joined #nixos
<catern>
because that's sometimes a lot more convenient
orivej has quit [(Ping timeout: 260 seconds)]
mekeor has joined #nixos
<catern>
in particular a pull model means that I just don't think about the deployment host C (which is performing the deployment), I only have to think about a single host H, and just invoke "pull" commands on that host H, such that it pulls in the things I want, from wherever it can get them (which would be C, automatically)
periklis has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peti pushed 2 new commits to master: https://git.io/vFzi7
<NixOS_GitHub>
nixpkgs/master b7f5acd Peter Simons: LTS Haskell 9.12
<NixOS_GitHub>
nixpkgs/master dfa0bcd Peter Simons: hackage-packages.nix: automatic Haskell package set update...
NixOS_GitHub has left #nixos []
srdqty has joined #nixos
<betaboon>
__monty__: is there any way to explicitly check for that? i followed the nixos manual in setting up this machine. i did not take any of the additional multi-user steps, mentioned in the manual.
<__monty__>
betaboon: I'm pretty sure nixos has a multi-user setup by default.
<catern>
similarly it would be nice to ssh-forward git... hmm
<__monty__>
betaboon: Maybe pgrep nix-daemon?
<betaboon>
returns a single PID
<__monty__>
So it's running, which means it's multi-user.
<Ankhers>
Does Nix and high siera work well together at this point?
<__monty__>
You'll have to set the permissions of the key properly, betaboon. So change the owner group and probably relax host key checking.
periklis has quit [(Ping timeout: 240 seconds)]
<MoreTea>
betaboon: I've not used fetchPrivateGit yet, so it's just a suggestion based on the error message that you got.
<betaboon>
__monty__: I'm already using `StrictHostKeyChecking=no`. the `id_rsa` file referenced in the ssh-config-file is owned by root:nixbld with 770 .
<MoreTea>
hyper_ch: got that rtorrent stuff sorted?
<hyper_ch>
MoreTea: yep :) works all fine :)
<MoreTea>
hyper_ch: cool
<betaboon>
MoreTea: thanks for the effort :)
<hyper_ch>
MoreTea: thx for the help :)
<__monty__>
betaboon: Where is this .ssh directory located?
zzamboni has quit [(Quit: Leaving.)]
<betaboon>
__monty__: i have a single directory containing a shell.nix, network.nix(nixops-deployment), nixops.state, ssh/id_rsa and ssh/id_rsa.pub. i then use nix-shell to enter that shell. the shell provides nixops and sets the NIX_PATH, which then contains nixpks=,... and ssh-config-file= .
MercurialAlchemi has quit [(Ping timeout: 258 seconds)]
<betaboon>
i hope that clarifies instead of distort your picture :P
<__monty__>
It does.
<__monty__>
The error you gave was with the current config?
<betaboon>
i'm pretty sure, but not 100%
zzamboni has joined #nixos
leat has quit [(Quit: WeeChat 1.9.1)]
asuryawanshi has quit [(Remote host closed the connection)]
ma27 has quit [(Ping timeout: 258 seconds)]
asuryawanshi has joined #nixos
<__monty__>
betaboon: Maybe chown the key file to the user calling the nix command.
<__monty__>
I'm out of ideas tbh.
patrl has joined #nixos
<betaboon>
i'll try chmod betaboon:nixbld later on. but I'm out of ideas as well.
<oleks>
I'm having trouble building Rider, a new IDE from JetBrains, it comes with its own JDK bundled in, but it needs libfontconfig on the host system - merely installing the fontconfig package doesn't help - any ideas?
w-spc-gir has joined #nixos
srdqty has quit [(Ping timeout: 248 seconds)]
darlan has joined #nixos
p4cman has joined #nixos
darlan has quit [(Client Quit)]
ryantrinkle has joined #nixos
scribbler has quit [(Ping timeout: 268 seconds)]
<__monty__>
manveru, Dezgeg, sphalerite, clever, rycee: And anyone I forgot to mention. Thank you. I now have a LAN cache cache, remote builds and nix is finally reading my nix.conf : D
scribbler has joined #nixos
freusque has quit [(Quit: WeeChat 1.7.1)]
jsgrant has quit [(Remote host closed the connection)]
<gchristensen>
anyone know what "error: setting synchronous mode: unable to open database file" means as root when running nix-collect-garbage?
<ldlework>
I was reading https://github.com/kamilchm/go2nix and it says "Note: You can also arrange a default.nix so that it can be used by nixpkgs and nix-shell but this is not covered here." Anyone know how to do that?
ertes-w has quit [(Ping timeout: 258 seconds)]
<infinisil>
ldlework: Just check out some examples in nixpkgs
<ldlework>
Are nixpkgs generally setup to be nix-shell-able?
<pierron>
hyper_ch: the presentation should be quite accesible, except for the part where I say that making a part is easy (from what I have been told)
<ldlework>
Like you can't nix-shell a default.nix that is formulated as a callPackages package right?
<hyper_ch>
pierron: well, I need to rewatch it again
<pierron>
s/making a part/making a parser/
<hyper_ch>
pierron: so, you're an overlay specialist? :)
<clever>
ocharles: that file handles declarative jobset management, it must return a set with just .jobsets (line 86) which contains a json file describing all jobsets in the project
<infinisil>
ldlework: Just do nix-shell -A attributename
<ocharles>
yea, that's the bit I'm stuck on
<pierron>
hyper_ch: I am the absent guys who only show up to make tiny changes with big impacts.
<ocharles>
clever: the docs just say "returns jobsets" but doesn't really say much about what that is
<ldlework>
I'm not sure how that helps but I'll contemplate it
<pierron>
hyper_ch: please refer to the video documentation :P
<clever>
ocharles: the one you linked has a dummy input defined on line 1, just run "nix-build jobsets/default.nix -A jobsets" and you can test it locally
<hyper_ch>
pierron: I need a more up-to-date version of zfs and I guess I have to do that with an overlay - somehow
steveeJ has joined #nixos
<infinisil>
pierron: tiny changes with big impact is much better than big changes with tiny impact!
<clever>
ocharles: it also relies on spec.json in the same directory
<infinisil>
this should probably be included in the video description
zraexy has quit [(Remote host closed the connection)]
simendsjo has joined #nixos
<clever>
ocharles: you point hydra to spec.json, and it will then run default.nix with the inputs defined on lines 12-17 of spec.json
<ocharles>
yea, I have hydra doing stuff, it's just defining the actual jobsets that I got stuck on
<clever>
ocharles: and then default.nix will return json with a list of objects like spec.json
<hyper_ch>
infinisil: I'll have a look at it when I have more time again
<clever>
ocharles: can you link your project on github?
<hyper_ch>
still need to figure out what kind of container I should use.... for the final thing
<ocharles>
clever: no, it's closed source I'm afraid
<clever>
ocharles: can you put a censored copy of the default.nix up on gist?
<hyper_ch>
and then I actually need to do some work... well, spent half my day on tuesday phoning everyone and asking for extensions
<ocharles>
but anyway, this all should give me enough to get going
<ocharles>
clever: don't worry, you've given me enough :)
<clever>
kk
<ldlework>
infinisil hmm I don't get it. how would -A allow me to use any ol nixpkg expression from nix-shell given that those packages are all formulated as functions expecting their specific build inputs as arguments?
<infinisil>
ldlework: well it just works for the defined packages
<infinisil>
in all-packages.nix
<hyper_ch>
srhb: bool(false) int(6)
<infinisil>
which contains smth like <attribute> = callPackage ../path/to/decl.nix { additional = stuff; };
<ldlework>
infinisil well yeah because those attributes are literally... yeah
<ldlework>
That's the point
<srhb>
hyper_ch: Ouch!
<srhb>
OK, not just me then.
<ldlework>
a default.nix sitting in a project can either be formulated as a package that needs to be called with callPackage, or it is a file that is calling a package with callPackage
<ldlework>
the author of that blog alludes that you can write up a default.nix such that it would work being called by callPackage, but also being called directly by nix-shell
<infinisil>
you don't need callPackage really, it's just something that makes overriding stuff easier
<hyper_ch>
srhb: int(1) int(0) - PHP 7.0.19-1 (cli) (built: May 11 2017 14:04:47) ( NTS ) on debian stable
<clever>
srhb: int(1) and int(0) when ran on the cmdline
__Sander__ has quit [(Quit: Konversation terminated!)]
<srhb>
hyper_ch: I think it's not just 7.1.11, it's all versions in nixpkgs.
<srhb>
Maybe something about the stack configuration.
<srhb>
clever: That's what it should be. Broken in httpd though
<srhb>
clever: (I get the same thing on the command line)
<clever>
ah
<clever>
and does phpinfo() show the same version?
<infinisil>
ldlework: I see, not sure how you'd do that
<srhb>
clever: Yes.
<srhb>
clever: Setting pcre.jit = false bypasses the problem entirely as well.
<srhb>
In php.ini
<srhb>
I think maybe the pcre library is setup wrongly.
<srhb>
But I don't see it mentioned explicitly anywhere...
<srhb>
clever: Well, it certainly affects the php in apache.
<srhb>
I'll dig some more.
<Dezgeg>
do the apache and php both link to the same version of pcre?
<michaelpj>
ldlework: I'm also not sure what to do there. I usually follow the nixpkgs convention and have my default.nix be a function, and then write a separate shell.nix which imports nixpkgs and calls the default.nix
<srhb>
Dezgeg: That sounds like a likely culprit.
<michaelpj>
but then I can't nix-build my default.nix directly, which feels wrong, I have to do `nix-build -E 'with import <nixpkgs> {}; ...'`
<clever>
michaelpj: i usually make something like a release.nix that handles that
<sphalerite>
michaelpj: it's a little confusing, but I'd say it's a good thing because it reflects very well that the package doesn't exist in a vacuum but has a bunch of dependencies
<gchristensen>
sphalerite: but should it bee `default`?
<sphalerite>
michaelpj: and nix-build '<nixpkgs>' -A foo works nicely once you've got that figured out
<michaelpj>
that doesn't work for packages in that form that aren't in nixpkgs, though (e.g. I have a couple in my nixos config repo that haven't made it into PRs yet)
<michaelpj>
I understand why things are the way they are, I just have a lingering feeling of dissatisfaction ;)
jgertm has joined #nixos
<ldlework>
It is certainly hard to explain in a technical intro to Nix talk
<ldlework>
Where I want to say "See, look how straightforward and elegant this is!"
pie__ has quit [(Ping timeout: 268 seconds)]
<michaelpj>
"do it like this, unless you're in a different context, in which case do it this other way" is always a bit embarassing
hotfuzz_ has joined #nixos
phanimahesh has joined #nixos
tgunb has quit [(Ping timeout: 248 seconds)]
snikkers has joined #nixos
<michaelpj>
for nix-build, I think the fact that it has a default file to build and everything suggests that it's aiming for a slick workflow where you just go to the project dir, run nix-build, and you're done
hotfuzz has quit [(Ping timeout: 240 seconds)]
<phanimahesh>
I'm not on nixos but using nix to build a docker image. There are some dependencies in scripts which I want to include.
<phanimahesh>
Which package(s) provides readlink, whoami and dirname?
<ldlework>
michaelpj yeah, the convention should probably be, that you add some second artifact to make it loadable by nixpkgs
<ldlework>
Like the callPackages wrapper, which is then utilized in all-packages
<ldlework>
but this would be a massive change I guess
<phanimahesh>
There's no easy way to know without some trial and error, can anyone on nixos do a which on them and find the package names?
<michaelpj>
phanimahesh: try nix-index, I'm pretty sure that searches for executable names
<ldlework>
hmm I think I got that backwards
<michaelpj>
ldlework: yeah, and it would be annoying for nixpkgs contributors to have to have a separate file for this use case that seems to be, in practice, rare
<phanimahesh>
michaelpj: Trying, thanks.
vcunat has joined #nixos
<ldlework>
It seems like it should be less rare though. You have some package. To do development on it, you use nix-shell and nix-build. Once its complete, it moves on to being a package.
<michaelpj>
phanimahesh: there's also `nix search` in nix 1.12, but I don't think it searches executable names, whereas I'm pretty sure nix-index does
<ldlework>
But I guess not all packages are developed with Nix
<phanimahesh>
This is going to load heaven knows what amount of stuff from the interwebz. There'sno prepopulated cache anywhere?
civodul has quit [(Quit: ERC (IRC client for Emacs 25.3.1))]
patrl has quit [(Ping timeout: 240 seconds)]
patrl has joined #nixos
goibhniu has quit [(Ping timeout: 260 seconds)]
<sphalerite>
My system seems takes an annoyingly long time to boot. This seems to be caused by several units taking just over 10 seconds to start, does anyone have any idea why this might be happening?
<clever>
iqubic: you can add zfs to /etc/nixos/configuration.nix and rebuild switch to get zfs
<iqubic>
How do I add zfs to the configuration?
<sphalerite>
but I wouldn't recommend zfs if you're not willing to take some time to learn about it
hellrazo1 has quit [(Quit: WeeChat 1.9.1)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] 4z3 opened pull request #31445: bash-completion: allow lazy loading where possible (master...bash-completion) https://git.io/vFgT5
NixOS_GitHub has left #nixos []
<sphalerite>
brb rebooting to see if disabling bluetooth helps
<iqubic>
What are the differences between ext4 and zfs?
<clever>
iqubic: zfs allows compression, and different volumes that have different settings
endformationage has joined #nixos
<srhb>
Building php against a different (the same as apache-httpd. Does that matter?) version of pcre removed the issue.
scribbler has quit [(Ping timeout: 248 seconds)]
<iqubic>
What is a volume in this case?
<phanimahesh>
It worked, and index building didn't take as long as I feared. Thanks michaelpj! Also, TIL nix-{index,locate}.
<clever>
iqubic: as an example / has automatic snapshots but /nix/ doesnt have backups
<iqubic>
I see.
<clever>
iqubic: and i could make /nix/ have more aggresssive compression (gzip-9) while / has just lz4
<clever>
gz9 costs more cpu, but nix is written to less, so it may balance out
<iqubic>
I don't think I need all of those power features.
<michaelpj>
phanimahesh: glad it worked :) I think the hope is for this kind of tooling to gradually become more baked in, but nix-index is pretty useful for now
<srhb>
I feel like testing this on a non-NixOS system, but ugghhh. :-)
w-spc-gir has quit [(Ping timeout: 260 seconds)]
<iqubic>
I think I want to just use ext4 because it has always worked for me in the past
pie_ has joined #nixos
<sphalerite>
clever: still the 10s delay on the other 3 units
<iqubic>
now, do I need any swap space?
<iqubic>
Because I only have 100 GB for swap + filesystem.
<sphalerite>
if you want to suspend to disk, you do
<sphalerite>
otherwise it depends on how much RAM you have and how you'll be using the system
<clever>
sphalerite: try disablign avahi-daemon next
<clever>
iqubic: i'm getting a compression ratio of 1.9x on my new laptop with zfs, so thats nearly doubling the disk space in the machine
<iqubic>
clever: Do you recommend the bash script?
<clever>
iqubic: you can use it more as a guide
<sphalerite>
srhb: :D
ma27 has joined #nixos
<iqubic>
I will.
<clever>
iqubic: that bash script will try to wipe the entire disk, so your dual-boot would go away
<iqubic>
I see.
<iqubic>
I want to keep my system.
<clever>
youll want to partition it differently then, either using gparted or normal fdisk
<clever>
and then adjust what partition names you pass to the rest of the commands
<iqubic>
Currently the way I have it set up is /dev/sda1 is /boot. /dev/sda2 is Windows FS. /dev/sda3 is a 20GB partition that I can Read and Write to from both OSes and /dev/sda4 is going to be my NixOS FS.
<iqubic>
So I only need to take /dev/sda4 and partition that.
<clever>
kk
<clever>
UEFI?
<iqubic>
Yes. It is UEFI
<iqubic>
I can just piggy-back on the boot partition that windows made.
<iqubic>
It is formatted correctly already.
<clever>
line 78-81 of justdoit will format $ROOT_DEVICE with zfs, you can skip 76
thblt has joined #nixos
<clever>
then 84 to mount the rootfs, and 87-88 to mount everything else (change the type on 88)
<clever>
then just nixos-generate-config and nixos-install
<iqubic>
Thank you for the detailed advice.
<iqubic>
Also clever I have had a previous nixos machine, so I'm just going to migrate my configuration.nix to the new machine.
<iqubic>
And this is the only time where it makes sense to change stateVersion.
<clever>
yeah
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<iqubic>
What does cryptsetup do?
<c0bw3b>
is there a nix-on-darwin user around? that would like to try and nox-review PR 31376 :)
<clever>
iqubic: encrypted root filesystem, its optional
<iqubic>
I don't want that.
<c0bw3b>
i'm not sure if the failed travis job was looping or just too long :/
<clever>
iqubic: its all inside an if statement on line 45
<iqubic>
I saw that.
<hyper_ch>
clever: howdy
<clever>
hyper_ch: afternoon
<hyper_ch>
clever: I see, you're busy again helping someone - this time not me :)
<clever>
iqubic: you also need to add /dev/sda4 at the end
<iqubic>
Right.
<adelbertc>
Is there anything I need to do to setup Nix distributed builds, environment wise, besides setting $NIX_REMOTE_SYSTEMS, $NIX_CURRENT_LOAD, and $NIX_BUILD_HOOK?
<hyper_ch>
also you can give it some options if you want :)
<iqubic>
And I can change rpool to be whatever name I want right?
<hyper_ch>
or if you manage massiv amounts of data like OCHs do.. .then dedup can come in quite handful
<clever>
i also made a dedicated /var/lib/docker pool, with dedup enabled
<hyper_ch>
what's the rule of thumb? Per 1 TB of data, another Gig of ram required (in addition to your normal ram usage)
<hyper_ch>
for dedup
<iqubic>
How do I disable dedup?
<gchristensen>
*gif of bart simpson walking around in a circle, banging a pot* if your dedupe table stops fitting in to ram, your data is hosed until you can buy more ram, and most people don't actually want dedupe, and it isn't enabled by default
<hyper_ch>
iqubic: its disabled by default
<iqubic>
hyper_ch: Good to know.
<clever>
hyper_ch: i'm getting a 4x ratio on docker
<hyper_ch>
clever: nice
Judson has joined #nixos
phreedom has joined #nixos
<adelbertc>
clever: whats the best way to inspect/modify the environment of nix-daemon?
Judson is now known as Guest91640
<clever>
adelbertc: not sure if darwin has an easy way to read the env
<clever>
adelbertc: but you can just try setting up /etc/nix/machines first, and see if it works, it may already be configured
phreedom has quit [(Client Quit)]
<iqubic>
Why do people name their zfs pools as tank, or fishtank or fishbowl or such?
<adelbertc>
clever: should that file exist already
<iqubic>
What is the right type of compression to enable when setting up my zpool?
<hyper_ch>
iqubic: most people chose lz4 becuase it's good compression but not very cpu intense.... there's of course better compression algos otu there but those require much more cpu
<clever>
iqubic: lz4 uses less cpu
<clever>
gzip-9 saves more, and has a noticable hit on performance
<adelbertc>
no go on /etc/nix/machines
<hyper_ch>
iqubic: if you want to enable compression, I'd go first with lz4
<clever>
one day, i set my desktop to use gzip-9, while systemd-coredump was enabled
<iqubic>
If I use compression, will I lose any data?
<clever>
then every time chrome had a segfault, it would lockup for nearly a minute, while the 2gig coredump was written to disk
<clever>
iqubic: nope
<hyper_ch>
no, it just gets compressed
ssmike has quit [(Ping timeout: 240 seconds)]
<hyper_ch>
text files, word documents etc. are good for compression
<hyper_ch>
video, mp3 etc. not so much
<iqubic>
As in, will compressing and uncompressing my data cause any loss of data.
<clever>
iqubic: its nothing like the c64 days where they messed with bit-rate of the drive to squeeze in more bits
<hyper_ch>
it's losless compression
<iqubic>
hyper_ch: I usually only store text files and .odt files on my machine.
thblt has quit [(Ping timeout: 240 seconds)]
<hyper_ch>
so you'll benefit from compression
<clever>
NAME RATIO USED LUSED COMPRESS
<clever>
tank/nix 2.04x 10.8G 20.2G lz4
<clever>
just /nix alone is compressed to half its size
<iqubic>
Wow.
<gchristensen>
iqubic: zfs is by oracle, oracle likees to not lose data, please do your own research,h #nixos is not #zfs-beginners
sigmundv__ has joined #nixos
<iqubic>
Is #zfs-beginners a thing?
<gchristensen>
no idea
jb55 has quit [(Ping timeout: 240 seconds)]
jgertm has quit [(Ping timeout: 268 seconds)]
<hyper_ch>
isn't it better to say zfs is by sun?
<clever>
NAME RATIO USED LUSED COMPRESS
<clever>
amd/nix 2.16x 69.0G 135G gzip-9
<infinisil>
probably videos and music can also benefit from compression
<clever>
iqubic: this dataset is a mix of lz4 and gzip, and also takes up less then half the space
<iqubic>
Cool.
<gchristensen>
iqubic: I think not actually? their containers are usually quite well compressed already
<gchristensen>
infinisil: ^
<hyper_ch>
gzip9 has better compress ratio but doesn't look so much better compared to cpu cycles required
<clever>
NAME RATIO USED LUSED COMPRESS
<clever>
naspool/nas 1.07x 1.20T 1.27T lz4
<clever>
this one is mostly video files
ssmike has joined #nixos
<iqubic>
Well, "-o atime=off" doesn't work, because atime is not a valid pool property.
<clever>
-O atime=off
<adelbertc>
how evil is modifying `/Library/LaunchDaemons/org.nixos.nix-daemon.plist` directly..
<clever>
-o for pool wide properties, -O for properties on the root dataset
<gchristensen>
adelbertc: "not so"
<clever>
adelbertc: is it a symlink?
<adelbertc>
yeah, into `/nix/var/nix/profiles/default/Library/LaunchDaemons/org.nixos.nix-daemon.plist`
<adelbertc>
gchristensen: ?
<infinisil>
gchristensen: Yeah, but still a compress ratio > 1
<adelbertc>
oh, not so evil
<clever>
adelbertc: editing anything in /nix will break things
nix-gsc-io`bot has joined #nixos
<ldlework>
If I'm using nix-shell, with a default.nix that returns a mkDerivation, do any of the build steps get run at all?
<hyper_ch>
iqubic: that was a thing I didn't get at first... when you create a zfs pool, at the same time you create a root data with the same name.... so you can apply settings to the pool and other settings to the dataset :)
nix-gsc-io`bot has quit [(Client Quit)]
<adelbertc>
bah
<adelbertc>
can i do an overlay or something to patch that plist?
ertes has joined #nixos
<clever>
ldlework: none of the steps run, it just drops you into a shell with the build deps, and a genericBuild bash function
<iqubic>
clever: I assume I should set ashift=12 and altroot=/mnt because that's what the nixos wiki tells me to do.
<clever>
iqubic: yeah
<ldlework>
clever ah and calling that would invoke the build steps
<clever>
adelbertc: yeah, you can probably replace it with an override over nix
<clever>
ldlework: yeah
<gchristensen>
adelbertc: make a package with it, placed at $out/Library/Launchdaemons/org.nixos.nix-daemon.plist andn set it to a higher priority package I think
<iqubic>
compression also needs the -O flag.
jtojnar has quit [(Quit: jtojnar)]
<iqubic>
Why is compression a property of the root data?
<hyper_ch>
iqubic: compression is set for datasets
<hyper_ch>
you can have datasets with compression
<iqubic>
Well, looks like zfs hates windows.
<hyper_ch>
and datasets without compression
fenedor has quit [(Ping timeout: 268 seconds)]
<hyper_ch>
pool properties apply to the whole pool
<iqubic>
I can't make my zpool because /dev/sda4 has filetype ntfs
<infinisil>
yee, windows is a no-go with zfs, although i just saw some project that tried to get zfs for windows
<clever>
iqubic: you may want to run wipefs on /dev/sda4 to remove traces of the old FS
<iqubic>
/dev/sda4 is empty though.
<hyper_ch>
als, when you create a child dataset, it will inherit the properies from its parents... so setting compression for the root dataset will ensure, that child datasets will also have compression on - until you turn it off for said child dataset
<infinisil>
iqubic: just add -f to the creation flags
<adelbertc>
clever: just FYI, poking into `/var/log/nix-daemon.log` shows that it attempts the distributed build only when I use /etc/nix/machines so that gives me something to work with :)
<adelbertc>
"unable to open SSH connection to 'ssh://nix-docker': cannot connect to 'nix-docker'; trying other available machines.." shows up in the logs
<clever>
ah, then its not configured well
<iqubic>
Does wipefs wipe all the data on the selected partition immediently?
<ldlework>
clever hmm, I overwrote the derivation's builder with builder = builtins.toFile "builder.sh" '' .... ''; but it just did the ol configure/make/make install dance...
<clever>
iqubic: i think so
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] aszlig pushed 3 new commits to master: https://git.io/vFgYq
<NixOS_GitHub>
nixpkgs/master 35f78e8 aszlig: paperwork-backend: Add myself to maintainers...
<NixOS_GitHub>
nixpkgs/master 8c5a3d8 aszlig: paperwork-backend: Move into its own file...
NixOS_GitHub has left #nixos []
<iqubic>
SHIT
<clever>
ldlework: i dont think nix-shell supports builder
phreedom has joined #nixos
<ldlework>
I see..
phreedom has quit [(Client Quit)]
<iqubic>
Well, /dev/sda4 was my windows and linux shared partition, which had data on it.
* hyper_ch
guesses someone wiped the whole disk instead of partition
<clever>
ldlework: its best to instead set the right phases
<ldlework>
OK
<hyper_ch>
I hope you have backups
<iqubic>
hyper_ch: No, I only wiped the one partition.
<iqubic>
hyper_ch: I don't
<hyper_ch>
oh... always have backups
<iqubic>
So there goes 2 school assignments I had already turned in electronically.
<srhb>
:(
<iqubic>
That's literally all the data I had on that partition.
<hyper_ch>
there's unformat tools for windows... not sure if those exist for linux as well
<infinisil>
iqubic: what the hell dude
<srhb>
Well, okay.
<clever>
2017-11-09 14:07:21 < iqubic> So I only need to take /dev/sda4 and partition that.
<clever>
2017-11-09 14:06:24 < iqubic> Currently the way I have it set up is /dev/sda1 is /boot. /dev/sda2 is Windows FS. /dev/sda3 is a 20GB partition that I can Read and Write to from both OSes and /dev/sda4 is going
<hyper_ch>
they can still try to read the actual data from the drive
<srhb>
At least already turned in!
<iqubic>
clever: I think I may have been wrong in saying that.
<infinisil>
and wipefs doesn't actually delete data, it just deletes the signatures such as where filesystems are
takle has quit [(Remote host closed the connection)]
<hyper_ch>
so some unformat tool probably can still restore the data
<infinisil>
Oh, apparently I'm lagging behind on irc, about 20 messages just flooded at once
jtojnar has joined #nixos
<iqubic>
One: wipefs did nothing.
<infinisil>
wipefs without any arguments does nothing
<iqubic>
infinisil: I ran "wipefs /dev/sda4"
<infinisil>
wipefs -a is the bad stuff
<infinisil>
well then nothing should've happened
reinzelmann has joined #nixos
<infinisil>
and please read what a command does before you run it
<infinisil>
`man wipefs`
<iqubic>
Two: the Nixos installer can't find the free partition on my hard drive.
<infinisil>
especially when handling disks and on root
<gchristensen>
#nixos is incredibly forgiving and willing to help people, you've been warned & kicked out about not _trying_ on your own. this is a warning: don't abuse #nixos.
<ldlework>
clever I've changed it to buildPhase = '' ... ''; but it just says, "no Makefile, doing nothing"
<iqubic>
Why is it that MBR can only have 4 primary partitions?
<BlessJah>
iqubic: MBR can only have 4 primary because historical reasons
<iqubic>
I see.
<iqubic>
That is really the best way to explain most stuff in computing.
<srhb>
Also bits.
<BlessJah>
One of them can be extended partition, inside extended you can have logical partitions, trick to have more than 4.
<srhb>
There are (were) only so many bits.
<iqubic>
Ah.
<clever>
originally, the bootloader, and partition table, had to live inside a 512 byte sector
<clever>
and the entire table was assigned bytes staticly
<clever>
so it just always had 4 slots
noam__ has quit [(Read error: No route to host)]
<clever>
and the bios had no notion of how to understand partitions, it just blindly loaded that 512 bytes into ram, and executed it
<iqubic>
Do I want to set up my zfs partitions with the mount=legacy option?
<srhb>
Yes.
<iqubic>
Why is that?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] c0bw3b pushed 2 new commits to master: https://git.io/vFg35
<NixOS_GitHub>
nixpkgs/master acd8172 Marius Bergmann: unifi: 5.6.19 -> 5.6.20
<NixOS_GitHub>
nixpkgs/master aec3326 Renaud: Merge pull request #31387 from mbrgm/upgrade-unifi...
NixOS_GitHub has left #nixos []
<srhb>
iqubic: Because that will make them behave like all other fstabby partitions, which can be dealt with in an ordinary fashion.
<iqubic>
I see.
<iqubic>
I like that.
<srhb>
I also think the alternative breaks NixOS :P
<srhb>
Because things just suddenly are mounted -- what!
<infinisil>
unlike every zfs nixos tutorial online, you can just set mountpoint=legacy on the root dataset, so you don't have to set it for every single one you create afterwards
<srhb>
Does anyone have an idea why I'm suddenly not seeing a login screen on nixos tests?
<srhb>
It's just black, after bootup.
<srhb>
I mean, when running them interactively.
<srhb>
They start out with lots of text scrolling in the qemu window though
<iqubic>
Is it possible to delete a thing made with "zfs create"?
<iqubic>
I accidetally created POOL_NAME/nixos instead of POOL_NAME/nix
<adelbertc>
needs entry in `/etc/nix/machines` (since that's the default it looks in, without mucking about with daemon config), and then you need to copy the public key into the known_hosts
<BlessJah>
gchristensen: what stops you? is code OS?
<adelbertc>
is there somewhere i can submit a PR to update/add this to docs?
Sonarpulse has joined #nixos
<BlessJah>
it's one line with git
<BlessJah>
damn it, 17.09.2029 iso fails to start login service
ryanartecona has joined #nixos
ryanartecona has quit [(Read error: Connection reset by peer)]
<srhb>
What do i need to do to set up a nixos-test I can feed arguments to at the command-line? I tried { foo, ... }: import <nixpkgs/nixos/tests/make-test.nix> { ... } but that gives me "expression does not evaluate to a derivation (or a set or list of those)"
<srhb>
Oh, callPackage maybe.
ryanartecona has joined #nixos
<srhb>
HAH. New error message I've never seen before. overrideDerivation not yet supported for functors.
<srhb>
I didn't even know we had functors. :-)
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] aszlig pushed 1 new commit to master: https://git.io/vFgn1
<NixOS_GitHub>
nixpkgs/master 0e790b9 aszlig: nixos/atlassian: Fix eval error in jrePackage...
NixOS_GitHub has left #nixos []
pxc has quit [(Ping timeout: 240 seconds)]
ma27 has quit [(Ping timeout: 240 seconds)]
<BlessJah>
clever: this one fixed wireshark: nix-env -e kodi-17.4 claws-mail obs-studio-20.1.0
romildo has joined #nixos
<BlessJah>
clever: VBox fails with Qt FATAL: This application failed to start because it could not find or load the Qt platform plugin "xcb"
<srhb>
I feel like import <nixpkgs/nixos/tests/make-test.nix ( { arg1, arg2, ... }: { ... } ) should be right, but I can't pass the arguments on the commandline then.
erictapen has quit [(Ping timeout: 248 seconds)]
ssmike has quit [(Ping timeout: 240 seconds)]
Henson has joined #nixos
ryanartecona has quit [(Quit: ryanartecona)]
<Henson>
Hi everyone, I'm trying to use the Nix package manager as a tool for making reproducible build environments. I'm trying to install OpenCV and have discovered that it's only installing libraries and not headers. This seems to be related to the "out" or "dev" output that is being installed. But even when doing something like "nix-env -iA nixpkgs.opencv.dev", it doesn't install the headers...
<Henson>
the headers are there in /nix/store, but aren't being symlinked into the default environment. Any suggestions?
<Henson>
I'm probably missing something simple, as I just started using the Nix package manager today, but this has been a very difficult problem to find the solution to using Google.
<srhb>
Henson: In general you don't want to install _anything_ for reproducible environments.
orivej has joined #nixos
<Henson>
srhb: oh, so maybe I'm using this in the wrong way. A friend of mine said the Nix package manager is great at making reproducible build environments, so I'm trying to figure out how to do it.
<srhb>
Henson: generally we tend to make disposable shells that we can enter and leave at will
ssmike has joined #nixos
<srhb>
Henson: And the packages for that specific environment only exist in that shell.
<ldlework>
I'm guessing that's what they mean by reproducable build environments....
<Henson>
srhb: ok, that sounds like what I want. Do I use "nix-shell -p opencv etc..." for that?
<ldlework>
Henson I'm working on the same problem but I'm finding that nix-shell is not very pure
<srhb>
Henson: That's a start! Check out env after doing that.
<eacameron>
Is it possible to tell nixops which ssh keys to use when connecting to a none backend?
<Henson>
srhb: will I still have the same missing headers problem in the shell?
<srhb>
Henson: More likely you'll want to write down some shell.nix files for various projects that instantiate complete environments for you. nix-shell will autoload the shell.nix of the current directory
<srhb>
Henson: They won't be symlinked anywhere, if that's what you mean.
<srhb>
Henson: Your builds should take their real path in the nix store into account.
thblt has joined #nixos
<srhb>
Henson: Perhaps we can help you set up things for your project, if you give us more details. :) (unfortunately I'm a bit low on time myself)
<Henson>
srhb: I just need to get set on the right track, and I'll be able to figure out a lot of stuff myself. So for the nix-shell environment, I shouldn't expect the OpenCV dev stuff to be put into a nice filesystem location for me, I should be working directly with things in the /nix/store directory?
<srhb>
Henson: Yes, and the normal builders tend to set up include paths and such for you nicely.
<srhb>
Henson: (via environment variables)
<srhb>
Henson: We generally don't set up a simulated FHS-like environment (except as a last ditch effort)
<Henson>
srhb: what does FHS-like mean?
<srhb>
Henson: The usual /usr/lib /usr/include shenanigans
<Henson>
srhb: ah, ok
<srhb>
Is this a python project? Maybe I can give you a good document to get started with.
<srhb>
(We tend to have language specific helpers)
<Henson>
srhb: so I shouldn't even use "nix-env -i" to install anything when using nix-shell? nix-shell will download the appropriate stuff to the /nix/store directory and instantiate a throwaway isolated environment for me?
<srhb>
Henson: Yes, that's right.
<srhb>
Henson: Once the shell is shut down, its dependencies will be garbage collected whenever the store is garbage collected.
romildo has quit [(Quit: Leaving)]
<Henson>
srhb: ok, I'll give this a try then. Thanks for your help!
<srhb>
(isolated implies --pure though, so your normal user and system environment doesn't pollute the shell)
<srhb>
Henson: Good luck and have fun :)
<iqubic>
So if i create my initial zpool with -O compression=lz4 I don't need to restate that for all children, right?
<srhb>
Right.
<srhb>
You can even probe the effective options of the children at any point
<srhb>
zfs get all
<srhb>
or zfs get pool/thing
<iqubic>
I really need a zfs tutorial
<srhb>
Oh wait, that's not how it works
<srhb>
get all will work
<srhb>
grep your way to goodness
<srhb>
iqubic: It's too much, cargo culting is the way to go. ;-)
* srhb
hides
<Henson>
srhb: wow, thank you for your advice. I just compiled a program using OpenCV I've never been able to compile before due to library problems!
<srhb>
Henson: :o !
ma27 has joined #nixos
<Henson>
iqubic: the datasets you create in a pool with compression enalbed should inherit the compression setting
<srhb>
Hey, more zfs experts. You should stick around, it's in great demand these days ;-)
<Henson>
srhb: ZFS is amazing, I've been running it for several years under zfs-on-linux, but am not very happy with its performance. So now I'm converting my servers over to FreeBSD, where ZFS is very, very nice
<srhb>
Guess we need a NixBSD... :-)
<ldlework>
clever I now have this, https://gist.github.com/dustinlacewell-wk/b9cd09370c1c60157099a93f8c11d36a but when I run $glidePhase, I always get the help for git config, its complaining about the args passed to it. But when I echo $glidePhase and then copy those commands into the shell they work fine.
<ldlework>
I think I'm confused about how $glidePhase is interpreted.
ylwghst has quit [(Remote host closed the connection)]
<iqubic>
How can I upload an image for you all to view?
<ldlework>
lol
<clever>
ldlework: the real hack, is that ssh-agent detects the "wrong" user (nixbld1) is connecting to your agent, and kicks them out for security reasons
<clever>
ldlework: so socat has to enter the mix and act as a proxy running as root
<ldlework>
jeeze...
ryanartecona has joined #nixos
civodul has joined #nixos
<clever>
ldlework: so now ssh-agent can only detect socat, running as root, and it trusts root
<clever>
then anybody that can access /tmp/hax (its best to chmod it right) can use the agent
<iqubic>
/dev/sda2 is a partition that windows created and is used by windows.
<clever>
that says sda2 is now formated with zfs
<ldlework>
clever if I simply have a secret in my environment, since that environment is inherited by nix-shell builder, does the secret ever get put in the store?
<ldlework>
assuming I never do something like foo = getEnv "foo"
<clever>
ldlework: if its just an env variable, it should be safe
<iqubic>
clever: I know it say that it's formatted with zfs.
<iqubic>
That's an issue.
<ldlework>
clever I feel like there should be a way to pass environment variables into the nix-build/nix-shell the same way
<clever>
iqubic: what if you instead try "blkid /dev/sda /dev/sda*" ?
<iqubic>
I never told my machine to format /dev/sda2 because I know that it is being used by windows.
<iqubic>
clever: I can try that
<clever>
ldlework: the would point of nix and nix-build, is that every env variable gets hashed, and any changes to them re-start the build
orivej has quit [(Ping timeout: 240 seconds)]
<clever>
whole*
<ldlework>
I mean, you're obviously also trying to get aroudn this problem
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<Henson>
srhb: is there a way to tell nix-shell to use a specific version of a package? I don't understand where the package list comes from, and for OpenCV for example, I opencv-2.4.13 as the only specific version I can install. But presumably that will eventually change to a higher version. But if I want to install that specific version some time in the future and use it in a shell, how do I do that?
<clever>
ldlework: fetchgitprivate is special, because its a fixed-output derivation
<iqubic>
Do you think my dual boot system will still work properly?
<clever>
ldlework: this allows certain variables to leak from the nix-build env down to the sandbox, without including them in the store or hashing
<iqubic>
Or have I somehow screwed over my windows partition?
<clever>
iqubic: only way to answer that is to try to boot
<ldlework>
that seems ideal yeah
<iqubic>
First I need to finish getting nixos installed.
orivej has joined #nixos
reinzelmann has quit [(Quit: Leaving)]
orivej has quit [(Read error: Connection reset by peer)]
orivej has joined #nixos
<iqubic>
Where do I put my hostID for zfs?
<clever>
in configuration.nix
<clever>
check the configuration.nix man page
sigmundv__ has quit [(Ping timeout: 248 seconds)]
proteus-guy has quit [(Ping timeout: 268 seconds)]
<BlessJah>
.3
<clever>
.4
proteus-guy has joined #nixos
<michaelpj>
does anyone know if there's an easy way to launch steam games from the console to make it easier to debug issues? Basically I want to drop into the environment that the steam app runs in
<iqubic>
What can I do if I have screwed up my windows partitions?
<clever>
michaelpj: i think steam-run is involved in that
<clever>
iqubic: restore from backup?
<michaelpj>
clever: yep, that works, thanks
<iqubic>
clever: I don't have a windows backup.
<iqubic>
So that's good.
<clever>
bbl
kuznero has joined #nixos
<kuznero>
Hi All, how does `newScope` work?
<ldlework>
clever thanks for all the help
<kuznero>
It is defined as `newScope = scope: newScope (self // scope);`... I don't understand that...
<kuznero>
Can somebody please explain?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] pbogdan opened pull request #31455: gcj: fix build with glibc 2.26 (master...gcj-fix-build) https://git.io/vFggL
NixOS_GitHub has left #nixos []
orivej has quit [(Ping timeout: 240 seconds)]
phanimahesh has left #nixos []
srdqty has quit [(Quit: WeeChat 1.9.1)]
<lejonet>
How do I do if I want to use ${} for bash variables inside '' '' strings? skip the curly brackets?
<symphorien>
the escaped sequence is ''${ iirc
<lejonet>
so I could do like derp=''${bashvar} ?
<symphorien>
yes
<lejonet>
Will try that, thanks
mahalel_ has joined #nixos
<tilpner>
michaelpj - IME the environment inside steam-run isn't quite the same as what Steam actually launches the games with. If steam-run is close enough, great! If it isn't... I registered my terminal emulator as a game in steam, then launch it from there, to get the exact same environment
<michaelpj>
tilpner: that's a great idea, thanks
pxc has joined #nixos
<tilpner>
Unfortunately you will have to close Steam after every session, because it won't let you run another game after your terminal session has ended, because it never gets notified that the terminal closed. Not sure why
<ldlework>
When I'm building a derivation is $src readonly?
<symphorien>
I don't think so
thblt has quit [(Ping timeout: 258 seconds)]
akapav_ has quit [(Quit: ZNC 1.6.5 - http://znc.in)]
akapav has joined #nixos
akapav has quit [(Client Quit)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vcunat pushed 4 new commits to master: https://git.io/vFgad
<NixOS_GitHub>
nixpkgs/master e26c52e Piotr Bogdan: libzip: 1.2.0 -> 1.3.0
<NixOS_GitHub>
nixpkgs/master 4173ec1 Piotr Bogdan: hhvm: fix build with glibc 2.26
<NixOS_GitHub>
nixpkgs/master fb00ed8 Vladimír Čunát: Merge #31449: hhvm: fix build with glibc-2.26
NixOS_GitHub has left #nixos []
<clever>
ldlework: it is
<clever>
ldlework: the default unpack phase copies $src to . and unpacks it
<guillaum3>
I'm just watching the video about the future features of nix 1.12. nix run --store local/path/foo appears to be a really nice way of packaging something and distributing it to a system where nix cannot be installed with root access. Does this feature needs userspace chroot or something similar?
<timokau>
Is here anybody that knows how npm and its integration with nix works? I have a package that is not installable with a simple `npm install`, but still needs dependencies from npm and is build using npm.
<timokau>
Is here anybody that knows how npm and its integration with nix works? I have a package that is not installable with a simple `npm install`, but still needs dependencies from npm and is build using npm.
<timokau>
Is here anybody that knows how npm and its integration with nix works? I have a package that is not installable with a simple `npm install`, but still needs dependencies from npm and is build using npm.
<timokau>
Sorry for the spam, was a client issue
<hyper_ch>
I always blame my kitty...
hiratara has quit [(Ping timeout: 258 seconds)]
ssmike has joined #nixos
digitus has quit [(Quit: digitus)]
markus1189 has quit [(Ping timeout: 264 seconds)]
hiratara has joined #nixos
ssmike has quit [(Ping timeout: 264 seconds)]
jb55 has joined #nixos
<Orbstheorem>
A couple of weeks ago I debootstrapped a server with the latest version of debian. I setup dropbear on the initramfs and luks+btfrs+custom btrfs subvolumes and mounting points (ie `/var/machines/hostname/subvolumes/.active/rootfs` is mounted on `/` by the initramfs; and similar for /var, /home, /Storage and a sepadafe ext4 partition for postgres on /Data. A couple of weeks after I switched from debian to nixos in my personal computer (where
<Orbstheorem>
I have a similar luks+btrfs subvolumes for mounting points layout). I installed nixos in a separate harddrive alongside with debian but use the primary's drive /home and /Storage. Declaring my filesystem layout in my configuration.nix seemed at first a little bit odd, but now I kind of really like the idea and I'm wondering if there any way to automate the boilerplate (that is create a partition table, luksFormat a given partition,
<Orbstheorem>
mkfs.btrfs, btrfs subvolume create, mkfs.ext4 on the second partition, ...). I know there's not enough information in (at least in my own) configuration.nix, but such tool would win a couple of hours of painfull procedure on each new host. (I'm thinking on installing nixos on ,y odroid-xu3 (arm) which I use for monitoring my monitoring, wifi router and git-annex storage).
kuznero has quit [(Quit: WeeChat 1.9.1)]
<samae>
Same feeling here
<samae>
I'd be really into a tool that can take a declarative description of a machine and set it up accordingly
<copumpkin>
aszlig had nixos-assimilate, and there have been other less ambitious attempts
<copumpkin>
nixos-infect, nixos-lustrate?
<copumpkin>
obadz: ?
<clever>
there is also my kexec tricks
Lisanna has joined #nixos
hexamod has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<copumpkin>
a thousand different infectors
<copumpkin>
we need 1001 to make the other 1000 irrelevant :D
radvendii has quit [(Ping timeout: 240 seconds)]
python476 has quit [(Ping timeout: 264 seconds)]
<LnL>
I've used clever's thing because I didn't want to search for a usb stick :)
<Orbstheorem>
I know we can kexec to a kernel and initrd in ram, woukdn't that be better than copying a bootloader on disk?
<Orbstheorem>
s/better/more reliable
<clever>
the rescue method in grub is more for repairing nixos later on
<clever>
you can boot back into an installer like env, and do whatever repair you want
<obadz>
copumpkin: am here
<sphalerite>
Orbstheorem: one thing to note with the odroid: unless it's armv8, no official binary cache.
patrl has joined #nixos
<sphalerite>
Orbstheorem: I run nixos on my armv7 Chromebook, and large updates (e.g. if glibc is touched) can take ~2 days to build
<obadz>
sphalerite: you're a saint for building
<sphalerite>
obadz: ??
<obadz>
I wouldn't be brave enough to use on a platform without cache
<sphalerite>
Oh right. I used Dezgeg's binary cache briefly but it doesn't contain all the software I need and I decided to switch to 17.09 from unstable to make the rebuilds less frequent
<Orbstheorem>
> Orbstheorem: I run nixos on my armv7 Chromebook, and large updates (e.g. if glibc is touched) can take ~2 days to build
<Orbstheorem>
Don't mind :)
<sphalerite>
And Dezgeg doesn't have 17.09 binaries at all
TweyII has quit [(Ping timeout: 240 seconds)]
<obadz>
wasn't there going to be an armv7/8 build farm at some point?
<sphalerite>
I'm planning to try and get my old phone running builds too though, which should improve it a fair bit
<sphalerite>
There are armv8 builders in the hydra build farm, thanks to gchristensen and packet.net iirc
<sphalerite>
They don't build for 32-bit arm though
<Orbstheorem>
I wouldn't mind hosting a mirror for armv7/nixos 17.09
<obadz>
sphalerite: be sure to write a blog post if you turn your phone into a build farm :-D
<sphalerite>
I'd gladly contribute builds to it. Would be nice to coordinate with Dezgeg as well
<sphalerite>
obadz: I'm already planning to do a blog post on getting it running on the Chromebook, since that was far from trivial in itself. As for the phone, that's still a way off
<WilliamHamilton>
hi, I'd like to ask a question relative to binaries in nixos: I'd like to run a binary, and so I patched it with `patchelf` as the internet suggested, then I studied the output of `ldd` to find all the needed runtime dependencies, and I'm invoking the program with `LD_PATH_LIBRARY=... executable`
simukis has quit [(Quit: simukis)]
<sphalerite>
WilliamHamilton: LD_LIBRARY_PATH
<sphalerite>
:)
<WilliamHamilton>
however, I get the error: `Unable to dynamically load libGL: Could not load EGL library` even if a directory containing it is in `LD_LIBRARY_PATH`
<schoppenhauer>
hi. is there a command coqWithPackages?
<WilliamHamilton>
Linus: thanks, I misspelled the name here but I wrote it correctly in my bash invocation
<clever>
WilliamHamilton: 32bit or 64bit?
<sphalerite>
WilliamHamilton: do you entirely replace the existing value of the variable? Try adding :$LD_LIBRARY_PATH at the end
<WilliamHamilton>
clever: my system is 64bit, and so is the executable: the output of the `file` command begins with ` ELF 64-bit LSB executable, x86-64,...`
<schoppenhauer>
well, apparently not, but is there something similar, some possibility to install coq with packages?
<clever>
WilliamHamilton: and what directory are you pointing LD_LIBRARY_PATH at?
vcunat has quit [(Ping timeout: 240 seconds)]
<WilliamHamilton>
Linus: I already do that; here is my export statement `LD_LIBRARY_PATH=/nix/store/7zffgqpas0hicfvxsiivs2z6bhr3i7la-alsa-lib-1.1.4.1/lib/:/nix/store/06l1hyd4m1v91y13h7gbl8qz5q72dzlw-gcc-6.4.0-lib/lib/:/nix/store/hb2vv82p8adanckksv60y49n7v728a5f-nvidia-x11-340.102-4.9.59/lib/:$LD_LIBRARY_PATH`
<WilliamHamilton>
by himself, `$LD_LIBRARY_PATH` would point to
hiratara has joined #nixos
<WilliamHamilton>
`/run/opengl-driver/lib`, which is in fact the folder containing the GL and EGL library that I would expect my executable to use
<WilliamHamilton>
however I tried with the similar invocation
<WilliamHamilton>
` strace -e open ./start`
<sphalerite>
Add -f
<WilliamHamilton>
where ./start is my starting script where I set variables etc, and I couldn't get any valuable information (but that's probably because I don't know how to use it)
<WilliamHamilton>
Linus: thanks, trying that
<sphalerite>
I don't know why strace doesn't default to following forks. It's almost always what I want
<clever>
it also helps to use "-ff -o logfile"
<clever>
then the threads wont interleave in the logs
<WilliamHamilton>
Linus: well, a lot more output came this time, what should I search for?
<clever>
usually the last page or 2
<clever>
once it fails, it will print an error and stop
<WilliamHamilton>
ok, I might have located the cause of the problem: in the strace output:
<WilliamHamilton>
`[pid 2691] open("/run/opengl-driver/lib/tls/libEGL.so.1", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)`
orivej has joined #nixos
<clever>
does it keep trying that filename in other locations?
<WilliamHamilton>
and in fact in that directory I only have `libEGL.so` and `libEGL.so.2`
markus1189 has quit [(Ping timeout: 240 seconds)]
<WilliamHamilton>
yes, it searches for that file in all the locations in my `LD_LIBRARY_PATH`
<clever>
what does ldd say?
jsgrant_ has quit [(Remote host closed the connection)]
<WilliamHamilton>
clever: so, a little preliminary problem: can I invoke `ldd <EXECUTABLE>` while prefixing a modification to `LD_LIBRARY_PATH`? Like in:
<WilliamHamilton>
if that invocation is right, ldd doesn't complain about any missing library
<clever>
does it say its looking for libEGL?
<WilliamHamilton>
and that's the point: I took care of the libraries marked as not found by `ldd` extending my `LD_LIBRARY_PATH`, but `libEGL` is not among them
<WilliamHamilton>
a thing that might be worthwhile to notice is that
[0x4A6F] has joined #nixos
markus1189 has joined #nixos
<sphalerite>
As the error message says, it tries to load libEGL dynamically
<sphalerite>
So it's normal for ldd not to detect it
<clever>
dlopen() isnt tested by ldd
<hyper_ch>
Orbstheorem: I use native encrypted root zfs :)
ylwghst has joined #nixos
<WilliamHamilton>
Linus: would it be possible to just do a symlink of `libEGL.so.0` to `libEGL.so.1`?
patrl has quit [(Quit: WeeChat 1.9.1)]
<WilliamHamilton>
btw, as I was saying before, it might be useful to know that I'm using nvidia340 video drivers
<sphalerite>
WilliamHamilton: certainly possible, but it probably wouldn't help. The reason shared objects have those numbers at the end of their names is to indicate version compatibility
<sphalerite>
I.e. libEGL.so.2 exposes a different binary interface from libEGL.so.1
Shados has joined #nixos
<sphalerite>
That means that the program you're trying to run will very likely crash if you do that
<sphalerite>
Of course, you can still try!
cement has quit [(Ping timeout: 248 seconds)]
<sphalerite>
Oh right, and since /run/opengl-drivers is a store symlink iirc it would probably be easier to create a directory in /tmp or whatever and put the symlink there, and add it to the LD_LIBRARY_PATH
<WilliamHamilton>
Linus: no, you're right: better idea, using `du -a /nix/store | grep libEGL.so.1` to locate a better `libEGL.so.1`
<WilliamHamilton>
Linus: side question: when I grep the nix store like this, I find a lot of libraries with the same name
<WilliamHamilton>
coming from different packages
<clever>
WilliamHamilton: some of those are buildEnv's merging many together
<WilliamHamilton>
is there the reasonable expectation that every one of these `libEGL.so.1` is equivalent in some sense, so that I can choose at random?
<clever>
WilliamHamilton: and thats also going to find a lot of old versions that may cause more problems
<clever>
WilliamHamilton: `nix-store -qR /run/current-system` will return all of the storepaths involved in the current nixos, which should help reduce the search space
<WilliamHamilton>
clever: exactly, that was my concern, but I somewhat hoped that being the same version number they would be the same
<WilliamHamilton>
that's a great tip!
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] LnL7 pushed 1 new commit to master: https://git.io/vFgyk
<NixOS_GitHub>
nixpkgs/master 63bac2c Daiderd Jordan: cf-private: overwrite headers from CF
NixOS_GitHub has left #nixos []
<WilliamHamilton>
clever: oh, that's for the packages, so I'd still to crossreference if I'm searching for a lib, right?
<sphalerite>
Also the libraries that are accessed via /run/opengl-drivers are specific to a given graphics driver.
tmaekawa has joined #nixos
<WilliamHamilton>
Linus: so if I'm using nvidia's graphic driver I should really use only the libEGL version from there?
<clever>
yeah
hexamod has joined #nixos
hexamod has quit [(Client Quit)]
<WilliamHamilton>
clever, so a version of `libEGL.so.1` from a `mesa` folder in my `/run/currentSystem` would'nt work if I'm using `nvidia` drivers?
<sphalerite>
^
<clever>
that would probably do CPU rendering
<sphalerite>
Yeah probably not.
<sphalerite>
Oh yeah
<sphalerite>
Or maybe render on onboard Intel graphics? Depending on your hardware setup
<boxofrox>
I'm trying to create a nix package for my rust project on NixOS 17.03 and the rustRegistry package is missing the latest versions of packages I use. Is a `nix-channel update` required, or is there another way to resolve the missing crates? I intend to eventually install this software on a NixOS 16.09 machine (the machine fails to boot 17.03).
<sphalerite>
I don't know enough about how this stuff works to say for sure
kai_w has joined #nixos
<sphalerite>
boxofrox: you should be using 17.09! 16.09 has many security issues by now
civodul has quit [(Quit: ERC (IRC client for Emacs 25.3.1))]
<sphalerite>
And 17.03 also isn't getting security updates anymore
ssmike has joined #nixos
<sphalerite>
We'd be glad to try and help you fix the boot for 17.09
roosemberth has joined #nixos
<kai_w>
I'm trying to improve a service module (kerberos.nix), and in the process have split it into kerberos/default.nix and kerberos/mit.nix.
<kai_w>
but I get error: getting status of ‘/home/scratch/kai/nixpkgs/nixos/modules/services/system/kerberos.nix’: No such file or directory
<kai_w>
how do I make nix aware of the rename (if that is the issue)?
<WilliamHamilton>
Linus: clever thanks for all the help guys, I'll try some symlinking and see if I can get things running!
ryanartecona has quit [(Quit: ryanartecona)]
roosemberth has quit [(Client Quit)]
hexamod has joined #nixos
hexamod has quit [(Client Quit)]
datakurre has quit [(Remote host closed the connection)]
c0bw3b has left #nixos []
datakurre has joined #nixos
<Orbstheorem>
hyper_ch9: I've been asking for native btrfs encryption for ages :( I guess that one of the features I use the most not available (to my knowledge) in zfs is btrfs-send/receive. I use snapper in all my systems and everyday I send over a snapshot to my backup server. (https://github.com/roosemberth/dotfiles/blob/master/.local/bin/backup-snapper.sh)
<sphalerite>
kai_w: adjust nixos/modules/module-list.nix to reflect the split and it should be fine
<sphalerite>
Orbstheorem: zfs send and zfs receive are a thing
<Orbstheorem>
:o
hexamod has joined #nixos
<boxofrox>
sphalerite: 16.09 is running on a server at a client with no remote console access. I didn't have enough time back in March to sort out the issues booting 17.03, so it's stuck on 16.09. I can't fiddle with it.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] sifmelcara opened pull request #31457: rstudio: fix build error about missing qt plugin (master...fix/rstudio-qt-build) https://git.io/vFgS7
NixOS_GitHub has left #nixos []
ylwghst has quit [(Quit: Lost terminal)]
ma27 has quit [(Ping timeout: 246 seconds)]
<euniarte>
clever: You spoke about your QtCreator workaround yesterday and mentioned having to use a qmake from /run/user (iirc).
<euniarte>
I got mine working by just launching qtcreator from a nix-shell with Qt available. This did even work for clion, too.
<clever>
strange, mine wasnt able to find any library when i did that
<clever>
i had to "type qmake" inside the nix-shell, then point creator to that binary
<slack1256>
Should "nix-shell -p gmp --command "gcc code.c -lgmp" " work on the get go?
<sphalerite>
boxofrox: that server is quite a liability. According to a quick browse through the git logs there are some 190 CVEs that have been fixed in 17.09 but not in 16.09
<euniarte>
clever: It's probably because I am using CMake instead of qmake, but QtCreator refuses to load my project if it doesn't find any toolkits. So QtCreator doesn't ignore the distribution and let CMake do its thing.
<clever>
slack1256: you may also need "-p gmp gcc"
<sphalerite>
Anyway I need to sleep. Gnight folks
<slack1256>
clever: OK, that is why my usual gcc wasn't picking the gmp library
<clever>
slack1256: the gcc you installed with nix-env/systemPackages is broken and wont be able to find any libraries
<clever>
it only works when loaded properly by nix-shell or nix-build
<slack1256>
got it
<slack1256>
what about of libraries that use pkgconfig as gtk?
<clever>
pkgconfig will also need to be added to the -p list
<slack1256>
oh right, I need the correct incarnation with pkgconfig
<clever>
yeah
<slack1256>
thanks
<clever>
and you need to escape the $( so it runs inside the shell, not outside
<clever>
at that point, your better off writing a nix expression for it
pie_ has quit [(Read error: Connection reset by peer)]
pie_ has joined #nixos
euniarte has quit [(Remote host closed the connection)]
drock2017 has joined #nixos
<kai_w>
that was it, thanks sphalerite
srdqty has quit [(Quit: WeeChat 1.9.1)]
<drock2017>
hi nixos-gurus. Q: i have a folder full of custom configs needed for a particular package. i can use a standard package, but how do i move the folder to the nixos target. currently just making mods to configuration.nix, but will be moving to nixops for deployment and remote update scenarios. Still I'm missing the trick to move folders then references them in .nix files.
<neonfuz>
cd etc
<neonfuz>
oops
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to master: https://git.io/vFgQ1
<NixOS_GitHub>
nixpkgs/master c910a08 Tuomas Tynkkynen: diffoscope: 86 -> 87
NixOS_GitHub has left #nixos []
Lisanna has quit [(Ping timeout: 260 seconds)]
<adelbertc>
has anyone hit a "Could not access KVM kernel module: Permission denied" erorr when trying to build a Docker image?
<adelbertc>
on Linux
<tilpner>
Do you want to load it, or asking why it needs to be loaded?
kai_w has quit [(Quit: WeeChat 1.9.1)]
<adelbertc>
tilpner: is that directed at me?
<tilpner>
Yes
<adelbertc>
afaict its part of building the image
drock2017 has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<ldlework>
Is haskell like Nix in that all you're ever doing is building up a description of some data structure, attrset, or list or whatever?
<dmj`>
They’re pretty different, both are lazy. Nix being dynamic and strong, haskell being static and strong.
<averell>
nix functions take exactly one arg, and i think the syntax is often at least heavily influenced
<ldlework>
I guess what I mean though, are they similar at least in the sense that the goal is building up some expression structure?
<ldlework>
I want to call them "functional modeling languages"
<ldlework>
in my talk
<averell>
i think a packaging DSL is not the same ballpark as a full-fledged programming language
<ldlework>
but I dunno if that's right. Nix kinda feels like a modeling language, if I understand the definition correctly
<ldlework>
Well certainly its more than a mere DSL
<ldlework>
But the goal of the language isn't so much to do anything, rather than describe what to do right?
<ldlework>
Like every nix "program" is really just an expression which resolves lazy data structure right?
erasmas has quit [(Quit: leaving)]
<ldlework>
(I'm mostly trying to understand, not say what's what :)
<averell>
i'm only a noob too, but from what i can tell nix expressions compile down to derivations which are basically json like descriptions of packages
<averell>
like inputs/outputs and such. i'm not even sure it's turing complete