<rotaerk>
because I end up needing to test nixpkgs commits that aren't fully in the binary cache
<trcc>
What is the purpose of running things inside the nix-shell instead of the terminal (using on mac)?
<rotaerk>
so I end up building GHC or GCC from scratch
<rotaerk>
trcc, nix-shell sets up your environment as specified in your shell.nix
<rotaerk>
for instance, you can set it up such that all your development tools are available within it
<rotaerk>
and all the libraries you depend on are available
tadni_ has quit [Ping timeout: 240 seconds]
<trcc>
rotaerk: thank you. Howeer, within the shell I am invoking cabal build. And this seems to use the regular cabal. I do not even have to add e.g. a haskell module to the shell.nix file
<trcc>
it is enough to have it in the cabal file
<clever>
rotaerk: i once did a bisect like that, and ran into all kinds of trouble
<trcc>
is that because it is using a general envrionment?
<clever>
rotaerk: it turns out, anything not in the cache worked, and anything in the cache failed, so the bisect just failed to locate the problem entirely
<clever>
rotaerk: the problem was that the nix sandbox broke the package, and i had the sandbox off at the time
<rotaerk>
nice
<clever>
the package was testing for things like /etc/mtab, to detect what your distro is doing
<clever>
the nix sandbox had none of the options
tmaekawa has joined #nixos
<clever>
so it hard-coded itself to open the path "unknown" at runtime
<rotaerk>
fortunately, I don't think I'm in a situation like that
<rotaerk>
I just get impatient waiting for each build
<rotaerk>
trcc, well, if you have cabal installed in your general environment, it will be accessible to nix-shell (well, by default; I think there's a way to hide it?)
<rotaerk>
but if you want a reproducible development environment that doesn't depend on what you have installed globally...
<rotaerk>
nix-shell is useful for that
<rotaerk>
heck you can have different nix-shell environments side-by-side, each with different versions of cabal accessible to tthem
tmaekawa has quit [Client Quit]
<trcc>
I am mainly interested in the a reproducable environment for the haskell packages. Not so much cabal
<rotaerk>
trcc, you mean, you want an environment in which a particular set of haskell packages, of specific versions, are accessible for building with cabal?
<trcc>
yes
<rotaerk>
yep, that's something you can setup with nix-shell
<rotaerk>
that's what I'm doing
<trcc>
do you have any repo publicly available?
<rotaerk>
I do, but it's not implemented in the simplest possible way, so might not be the best to learn from
sk8forether has quit [Quit: Connection closed for inactivity]
jtojnar has joined #nixos
<rotaerk>
let me know if you have any questions about it
mroutis has quit [Ping timeout: 268 seconds]
<trcc>
Thank you rotaerk.
<trcc>
I probably will later on
<trcc>
What is the process of creating a new project? Cause now I did cabal init, and then created shell.nix from cabal init. But this leaves me with a global cabal and global packages
<rotaerk>
"created shell.nix from cabal init"? how
<trcc>
cabal2nix
mroutis has joined #nixos
<trcc>
the cabal file genereated from cabal init
<rotaerk>
oh k
<trcc>
But this seems to be the wrong way around
<rotaerk>
well, all of my nix files were written manually. cabal2nix is something that I call dynamically from within them
<rotaerk>
rather than storing the cabal2nix result itself in my source
<rotaerk>
note that the main folder contains nothing nix-related; it's just a haskell project
<rotaerk>
well, the general approach I take is to pick a version of nixpkgs as a foundation and reference that specific version (as opposed to using <nixpkgs>, which relies on whatever version I happen to have installed)
<rotaerk>
and then I apply overrides on top of that
<rotaerk>
i don't currently override the cabal version, so I'm using whatever version comes with this version of nixpkgs
<trcc>
okay. Thank you. I am trying to set up my project using nix for the frist time. So the development environment is being set up.. takes a bit of time
<rotaerk>
it's got a learning curve
pie_ has joined #nixos
pie__ has joined #nixos
pie__ has quit [Client Quit]
trcc has quit [Remote host closed the connection]
freeman42x[nix] has quit [Ping timeout: 276 seconds]
fractalcat has joined #nixos
<fractalcat>
#pipe
<fractalcat>
gah
andymandias has quit [Ping timeout: 244 seconds]
andymandias has joined #nixos
wgas has joined #nixos
thc202 has quit [Ping timeout: 244 seconds]
tadni has joined #nixos
tadni_ has joined #nixos
zearen has joined #nixos
tadni has quit [Remote host closed the connection]
ericsagnes has joined #nixos
drakonis has joined #nixos
Supersonic has quit [Disconnected by services]
Supersonic112 has joined #nixos
Supersonic112 is now known as Supersonic
zearen has quit [Ping timeout: 248 seconds]
Guest23226 has quit [Remote host closed the connection]
<tnks>
I've been using cabal new-build with Nix for a while, but there's something that's been bothering me. I can use Nix to lock down all my Haskell dependencies. But there's a few tools that Cabal insists on pulling down to compile like alex, happy, c-language, and syb.
chessai has joined #nixos
catern has quit [Excess Flood]
catern has joined #nixos
sir_guy_carleton has joined #nixos
sigmundv has quit [Ping timeout: 240 seconds]
voiceftp has joined #nixos
markus1199 has joined #nixos
seafood has quit [Quit: seafood]
qyliss^work has quit [Ping timeout: 276 seconds]
joepie91 has quit [Excess Flood]
joepie91 has joined #nixos
joepie91 has quit [Changing host]
joepie91 has joined #nixos
sam has joined #nixos
qyliss^work has joined #nixos
lnikkila_ has quit [Ping timeout: 256 seconds]
sam is now known as Guest2906
markus1189 has quit [Ping timeout: 256 seconds]
Guest17457 has quit [Ping timeout: 264 seconds]
<joepie91>
,locate libX11.so.6
<{^_^}>
Found in packages: xlibs.libX11
freeman42x]NixOS has quit [Ping timeout: 245 seconds]
<kalbasit>
I'm having an issue with my locale, when `LC_ALL` is not set, `rofi` exits with the error `Failed to set locale` but I set `LC_ALL=$LANG` I get a warning instead: `setlocale: LC_ALL: cannot change locale (en_US.UTF-8)`. Running on NixOS 18.03 with `i18n.defaultLocale = "en_US.UTF-8"`
<kalbasit>
I have no mention of `LANG` or `LC_*` in my home dir, just that system-wide defaultLocale setting
muzzy__ has quit [Read error: Connection reset by peer]
blob_ has joined #nixos
<{^_^}>
[nixpkgs] @vaibhavsagar opened pull request #43271 → all-cabal-hashes: update snapshot to Hackage at 2018-07-03T07:36:18Z → https://git.io/fNUmT
stepcut has joined #nixos
jasom has joined #nixos
blob_ has quit [Quit: Leaving]
mroutis has quit [Ping timeout: 260 seconds]
reinzelmann has joined #nixos
lnikkila_ has joined #nixos
jackdk has joined #nixos
muzzy__ has joined #nixos
lnikkila_ has quit [Ping timeout: 244 seconds]
Lisanna has joined #nixos
<Lisanna>
What order does hydra evaluate jobs in?
<Lisanna>
I've got a bunch of tests where each test is a job, and those test drvs are also constituents in some other jobs. But hydra starts building the aggregate jobs first, which makes it hard to keep track of the job progress
<Lisanna>
since the test jobs don't get their result marked until the very end
stepcut has quit [Remote host closed the connection]
reinzelmann has quit [Quit: Leaving]
stepcut has joined #nixos
doyougnu has joined #nixos
stepcut has quit [Remote host closed the connection]
sbdchd has joined #nixos
<Lisanna>
ah, well, not the *very* end, but it does seem to be a bit scattered
sbdchd has quit [Ping timeout: 240 seconds]
<jasom>
can I install a 32-bit version of firefox-52 ESR (for the purposes of running Java applets)
<Lisanna>
seems like hydra will only associate a derivation as a built step of a single job, rather than all the jobs that actually have it as a build step
vmandela has joined #nixos
doyougnu has quit [Ping timeout: 240 seconds]
andymandias has quit [Ping timeout: 256 seconds]
dvim has joined #nixos
andymandias has joined #nixos
MercurialAlchemi has joined #nixos
jb55 has quit [Quit: WeeChat 2.1]
`_ has joined #nixos
<`_>
so we have two people using nixops to deploy to the same box
<`_>
every time person A deploys, person B cannot because ssh gets rejected
<`_>
vise versa
<`_>
ideally we can both deploy, without having to do a `destroy` `delete` `create` to recover
<`_>
is it possible to share the initial ssh keys?
<`_>
by the way this is 2 people sharing a tmux session back to spice up your life
tadni has quit [Remote host closed the connection]
Guanin has quit [Ping timeout: 244 seconds]
seafood has joined #nixos
bmikedesk has quit [Ping timeout: 276 seconds]
Lisanna has quit [Quit: Lisanna]
Ariakenom has joined #nixos
Guanin has joined #nixos
reinzelmann has joined #nixos
reinzelmann has quit [Client Quit]
igo95862 has quit [Quit: igo95862]
blonkhart has quit [Quit: WeeChat 1.9.1]
seafood has quit [Quit: seafood]
slack1256 has joined #nixos
slack1256 has quit [Read error: Connection reset by peer]
wgas has quit [Read error: Connection reset by peer]
rprije has quit [Ping timeout: 244 seconds]
fendor has joined #nixos
fendor has quit [Client Quit]
lnikkila_ has joined #nixos
trcc has joined #nixos
tmaekawa has joined #nixos
deanman has quit [Quit: EliteBNC free bnc service - http://elitebnc.org - be a part of the Elite!]
lnikkila_ has quit [Ping timeout: 260 seconds]
mariatsji has joined #nixos
mariatsji has quit [Remote host closed the connection]
mariatsji has joined #nixos
crmlt has quit [Remote host closed the connection]
lfish has joined #nixos
<`_>
Is there a way to simply check that the syntax of a nix expression is valid?
<`_>
Without actually evaluating it?
mariatsji has quit [Ping timeout: 240 seconds]
sbdchd has joined #nixos
bkchr has quit [Ping timeout: 260 seconds]
Lisanna has joined #nixos
lord| has quit [Read error: Connection reset by peer]
mariatsji has joined #nixos
lord| has joined #nixos
<elvishjerricco>
`_: `nix-instantiate --parse`
<`_>
elvishjerricco: thank you
johann__ has joined #nixos
<angerman>
hey elvishjerricco shouyld we start looking into merging your stuff into nix-tools?
crmlt has joined #nixos
<elvishjerricco>
angerman: If you think so. Only reason I haven't opened a PR is because of that hackage-db fork
<elvishjerricco>
Just haven't gotten around to addressing peti's concerns
<angerman>
well, open a PR and we can discuss the progress there?
<elvishjerricco>
Sure
<angerman>
I'd just hate to have this lost :D
Ariakenom has joined #nixos
<angerman>
elvishjerricco: I'm also currently building it with `cabal new-build --allow-newer --constraint="hnix==0.4.0"`, which feels rediculously stupid :-)
<lfish>
hello. I installed a texlive.combined.scheme-basic with nix-env and texlive.combined.scheme-full with /etc/nixos/configuration.nix. The linked programs are those installed with nix-env, how do I get things to be the other way?
<kalbasit>
I added Kubernetes to my NixOS-18.03 with `services.kubernetes.roles = [ "master" "node" ];`, however k8s is not starting and I see this in the logs: `kube-proxy[1101]: E0710 00:43:08.638543 1101 proxier.go:1667] Failed to execute iptables-restore: exit status 1 (iptables-restore: invalid option -- '5'`
crmlt has quit [Read error: Connection reset by peer]
tilpner has quit [Ping timeout: 240 seconds]
nlytend has quit [Ping timeout: 240 seconds]
dejanr has quit [Ping timeout: 240 seconds]
neof0x has quit [Ping timeout: 245 seconds]
phreedom has quit [Ping timeout: 250 seconds]
phreedom_ has joined #nixos
nschoe has joined #nixos
<nschoe>
Hello everyone, I'm having an issue with a Qt Project. I'm having difficulties getting used to developping with nixOS, I thought this would make my life easier, but it doesn't :/ ^^ So I have a shell.nix file, with a simple "buildInputs = [ pkgconfig cryptopp qt5.full qtcreator gdb ];". The idea is to enter the environment with `nix-shell` and from there launch qtcreator and dev fron there.
<nschoe>
This worked for a bit, but recently, I ran some `nix-channel --update` and some `nix-collect-garbage` (not sure if this is related, but I think so). Yoday I enter my `nix-shell` dev env as usual, this time it downaloded / unpackqd afew packages, which seemed odd, but "why not"
<nschoe>
And now when I try to run `qtcreato` it says "Cannot update Qt version information: /nix/store/wi32b30w1an36l7r19vi4v9g2liznspg-qt-full-5.10.1/bin/qmake cannot be run." And when I run `which qmake` from inside the dev environment, I get "/nix/store/v521qbyfwv3png4bliwgz693v2534ynd-qt-full-5.10.1/bin/qmake". Notice that it's not the same hash.
nlytend has joined #nixos
<nschoe>
Then I launched a few `grep` command to find the first hash, and it appears all over the "QtProject" directory, which is where the qtcreator's settings are saved.
humanoyd has joined #nixos
<nschoe>
I have lots and lots of hits about the old hash. I believe that my `nix-collect-garbage` destroyed the old qt-full-5.10 hash, form that it seems to me that it's all very fragile in my case: my dev environement depends on me not running the garbage collector. How is this possible, I'm sure I'm missin something.
reinzelmann has joined #nixos
<nschoe>
(Sorry this was a long message, but I wanted to explain the issue clearly ^^)
<nschoe>
The TL;DR is that I have 2 questions: 1) what should I do to have a reliabe dev environement? Apparently shell.nix + nix-shell doesn't seem to be safe (do I have to nix-build in order to get a result/ that points to the hash?) and 2) how can I fix the fact that tcreator doesn't find the libraries / qmake executable anymore?
<nschoe>
Thanks in advance :)
<adisbladis[m]>
nschoe: Yes, store paths without a garbage collection root will be collected
<adisbladis[m]>
You can manually add a gc-root
ericsagnes has quit [Ping timeout: 244 seconds]
Guest26247 has quit [Quit: Guest26247]
johann__ has quit [Quit: Leaving.]
<nschoe>
adisbladis[m], okay this is what I feared: so running `nix-build` and keeping the `result/` directory should help?
<adisbladis[m]>
nschoe: Yes that's enough
<adisbladis[m]>
You may also want to look into nix-shell/nix-store --add-root
<adisbladis[m]>
I have some direnv + nix magic to add roots for all nix-shells I enter
<adisbladis[m]>
And then you can use `nix-collect-garbage --older-than`
<adisbladis[m]>
`nix-collect-garbage -d` will still remove "everything" regardless of roots
trcc has quit [Remote host closed the connection]
<nschoe>
adisbladis[m], okay thanks for the tip. I never use the `-d` option, it's too nuclear for me :P
<nschoe>
adisbladis[m], so from now, I'll run `nix-build`. But now the tricky part: if I do this, I'm stuck with thi sfrozen version.
<adisbladis[m]>
nschoe: nix-shell -add-root will help with that :)
<nschoe>
I can't update or anything. And I thought it was the purpose of nix and nixOS to allow those kinds of things. But here the issue is that (the way I understand it), qtcreator store the fullpath of the libraries / executables it uses, so the momen tI change version, the hash will change and nothing ill everg work.
<{^_^}>
[nixpkgs] @xeji pushed commit from @tilpner to master « gomuks: 2018-05-14 -> 2018-7-10 (#43277) »: https://git.io/fNUz1
rprije has joined #nixos
<sphalerite>
nschoe: you may need to remove stale build files that nix is unaware of/which aren't aware of nix
dejanr has joined #nixos
<sphalerite>
I don't know which build system you're using, but typically this will be stuff in your gitignore
<{^_^}>
[nixpkgs] @offlinehacker opened pull request #43278 → kail: init at 0.6.0 → https://git.io/fNUzy
<sphalerite>
or your build directory if one exists
<sphalerite>
you should only need to do this when you've updated your nixpkgs
<nschoe>
sphalerite, the issue doesn't seem to be stable build files, but qtcreator's preferences.
<nschoe>
sphalerite, the problem is that when I try to open qtcreator it says "/hash-to-qmake" cannot be run.
<nschoe>
It complains that qmake is not found.
<sphalerite>
ah that's a pain. It should just look it up in $PATH rather than using an absolute path :(
tilpner has joined #nixos
<sphalerite>
maybe it will work if you set it to just "qmake" rather than an absolute path in the settings?
<nschoe>
sphalerite, I can try. The issue is that I grepped the QtProjects directory and there are *tons* of references to /nix/store/... so.. I'm afraid nothing will work :/
<sphalerite>
ah, what a pain
__monty__ has joined #nixos
<sphalerite>
yeah in that case, at least with qt creator, nix will not be fun :(
<nschoe>
Damn ^^
<nschoe>
sphalerite, this is even weirder: when I launch qtcreator and go to the settings, it detects that the current Qt version is invalid (the path disappeared) but I can't change it, so it doesn't detect the new version of Qt :/ Is smth broken here?
<{^_^}>
[nixpkgs] @xeji opened pull request #43279 → [18.03] linux-hardened: Adjust for Linux 4.17 → https://git.io/fNU2Z
<{^_^}>
[nixpkgs] @jtojnar pushed to master « nixos/udisks2: use upstream unit »: https://git.io/fNU2c
Coyin has quit [Ping timeout: 256 seconds]
camsbury has joined #nixos
Coyin has joined #nixos
<{^_^}>
[nixpkgs] @jtojnar merged pull request #43222 → nixos/udisks2: use upstream unit → https://git.io/fNvAs
nlytend has quit [Read error: Connection reset by peer]
camsbury has quit [Ping timeout: 244 seconds]
nlytend has joined #nixos
sbdchd has joined #nixos
init_6 has joined #nixos
orivej has joined #nixos
mmercier_ has quit [Quit: mmercier_]
sbdchd has quit [Ping timeout: 260 seconds]
mmercier_ has joined #nixos
logzet has joined #nixos
mariatsji has joined #nixos
wucke13 has joined #nixos
nschoe has quit [Ping timeout: 240 seconds]
mariatsji has quit [Remote host closed the connection]
<wucke13>
Why does the first thing complain about feature not available in stable version but the second one works just fine? https://pastebin.com/xvswCK8v
<hodapp>
hmmm. Blender has a lot of add-on packages and some of them need paths to (in theory) other things that would be in the Nix store... I wonder if anyone has tried to make Nixpkgs handle this
timon37 has joined #nixos
<hodapp>
though I guess this isn't really a big problem since it's still possible to install them directly from Blender
<Yaniel>
mm... blender.withPackages when
wgas_ has left #nixos ["Leaving"]
<hodapp>
you a Blender user?
<Yaniel>
casual
<Yaniel>
but yes
<hodapp>
I bounce between avoiding it and using it...
disasm has joined #nixos
<hodapp>
if Blender's method of installing add-ons (if you do it manually) is just dumping the files in some directory this would seem to be feasible to make Nix do
phdoerfler has quit [Quit: Leaving.]
<hodapp>
and a lot of them have configuration that may depend on specific packages (e.g. Blenderseed needs to know where Appleseed is), so there'd need to be some kind of custom configuration it could set
iyzsong has joined #nixos
<Yaniel>
I haven't used extra addons much
<Yaniel>
so far it's been enough to dump them in ~/.config/blender/add-ons
freeze_ has joined #nixos
<Taneb>
Is it possible to start a nix shell that has a shell.nix's stuff plus some extra packages?
<hodapp>
oh, bleh, I think it stores add-on configuration in ~/.config/blender/.../config/userpref.blend which is one of Blender's usual monolithic binary blobs
<hodapp>
so maybe this won't work for any add-on requiring configuration :[
<freeze_>
hi, I'd like to use the nixpkgs/pkgs/applications/editors/vscode/with-extensions.nix expression to install different extension sets in nix-shells (i tested, it worked), but all the wrapper scripts are called "code", I'd like to customize this file name and also add a --user-data-dir param in it. the solution I imagine is patching the .nix file to add those two things, but I don't know if there is some pattern/guideline for this ?
<freeze_>
adisbladis[m]: so to sum up: I create a new derivation depending on with-extensions and call wrapProgram to inject my new param and then I don't really need to rename it, because it will take the name of my new derivation ?
<freeze_>
oh it is simpler
<adisbladis[m]>
It's much simpler :) You override vscode-with-extensions, and overrides can be chained :)
<freeze_>
adisbladis[m]: then I just bundle it in some function "mkCustomVscode" in my overlay, and I can use it for my different needs
<dami0>
hi, i've tried setting the HTTP_PROXY, HTTPS_PROXY, http_proxy, https_proxy and CURL_NIX_FLAGS, but i can't do a nix-rebuild switch to install packages. this worked during install, is there some more correct way of setting the http proxy?
<worldofpeace>
Can someone review/merge :P #43166 and #43167 ?
MercurialAlchemi has quit [Ping timeout: 244 seconds]
<freeze_>
worldofpeace: adisbladis[m] : is there a reason why there are only 3 extensions in the nixpkgs repo ?
jperras has joined #nixos
<worldofpeace>
freeze_: Packaging all of the vscode extensions would be pointless
<freeze_>
worldofpeace: yes but maybe the most popular ? hackage, pip &al contain more stuff for example
<worldofpeace>
freeze_: the nix syntax one makes sense. I'm sure no one would mind adding a few more as long as they're maintained.
<freeze_>
ok I'll think about it then ;)
<etu>
freeze_: PR's are in general welcome :)
fendor has joined #nixos
<freeze_>
adisbladis[m]: I have almost a working solution for the vscode stuff ... I try to add a call to substituteInPlace but can't figure out where to put it in the override chain
nuncanada has joined #nixos
nuncanada has quit [Remote host closed the connection]
tzemanovic has quit [Remote host closed the connection]
<freeze_>
etu: I just need to take the time to scrash my main machine and install it with nixos, atm I just play with a few vms to feel more comfortable with this lovely beast
sbdchd has joined #nixos
freeman42x]NixOS has joined #nixos
FRidh has joined #nixos
rschm has joined #nixos
ckauhaus has quit [Ping timeout: 260 seconds]
<{^_^}>
[nixos-org-configurations] @zimbatm opened pull request #43 → add graham and zimbatm to the bastion host → https://git.io/fNUSF
<nschoe>
Hi, is there any chance someone here used nixOS, ROS and Qt at the same time? :D (I remember acowley helped me with nixOS and ROS, maybe I have my chances?)
<{^_^}>
[cabal2nix] @peti pushed to master « bump version number to 2.9.3 »: https://git.io/fNU9h
<{^_^}>
[cabal2nix] @peti pushed 0 commits to refs/tags/2.9.3: https://git.io/fNU9j
nlytend has quit [Read error: Connection reset by peer]
nlytend has joined #nixos
<worldofpeace>
There's supposed to be periods at the end of meta descriptions?
hxrts_ has quit [Remote host closed the connection]
armin has quit [Quit: Without a solid grasp of theoretical physics, most of the jokes will go over a typical user's head.]
Coyin has quit [Ping timeout: 256 seconds]
tzemanovic has joined #nixos
sk8forether has joined #nixos
Coyin has joined #nixos
ckauhaus has quit [Ping timeout: 240 seconds]
coot has joined #nixos
ckauhaus has joined #nixos
<{^_^}>
[nixpkgs] @Ma27 opened pull request #43282 → nixos/ohMyZsh: allow multiple derivations in `ZSH_CUSTOM` → https://git.io/fNUdD
tzemanovic has quit [Ping timeout: 244 seconds]
andymandias has quit [Ping timeout: 244 seconds]
ckauhaus has quit [Remote host closed the connection]
ckauhaus has joined #nixos
<{^_^}>
[cabal2nix] @peti pushed to master « git-annex: drop uncompilable support for Webapp and S3 »: https://git.io/fNUFm
<{^_^}>
[nixpkgs] @offlinehacker closed pull request #43135 → dockerTools.buildImage: passthru image name and tag → https://git.io/fNvYZ
ckauhaus has quit [Ping timeout: 244 seconds]
trcc has quit [Remote host closed the connection]
palo_ is now known as palo
trcc has joined #nixos
andymandias has joined #nixos
trcc has quit [Remote host closed the connection]
trcc has joined #nixos
trcc has quit [Remote host closed the connection]
trcc has joined #nixos
<timokau[m]>
How can I force a rebuild with `nix build`? I'm almost certain there was a flag for that, something to do with validating reporducibility .
<gchristensen>
--check
<symphorien>
--check
<gchristensen>
symphorien++
<{^_^}>
symphorien's karma got increased to 1
<symphorien>
gchristensen++
<{^_^}>
gchristensen's karma got increased to 12
<symphorien>
for symmetry :)
<gchristensen>
:D
sir_guy_carleton has joined #nixos
knupfer has joined #nixos
trcc has quit [Ping timeout: 240 seconds]
<timokau[m]>
Thanks :D Although that only works with `nix-build`, not with nix 2.0
<timokau[m]>
Thanks :D Although that only works with `nix-build`, not with nix 2.0
<timokau[m]>
(Which apparently still lacks a manpage?)
<gchristensen>
`nix build`'s interface is still beta and subject to change
<LnL>
that's somewhat intentional since the interface isn't stable yet
justbeingglad has joined #nixos
armin has joined #nixos
justbeingglad has left #nixos [#nixos]
ckauhaus has joined #nixos
<{^_^}>
[cabal2nix] @peti pushed to master « Remember to run tests without $HOME defined. »: https://git.io/fNUNY
<{^_^}>
[nixpkgs] @matthewbauer merged pull request #43274 → androidndk: Use `callPackage` to support overriding the `fullNdk` option → https://git.io/fNUnu
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/fNUNl
ckauhaus has quit [Ping timeout: 264 seconds]
kamil- has joined #nixos
<{^_^}>
[nixpkgs] @OliverEvans96 opened pull request #43284 → Typo in nixos/tests/taskserver.nix → https://git.io/fNUNp
<timokau[m]>
I didn't know that `nix build` is still in beta, I thought all of nix 2.0 was stable
<timokau[m]>
Although its probably a very good idea to stabilize and ship it for one version before setting the exact interface in stone
Coyin has quit [Quit: Leaving]
<gchristensen>
timokau[m]: unfortunately it is hard to communicate that
<timokau[m]>
gchristensen: Re expermental: Maybe the "nix 2.0 is now stable" blog post didn't make that point clear enough. Maybe I'm just misremembering though. Either way, its a good idea.
<gchristensen>
because, then, too, much better signal can be provided: specifics on how the closure size has changed, if it could build before / could build now, build time differences, build closure changes, etc
<{^_^}>
[nixos-org-configurations] @zimbatm opened pull request #44 → terraform: introduce DNS resource definitions → https://git.io/fNUhj
hyper_ch2 has quit [Quit: Page closed]
<ryantm>
gchristensen: You're saying even if it doesn't 404 it might not be a good signal, right?
<gchristensen>
yeah
<gchristensen>
because then you have to know what you're looking for and how to find it, and then to know what it means
<ryantm>
gchristensen: well, building before and after sounds great, and build caching might mean it's often pretty quick.
<ryantm>
gchristensen: I'll go ahead and close that PR.
<gchristensen>
true
<gchristensen>
maybe we can fetch data from the Hydra API (or add data to the Hydra API so it is useful)
rardiol1 has joined #nixos
drakonis has joined #nixos
kamil- has quit [Quit: Leaving...]
<timokau[m]>
I'd like linking to the hydra jobset for another reason: The PR creator is very likely interested in the package and might want to subscribe to the hydra job. And navigating hydra and finding the job yourself is very confusing.
<gchristensen>
that should probably be solved in hydra, then
<timokau[m]>
Then again it would be even better to automatically email maintainers about hydra failures (wasn't that the case? Is it still? I haven't gotten an email for a while)
<gchristensen>
ofborg shouldn't be trying to solve Hydra's UI issues
<timokau[m]>
True, that'd be a hack
<gchristensen>
doing some 404 checking, it looks like S3 is the best backend out of S3, CloudFront, GCP, Azure
simukis has quit [Ping timeout: 244 seconds]
<gchristensen>
for a nixos cache
kamil has joined #nixos
Ariakenom has quit [Read error: Connection reset by peer]
<srk>
[6~\
<adisbladis[m]>
gchristensen: Latency wise or for some other reasons?
Ariakenom has joined #nixos
Ariakenom has quit [Read error: Connection reset by peer]
<gchristensen>
mostly, yeah
<gchristensen>
Nix issues lots of requests and the faster the 404 the better
Ariakenom has joined #nixos
<adisbladis[m]>
That's currently only true for us/eu
<adisbladis[m]>
Asia/australia/nz has pretty horrible nixos cache latencies currently
<gchristensen>
yeah, it is naturally a region-specific question/answer. best would be a s3 bucket per region your user has. second to that, cloudfront in every region they offer
nlytend has quit [Read error: Connection reset by peer]
alex`` has joined #nixos
nlytend has joined #nixos
<{^_^}>
[nixpkgs] @orivej opened pull request #43290 → qt511.qtbase: fix darwin build → https://git.io/fNTfe
shiver32 is now known as shiver
<ryantm>
I'm probably going to add a link to Hydra in nixpkgs-update PR messages. It seems especially relevant for updating packages (rather than just adding them), and I can check if the page is a 404 before including the link.
revtintin has quit [Quit: WeeChat 2.1]
<gchristensen>
ok, cool
<{^_^}>
[nixpkgs] @adisbladis merged pull request #43288 → plasma-workspace: Add patch for missing systemmonitor exec line → https://git.io/fNUhU
<cocreature>
Hey, when I install firefox-devedition using "nix-env -iA nixos.firefox-devedition-bin", I get "firefox-devedition" symlinked to ~/.nix-profile/bin but also "firefox" which seems to come from the unwrapped derivation. a) why is that the case? I don’t think dependencies are symlinked into the environment in general. and b) is there a way to avoid this? I would like to have both regular firefox and
<cocreature>
firefox-devedition installed side by side
<ryantm>
Whether or not it actually contains super useful information, I think the PR reviewers are looking at it, so I'd like to make it easier for them to get to it.
phdoerfler has joined #nixos
Guest75007 has quit [Ping timeout: 240 seconds]
<Taneb>
What's the best way to define a Hydra project that builds a branch of a Github repo that is not associated with a PR?
<gchristensen>
Taneb: after the repo URl in the input, just add a space, then the branch name
<tilpner>
gchristensen - Clicking on linenumbers in search.nix.gsc.io opens a GH URL. Pasting those into chat may result in logs bitrotting over time as the repo evolves. Permalinks would improve that situation
<gchristensen>
oh, so linking to the commit-version instead of branch-version?
<tilpner>
gchristensen - But I was assuming search.nix.gsc.io was a self-written thing, which may not be the case
<dami0>
tilpner: thanks, i think i'm starting to get it. so considering only vimrc is defined there but no other stuff, would i need to extend the vim-utils.nix file to then make a derivative package of vim that specified custom filetype.vim files?
stepcut has quit [Remote host closed the connection]
<gchristensen>
tilpner: its by Etsy, actually: https://github.com/etsy/hound but maybe somebody wrote a patch for that
stepcut has joined #nixos
<gchristensen>
unfortunately the author of Hound no longer works at Etsy and Etsy doesn't do active maintenance, so it is half-way to abandonware
<gchristensen>
but it is _very good_ so worth manually applying PR patches :)
* tilpner
didn't find such a PR, or an issue for it
<gchristensen>
me either
stepcut has quit [Ping timeout: 240 seconds]
tzemanovic has joined #nixos
nlytend has joined #nixos
<tilpner>
dami0 - If you want to keep a consistent interface, that would be the easiest option. You could then PR that change so you don't have to maintain a copy of the vim expr
<tilpner>
(AFAICT)
chessai has quit [Ping timeout: 240 seconds]
tzemanovic has quit [Ping timeout: 244 seconds]
<kalbasit>
srhb: I switched to the unstable channel, when I install kubernetes now (overlay removed of course) I see the error `/nix/store/62fqqskmfgiihzq772rsx3d3bl60z6ph-kubernetes-1.10.5/bin/.kube-addons-wrapped: line 158: awk: command not found`
nschoe has quit [Quit: Program. Terminated.]
<{^_^}>
[cabal2nix] @domenkozar pushed 2 commits to refactor-cabal2nix: https://git.io/fNTIP
<kalbasit>
and I get `kubelet.go:1359] Failed to start ContainerManager failed to get rootfs info: failed to get device for dir "/var/lib/kubernetes": could not find device with major: 0, minor: 28 in cached partitions map`
shapr has joined #nixos
<shapr>
greetings! I installed another drive in my laptop, is there an easy way to install nix on a separate drive from within my existing ubuntu?
<shapr>
er, nixOS specifically?
<shapr>
I'd like to dual-boot
<clever>
shapr: uefi or legacy booting?
<shapr>
uefi, figured I'd chain from /boot on my primary drive
<clever>
if using uefi, then you can just entirely ignore the ubuntu drive, give nixos its own efi system partition and rootfs, and then youll have an option for both in the UEFI menu
<shapr>
oh, that does sound simpler
<shapr>
that would also let me swap this drive to another system and it'd work fine
<clever>
it relies in the efi vars in the nvram being setup right
* shapr
reallocates partitions
<clever>
so another machine will refuse to boot it by default
<shapr>
oh?
<{^_^}>
[nixpkgs] @eliasp opened pull request #43293 → hetzner-kube: init at version 0.3.1 → https://git.io/fNTLI
<clever>
uefi is a pain like that
<shapr>
:-/
<clever>
but there is a removable option to get around it
<clever>
shapr: id also recomend that the efi sys partition be at least 512mb, maybe 1gig
<shapr>
that's fine, I'm installing on a terabyte drive
redj_ has joined #nixos
<shapr>
does the nix install process do partition creation or formatting?
<clever>
shapr: that has to be done manually
dbmikus_ has joined #nixos
<clever>
you need to partition, format, and mount everything under /mnt, then run nixos-generate-config --root /mnt, edit /mnt/etc/nixos/configuration.nix, and then nixos-install --root /mnt
<shapr>
ok, works for me
humanoyd has quit [Quit: WeeChat 2.1]
redj has quit [Ping timeout: 268 seconds]
jensens has quit [Ping timeout: 248 seconds]
mariatsji has joined #nixos
<betaboon>
and then you have to checkout your awesome git repo, with all your awesome nix-expressions that create your system just the way you always wanted :D
<acowley>
Does anyone have an approach they're happy with for emacs packages given that they don't get updated in nixpkgs very often anymore?
<acowley>
I have a few specific definitions that I create manually when I know I need an update, but I'd like it to be somewhat more automated.
<gchristensen>
we should update them
<acowley>
I think there's lingering uncertainty about how to manage emacs packages
<infinisil>
They aren't updated? I thought there was some auto-update thing going on
<infinisil>
auto-generated i mean
<acowley>
There was for quite a while rather nasty conflict between melpa-stable and melpa, and things sort of fizzled from those discussions when everyone became marginally satisfied with their current setups.
<acowley>
At least that was my impression.
<acowley>
We have auto-generated package lists, but they're several months old.
doyougnu has joined #nixos
<infinisil>
Ah that one, yeah I've been annoyed by that too previously, but how it's now I like it
<infinisil>
melpa is preferred over melpa-stable now
<acowley>
Yeah, I got mine squared away, and didn't help prod things along
dbmikus__ has joined #nixos
<acowley>
But I think Thomas mentioned on twitter a while back that he's not entirely happy with things the way they are.
<ldlework>
acowley: yeah, I use straight.el instead
<ldlework>
I don't like the idea of specifying my packages in Nix, and then respecifying them in my emacs config
<acowley>
I think there are lingering issues with packages that have executable components (most prominently pdf-tools).
<acowley>
ldlework: I find it's okay because my nix specifications are very terse
pie_ has joined #nixos
<ldlework>
still always two places to update
<acowley>
ldlework: In nix, I for the most part just name every package. But that naming provides a convenient opportunity for applying patches. Then I use use-package in my emacs.el to configure things. It's slightly redundant, but I quite like the snapshot aspect nixpkgs provides.
mariatsji has quit [Remote host closed the connection]
logzet has quit [Remote host closed the connection]
<acowley>
I don't update the use-package parts
<johnw>
acowley: same here
<acowley>
I don't actually update anything manually
<acowley>
Except for the custom patches part
<ldlework>
straight.el is like nix for emacs, but in elisp. it uses straight.el to get the packages from melpa and the like
<acowley>
But that seems essential
<johnw>
it wouldn't be very hard though to come up with a Nix module that allows you to specify use-package declarations using Nix
<ldlework>
I don't understand how you dont update the use-packages part
<ldlework>
if you add a new package, you have to tell nix and emacs about it
<acowley>
ldlework: Oh, for adding things, yes. I thought you meant package updates.
jperras has joined #nixos
<ldlework>
where as if you use straight.el, use-package will cause the package to be installed
<acowley>
ldlework: Yes, that duplication is an irritant.
ajs124 has left #nixos ["Machine going to sleep"]
<ldlework>
acowley: if you're already using using-package, I don't see the point in having nix install the emacs packages
<infinisil>
Personally I'm using my own emacs modules using the modules system, they work without NixOS
<ldlework>
like, not in terms of preference, but I don't see any utility whatsoever
<acowley>
johnw: I actually like the separation as it makes it so easy to switch how I provide the packages!
kamil has quit [Quit: Leaving...]
<acowley>
ldlework: Compared to use-package installing from melpa, I enjoy the aforementioned opportunity for applying patches during the installation phase. Compared to mangaging my own clones, I like being able to rollback everything if something goes sideways.
<ldlework>
i have never needed to apply a patch to an emacs package I guess
gsc70cstr has quit [Quit: leaving]
<acowley>
I'm not disputing there are many ways to do all of this. But I was wondering if anyone who is using nix for emacs packages has a personal workflow that involves creating their own melpa snapshots.
<acowley>
ldlework: I have several! Sometimes it's a patch that hasn't made it to master yet, sometimes it's to customize something the package doesn't expose in a convenient way... like tweaking a theme or something.
<acowley>
But if something goes wrong there, I'm in a bit of trouble because I don't have a hash that identifies my entire package set.
<acowley>
infinisil: What Thomas mentioned before was the idea of having a melpa2nix or something that takes a user's list of packages and generates nix expressions from a given melpa checkout.
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
<acowley>
infinisil: I don't think either of us ever followed up on the idea. It seems like a way of relieving anyone from having to update all snapshots. But if any patching is needed to address nix concerns (e.g. pdf-tools), things may get complicated.
<acowley>
ldlework: It's fine, I've been through many iterations of my emacs config over the years and expect to go through many more!
<infinisil>
acowley: Something like we do for hackage could work
<jasom>
andi-: thanks, I'll try that tonigh
<acowley>
infinisil: I think so, yes
<infinisil>
Having the base layer autogenerated, but allow for different mods on top of that
<dami0>
tilpner: thanks, when i understand nix enough, i'll surely do it. i'm surprised noone did a .nix filetype for vim already
<acowley>
ldlework: But I've found that the most useful thing in terms of sharing with other folks is sharing snippets rather than the whole config. I do the same in the reverse direction, too. You will find various johnw-isms in my configuration files.
<acowley>
ldlework: Org's include feature is a little iffy for parameterization. Things like macros don't always work as one might hope.
stepcut has joined #nixos
<acowley>
ldlework: That said, it's org so it's fixable.
andymandias has joined #nixos
<ldlework>
acowley: we have just been using setq-default in the snippet
<ldlework>
acowley: you do setq before the import if you wanna override
<ldlework>
I'm not an elisp genius...
nlytend has quit [Quit: disconnected.]
<ldlework>
but that -does- work :)
<acowley>
That's at *least* half the battle :P
init_6 has quit [Ping timeout: 264 seconds]
<ldlework>
acowley: there were two people who were constantly tracking changes from my emacs config who were complaining that I was constantly iterating the parts they had stolen. they don't complain anymore because that's solved.
<ldlework>
if johnw used this, it would be a revolution in emacs config sharing :)
<ldlework>
I'm always hesitant to take things from people like johnw because no doubt they will improve it later and the chances of me noticing and updating my own copy is very very low
<sphalerite>
what's a good program I can use to just display a live feed from a V4L2 device and allow snapshotting it?
<sphalerite>
Tried cheese just now but some weird issues, the window it opens doesn't display anything and I get a bunch of error messages
<goibhniu>
vlc?
<clever>
sphalerite: i think vlc can do it
<clever>
mplayer too
<sphalerite>
will try, thanks!
Izorkin has joined #nixos
__monty__ has joined #nixos
reinzelmann has quit [Quit: Leaving]
<sphalerite>
yep it works, thanks!
nlytend has quit [Ping timeout: 260 seconds]
stepcut has quit [Ping timeout: 260 seconds]
reinzelmann has joined #nixos
igo95862 has joined #nixos
<gchristensen>
hey so how do I get cabal?
<clever>
gchristensen: are you in a shell from haskellPackages.mkDerivation?
<nixdummy>
I'm trying to hackily manage windows programs/settings with nix. Is there an easy way to specify that a package is for WSL instead of regular linux? I don't see any flags in stdenv.hostPlatform that would differentiate the two.
mroutis has joined #nixos
<gchristensen>
clever: I think so, its a shell.nix generated by cabal2nix
<clever>
gchristensen: that only has the cabal library, not the binary, you need to run Setup.hs
<gchristensen>
I want to run `cabal test`
<clever>
either `runhaskell Setup.hs configure` or compile it with ghc, then run it
<clever>
`runhaskell Setup.hs test`
<cocreature>
I tend to just install cabal globally
<infinisil>
clever: You sure he doesn't just want to install `cabal-install`?
<clever>
that can break things, because it wont be the same version as what the nix expressions want
<bpye>
Agh, I asked this the other day. I'm trying to use import to seperate out some config values that need to be referenced multiple times, how do I get this working correctly? Currently it seems they can't be seen in my config
<infinisil>
bpye: With your nixos config?
disasm has joined #nixos
<cocreature>
clever: the version of cabal-install that you use, does not determine the version of Cabal that will be used to compile your Setup.hs file
<gchristensen>
thank you a lot, clever!
<bpye>
Wait I'm silly, I worked it out, yeah infinisil
<clever>
cocreature: but if you ignore the Setup.hs entirely, and run `cabal configure` + `cabal build`, it may be the wrong version of cabal
<cocreature>
clever: are you sure about that? afaik for a build-type Simple, you are basically saying that the version of Cabal doesn’t matter and for build-type: Custom, it will actually compile your Setup.hs with the correct version (i.e. the one that is brought in scope by your nix shell)
<cocreature>
at least this has literally never caused problems for me but maybe I’ve just been lucky :)
<clever>
cocreature: i have seen it break for others before, when they manually installed cabal-install with nix-env, and then tried to manually build in a shell
<cocreature>
hm weird
<clever>
cocreature: id expect it to only break a few years down the road, when the cabal in nix-env has aged, and your using a much newer nixpkgs for the nix-shell
<makefu>
gchristensen: i continued a bit regarding the intercepting proxy thing, it looks like the version i get with nix (after nix-shell -p curl) is not able to connect through the proxy. This is curl -vvvv of the working call https://paste.krebsco.de/NKkQ7TXt and this is how it looks with the nix version: https://paste.krebsco.de/aifqU3jY
<cocreature>
oh right, if the version of cabal-install is _older_, I could imagine that you run into problems
<cocreature>
I’ve only tried with newer or equal versions of cabal-install
<clever>
cocreature: but Setup.hs will always use the library that the nix-shell provided, so its always perfect
<cocreature>
sure but interacting with Setup.hs manually can be annoying for more complex workflows (e.g. multipackage projects)
<{^_^}>
[cabal2nix] @peti pushed 0 commits to refactor-cabal2nix: https://git.io/fNTWB
<lewo``>
jD91mZM2: there are tools dep2nix and go2nix v2
muzzy__ has quit [Ping timeout: 260 seconds]
<jD91mZM2>
Oh, awesome, thanks!
rindolf has joined #nixos
<rindolf>
Hi all! nix takes 71% of my VM's disk. How can I trim it?
<gchristensen>
nix-collect-garbage
<maurer>
rindolf: nix-store --gc
<maurer>
or that
<rindolf>
thanks
<dtz>
Sonarpulse: from another channel, but thought you'd chuckle: > the problem is that simple questions like "what target tuples does the compiler support?" and "what is the correct target tuple for me to use?" can only be answered by reading the compiler source code, or by copying what other people are using and hoping that's right.
<dtz>
haha actually nevermind it's even less funny after copying it
<dtz>
:(
<rindolf>
maurer: gchristensen : both cmds didnt help me
<maurer>
rindolf: That probably means that you either have roots still pointing to old generations or old builds
<maurer>
are you using the system for dev, or is it just being used a normalish machine?
<manveru>
i really wonder where those memory gcroots come from...
<LnL>
stuff that's still running :)
dami0 has quit [Remote host closed the connection]
<manveru>
ah :D
dsx has quit [Ping timeout: 260 seconds]
<manveru>
is that new?
<clever>
manveru: nix1 did it, but silently didnt tell you about them
<LnL>
don't think so
<manveru>
ok
<clever>
which lead to confusion, nix-store --query --roots said a path had no roots
<clever>
but --delete said it had roots keeping in in-use
<srk>
cool, was wondering about that yesterday as well
<manveru>
LnL: btw, do you have some kind of presentation of nix for macos users? gonna do a talk in our company next week
<manveru>
i'm pretty worried that the nix setup might go wrong in very strange ways again :|
<nixdummy>
Is there an easy way to specify that a package is for WSL instead of regular linux? I don't see any flags in stdenv.hostPlatform that would differentiate the two.
<LnL>
or replace it with some kind of 'upgrade to nix-daemon' workflow, but that might be a bit harder to implement
<manveru>
for now just using --no-daemon seems fine for me
<samueldr>
it may be a good idea to save installer settings (monotonically increasing version, parameters) to help migrating installations in the future?
<samueldr>
may make it safer to do some assumptions?
<worldofpeace>
cocreature: You didn't have to do that :P
<rindolf>
nix-env -e foi does not give an error
reinzelmann has quit [Quit: Leaving]
<rindolf>
nix-env -e kde does nothing too
<cocreature>
worldofpeace: sorry, I thought I might make life for you a bit easier :)
<clever>
rindolf: only names listed in `nix-env -q` can be removed with `nix-env -e`
<nixdummy>
samueldr + manveru: unfortunate. thanks for the help, though.
<rindolf>
clever: i see
<rindolf>
clever: how do i purge qt/kde?
<clever>
rindolf: remove every package that depends on QT
<clever>
rindolf: can you gist the output of `nix-store -qR ~/.nix-profile` ?
<PyroLagus>
so, i see on howoldis that unstable is 3 days old because of an osx test. does that just block updates for osx or also for linux?
reinzelmann has joined #nixos
<clever>
PyroLagus: thats nixpkgs-unstable, which is used outside of nixos (ubuntu, debian, darwin, and so on)
<clever>
PyroLagus: nixos-unstable only does linux testing, and tests more OS related things, to ensure nixos can still boo
<clever>
t
<PyroLagus>
ah, so for nixos installs only nixos-* is relevant?
<clever>
PyroLagus: yeah
<clever>
PyroLagus: and if you install nixos from nixpkgs-unstable, you risk breaking the install
<PyroLagus>
okay, cool
<clever>
about a year ago, grub was broken, and corrupted the grub config file
<clever>
nixos-unstable correctly refused to update
<PyroLagus>
i thought the nixpkgs-* stuff also applied to nixos systems
<clever>
then all of the users with nixpkgs-unstable came out of the woodwork :P
<clever>
that bug also broke the ability to rollback via the grub menu
<manveru>
the last part is mostly interesting for you i think
<manveru>
i.e. just keep the names of all packages you want in your file, and install that, when you want to update, just install it again
betaboon has quit [Ping timeout: 256 seconds]
tzemanovic has joined #nixos
<cocreature>
is there a more granular version of require-sigs? i.e. something that allows me to not require sigs for certain caches without turning it off completely?
<clever>
cocreature: why cant you just sign things on the cache?
tmaekawa has joined #nixos
<mkaito>
I'm writing a derivation for a ruby app. I'm using lib.cleanSource, but there are a few specific folders that I want to exclude from the derivation. How would I do that? There doesn't seem to be a function for that.
<cryptocat1094>
So, M4 has the right artifact name but not the right package name. What would be the best fix?
<gchristensen>
I think you can pass ?trusted=1 to the end cocreature
tmaekawa has quit [Client Quit]
<cocreature>
clever: because it’s not my cache
<clever>
mkaito: i think builtins.cleanSourceWith lets you extent the filter
<cocreature>
gchristensen: thanks, I’ll try that!
<{^_^}>
[cabal2nix] @peti pushed to master « cabal2nix.cabal: tentatively bump version to 2.10.0. »: https://git.io/fNT0B
<mkaito>
clever: is there an example somewhere of how to use that? sorry, I'm dense.
<pip3000>
@clever what is the recommended way to manually track changes
<clever>
srk: it pre-dates the network support in the initrd, so it may need updating
<srk>
and for iscsi server?
<clever>
pip3000: i keep a git repo at /etc/nixos/nixcfg/ with 1 file for each host, and then configuration.nix is just the bare-minimum required to boot, and imports = [ ./nixcfg/hostname.nix ];
<srk>
my previous farm was running of nbd with patched nbdserver..
<clever>
srk: i ran tgtd on gentoo, i tried to do some fancy things with the tgtd nixos module but it never worked 100% right
<{^_^}>
[nixpkgs] @peti pushed to master « hackage2nix: cosmetic (avoid use of tabs) »: https://git.io/fNTuE
<srk>
right, I might even settle for nfs as that's very easy to export and manage from zfs
<clever>
srk: zfs/xfs totally break 32bit nixos over nfs
<srk>
oh wow
<clever>
srk: the inode numbers dont fit into a 32bit field, and half the packages on nixos tread EOVERFLOW as an empty directory
<clever>
srk: so all of your builds claim file not found, when the file clearly exists
<clever>
it only happens when a 64bit capable FS is shared to a 32bit client over NFS
<srk>
I see
<clever>
for xfs, it happens with any volume over 2tb in size
<clever>
pip3000: that saves a complete copy of the directory containing core.nix at /run/current-system/nixcfg
<clever>
pip3000: so after a rollback, i can look in there to see what the old config was
<clever>
pip3000: and in addition to that, i also have zfs snapshots enabled, so i can diff any file on the entire machine, over almost any time-period
phdoerfler has quit [Quit: Leaving.]
<clever>
srk: and here is the incantation i used for the not-os image: [root@router:/tftproot/try2]# nix-env -p /nix/var/nix/profiles/per-user/root/rpi3-netboot -f not-os/release.nix -A rpi_image -I nixpkgs=./nixpkgs/ --set
<clever>
srk: the above command will build the rpi_image derivation of release.nix, then create a new generation of the rpi3-netboot profile, pointing to the result
<srk>
so even on pi3 boot without sd card is not possible without bootcode.bin, right?
<cryptocat1094>
Hadn't that been RE'd?
<clever>
the docs claim that is possible, but a bug in the boot rom stops it from working on mine
<srk>
yeah
<srk>
if they could just put eeprom on the board so we could reflash it /o\
<clever>
if you have an sd card with ONLY bootcode.bin (it must not have ANY other file), then it will act like it has no card at all, and try other enabled modes
<clever>
srk: it does support booting from a serial flash chip
<clever>
srk: the compute module saves you from having to deal with the fine pin-pitch and picky signal routing of ram lines, so you can just work with "low-bandwidth" usb and hdmi
<srk>
ls1046a looks quite nice and it's 64bit, I like the dimm form factor of such boards
<clever>
srk: the connector on the compute modules gives you access to hdmi, gpio, the camera and lcd connectors, a single usb port, jtag and composite video
<srk>
still limited to USB2 until you pull-out some craziness like HDMI networking :D
<clever>
srk: i have considered abusing the LCD and camera connectors as high-bandwidth
<srk>
we did as well! few years ago :)
<clever>
srk: there is something else, let me find it
<srk>
but why dance around crappy hardware when you can build solid boards on your own
<clever>
srk: i think the SMI was a 12 bit wide memory region, meant for driving certain types of displays
<grp>
I've fixed an issue in a module. Before going through the hassle of testing every combination to see if this change doesn't break anything, I'd like to ask: is there some test suite I can run against it? I've seen nixpkgs/nixos/tests but I still haven't found how to run them
<clever>
so you could address the framebuffer on an external chip
<clever>
grp: run `nix repl nixos/release.nix` and try to evaluate the tests attribute set, and navigate thru it
xy2_ has joined #nixos
<srk>
clever: nice
<clever>
srk: once you know the attribute path, you can just `nix-build nixos/release.nix -A tests.foo`
<srk>
but SMI is not exposed, right?
<clever>
srk: its on the main gpio header i believe
<{^_^}>
Use `,` to list all commands, `,foo = Foo!` to define foo as "Foo!", `,foo =` to undefine it, `,foo` to output "Foo!", `,foo somebody` to send "Foo!" to the nick somebody
<clever>
srk: and a diff confirms that the only change is in cmdline.txt
<srk>
cool
<clever>
so it makes no sense for it to have such a different file pattern being loaded
<bpye>
I'm a little confused, how do I set the channel I want to be on when using nixops? I appear to be deploying packages from nixpkgs-unstable unintentionally?
<srk>
like, I was struggling a bit with dtbs recently, recompiling kernel to be able to use pps input and /dev/spidevs
<clever>
bpye: i lookup the rev of a channel i want to use, and then run this with the rev
nick_h has quit [Ping timeout: 244 seconds]
<srk>
would be nice to have this per machine! :)
<clever>
srk: changing absolutely nothing on the netboot server, replugging the the rpi, and watching tftp, results in it loading the 1st file pattern again
<bpye>
Is there some way to set that in the deployments file itself?
<{^_^}>
#665 (by grahamc, open): Support multiple versions of nixpkgs in one network
<srk>
off by one
<{^_^}>
[nixpkgs] @dtzWill opened pull request #43306 → ghc{843,861}: add musl variants of "llvm-targets" → https://git.io/fNTos
<srk>
infinisil: I wanted to implement factoids for lambdabot similar to how gthx or kthx does it on #reprap channel but no luck with lambdabot so far :D
<bpye>
Oh wow I see
<bpye>
This has been a missing feature for some time :(
<clever>
srk: got distracted, then i found the pi sitting at kgdb (kernel debugger), with a thermometer in the corner
<clever>
srk: its overheating to the point of malfunctioning, lol
<clever>
its malfunctioning, before it can even load the kernel
<srk>
ah :D
seafood has joined #nixos
tenten8401 has joined #nixos
<bpye>
The release-18.03 branch should be current stable right?
<clever>
bpye: no
nick_h has joined #nixos
<tenten8401>
Pretty sure you have to go to nixpkgs-channels
<clever>
bpye: that is what hydra is testing, and may become the current stable
<clever>
tenten8401: correct
<clever>
bpye: the branches on the nixpkgs-channels repo mirror the channels exactly
<bpye>
Had no idea that repo existed :)
<tenten8401>
Also, would anyone know what's up with PR #42075 by chance? Been waiting for Lutris to get a package for a while and it seems to be failing because "taskset/cpuset not available on your platform", despite working on other distros.
<ambro718>
My system fails to boot when I power it on without a monitor connected, and if I connect the monitor subsequently there is nothing to see. It used to work some time ago. Is it possible that a nixos upgrade broke it?
<ambro718>
it is quite an issue since this is supposed to be a router
phdoerfler has quit [Client Quit]
<ambro718>
I know there is quite some disk activity so I am guessing that the OS is starting to some point
<clever>
ambro718: hdmi or vga?
<clever>
or dvi
nick_h has quit [Ping timeout: 268 seconds]
<ambro718>
clever: connecting a dvi monitor and powering on solves it, but there is a vga connector on the motherboard too
<ambro718>
I have no way of diagnosing what happens since the network doesn't even start
stepcut has quit [Remote host closed the connection]
<ambro718>
and nothing to see on monitor
<clever>
ambro718: after it fails to boot, connect a monitor and reboot, then run `journalctl -b -1`
phdoerfler has joined #nixos
<clever>
that will show the logs from the previous bootup
<ambro718>
yeah just got that idea
<ambro718>
tomorrow though, for now I'll just leave it running
stepcut has joined #nixos
lnikkila_ has joined #nixos
Guest26247 has joined #nixos
<clever>
srk: i put the rpi ontop of the desk fan, since its overheating before it even gets to start.elf
<clever>
srk: lets take this to #nixos-aarch64
phdoerfler has quit [Client Quit]
phdoerfler has joined #nixos
lnikkila_ has quit [Ping timeout: 244 seconds]
stepcut has quit [Ping timeout: 244 seconds]
neof0x has left #nixos ["User left"]
Edes has joined #nixos
phdoerfler has quit [Client Quit]
phdoerfler has joined #nixos
<Edes>
hey, does anyone have any experience with replacing arch with NixOS while not destroying an EFI partition with windows on it
<{^_^}>
[nixpkgs] @xeji pushed commit from @vbgl to master « ocamlPackages.xtmpl: 0.16.0 -> 0.17.0 (#43304) »: https://git.io/fNT9Q
<vajralambda>
in the nix manual, it refers to all-packages.nix. I couldn't find a link in my home, but there are two versions available on /nix/; how can I get to the correct version?
<clever>
vajralambda: you cant edit either one, the entire /nix/store is read-only
<clever>
vajralambda: if you want to edit things, you need to clone nixpkgs from github
stepcut has quit [Remote host closed the connection]
<countoren>
clever: well i can just take the my_vim.text and get it from there but it does not feel right
<infinisil>
countoren: What's your attempt?
jluttine has quit [Ping timeout: 256 seconds]
jperras has quit [Ping timeout: 244 seconds]
<clever>
countoren: the derivation returned by `pkgs.vim_configurable.customize { ..}` is just a pkgs.writeTextFile call, to generate a $out/bin/vim with contents like "#!/bin/sh\nexec /nix/store/9p4kzfi2kg5hfhh1j3dbjz96arpcd4b3-vim_configurable-8.0.1451/bin/vim -u /nix/store/39wqqljfmvhv2v59k59jkn80pa4n02k9-vimrc \"$@\"\n"
stepcut has quit [Remote host closed the connection]
Ariakenom has quit [Read error: Connection reset by peer]
<clever>
countoren: so its not really possible to access the config derivation in an automated manner
telent has quit [Ping timeout: 240 seconds]
<countoren>
there is a function that i can use to create this vimrc?
<clever>
countoren: you would need to look into how nixpkgs generates it
orivej has joined #nixos
<countoren>
btw is it possible to build macvim with vimUtils?
<countoren>
in darwin
spear2 has joined #nixos
jluttine has joined #nixos
UltraPhil has joined #nixos
xy2_ has quit [Ping timeout: 240 seconds]
UltraPhil has quit [Quit: Off to a better world]
zearen has quit [Quit: WeeChat 1.9.1]
coot has quit [Quit: coot]
jackdk has joined #nixos
ericsagnes has quit [Ping timeout: 240 seconds]
<clever>
countoren: pkgs.vim_configurable.customize can still build on darwin
garbas has quit [Quit: WeeChat 1.9.1]
fractalcat has joined #nixos
<countoren>
clever: yes but i was not able to run GUI vim/macvim from there