fendor has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @zimbatm merged pull request #75309 → defaultGemConfig: change 'infocmp' reference from 'ncurses.dev' to 'ncurses' → https://git.io/JeS8M
<{^_^}>
[nixpkgs] @zimbatm pushed commit from @buckley310 to master « defaultGemConfig: change 'infocmp' reference from 'ncurses.dev' to 'ncurses' (#75309) »: https://git.io/JeSjl
<kvda>
how would you go about having services start (ie. postgres) on running nix-shell?
<kvda>
with global NixOS configuration it seems you can add it to services.postgresql, is something similiar possible in nix-shell or do you need a shellHook?
<infinisil>
kvda: shellHook is the only way to do such things
fusion809 has quit [Remote host closed the connection]
<infinisil>
It's just a shell, not a full os after all
<kvda>
mm yes thanks infinisil , abstractions don't reach that far down i guess
boxscape has quit [Remote host closed the connection]
lovesegfault has quit [Ping timeout: 268 seconds]
evanjs has joined #nixos
<evanjs>
Finally back up.... Though I guess I won't know why my login broke earlier today... One thing on my wishlist now though is something that helps flag kernel rebuild-requiring features. Is there a way to tell what options need the kernel rebuilt at a glance?
<evanjs>
Spent all day with my /tmp filling up, only to find out my kernel needed to be rebuilt for `boot.crashDump.enable`. I know it's in the description for the option... But still ...
<o1lo01ol_>
I'm working on a derivation that puts a few things in $out/share/foo and $out/build/bar/baz/. A handful of those artefacts will need to be accessable by some child dependencies. I know I can put the whole derivation in propagatedBuildInputs but that doesn't put the aformentioned handful on the path where they can be found. Can I do this just by `export LIB_FOO = $out/blah/blah`?
<keithy[m]>
Hi guys just curious about BASH_VERSINFO[5], expected to see "nixos" in there ?
<notgne2>
steam likes to update on it's own, bringing in more dependencies in the process, so you will probably want to pull steam at least from the unstable channel
<notgne2>
`sudo nix-channel --add https://nixos.org/channels/nixos-unstable unstable && sudo nix-channel --update` - then you can use `import <unstable> {}` in your config
<notgne2>
you could do `(import <unstable> {}).steam` if you only really want to do this once, or preferably do `let unstable = import <unstable> {}; etc etc`
<Qretin>
Awesome, thankyou!
<notgne2>
Qretin: actually one more thing, since steam is unfree, you will need to do `unstable = import <unstable> { config = config.nixpkgs.config; };` so it uses your normal nixpkgs config (where you have already told it to allow unfree)
endformationage has quit [Ping timeout: 276 seconds]
Qretin has quit [Remote host closed the connection]
bvdw has quit [Read error: Connection reset by peer]
Lears has joined #nixos
bvdw has joined #nixos
[Leary] has quit [Ping timeout: 250 seconds]
<colemickens>
weird, I can boot a 19.09 iso on my xps 9370 but not my new 7390.
<colemickens>
I have SecureBoot disabled, and I was able to boot windows install media
lovesegfault has joined #nixos
<colemickens>
if I boot manual entries in for bootx64 it does the same hanging behavior. if I set a manual entry for refind, the menu loads, but the entries don't boot. really strange
<samueldr>
colemickens: updated firmwares?
<samueldr>
a dell model (or multiple?) had recent firmware updates related to booting from usb iirc
<colemickens>
That's the first thing I did. Had to do it twice of course, the ME update silently failed to apply.
<samueldr>
tried usb ports on the other side of the computer?
<samueldr>
at one point a (multiple?) models of dell had issues on one side but not the other
<colemickens>
I'll be damned, the nixos-unstable img works.
<samueldr>
grub updated
<colemickens>
That's what I get for trying to use a stable img, pfft.
* colemickens
nods
<samueldr>
good to know as a new troubleshooting point
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « python38Packages.Rtree: 0.8.3 -> 0.9.1 »: https://git.io/Je9Ux
PyroLagus has quit [Ping timeout: 246 seconds]
is_null has joined #nixos
scassyy has joined #nixos
mexisme has joined #nixos
ShaRose has quit [Quit: I appear to have left for some reason.]
is_null has quit [Ping timeout: 240 seconds]
mexisme has quit [Ping timeout: 252 seconds]
bahamas has joined #nixos
oida has quit [Remote host closed the connection]
oida has joined #nixos
ShaRose has joined #nixos
<notgne2>
does anybody here run NixOS on an NVME using a BIOS that doesn't support NVME drives, I'm not really sure what the right way to go about resolving this would be (does GRUB support NVME if the BIOS doesn't?)
<clever>
notgne2: grub relies on the efi firmware to provide nvme drivers
<clever>
notgne2: your only option is to put the entire /boot on a non-nvme disk
<notgne2>
will that just be as simple as during the UEFI installs where I can just mount `/mnt/boot` to a seperate disk and have `nixos-generate-config` figure it out?
<clever>
notgne2: yep
dingenskirchen has quit [Quit: dingenskirchen]
dingenskirchen1 has joined #nixos
drakonis has quit [Ping timeout: 265 seconds]
dingenskirchen1 is now known as dingenskirchen
scassyy has quit [Ping timeout: 250 seconds]
<{^_^}>
[nixpkgs] @jonringer opened pull request #75420 → python3Packages.docker: fix darwin tests → https://git.io/Je9Tg
<kolbycrouch>
Does anyone know how I can get a toolchain without tuple prefixes in a stdenv using pkgsCross. Several packages fail to build because the don't respect env variables for $AR, $LD etc..
<angerman>
does anyone have an in-depth nix-perf-optimization walk through? I know there are env vars I can use the the start/stop logic, but some hands one tutorial would be cool if there was one.
chloekek has joined #nixos
hyper_ch2 has joined #nixos
dingenskirchen has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @kalbasit pushed commit from @r-ryantm to staging « go: 1.13.4 -> 1.13.5 (#75236) »: https://git.io/Je9IX
<fuzen>
Anyone know why installing Firefox-bin or nerdfonts hangs?
<tilpner>
fuzen: nerdfonts downloads multiple GB, which might appear to hang if your connection isn't very fast
<fuzen>
Oh okay, just don’t see an active status bar
smatting has joined #nixos
joshuagl has joined #nixos
<raboof>
as a developer, nixos can sometimes be frustrating because tools might need fiddling to work with nix (patchelf'ing, finding dependencies). on the other hand, I'm really starting to appreciate having a `shell.nix` for each project and having them not interfere with each other.
<fuzen>
I have been using nix for developing for far longer than I have been on NixOS for that reason
<bahamas>
how can I start ghci with a specific package available?
mmercier has joined #nixos
<bahamas>
I figured it out: `nix-shell -p "ghc.withPackages (pkgs: [ pkgs.hex ])"`
__Sander__ has joined #nixos
sudoforge has quit [Ping timeout: 265 seconds]
<mmercier>
Hi! I'd like to package python application using the Nuitka python compiler. I manage to do it manually, but I'm not sure how to create a package using it. Do I need to override the buildPythonPackage builder?
<mmercier>
If anyone as an example on using Nuitka through Nix I'll be very happy to see it :)
cjpbirkbeck has quit [Quit: Quitting now.]
knupfer has joined #nixos
ThatDocsLady has joined #nixos
noudle has joined #nixos
FRidh has quit [Quit: Konversation terminated!]
<{^_^}>
[nixpkgs] @peti pushed 3 commits to haskell-updates: https://git.io/Je9L5
sigmundv has joined #nixos
<{^_^}>
[nixpkgs] @peti opened pull request #75429 → LTS Haskell 14.17 (plus other fixes) → https://git.io/Je9tm
<dirkx>
I am pulling in a .tar.gz in my nixos package that always has a digital signature with clear, long term governance. Should I verify that key rather than the SHA256 of the result being pulled in ? Or should a `default.nix' fetch ALWAYS contain a SHA256 in preference to all else ?
<tilpner>
dirkx: Nix only believes you that your network-during-build derivation will always build to the same, if you give a hash
<niksnut>
dirkx: just the sha256
<dirkx>
tilpner: Ok - so a hash only it is then.
knupfer has quit [Ping timeout: 265 seconds]
<dirkx>
niksnut: ack - dank
<niksnut>
you can verify the signature before updating the hash of course
<dirkx>
niksnut: we actually generate the default.nix at the same time we sign & check. So that part is fine.
<qubasa>
.
<qubasa>
Does someone know how to get the c stdlib for 32bit? If I try to compile something with clang I get the error "gnu/stubs-32.h' file not found"
<dirkx>
b.t.w. if I have 1 main package and 6 packages that need that main one anyway — is there a neat way to refer to a convenice fetcher already defined in that main package ? Can you do something like src = mainpgname.myfetcher ….. when mainpgname is already in the {} on the first line ?
<Taneb>
When I cancel a build using Hydra's web interface, I can't work out how this propagates to actually aborting the build. How does this happen?
<niksnut>
this will probably go away though, I recently rewrote hydra-queue-runner to use the Store abstraction rather than connecting to remote builders directly
<niksnut>
so it won't know about ssh processes to kill anymore
bahamas has quit [Quit: leaving]
reactormonk has joined #nixos
<reactormonk>
Are .so files supposed to be linker files? Got some software which seems to choke on it.
<reactormonk>
Ah no, the linker script seems to work just fine. can't load .so/.DLL for: /nix/store/pnd2kl27sag76h23wa5kl95a76n3k9i3-glibc-2.27/lib/libpthread.so (/nix/store/pnd2kl27sag76h23wa5kl95a76n3k9i3-glibc-2.27/lib/libpthread.so.0: undefined symbol: __libc_vfork, version GLIBC_PRIVATE)
<reactormonk>
Aka it correctly links in the .0 file
<reactormonk>
nm seems to find that symbol though - U __libc_vfork@@GLIBC_PRIVATE
<reactormonk>
Huh, U means undefined.
<Yaniel>
a
<reactormonk>
I'm on an Arch currently, and no vfork here... nm /usr/lib/libpthread.so.0 | grep vfork
<reactormonk>
But multi-project HIE doesn't work with stack, so I'm in cabal land again
sigmundv has quit [Ping timeout: 265 seconds]
<Shoubit>
reactormonk, do you call nix-shell with the --pure flag?
<reactormonk>
Shoubit: to be a bit more clear, I'm loading it in via vscode nix integration. And no, I don't think so.
<Shoubit>
System level libraries can pollute your nix-shell environment if you don't call it with --pure on non-NixOS systems iirc
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
philr has joined #nixos
<DigitalKiwi>
i haven't used nix on it but they've always packaged haskell in an...well *tinfoil hat* i think they don't like it and are actively antagonistic of it in their packaging :|
<{^_^}>
[nixpkgs] @teto pushed commit from @minijackson to master « vimPlugins.LanguageClient-neovim: 0.1.154 -> 0.1.155 »: https://git.io/Je9md
EdLin has joined #nixos
EdLin has quit [Client Quit]
<dirkx>
in a dependency I've got a nice stdenv.mkDerivation rec { … } that can be reused for this sub-module (with just a different name, description, url) — is there an easy way to simply 'overwride' it — e.g. do something like '{stdenv, coreutils, mothership }; with mothership; stdenv.mkDerivation rec { name = "extrabit
<dirkx>
and then get all the configure and other flags/settings ?
<reactormonk>
So I did some fuckery with putting the libpthread from arch into the nix repo, now I'm getting another error - I thing somehow the build of the IDE tool is messed up because it doesn't seem to hardlink the libraries it needs properly, but rather dynamically loads them from the path on startup, and they don't seem to be the ones it's built against - is that possible?
<Jarva[m]>
Has anyone used dm-integrity with nixos?
Mateon3 has joined #nixos
Chiliparrot has joined #nixos
<{^_^}>
[nixpkgs] @FRidh pushed to master « python.pkgs.nidaqmx: init at 0.5.7 »: https://git.io/Je9Yt
Mateon1 has quit [Ping timeout: 252 seconds]
Mateon3 is now known as Mateon1
cosimone has joined #nixos
viric has quit [Read error: Connection reset by peer]
<Synthetica>
Is there a way to get SCons to run under Python3 in Nixpkgs?
EdLin has joined #nixos
<{^_^}>
[nix] @edolstra pushed to master « EvalState::callFunction(): Make FunctionCallTrace use less stack space »: https://git.io/Je9Ok
Maxdamantus has quit [Ping timeout: 240 seconds]
<stammon>
Hi I need someone with more experience with node and nix to help me. I want to run a npm script in a nix package I'm building. More specifically: I try to build the vim plugin for vscodium from source. I have this repo for the vimplugin https://github.com/stammon/Vim and I want a nix expression that turns it into a built plugin. Something like https://pastebin.com/BtAFxnRj but without the npm ci in the build step since that
<stammon>
is impure and requires internet. As far as I understand, I can use node2nix to get the npm packages as nix packages, but how I can tell my derivation to use these instead of `npm ci` and then use the vsce binary from the converted nix packages to do the `npm run build-package` is unclear to me
<fresheyeball>
it's there but I just replaced it with fetchGitHub
<fresheyeball>
lets see if that works
<dooms>
im not super familiar with python packaging, but it looks like the error wants a wheel with 'none' arch, but pypi contains builds for 'manylinux' and 'macosx'
philr has quit [Ping timeout: 240 seconds]
justanotheruser has quit [Ping timeout: 240 seconds]
<fresheyeball>
me niether
<fresheyeball>
but its building now, so we will see what happens
civodul has quit [Read error: Connection reset by peer]
<pistache>
can I get access to the hash of a derivation from within Nix ?
civodul has joined #nixos
<gchristensen>
what for, pistache?
<raboof>
right now as an 'outsider' it is pretty hard to figure out what's going on, is anyone working on fixing it, can I help?
hyper_ch2 has quit [Remote host closed the connection]
<pistache>
gchristensen: I want to trigger some actions in a script, but only when a derivation's contents have changed
<pistache>
best I could find is using the store path
<pistache>
(yes, I'm doing... impure stuff)
<gchristensen>
sounds like you shouldn't :)
<pistache>
I want to recreate an ephemeral LXD container, but only when the derivation used to create its base image has changed
<pistache>
otherwise, the container is going to be regenerated at each system boot, even if the base image has not changed
<gchristensen>
probably shouldn't do this in a nix-build, but an action taken by the output of a nix-build
<pistache>
it's done in a systemd service
<gchristensen>
what does the nix-build need to do then which is impure?
<pistache>
nothing really, the impurity is in the systemd service
sigmundv has joined #nixos
<gchristensen>
ah
<pistache>
but I have to provide it with a way to know if the base image has changed
<gchristensen>
well the output of the nix-build could include a comparison
<gchristensen>
so put something like: if ! the-lxd-containers-root-is-from "$out"; then ...
<pistache>
currently I just use the base image derivation's path, save it in the container config at creation, and compare it with the current derivation's path when starting the service
<pistache>
ghuntley: yes that's basically what the systemd service does
<pistache>
I was wondering if there's something better than outPath for this
veske has quit [Quit: This computer has gone to sleep]
chimay has joined #nixos
<pistache>
but anyway, outPath - or a hash of it - should be fine
<gchristensen>
$out :)
fresheyeball has quit [Read error: No route to host]
veske has joined #nixos
turion has joined #nixos
<pistache>
gchristensen: the image is a derivation, but the systemd script is just specified as a NixOS systemd service (in config), so I can't use ".out" there (as pkg.out is pkg)
justanotheruser has joined #nixos
<pistache>
pkgs.hello.out is pkgs.hello, while pkgs.hello.outPath is a proper path
<angerman>
gchristensen: that one dies horribly on large datasets.
<angerman>
gchristensen: also, I'd still have to go back to the source. The neat thing with xcodes instruments is that it shows you the timing (and number of times that line was hit) in-line.
<turion>
lordcirth_: do I have to bump any versions, or delete any entries from configuration-hackage2nix.yaml?
<lordcirth_>
turion, what package? There should be a version explicitly defined already, that you just need to bump. Unless haskellPackages is different.
<turion>
For example, clay
<turion>
But also other packages like essence-of-live-coding, ghcide, ...
<turion>
What do I do so I don't have to do any manual steps beyond releasing on hackage in the future?
<turion>
pbogdan: Ah that's great thanks :)
<{^_^}>
[nixpkgs] @etu opened pull request #75444 → php74: Enable zip support → https://git.io/Je9cu
chimay has quit [Quit: WeeChat 2.6]
<gchristensen>
etu: would you be interested in a php.withPackages (phppkgs: with phppkgs; [ imap openssl ldap ]) <- where these are built-in extensions?
<turion>
lordcirth_: Is that the change you would expect?
<turion>
It would be great if someone who understands how broken haskell packages work, writes down something in the nixpkgs manual
sudoforge has joined #nixos
<turion>
It's nice that there is a Youtube video, but it doesn't really replace documentation
__Sander__ has quit [Quit: Konversation terminated!]
<etu>
gchristensen: sure!
Tucky has quit [Quit: WeeChat 2.7]
<lordcirth_>
turion, I would expect that you would bump the version to whatever version is fixed, remove it from broken, and test
<turion>
The version was already bumped in haskell-updates. I can build it locally, but I didn't execute hackage2nix in order to check whether that generates the correct .nix file
smatting has quit [Ping timeout: 268 seconds]
cosimone has quit [Quit: Terminated!]
<turion>
Why is it necessary to keep versions around at all? Isn't hackage & stackage enough for keeping book on which versions work? Why do we need to track them in a text file?
o1lo01ol1o has joined #nixos
drakonis has quit [Ping timeout: 268 seconds]
<o1lo01ol1o>
Are there any resources for using awslog or similar to get aws `systemd` logs out of an EC2 instance and into something that can use them (ie, cloudwatch)?
drakonis has joined #nixos
dsx has quit [Quit: dsx]
shibboleth has joined #nixos
<{^_^}>
[nixpkgs] @orivej-nixos pushed commit from @orivej to master « ifm: init at 2015-11-08 »: https://git.io/Je9cF
<gchristensen>
etu: gustavderdrache is the person I was talking about. they have made a bare bones PHP which they link together with all the plugins they need, for minimal docker images
boxscape has joined #nixos
ThatDocsLady_ has joined #nixos
civodul has quit [Quit: ERC (IRC client for Emacs 26.3)]
<gchristensen>
gustavderdrache: can you share code?
<gustavderdrache>
one sec, making a gist
<{^_^}>
[nixpkgs] @KoviRobi opened pull request #75450 → wrapGAppsHook: don't add empty variables (see also #75443) → https://git.io/Je9Cn
ThatDocsLady has quit [Remote host closed the connection]
poprox has joined #nixos
<emilsp>
has anyone here had any luck with running an android emulator on nixos lately?
<gustavderdrache>
disclaimer: 1. this isn't all of the extensions (i hadn't had time to finish the PDO/XML-based extensions, for example) and 2. i'm not sure i actually know what i'm doing with nix so please forgive the likely plentiful newbie mistakes :)
<gchristensen>
okay, intro made. in 30min I'm heading out for about 1h
<exarkun>
It uses the bare Python interpreter derivation's Python executable
<exarkun>
which is contrary to the way you're supposed to use Python from Nixpkgs
<exarkun>
instead it should be doing something like `python.withPackages (buildInputs + checkInputs)` I guess
<nilsirl[m]>
I have a systemd timer and for some reason it becomes inactive after being executed, it uses the OnCalendar setting
<elay>
Anybody got a good example for a developer environment for elixir running on NixOS 19 suitable new elixir / nixos user.
<exarkun>
nilsirl[m]: What OnCalendar setting?
<nilsirl[m]>
I think I got it wrong, I had wantedBy set for the service to run not the timer
<nilsirl[m]>
Thank you anyway
[Leary] has joined #nixos
erasmas has joined #nixos
Lears has quit [Read error: No route to host]
<etu>
gustavderdrache: Oh, interesting!
<etu>
gustavderdrache: Very nice :)
captn3m0- has quit [Ping timeout: 250 seconds]
captn3m0 has joined #nixos
<gustavderdrache>
etu: thanks :)
<gustavderdrache>
the use case i'm working to support is paring down a php to use in an FPM-based environment for drupal/wordpress sites, which don't normally need large swaths of extensions
<{^_^}>
[nixpkgs] @FRidh pushed commit from @r-ryantm to staging « vimHugeX: 8.1.2237 -> 8.1.2407 »: https://git.io/Je9Cx
<gustavderdrache>
and i figured it'd be easier to have each extension as its own item to pull in rather than rebuild the source if i wanted different ones
<gustavderdrache>
etu: i don't know much about contributing to nixpkgs (i'm a pop!_os user myself), but i'm willing to try porting this over if i can get some help from the community getting it finalized
<{^_^}>
[nixpkgs] @FRidh pushed 69 commits to staging-next: https://git.io/Je9WE
<fresheyeball>
thoughtpolice: are you around?
<{^_^}>
[nixpkgs] @FRidh opened pull request #75452 → Staging next → https://git.io/Je9Wg
<{^_^}>
[mobile-nixos] @samueldr pushed 5 commits to master: https://git.io/Je98u
viric_ has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
viric has quit [Read error: Connection reset by peer]
knupfer has quit [Quit: knupfer]
cosimone has joined #nixos
o1lo01ol1o has joined #nixos
drakonis has quit [Ping timeout: 268 seconds]
drakonis has joined #nixos
unnamed55355 has joined #nixos
justanotheruser has quit [Ping timeout: 268 seconds]
unnamed55355 has quit [Quit: Quit]
kalipso has joined #nixos
<kalipso>
hey. im firstly installing nixos on a second machine. i need git on installation but
<kalipso>
"nix-env -i git" takes forever so i used nix-shell -p git to clone my repository. now on nix-install (outside of the nix shell) git is missing. can i call nix-install inside a nix-shell?
<cransom>
nix-env -iA nixos.git will be far faster.
<zeta_0>
vaibhavsagar: last time we chatted you said you were only supporting nixos-stable, anyways, i switched to the nix-stable(19.09) channel and i was able to install ihaskell but ihaskell is not working in emacs anymore, when running ein(emacs ipython notebook) i am getting this error: kernelspec haskell not supported, i have no idea what is wrong?
<turion1>
I have two machines, both identical configurations and identical home manager configurations. I have KDE on both, and installed an unusual keyboard layout (neo2). On both, I've declaratively installed vscodium through home manager. On one machine, everything works fine. On the other one, vscodium interprets all keyboard _shortcuts_ (like Ctrl + V for paste) with the QWERTZ layout. Wat.
levdub has joined #nixos
levdub has quit [Client Quit]
<infinisil>
I guess the application preferences can differ
kleisli has joined #nixos
chloekek has quit [Ping timeout: 245 seconds]
<o1lo01ol1o>
Dec 10 20:48:02 ec2-3-84-24-245.compute-1.amazonaws.com kernel: refused connection: IN=eth0 OUT= ... <- this shows up in journald when I direct a browser to where my website should be. What is it?
<clever>
o1lo01ol1o: the firewall, networking.firewall.allowedTCPPorts i think its spelled
<o1lo01ol1o>
clever: [ 80 443 ]
<clever>
o1lo01ol1o: what is the full message from the logs?
<clever>
o1lo01ol1o: have you rebuild since allowing 443?
<o1lo01ol1o>
clever: the system is provisioned useing copy-closure and installing the derivation to the system profile. after which we run: /nix/var/nix/profiles/system/bin/switch-to-configuration switch
domogled has quit [Ping timeout: 252 seconds]
<clever>
o1lo01ol1o: nix-env --set or -i?
<o1lo01ol1o>
clever: nix-env --set
<turion1>
infinisil: Apparently it does some automatic configuration on startup...
<clever>
o1lo01ol1o: and your sure you have the latest version deployed?
<clever>
o1lo01ol1o: does the /nix/store/75bb8q3idi8swamrss36cy2z5wvkc8ym-firewall-start/bin/firewall-start file open port 443?
selfsymmetric-pa has joined #nixos
<o1lo01ol1o>
clever: interesting, it does not
<clever>
o1lo01ol1o: compare the path of your latest build with `ls -l /run/current-system`, do they match?
<o1lo01ol1o>
clever: yes, they do
<clever>
o1lo01ol1o: does `result/etc/systemd/system/firewall.service` open 443?
<o1lo01ol1o>
clever: result/etc/ has no systemd entry
<clever>
o1lo01ol1o: is result from the latest build of that machine?
<clever>
o1lo01ol1o: try just building it again with nix-build and check its path
<o1lo01ol1o>
clever: Ah, ok, one sec
cosimone has quit [Remote host closed the connection]
<o1lo01ol1o>
clever: unfortuneatly I can't build the derivation locally as I'm on darwin.
<o1lo01ol1o>
clever: the above logs are from our gitlab runner
<etu>
gustavderdrache: I'm willing to do the work for that
<clever>
o1lo01ol1o: but you could still evaluate it and compute the $out, using nix-instantiate
<etu>
gustavderdrache: I do quite a lot of things with php in nixpkgs
cosimone has joined #nixos
<gustavderdrache>
etu: now i know who'll get sick of my questions as i play with nixpkgs more :)
jgt2 has quit [Ping timeout: 252 seconds]
<gustavderdrache>
etu: also i have already completely forgotten how i phrased my reply - which one of us is getting started on this?
<o1lo01ol1o>
clever: I don't have access to some of the configuration variables used in the deployment so I suspect the derivation will differ.
<clever>
o1lo01ol1o: ah yeah, dang
leotaku has quit [Ping timeout: 265 seconds]
<clever>
o1lo01ol1o: it sounds like its just running the wrong build
<etu>
gustavderdrache: I can do the work, no problem. But I won't have time until the weekend :)
mexisme has joined #nixos
<gustavderdrache>
etu: i can most certainly wait that long
<gustavderdrache>
is there anything more beyond the github gist that you'd need to get started?
leotaku has joined #nixos
<etu>
gustavderdrache: Free time :D
EdLin has joined #nixos
<gustavderdrache>
just hop in my delorean :D
<o1lo01ol1o>
clever: There's step here that's suspect: the build derivation step generates more than one derivation, the first is the /nix/store/...nixos-system-ec2-3-84-24-245.compute-1.amazonaws.com-18.09pre-git, the other two are named, if memory serves, nixos-vm. We're only `nix-env -p --set`ing the nixos-system derivation.
<clever>
o1lo01ol1o: what args do you pass to nix-build?
<o1lo01ol1o>
clever: build_path="$(nix-build deploy-server.nix --no-out-link -A server --argstr aBunchOfArgs)"
poprox is now known as syzygyzer
<clever>
o1lo01ol1o: and what is server set to in the expr?
<o1lo01ol1o>
clever: server is a base ec2 image and several services (acme, nginx, and others)
<clever>
o1lo01ol1o: what is that expr doing? does it return a single derivation or a set?
<clever>
o1lo01ol1o: if you eval it in nix repl, what do you see?
<o1lo01ol1o>
clever: well, the output of nix-instantiate deploy-server -A server is 3 derivations
<{^_^}>
[nixpkgs] @mmilata opened pull request #75459 → nixosTests.haproxy: port test to python → https://git.io/Je9R5
<clever>
o1lo01ol1o: that would likely be part of your problem, try -A server.system ?
<o1lo01ol1o>
clever: nix-instantiate deploy-server -A server.system is one derivation: [...] amazonaws.com-18.09pre-git.drv
<clever>
o1lo01ol1o: youll need to modify deploy-server.nix, and add a .system at the right spot, then see what happens when you deploy
lovesegfault has joined #nixos
lovesegfault has quit [Client Quit]
<o1lo01ol1o>
clever: could part of the issue be that the system is using a pinned nixpkgs 18.09 and the deployment target and build machines are 19.09?
<raboof>
is there something like `pkgs.writeTextDir` that rather than writing text to a new file in a directory, copies an already-existing file to that directory?
EdLin has quit [Quit: Going offline, see ya! (www.adiirc.com)]
<o1lo01ol1o>
clever: looks like redeploy didn't get the correct derivations: Dec 10 21:56:08 ec2-3-84-24-245.compute-1.amazonaws.com kernel: refused connection:
<lovesegfault>
worldofpeace: Around?
davidcl has quit [Quit: Ping timeout (120 seconds)]
<fresheyeball>
gustavderdrache: as far as I can tell clever is the only person who really understands this nix thing
<gustavderdrache>
luckily for me, the bar is really low so pretty much everybody looks like they understand nix more than i do :)
octe has quit [Ping timeout: 246 seconds]
werner292 has joined #nixos
mexisme has quit [Ping timeout: 245 seconds]
scasc has quit [Ping timeout: 276 seconds]
werner291 has quit [Ping timeout: 268 seconds]
werner292 is now known as werner291
octe has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
viric has joined #nixos
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
knupfer has quit [Ping timeout: 250 seconds]
Gilfoyle- has quit [Quit: WeeChat info:version]
Olgierd has quit [Ping timeout: 276 seconds]
Church- has joined #nixos
viric_ has quit [Ping timeout: 276 seconds]
magthe has joined #nixos
jgt2 has joined #nixos
<magthe>
Is there some variable that holds the "current ghc", so I don't have to hardcode a specific version when I do `pkgs.haskell.packages.gch865.override`, i.e. can I replace that `ghc865`?
<clever>
betawaffle: not-os uses a heavily stripped down nixos config, and generates a ~40mb squashfs for the nix store
<clever>
betawaffle: haskell-init generates a single 2mb executable, which runs as init in the initrd, thats it
<betawaffle>
how long until we build a single-binary userspace out of nix-generated rust code?
<clever>
betawaffle: have fun with upgrading! lol
<clever>
chromium alone is a 200mb executable
<betawaffle>
i don't mean for running chromium...
<clever>
ah yeah, rewriting in rust would be tricky there, lol
<clever>
haskell makes it much simpler to just jam 2 unrelated programs into one binary
<betawaffle>
interesting
eacameron has quit [Quit: Connection closed for inactivity]
mexisme has joined #nixos
<fuzen>
I'm trying to install Cryptomator, I'm not sure how I should go about it. I checked the AUR to see how it's done there, but they are using AppImages, which doesn't seem to work on NixOS
<fuzen>
How should I go about creating a package for this?
<fresheyeball>
well I am making progress toward torchvision on nix
<clever>
fuzen: id unpack the image, and use patchelf
<fresheyeball>
by setting doCheck = false; and pulling the src from github it builds
<fresheyeball>
but not with cuda support
<fresheyeball>
I set an env var to make to build with cuda support and added cudatoolkit and linuxPackages.nvidia_x11 tot he nativeBuildInputs
<fresheyeball>
and now it fails like this
<fresheyeball>
fatal error: c10/cuda/impl/cuda_cmake_macros.h: No such file or director
<clever>
fresheyeball: is that directly in the $out/include of cuda?
<fresheyeball>
clever: I don't know
<clever>
fresheyeball: use nix-build to find the $out of cuda, and then ls it
<fresheyeball>
clever: ok where I am looking in here?
<fresheyeball>
there is a bunch in this `result`
<clever>
fresheyeball: ls -l result/include/
<o1lo01ol1o>
clever: sorry, I had to go into a meeting. The hash generated by CD appears to be the same as the hash that is installed, however, neither [...]wvkc8ym-firewall-start/bin/firewall-start nor result/etc/systemd/system/firewall.service seems to have any reference to port 443
<fresheyeball>
clever: that pooped out a ton
<clever>
o1lo01ol1o: sounds like gitlab isnt building the right thing