jb551 has quit [Remote host closed the connection]
noudle has joined #nixos
<{^_^}>
[nixpkgs] @teto opened pull request #69400 → zfsStable: don't throw with incompatible kernels → https://git.io/JeGsL
jaeckel has joined #nixos
growpotkin has joined #nixos
alanz has joined #nixos
<ddima>
do we have some sort of own repo somewhere for 3rd party archives with unreliable sources/retention?
iqubic` has joined #nixos
iqubic` has quit [Client Quit]
iqubic has quit [Ping timeout: 245 seconds]
nikivi has joined #nixos
iqubic has joined #nixos
palo1 has joined #nixos
palo has quit [Ping timeout: 240 seconds]
palo1 is now known as palo
<evils>
iqubic: this PR just opened, https://git.io/JeGsL, you may want to try it out
captn3m0- has quit [Ping timeout: 265 seconds]
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Ralith>
ddima: man, I wish
drakonis_ has joined #nixos
kvda has joined #nixos
drakonis has quit [Read error: Connection reset by peer]
nikola_i has quit [Read error: Connection reset by peer]
vonfry has joined #nixos
captn3m0 has joined #nixos
zupo has joined #nixos
drakonis1 has quit [Ping timeout: 264 seconds]
zupo has quit [Client Quit]
zupo has joined #nixos
zupo has quit [Client Quit]
nikola_i has joined #nixos
octe has quit [Ping timeout: 246 seconds]
adamt has joined #nixos
adamt is now known as Guest70734
<danderson>
is there a way to tweak a couple values of a derivation in nixpkgs, without having to copy the entire derivation? Case in point, I'd like to just update the versions/hashes in servers/plex/raw.nix of my nix-stable server, without duplicating the entire recipe.
<LnL>
with the previous release the module also changed, that required a few extra lines to also update that
bendies has joined #nixos
<danderson>
ah, interesting. Well, 1.17.something is out now, hopefully it'll be an easy update
srtsrtsrtsrt has joined #nixos
<bendies>
Hello! I just ran `nixos-rebuild switch` and got a "error: Linux v5.3 is not yet supported by zfsonlinux v0.8.1." - which surprised me since I don't have the string "zfs" anywhere in my nix configs. How can I find out what's causing zfs to be built?
<tetdim>
worldofpeace, thats how far it builds to atm
<rndd>
ikwildrpepper: not really, will try this way. thanks
__Sander__ has joined #nixos
veske3 has quit [Quit: This computer has gone to sleep]
joehh has joined #nixos
<ikwildrpepper>
rndd: that's what works for me
<ikwildrpepper>
just add `export PS1=....` in there
hmpffff has joined #nixos
liberiga has joined #nixos
<rndd>
ok, have another question. In some games I installed through steam there are problems with cpp libs. In arch I simply install all these dependencies and games become playable. Don't really know how to do right in NixIs
<graf_blutwurst>
Hi I was wondering, how long does it generally take for a package to be available on unstable after the PR being merged? I'm anticipating playing around https://github.com/NixOS/nixpkgs/pull/68453
<sevcsik[m]>
hi all, when I run `nixos-rebuild switch`, I get a warning that the new configuration won't take effect until I reboot the system. After I reboot, I still have tne old configuration. Running rebuild again gives me the same message. Just realised that I'm stuck with an old configuration for a month now :) Any ideas where should I look?
<{^_^}>
[nixpkgs] @Mic92 pushed 4 commits to release-19.09: https://git.io/JeGnI
<graf_blutwurst>
sevcsik[m]: did you change anything in your config regarding boot settings / grub etc? I seem to faintly remember I had similar issues when I tried nixos grub to detect my windows install.
ubert has joined #nixos
<sevcsik[m]>
The first edit I remember not activating is removing an entry from fstab. previously I migrated my root fs to a new drive
Neo-- has joined #nixos
<graf_blutwurst>
hmm right. Sorry I don't know what that could be.
Zer0xp has joined #nixos
<sevcsik[m]>
graf_blutwurst I might have screwed up something with my /boot partition, I'll check that out. Thanks for the idea
<{^_^}>
[nixpkgs] @fpletz closed pull request #35364 → nixos: make systemd-nspawn work out of the box on a NixOS root filesystem → https://git.io/vAgbe
psyanticy has joined #nixos
oida has quit [Remote host closed the connection]
oida has joined #nixos
<cinimod```>
Hmmm I change editors to editrs inside examples.nix but nix-build '<nixpkgs>' -A dockerTools.examples.editors still runs???
<musteresel>
Hi ... I'm a bit confused atm: I'm on x86-64 and have a library that needs to be built in 32bit mode. Is there a stdenv or what-not which gives me a C/C++ compiler producing 32bit (linux) executables by default?
judson_ has quit [Ping timeout: 245 seconds]
<LnL>
you can build with --option system i686-linux
<cinimod```>
docker run -it editrs:5a30zbxwc2qnb65kq2z0lkliipw9c3y2 R
<musteresel>
LnL: thanks ... trying atm. Looks as if this causes all dependencies (e.g. CMake) to be fetched again, though, so I'd assume that I've now also a "32 bit CMake"?
<{^_^}>
[nixpkgs] @globin pushed commit from @kmein to release-19.09 « tamarin-prover: mark as broken because upstream is broken »: https://git.io/JeGc7
rawtaz has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @globin pushed 2 commits to release-19.09: https://git.io/JeGcb
<equivrel>
It feels like nix-env --delete-generations is not working for me: I can still see the links in /nix/var/nix/profiles, and I see stuff in my store which is not being gc'd which I think belongs to old generations... am I misunderstanding how it works?
<Shell>
if you want to delete all root's generations.
iyzsong has joined #nixos
<equivrel>
Shell: aaaah, I see! didn't realise. So I guess nix-store --gc needs to be run both as user and as root, to gc both my global packages and the ones I built as user?
pie_ has joined #nixos
<tetdim>
grrrrrrrrrrrrrrrrrrrr goddam config
<tetdim>
i built to the wrong directory and was getting errors
<tetdim>
fixed now
<adamt>
equivrel: The main difference when running as root is that it can also remove old generations, which accidentally also causes gc to remove more packages.
<equivrel>
adamt: ok, so is nix-store --gc equivalent when run as user or as root, or not?
johanot_ has joined #nixos
orivej has quit [Ping timeout: 268 seconds]
veske has joined #nixos
<adamt>
equivrel: Oh, sorry, I misread, I thought you were talking about something like `nix-collect-garbage --delete-older-than 7d`
__Sander__ has quit [Ping timeout: 276 seconds]
<adamt>
(which causes generations to also get cleaned up when run as root, which obviously can't happen when run as a normal user)
johanot has quit [Ping timeout: 240 seconds]
<equivrel>
Oh, I was, I just moved on to my next question :D I realise now that to delete system generations I need to run nix-env --delete-generations (or one of its wrappers) as root
marianobqp[m] has joined #nixos
ilyagotfryd has joined #nixos
johanot_ is now known as johanot
marianobqp[m] has left #nixos [#nixos]
nikivi has quit [Ping timeout: 250 seconds]
alanz has quit [Ping timeout: 245 seconds]
misuzu has quit [Remote host closed the connection]
misuzu has joined #nixos
hmpffff has quit [Quit: nchrrrr…]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
o1lo01ol1o has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<clever>
equivrel: `nix-store --gc` doesnt matter what uset its ran as, and can only delete things without roots
<equivrel>
clever: I see, thanks. that agreed with my intuition
<clever>
equivrel: `nix-collect-garbage --delete-older-than 7d` will delete the roots for profiles it has access to (root can delete from all), then delete the garbage that it exposed
hmpffff has joined #nixos
srtsrtsrtsrt has left #nixos [#nixos]
hyper_ch2 has joined #nixos
iyzsong has quit [Read error: Connection reset by peer]
adamt has quit [Ping timeout: 265 seconds]
adamt has joined #nixos
adamt is now known as Guest7377
chr15 has quit [Quit: WeeChat 2.4]
Zer0xp has quit [Quit: Bye!]
kenshinCH has joined #nixos
iyzsong has joined #nixos
ilyagotfryd has quit [Ping timeout: 245 seconds]
johanot has quit [Ping timeout: 240 seconds]
<evils>
i'm having trouble getting my derivation's binaries in my path, also the man pages aren't being found, how is this supposed to work?
<dminuoso>
Mmm, I have a nixos option that requires an attribute set of paths. How can I generate a file in the store and obtain a nix path to that?
nikivi has joined #nixos
alanz has joined #nixos
<cinimod```>
I am trying to create a docker container using buildImage
<cinimod```>
The image just contains R but when I try to run it I get: Fatal error: cannot create 'R_TempDir'
zupo has joined #nixos
<gchristensen>
has systemd and journalctl changed how people do centralized logging?
<cinimod```>
Has anyone had any success with e.g. python or haskell? Are there any examples I can look at?
zupo has quit [Client Quit]
niso has quit [Ping timeout: 240 seconds]
<cinimod```>
Adding config = { WorkingDir = "/home/sundials"; }; didn't help - I still get the same error
<cinimod```>
niso has joined #nixos
<cinimod```>
Maybe extraCommands = '' mkdir -p tmp ''; will help
<cinimod```>
twoolie has joined #nixos
<twoolie>
hi all, is this an ok channel to ask questions about writing derivations, or is there a better venue?
<Ashy>
twoolie: ask away
<makefu>
gchristensen: last time i checked there was no builtin solution for forwarding logs unfortunately. however it is almost a year ago where i last checked. but i'd love to know how people are doing centralized logging as well
<fpletz>
gchristensen: only how log shipping is done, and journald exposes much more metadata
<fpletz>
gchristensen: graylog2 works particularly well because it stores an arbitrary amount of metadata fields and has its down log shipping format for that, and there's https://github.com/systemd/journal2gelf to ship the logs
<cinimod```>
BTW the nix to docker stuff does *not* work on macos
<gchristensen>
fpletz: oh cool
<Ashy>
there's also journalbeat if you have an elasticsearch cluster to push it all to
<cinimod```>
I guess I ought to report this
nikivi has quit [Client Quit]
<gchristensen>
cinimod```: it works for me as long as I have a linux remote builder configured
alanz has joined #nixos
<makefu>
Ashy: journalbeat is what i used until now
nikivi has joined #nixos
Guest7377 has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to gnome-3.34: https://git.io/JeGWC
veske has quit [Quit: This computer has gone to sleep]
Guest7377 has joined #nixos
<twoolie>
@Ashy I'm trying to package a recent version of 389ds, which comes with a python3 module in a subdirectory. Is there any way to make a sub-derivation that uses the python3Package machinery to build the python lib, and put it at pkgs.389ds.lib? how can i make it do all the right stuff, but in a subdirectory of src?
<gchristensen>
fpletz: say I wanted to setup central logging for all the hydra infra, is graylog the way to go?
mattyu has joined #nixos
cfricke has joined #nixos
zupo has joined #nixos
nexgen has quit [Quit: Leaving]
<mattyu>
hi all - I just upgraded from 18.03 to 19.03 and something is off when using sddm
ilyagotfryd has joined #nixos
<mattyu>
eg. when I open chrome in 19.03 with sddm, the screen doesn't render (all grey) and then I get "unresponsive" message. I rebuilt without sddm and it works fine.
<mattyu>
anyone else seen this?
rndd has joined #nixos
<cinimod```>
gchristensen: thanks - that's way beyond my skill level
Zer0xp has joined #nixos
<cinimod```>
clever: thanks for your help yesterday - I am thinking that rather than trying to copy all the deriviations across to the secure computer, I am better off using a docker image
eisbaer-north has quit [Ping timeout: 265 seconds]
<twoolie>
@Ashy is it possible to do something like 389ds = mkDerivation { src = fetchurl {...} ; lib = buildPythonPackage { inherit src ; setupPhase = "cd src/lib389" }; }; ?
<jg20>
Hi guys, has anyone considered using Nix for running Continuous Integration/Continuous Delivery pipelines? E.g having a derivation that sets up an environment with AWS/Azure credentials, then runs terraform init + terraform apply etc. I guess this would be considered an "impure" derivation as it is nondeterministic. I'm not sure if this idea is abusing Nix or whether it is a legitimate usage but I think it could work quite well as Nix can nicely define
<jg20>
dependencies between derivations and be used for populating variables for different CI/CD pipelines, even if the derivations themselves are "impure"
<exarkun>
Hydra?
<emily>
sounds more like nixops?
<emily>
I know of a few people using Terraform as a replacement for nixops to deploy NixOS machines but that's probably not quite the layering you're thinking of
hyper_ch2 has quit [Remote host closed the connection]
<dminuoso>
Mmm, okay Im not sure how this is supposed to work. Inside nixos, can you create a file such that you get a nix path back?
<dminuoso>
writeText appears to be for derivations
<Ashy>
twoolie: no idea, but someone else in the channel might know, alternatively you might have to go grepping through nixpkgs and see if any other derivations do something similar
<Ashy>
twoolie: i find ripgrep really useful for that
<exarkun>
emily: kinda hard to say without more details, I guess.
cfricke has quit [Quit: WeeChat 2.6]
<jg20>
of each tool available in the PATH), and also would be good to pass variables between different stages
<jg20>
emily: So I've had a look at NixOps, but that seems more like a replacement for Terraform right? I'm looking for a layer above Terraform in a way which coordinates running different scripts. So I might have part of a CD pipeline that uses Terraform, part of it might use AWS CLI commands directly... but I was thinking Nix could be a nice way to coordinate running all of this by setting up the environment for each bit (environment variables, correct versions
<cinimod```>
Can I just say that error: cannot coerce a function to a string is really confusing error message for when you should have put an expression inside parentheses
<exarkun>
a lot of nix error messages take some real practice to make sense of, it's true
mexisme has quit [Ping timeout: 245 seconds]
<jg20>
Out of the box Nix does this just fine, I've got a proof of concept where it works fine if I run nix with sandbox set to relaxed and derivations have network access, but it seems like a bit of a hack even though it works well. I saw the Github issue for impure derivations which seems like it's more for the usecase I'm looking for (https://github.com/NixOS/nix/commit/647291cd6c7559f68d49a5cdd907c2fd580790b1)
<exarkun>
jg20: seems better to use Nix to create the environment where you can run the side-effectful operation and then run the side-effectful operation outside of the context of building a derivation.
cfricke has joined #nixos
<jg20>
exarkun: Hmm maybe, but if I have let's say 10 different stages to a pipeline, let's say each one uses Terraform, with Nix I can nicely define those as derivations and pass variables between those derivations based on outputs (by using import from derivation) and then all of those stages run nicely with a single `nix-build` call. If I use Nix to just set up the environment then I lose this ability and would have to use e.g bash to achieve something similar
<jg20>
It's just an idea - I've found Nix solves this problem really well of coordinating a large CI/CD pipeline by using derivations, and whilst it might be impure and technically not a valid use case of Nix, it solves a lot of problems as otherwise I'd be writing a load of bash to do this, unless anyone knows of any other tools that could do something like this instead?
<emily>
I think using Nix to generate bash scripts is basically your best option here
<emily>
a la what NixOS does
<emily>
but yeah, it reduces flexibility
<emily>
it would be nice if Nix was general enough for these more impure coordination tasks but it's pretty fundamentally based on the pure store model so I don't think it's ever going to be "approved" of.
rawtaz has quit [Changing host]
rawtaz has joined #nixos
twoolie has quit [Quit: leaving]
ilyagotfryd has quit [Ping timeout: 245 seconds]
mattyu has quit [Ping timeout: 240 seconds]
<jg20>
Yeah fair enough, thanks @emily
<jg20>
Maybe rather than explicit bash scripts, generating bash scripts using Nix like you said would work, that's something I'll take a look at :)
<clever>
sondr3: `nix-shell -p` will provide you with make
init_6 has quit []
<sondr3>
clever: yes, but that is GNU Make, right? I have an issue I where a Makefile works on macOS and not on NixOS
<sondr3>
and I have a feeling it might be because of differences between pmake and gmake
<clever>
sondr3: i think it will always be gnu make, you can check `make --version` to confirm
<sondr3>
clever: yes, but I want BSD Make, not GNU Make :)
zacts has quit [Ping timeout: 246 seconds]
<clever>
i dont think i see bsd make packaged
<clever>
sondr3: what is the error?
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<sondr3>
clever: its for a university assignment where we need to compile our C++ with a Makefile, it fails with a `could not find file`, but running the same `g++` command works
<sondr3>
and it does some weird magic to find the files and their headers, I asked a TA and he said it worked on his macOS machine
<clever>
sondr3: can you pastebin the Makefile contents and the exact output it gives when failing?
<clever>
sondr3: are you using nix-shell or did you install a compiler?
<sondr3>
clever: gist.github.com works as well?
<sondr3>
and no, not yet, everything works _besides_ running make
<gchristensen>
since switching to 19.09, `wl-copy` doesn't seem to actually put thetext in to my pasting clipboard. `echo -n "foobar" | wl-copy` then `wl-paste` doesn't print "foobar", nor does pasting anywhere else. if I do `echo -n foobar | wl-copy --primary`, then `wl-paste --primary` does print "foobar", but Ctrl-V still doesn't paste foobar.
<gchristensen>
any ideas?
<clever>
sondr3: gist works good, and nix-shell should fix everything
<sondr3>
clever: nix-shell still fails, gimme a sec
<tetdim>
Ericson2314, did you already port this ?!?
<Ericson2314>
I would really take my stuff and keep on going
<Ericson2314>
No
<Ericson2314>
I got part way
<tetdim>
Ericson2314, part wat?
<tetdim>
did you get stuck or something
domogled has quit [Ping timeout: 240 seconds]
<Ericson2314>
It got pulled to other things
<Ericson2314>
Though slightly stuck on some thing too o think
<tetdim>
oof dont say that
<Ericson2314>
Haha ok
<tetdim>
this has been going smooth af so far
<Ericson2314>
Yeah the main thing is switching
<tetdim>
what build object did you get to?
<tobiasBora>
Hello,
<Ericson2314>
I had some big yaks that were earlier on the docket
<Ericson2314>
Tetdim merely cross referencing mine, yours, and the original will probably find a bug or two in mine
<tobiasBora>
I'm not sure to understand what's wrong in my python derivation: http://paste.debian.net/1102533 . Basically, I get an error during build about "Collecting chompack>=2.3.2 (from smcp==0.4.6) ERROR: Could not find a version that satisfies the requirement chompack>=2.3.2"
<Ericson2314>
And that could well fix things
<tetdim>
Ericson2314, one thing i will say is my code is way more readable than yours :^)
<tobiasBora>
However, I did included in propagatedBuildInputs a chompack line... But it seems that when I compile this file (with nix-build -E "with import <nixpkgs> {}; python3Packages.callPackage ./smcp.nix {}"), it does not even try to build firt chompack... Any idea why?
<Ericson2314>
@tetdim that's fine, I did all the configure stuff line by line more or less
zacts has quit [Quit: WeeChat 2.4]
<Ericson2314>
And then started on the make
<Ericson2314>
It was just a grind
<tetdim>
ah im not doing that
<tetdim>
im doing it lib by lib]
<tetdim>
configure and make one lib, move to next
<tetdim>
do it in dependency order
<Ericson2314>
Well once I got to the make then yes line by line
<Ericson2314>
Sorry lib by lib
<tetdim>
Ericson2314, also you arent using standard names for the meson stuff :^)
<Ericson2314>
configure dependencies are famously unclear, so I thought line by line was easier for that
<tetdim>
libutil nahnah nixutil_lib yeee
<tetdim>
_lib _dep _src _inc all those
<tetdim>
its like type setting but variable names lmao
rndd has quit [Ping timeout: 245 seconds]
<Ericson2314>
Hehe so I've contributed like 50 commits to meson but I have no clue what the ideomatic shit is
<Ericson2314>
I have developed it far more than I've used it
Ariakenom has quit [Quit: Leaving]
<tetdim>
Ericson2314, ive spent a grand total of a week crash coursing meson
<tetdim>
prolly 12 hours a day? idk
<Ericson2314>
Well, that's a good deep dive :)
<tetdim>
ive actually learned more about auto tools than meson tbh
<tetdim>
meson is very python like and ive been doing python for s o l o n g
<Ericson2314>
Hahahaha that's often how it needs to be
<Ericson2314>
I need to hop on a train, but good luck, and please open even a WIP PR as soon as you feel comfortable
rndd has joined #nixos
<Ericson2314>
The fact that there are/were two independent efforts well help demonstrate the interest in this
<Ericson2314>
People's comments will also be extra motivation to finish :D
<tetdim>
Ericson2314, correct me if im wrong here
<tetdim>
yoou shouldnt need to include the hh file in the build dir?
<tetdim>
you put it in the library() sources
<tetdim>
you can just include_directories('.') no?
mattyu has joined #nixos
<Ericson2314>
@tetdim I'm not sure but meson will not go from . To any dependencies
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
<wucke13>
I have a python package, which only works with python3. Should I write anything in the default.nix to mark it as python3 only?
<{^_^}>
[nixpkgs] @vbgl opened pull request #69420 → ocamlPackages.patience_diff: init at 0.12.0 for OCaml ≥ 4.07 and related additions → https://git.io/JeG4j
<gchristensen>
fpletz: graylog used to have spunk, but they really boringed up their logo
<tokudan[m]>
inkbottle: sorry, saw your message just now. only enable one of them. if you want to use networkmanager (probably makes sense for a laptop), then set networking.wireless.enable = false
<inkbottle>
no worries
niksnut has quit [Quit: Lost terminal]
<{^_^}>
[nixpkgs] @Ma27 opened pull request #69421 → clipman: init at 1.0.1 → https://git.io/JeGBL
<tetdim>
clever, fixed the linking errors on a few libraries
growpotkin has quit [Ping timeout: 265 seconds]
niksnut has joined #nixos
arjen-jonathan has joined #nixos
<wucke13>
LnL: I have inserted the line in the default.nix, however, I can still build it with python2Packages.<p-name> (and the build fails due to python3 deps).
wfranzini has quit [Remote host closed the connection]
Ariakenom has joined #nixos
<LnL>
that should make it bail out before the build, are you sure it's not a dependency that fails?
<exarkun>
anyone want to explain the nixos test logging format to me? It seems to be some XML but I'm not sure what tags etc are involved. What does a line of log output in this format look like?
selfsymmetric-mu has quit [Remote host closed the connection]
<exarkun>
if I read LOGFILE from the environment in testScript I get a path like /nix/store/ac7ab4hdmabkmfnfi6175sbi1zdcm68i-vm-test-run-unnamed/log.xml
<exarkun>
kinda hard to write to a path in /nix/store isn't it?
<exarkun>
clever: which seems to be part of some kind of awkward connection loss detection logic?
alex`` has joined #nixos
<clever>
exarkun: also behind the scenes, it uses a qemu hypervisor serial port, to get a backdoor shell into the guest
<exarkun>
kind of a shame there's no some real (if simple) rpc there instead
bitmapper has quit [Ping timeout: 245 seconds]
<clever>
exarkun: that is something i have wanted to solve
<exarkun>
instead of shell hacks piled on perl hacks piled on qemu hacks
<clever>
exarkun: what i want, is a haskell library, with a muxed RPC over that serial console
<exarkun>
+100
<Ericson2314>
tetdim: thanks! nice catch!
<clever>
exarkun: with haskell code on both the client and server side (host and guest)
<clever>
exarkun: so you can have haskell code in the host, acting as a conductor
<exarkun>
except no one wants haskell in the build tool chain because the compiler takes forever and the dependencies are huge? :)
<clever>
exarkun: and then haskell code in each guest, doing actions, and reporting status over the rpc
<clever>
exarkun: the issue, is that i basically need an RPC channel, that can act over a byte stream, and is mux-able (can deal with parallel concurrent streams)
<clever>
does haskell have anything close to that?
<exarkun>
I bet it has several
<emily>
rust seems like a better fit here
<exarkun>
Technically HTTP/2 fits the bill, though whether that would be anyone's first choice or not I dunno
<gchristensen>
I was going to say HTTP/2 :)
<emily>
similarly bad in terms of compilation toolchain weight, but certainly lighter for the resource-constricted VMs
<clever>
network and tcp isnt going to be available
<clever>
its just a single raw byte stream
<emily>
not that haskell would work badly either
<gchristensen>
HTTP/2 over a byte stream
<emily>
just feels like the slightly more awkward choice
<exarkun>
You can still do HTTP/2 over that
<emily>
gRPC over stdio...
<emily>
truly beautiful
<exarkun>
Maybe some irrelevant corners would break, like don't return a redirect... but so what
<gchristensen>
emily: bingo
<exarkun>
But yea, gRPC might be better
<hodapp>
say, is it fairly doable to take an existing NixOS system and basically run the installer from it onto a separate disk? I *could* just burn some install media, but I'm sort of at $client and trying to avoid taking the system down in the meantime until I need to
bonnes has left #nixos [#nixos]
<emily>
I think there are, like, lighter-weight solutions to this problem, but I guess gRPC is batteries-included.
bonnes has joined #nixos
<hodapp>
would use CloneZilla, but there are some complications
bonnes has left #nixos [#nixos]
<exarkun>
it might be nice to use a protocol you can read with your eyes, for easy of debugging. bandwidth/cpu costs aren't really an issue. so possibly some flavor of json-rpc.
<exarkun>
but with the right tooling you can probably make gRPC just as usable
<exarkun>
Also, last week I read a blog post about using Servant (which you might know from Haskell HTTP'ing) for defining type-safe arbitrary protocols.
<clever>
exarkun: for this test for example, i just want the host side to be a 2 step process
cinimod``` has joined #nixos
<clever>
exarkun: step 1, boot the vm, step 2, tell the guest half of the code to do everything else
<exarkun>
yea, that would be a lot better
<exarkun>
and that's basically what I'm doing right now - by hand
<clever>
exarkun: the guest side can then do lines 34-55, and spit out a json blob? with status
<clever>
and also stream logs over, in real-time
<exarkun>
rewriting 50 perl steps as 1 perl step that runs a python program on the guest
<infinisil>
hodapp: Well you can copy only the configuration.nix over, then nixos-install
<clever>
the reason i was thinking haskell and rpc, is that the guest could create any kind of object, serialize it, and fire it off to the host, which deserializes
<clever>
and then you can do whatever you want
<exarkun>
speaking of which ... "ImportError: No module named site" :/
<infinisil>
hodapp: Plus all the data you need
<hodapp>
infinisil: yeah, data should just be homedir so that's no big deal. I need to figure out LUKS though
<exarkun>
clever: I definitely like that idea more than ad hoc gluing together random programs.
<infinisil>
hodapp: That you need to reconfigure manually
<hodapp>
not sure what you mean
<clever>
hodapp: nixos-install works on any nixos machine, and i do prefer the idea of just installing to a usb stick
<clever>
hodapp: because then changes you make to that stick persist, and you can leave a clone of your config repos, and ssh priv keys on it
<infinisil>
hodapp: If you want luks on a machine you need to configure it, NixOS can't do that
<infinisil>
At least as long as nixpart isn't finished :)
<lassulus>
nixos-generate-config detects LUKS if there is no layer (like lvm) in between
<gchristensen>
(and may we never finish it)
<clever>
infinisil, gchristensen: did you see my nixops issue, that could use nixpart features?
<exarkun>
Running python3 on this vm works. Running python2 on this vm works. Running `tahoe` (from `pkgs.tahoe-lafs` in systemPackages) works fine. Running `python3 -c 'import subprocess; subprocess.check_output(["tahoe"])'` fails. :/
<clever>
which copies to /mnt/nix/store on a remote box
<clever>
which installs the bootloader for an already copied nixos
<clever>
then you just reboot, and its live!
alex`` has joined #nixos
<clever>
gchristensen: my idea, is that steps 2/3/4 are all a common library, that any nixops backend can use
<clever>
gchristensen: and the kexec is an optional function, that some backends can choose to use, but the backend must supply a way of getting root on a rescue shell or ubuntu image
<infinisil>
edef: Regarding the commit you just pushed, you should use literalExample with a string to have it show up correctly in the manual
<edef>
infinisil: hm, elaborate?
<infinisil>
edef: This is a good example of why we should have every change be a PR :)
<clever>
gchristensen: aws would just have an ami, that is basically the installer iso with ssh enabled, to let you in
<edef>
infinisil: it shows up in the manual properly here
<clever>
gchristensen: packet.net could be a real pxe boot, with the installer image
<infinisil>
edef: With how you have it now, the whole nix expression is on a single line (pretty sure)
<edef>
infinisil: this is true, but it's tiny
<infinisil>
edef: But still, it would look nicer with multiple lines, that's why literalExample exists
liberiga has quit [Ping timeout: 260 seconds]
<infinisil>
I think all multi-line examples should use it
<edef>
fair
<edef>
i'm somewhat averse to making PRs for Literally Everything because, well, we have tonnes of them and review is always far behind
<gchristensen>
I do a PR for literally everything, and just wait for ofborg to agree prior to merging
<edef>
mhm
gxt has quit [Ping timeout: 260 seconds]
<infinisil>
Yeah even for that it's beneficiql
<edef>
i'm not particularly used to making PRs for things i merge myself instantly, and at that point the likelihood of gathering commentary is fairly tiny
<gchristensen>
yeah fair enough, I don't do it for commentary, I do it so I don't break the build for the next PRs
<gchristensen>
I've been burned on PRs I thought were 100% definitely safe and weren't, so I just don't anymore
<edef>
right, i do check that everything builds before shipping it out
<edef>
i should probably have a more pleasant CLI github experience
<gchristensen>
the thing that trips me up sometimes is evaluations for the different platforms, plus making sure the nix-env output is clean
<andi->
I share the pain... I go through PRs most of the time but I'd rather do it all from the CLI
drakonis_ has quit [Quit: Leaving]
<edef>
i've moved onto git email flow for collaborating with friends
<gchristensen>
nice
<edef>
for which a decent CLI setup is pretty trivial
drakonis has joined #nixos
rotaerk has joined #nixos
<edef>
and now github drives me up the wall constantly
o1lo01ol1o has quit [Remote host closed the connection]
<andi->
+1
<edef>
gchristensen: define "making sure the nix-env output is clean"?
o1lo01ol1o has joined #nixos
<gchristensen>
ensuring this only outputs what we expect, no asserts / warnings / traces: nix-env -qa --json --file .
<edef>
mmm
<edef>
does ofborg check that?
<gchristensen>
yeah
<gchristensen>
`hub` is pretty nice
<edef>
it has bitten me before
<gchristensen>
hub?
mexisme has joined #nixos
<edef>
if i create a PR with hub, and i haven't manually pushed the branch to edef1c/nixpkgs, it pushes it to NixOS/nixpkgs
<gchristensen>
nice :|
rotaerk has quit [Ping timeout: 245 seconds]
o1lo01ol1o has quit [Ping timeout: 245 seconds]
domogled has quit [Ping timeout: 265 seconds]
<slabity>
Hey guys, I' trying to get nix-plugins working on my system. Does anyone have experience with that? I'm not quite sure what needs to be added to my nix configuration.
<slabity>
I have `nix.extraOptions == "plugin-files = ${nix-plugins}/lib/nix/plugins/libnix-extra-builtins.so"` and that works fine. But it only works if I pass `--option extra-builtins-file` when running it.
<sphalerite>
slabity: if you add your extra-builtins-file setting to nix.extraOptions it should work
<sphalerite>
AFAIK
<slabity>
Oh... That makes a lot of sense...
<clever>
slabity: --option just lets you override any nix.conf entry
<clever>
slabity: `nix show-config` also shows the result of parsing the config
<slabity>
Thanks. I'll see if this works
<sphalerite>
I want to set up bind as a local nameserver in order to forward DNS requests for some domains to a custom nameserver. Is there a way to have bind use the upstream DNS server acquired via DHCP?
<clever>
sphalerite: ive done that before, let me see
ris has joined #nixos
SyrupThinker has quit [Quit: WeeChat 2.5]
<clever>
sphalerite: ah right, i did that on an old gentoo machine, let me see what the exact method was...
<clever>
sphalerite: `Hooking into events` in `man dhcpcd`, it will run some shell scripts when it does things
domogled has joined #nixos
<clever>
sphalerite: `man dhcpcd-run-hooks`
<clever>
networking.dhcpcd.runHook
aveltras has joined #nixos
<clever>
Shell code that will be run after all other hooks. See `man dhcpcd-run-hooks` for details on what is possible.
<sphalerite>
oh, hm, I'm using network-manager though
<clever>
sphalerite: the dns servers are in an env var when that hook gets ran
kleisli has joined #nixos
<clever>
and i then echo'd them to a file, in the right format, to include into the bind cfg
<clever>
Type: one of "dhclient", "dhcpcd", "internal"
<Ilya_G>
pypi2nix question: I have a so called unstable package I am trying to include in the mix. I am being told that it can only import latest stable package which is many versions behind. Any suggestions of what I should be looking for?
Jackneill has quit [Read error: Connection reset by peer]
<clever>
On after defining the make_resolv_conf function, the client script checks for the presence of an executable /etc/dhcp/dhclient-enter-hooks script
<clever>
sphalerite: so you can either write a dhclient hook, or switch to dhcpcd, and use a dhcpcd hook
rotaerk has joined #nixos
justan0theruser has quit [Ping timeout: 245 seconds]
<fpletz>
sphalerite: or you could use networkd and resolved where you can configure what domains you want resolved from which interface's nameserver :)
<sphalerite>
fpletz: hmmmm :D
<fpletz>
sphalerite: man systemd.network, grep for Domains=
orivej has joined #nixos
cinimod``` has quit [Ping timeout: 265 seconds]
<hodapp>
hmm... if I set boot.loader.grub.device, run nixos-install, and the device name changes at the next boot (due to me cloning to a new disk, planning to remove the old one), is that still okay?
<clever>
hodapp: nixos will only use that device setting when it has to re-install the bootloader to the MBR
<clever>
hodapp: which only happens when the grub version changes
<clever>
so it can still boot, but you will likely want to fix it, or you risk overwriting the MBR of the wrong disk
<clever>
personally, i usually just set it to "nodev" after i install
<clever>
then the MBR just never updates
<hodapp>
hmmmm
<clever>
ive yet to break a machine, lol
<clever>
but grub also almost never updates
<clever>
EFI eliminates that issue
<Ilya_G>
sorry reposting ---> pypi2nix question: I have a so called unstable package I am trying to include in the mix. I am being told that it can only import latest stable package which is many versions behind. Any suggestions of what I should be looking for?
zupo has joined #nixos
<spazzpp2>
hi
<spazzpp2>
I get `'/nix/var/nix/db/schema' is corrupt`
<spazzpp2>
what should I do. Firstly, is it correct that I copied the img to my sd and booted from it?
<spazzpp2>
Shouldn't the sd be rw?
<clever>
spazzpp2: what is the contents of that file?
<sphalerite>
spazzpp2: for raspberry pi?
<spazzpp2>
yes for raspberry
<sphalerite>
spazzpp2: your sd card might be bad
<exarkun>
a python program is misbehaving. can I make it pass `-v` to Python to assist with debugging? how?
<spazzpp2>
clever, it's empty
<clever>
spazzpp2: that would be the problem, it shouldnt be empty
<{^_^}>
[nixpkgs] @ttuegel opened pull request #69426 → Reduce closure size of Qt applications → https://git.io/JeG0b
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
cinimod``` has quit [Ping timeout: 276 seconds]
<spazzpp2>
has anyone here been at the datenspuren, too?
<clever>
gchristensen: i'm using buildLayeredImage and am winding up with 3gig layers, got any tips on how to debug it, or should i just increase the layer count more?
<{^_^}>
[nixpkgs] @ttuegel opened pull request #69427 → Reduce closure size of Qt applications (backport) → https://git.io/JeGEt
<exarkun>
Figured out what is going on with my "ImportError: no module named site"
<exarkun>
nixos python wrapper sets PYTHONHOME in the environment
<exarkun>
If you try to run a _different_ python then it respects that PYTHONHOME and breaks
<exarkun>
because it's not for that python
<hodapp>
hey, slick. I was enough of an idiot to use a crappy SATA to USB adapter which then dropped the drive right off the bus in the middle of a nixos-install, but it appears to have resumed okay post-reboot
<clever>
exarkun: ive run into related problems, if you `nix-shell -p nixops`, it will leak a PYTHONHOME
<exarkun>
Seems like either the python wrapper shouldn't set PYTHONHOME or the python script wrapper should unset/reset it
<clever>
exarkun: if you then `nix-build -A nixops` and ./result/bin/nixops, it will use the nixops from nix-shell, not from result
<clever>
exarkun: that leads to you spending 6 hours trying to get your changes to print anything at all :P
zupo has joined #nixos
<exarkun>
ouch
<hodapp>
hopefully nothing was left in a partly-copied state in /nix/store, as I could only imagine that would cause some subtle problems
<exarkun>
at least I figured this out after only about 3 hours
SyrupThinker has quit [Quit: WeeChat 2.6]
acarrico has quit [Ping timeout: 268 seconds]
<clever>
hodapp: nix wont register a path as valid in db.sqlite until it has finished writing it
<clever>
hodapp: but ext4 can commit things out of order, and i have seen files get truncated after nix flagged them as valid, after an improper shutdown
<clever>
hodapp: nix-store --verify --check-contents, will validate everything
<hodapp>
clever: cool, I will do that once it's off of USB and onto actual SATA
<hodapp>
this wasn't so much an improper shutdown as an improper "drive fell off the bus but was still powered"
<clever>
hodapp: could have similar effects though
<clever>
since ext4 was interupted mid-write
<hodapp>
yeah... I'll run the verify
<hodapp>
how did nixos-install know to resume? did it refer to that DB?
<clever>
hodapp: nixos-install is basically just a helper script, that runs nixos-rebuild in a chroot
<hodapp>
oh!
<clever>
hodapp: and nixos-rebuild is just a helper script that runs nix-build
<hodapp>
oh noes... "builder for '/nix/store/b9anbli2ydqfvqkbhz3bq56g0sycd7i9-texlive-dviincl-1.00.drv' failed with exit code 2; last 4 log lines:"
<Taneb>
Is there a good way to do a git bisect run with nix, so it doesn't repeatedly try to build the same failing derivation? (assuming the derivation doesn't change every commit...)
<Ericson2314>
I've done a bunch of work this year on the haskell compiler, and earlier this year wrapped up my big changes to meson itself, and before that was really busy with work, and before that did this
<Ericson2314>
something like that
<Ericson2314>
tetdim: I will make an offering to the C++ gods for you
<hodapp>
hrmph. I appear to have corrupted files that are keeping nixos-install from proceeding
<Ericson2314>
I pushed the bug you found ot my branch too, just for safe keeping
<Ericson2314>
*bugfix
tertl3_ has joined #nixos
drakonis has quit [Read error: Connection reset by peer]
drakonis has joined #nixos
tertl3_ is now known as tertl3
<hodapp>
but I don't know how to find out what files are the problem...
<{^_^}>
Found in packages: libGLU, libGLU_combined, xorg_sys_opengl
<jabranham>
,locate libGL.so.1
<{^_^}>
Found in packages: libglvnd, primusLib, libGLU_combined, xorg_sys_opengl
<jabranham>
,locate libstdc++.so.6
<{^_^}>
Found in packages: robo3t, gcc-unwrapped.lib
<jabranham>
,locate libz.so.1
<{^_^}>
Found in packages: gap, zlib, gap-full, scilab-bin, gap-minimal, jetbrains.clion
<jabranham>
,locate libuuid.so.1
<{^_^}>
Found in packages: libuuid, utillinux
<werner291>
ivan: Aha, interesting. I'm trying to reset the state of matrix-synapse. Tried deleting the database, but now it's complaining about a missing db. Doesn't nixos-rebuild run some initialization scripts?
<jabranham>
,locate libstdc++.so.6\
<jabranham>
,locate libstdc++.so.6
growpotkin has joined #nixos
<{^_^}>
Found in packages: robo3t, gcc-unwrapped.lib
<{^_^}>
Couldn't find in any packages
<infinisil>
jabranham: I recommend using #bottest or PMing the bot for lots of such queries
<jabranham>
infinisil: thanks, will do in the future
Ariakenom has quit [Quit: Leaving]
<ivan>
werner291: system configuration can include modules which include systemd services that do things like initial setup, but there's no standard way to blow away all the state for a thing
<gila>
Im trying to compile a package on Nixos (19.x but so desperate now on unstable) but I keep getting libtool error unrecognised option: '-f'. Im using the autoreconfHook and looked at other distro's and they dont seem to do any particular magic either: https://aur.archlinux.org/cgit/aur.git/tree/PKGBUILD?h=isa-l
<tokudan>
gila, can you show the derivation that you're trying to build?
aveltras has quit [Quit: Connection closed for inactivity]
<gila>
tokudan different as in older version of gcc?
<tokudan>
maybe for an intel specific compiler? not sure tbh
<tokudan>
though gcc is listed as supported
<yorick>
tokudan: libtool --force
<yorick>
oh, that's libtoolize
drakonis has joined #nixos
<jtojnar>
Weird, if I run `nix-shell -p nodejs` in `direnv` directory, I just get `bash: /nix/store/kgx47wbbfkk8kylk2w5a6f5rbw4ymg5m-shell/varmap-v1: No such file or directory`
<gchristensen>
a bug with lorri
<gchristensen>
I think if you upgrade lorri it will be fixed
<tokudan>
gila, sorry, no idea what exactly is going on there. apparently the configure script creates a libtool and that version somehow gets called with -f, though I cannot figure out what exactly is causing that
<jtojnar>
gchristensen++
<{^_^}>
gchristensen's karma got increased to 152
<gchristensen>
:)
<jtojnar>
thanks, recompiling
<gila>
tokudan yeah wierd indeed .. thanks for have a look though
Soft[m] has joined #nixos
<{^_^}>
[nixpkgs] @worldofpeace opened pull request #69434 → dbus: set datadir again → https://git.io/JeGul
sondr3 has quit [Quit: WeeChat 2.4]
<worldofpeace>
Jan Tojnar: turns out we did subtly break dbus #69434 🤣
<cransom>
also the 19.03 to 19.09, it's just a set of tags. they are just updating to the next git release and have no care about what version it has been labeled.
Syrup has quit [Quit: WeeChat 2.6]
alex`` has quit [Ping timeout: 240 seconds]
alex`` has joined #nixos
iqubic has quit [Ping timeout: 240 seconds]
spazzpp2 has quit [Quit: Leaving]
uranther has joined #nixos
<{^_^}>
[nixpkgs] @matthewbauer merged pull request #69111 → Fix libcanberra-gtk{2,3} on Darwin → https://git.io/Je308
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/JeGzY
growpotk- has quit [Ping timeout: 268 seconds]
Jackneill has joined #nixos
o1lo01ol1o has joined #nixos
<tetdim>
emily, thats a cowards opinion who needs security just ship it :^)
hmpffff has quit [Quit: nchrrrr…]
waleee-cl has quit [Quit: Connection closed for inactivity]
o1lo01ol1o has quit [Remote host closed the connection]
Jackneill has quit [Remote host closed the connection]
Izorkin has quit [Ping timeout: 276 seconds]
o1lo01ol1o has joined #nixos
drakonis has quit [Ping timeout: 276 seconds]
civodul has quit [Quit: ERC (IRC client for Emacs 26.3)]
<iqubic>
And ~/.config/nix/nix.conf doesn't exist.
<infinisil>
Okay so what nix-shell command are you running exactly?
<iqubic>
I feel silly
<iqubic>
I was running nix-shell with --run, but I forgot quotes around the command to be run inside the environment.
<iqubic>
Issue solved.
<niso>
the initial nixops deployment to hetzner fails due to "`GLIBC_2.28' not found"? (due to nixkickstart/libudev) by chance someone was also running into this problem?
philr has joined #nixos
wildtrees has quit [Quit: Leaving]
CrazyMind102 has quit [Read error: Connection reset by peer]