<{^_^}>
[nixpkgs] @xeji pushed commit from @dtzWill to master « fwupdate: 10 -> 12 (#44710) »: https://git.io/fAmnO
<lejonet>
For me it was actually nixops that caught me :P
<genesis>
i was fed up with conflict dependancies
<genesis>
upgrade become impossible and no binaries, and portage become a mess
<lejonet>
Know what you mean
<genesis>
and i won't tell about packaging stuff and receive an answer 10 years ago
<lejonet>
Especially if you're doing weird stuff like I did, with investigating weird binaries with odd library dependencies and such
<genesis>
yes, i'd like that, i package mbrola , dead since 2005, for nixos
<genesis>
and look at that
<genesis>
nix-build nixops, nix-locate, and much tools we have
<genesis>
how effective is it to write stuff.
<genesis>
people wait so long to have their stuff include that every gentoo power user get his own overlay
<genesis>
and there is website like http://gpo.zugaina.org/ where we searched desasperatly for an ebuild
<lejonet>
Yeah and whole layman and stuff
<genesis>
get stuck with their svn for too long...
<lejonet>
hehe :P
<infinisil>
Is there a reason one would prefer abort over throw? Profpatsch maybe you know
<genesis>
my friend we get on gentoo 15 years ago switch for arch after try nixos no enough :)
slack1256 has joined #nixos
<samueldr>
infinisil: throw s
<samueldr>
Throw an error message s. This usually aborts Nix expression evaluation, but in nix-env -qa and other commands that try to evaluate a set of derivations to get information about those derivations, a derivation that throws an error is silently skipped (which is not the case for abort).
<callipygous>
but I'm getting some enviroment variable has been set twice
<callipygous>
to do with kde
<Ashy>
actually i guess according to kernel.org what i really want is a linuxPackages_mainline package
<samueldr>
gchristensen: generally it's backported since the penultimate release (in this case 4.17) gets dropped before the next nixos release updates
<gchristensen>
ah
<samueldr>
(you don't want _latest which referred to 4.16 brought back to 4.14, the LTS)
<samueldr>
(and then 4.16 isn't maintained)
<samueldr>
I know 4.17 was hung up by the plumbing changes (did the backport at the time)
<callipygous>
what am I supposed to do? I have checked the nix config file, and it has no instance of kde/plasma
<callipygous>
but the virtualbox image has plasma installed
<{^_^}>
[nixpkgs] @aanderse opened pull request #45638 → incron: init at 0.5.12 → https://git.io/fAmCG
<samueldr>
*puts psychic hat on* callipygous those come from conflicting options from the imported files in imports
<samueldr>
(your last sentence about the virtualbox image I think is the key here)
<callipygous>
files in imports?
ris has quit [Ping timeout: 252 seconds]
<samueldr>
let me check how the file looks like in the VM
Piece_Maker has joined #nixos
<callipygous>
I'll get you the exact error, 1 sec
<callipygous>
error: The unique option `environment.variables.GDK_PIXBUF_MODULE_FILE' is defined multiple times, in `/nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs/nixos/modules/services/x11/desktop-managers/plasma5.nix' and `/nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs/nixos/modules/services/x11/desktop-managers/xfce.nix'.
aanderse has quit []
<callipygous>
(use '--show-trace' to show detailed location information)
<callipygous>
all it had in it when I edited it was :
<callipygous>
error: The unique option `environment.variables.GDK_PIXBUF_MODULE_FILE' is defined multiple times, in `/nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs/nixos/modules/services/x11/desktop-managers/plasma5.nix' and `/nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs/nixos/modules/services/x11/desktop-managers/xfce.nix'.
Acou_Bass has quit [Ping timeout: 252 seconds]
Piece_Maker is now known as Acou_Bass
<callipygous>
(use '--show-trace' to show detailed location information)
<callipygous>
I'm sorry for that paste
<callipygous>
accident
<samueldr>
oh, then I missed something, when you said "the virtualbox image" were you talking about the nixos ova?
<callipygous>
Where do I find nixpkgs/nixos/modules/installer/virtualbox-demo.nix?
<samueldr>
you could realistically, instead, add `services.xserver.displayManager.plasma5.enable = lib.mkForce false` and make sure at the top of configuration.nix `lib` is in the parameters (something like {pkgs, lib, ...})
<samueldr>
callipygous: when you see a path within angled brackets (<>), nis will use $NIX_PATH to figure out where to find those... in that situation, it will find a local copy of the nixpkgs repo that was either built-in the ova or downloaded from the channels
<samueldr>
(cannot say fore sure, depends on whether you updated the channels)
<samueldr>
the bot here has the NIX_PATH pointing to a nixpkgs checkout in its state folder (just for explaining)
<callipygous>
ok
<callipygous>
I could copy some of the stuff over from that file, and delete the import?
<callipygous>
import reference that is
<samueldr>
yes you could
seanparsons has quit [Read error: Connection reset by peer]
<samueldr>
everything in the `imports` attribute will be recursively merged when building the system
seanparsons has joined #nixos
<samueldr>
(so imports in imports work)
Supersonic112 has joined #nixos
Supersonic has quit [Disconnected by services]
Supersonic112 is now known as Supersonic
<callipygous>
3those paths don't exist
<callipygous>
I'll look harder
<samueldr>
you won't find it under /var/ in your system, that's really specific to how the bot works
<samueldr>
you could and probably should use the github repo to find the contents in this particular instance
<samueldr>
oh uh
* samueldr
thinks
<samueldr>
`nix-instantiate --eval -E '<nixpkgs>'` would show the location of the files
<samueldr>
(I would bet it'll be in the store)
<samueldr>
oh, and callipygous, don't hesitate to ask more questions if you have doubts
pxc has quit [Ping timeout: 252 seconds]
<callipygous>
there are a few instances of the file in /nix/store
<samueldr>
yeah, that's entirely possible and probably normal; they would be the different generations created by nix-channel I guess
<callipygous>
weird
<callipygous>
I did a diff on them, they're all the same
pxc has joined #nixos
<callipygous>
it has a bunch of imports, too, just like I think you said
<samueldr>
the file isn't one that change often, but the store paths /nix/store/.../ are all different (git) revisions of the whole nixpkgs tree... so even if only one file changes, you will see a whole copy of the tree
Acou_Bass has quit [Ping timeout: 252 seconds]
<infinisil>
samueldr: `nix-instantiate --find-file nixpkgs` works too btw
<samueldr>
thanks infinisil
<samueldr>
(haven't really used it in the past)
Acou_Bass has joined #nixos
<callipygous>
hmm, I guess there isn't an easier way to disable plasma?
<samueldr>
[21:25:25] <samueldr> you could realistically, instead, add `services.xserver.displayManager.plasma5.enable = lib.mkForce false` and make sure at the top of configuration.nix `lib` is in the parameters (something like {pkgs, lib, ...})
<samueldr>
(added the semi-colon which would be required)
<samueldr>
lib.mkForce makes conflicting options "more important"; nixos options only fail when they are of the same importance
<callipygous>
ok :) thank you
<samueldr>
with hindsight, I would have built the internal profiles from nixpkgs with just a bit lesser importance, allowing more natural overriding for users using the ova
<samueldr>
(I did not build any of that)
pxc has quit [Quit: WeeChat 2.1]
<callipygous>
error: The option `services.xserver.displayManager.plasma' defined in `/etc/nixos/configuration.nix' does not exist.
<samueldr>
I may have mistyped a part
<samueldr>
oh, or you, plamsa_5_
<samueldr>
oh, even worse
<ldlework>
I have these three Nix files, https://gist.github.com/dustinlacewell/4aa025bfb301d433684f6ccf16af073b and yet when I try to build home-manager with them, i do not get a "recap" in a bin path. I can however "nix-env -i -f default.nix" and it gets installed properly. What am I doing wrong?
Dedalo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Travankor has joined #nixos
Travankor has quit [Client Quit]
dmj` has quit [Quit: Connection closed for inactivity]
mzan has joined #nixos
balsoft has joined #nixos
goibhniu has joined #nixos
<emily>
hi, I can't "nix run nixpkgs.purescript" because the spdx build fails, saying it requires base <4.11 (i.e. GHC <8.4); is this expected, and how can I override the GHC version so that it works out okay?
<srhb>
emily: If you grab it directly from the haskell package set, you can select the version easily
<emily>
I guess the only remaining question is how expected this is, then
<srhb>
How expected that things like this happens?
<emily>
right; especially since I'm just trying to use purescript as an end user rather than as a Haskell library per se, I was surprised to find the build failing on a random dependency due to the GHC version (that I didn't even pick)
<srhb>
I only just realized we have bumped to hc 8.4.3, and I imagine _many_ packages are not compatible yet, so I think it's pretty expected.
<srhb>
So, the top level attribute is an alias for haskellPackages.purescript
<emily>
I guess my assumption was that nixpkgs.purescript would be defined to use the correct GHC version
<emily>
*nods*
<srhb>
haskellPackages in turn is an alias for haskell.packages.ghcxyy
<srhb>
It's been... Many months since the default was bumped here, and I imagine it caused similar breakage
<srhb>
So I guess the answer is: Pretty expected during major ghc bumps :)
<srhb>
I think having even top level aliases point to specific ghc versions is very costly, since it might cause hydra to have to build a lot of hackage for that ghc version alone
<srhb>
Losing the sharing when using the same ghc version across the board
* emily
nods
<emily>
makes sense
<emily>
I mean, I was surprised it wanted to build spdx at all, it was fine with binaries for most of the dependencies
<srhb>
Right.. :)
<emily>
I suppose it wanted to exactly because it fails on hydra
<srhb>
Yeah.
simukis has joined #nixos
<emily>
I see, the tradeoff of using a package that works is that I have to build everything T_T
<srhb>
Yep. :P
<srhb>
emily: you could also use the version from nixos-18.03
<srhb>
We're entering the last sprint of unstable -> 18.09 so it might be a bit messy for a little while :-)
<srhb>
nix run -f channel:nixos-18.03 purescript
<emily>
cool, thanks!
<emily>
I'm fine dealing with mess -- Nix makes it so easy after all
<{^_^}>
[nixpkgs] @FRidh pushed to master « docs python: `python.buildEnv` does not include `buildPythonApplication` modules »: https://git.io/fAmEN
<srhb>
ixxie: ah :)
<ldlework>
Anyone here handy with packaging Python applications?
<srhb>
ldlework: I've done a few, not a master of it.
<ldlework>
I'm a bit confused about how to include and access data files.
<ldlework>
Like an example configuration file
<srhb>
Stick it in share somewhere?
<ldlework>
I have data_files=[('data', ['data/config.toml'])],
<ldlework>
in my setup.py
<ldlework>
In my actual application I have a command to copy the example to the user's XDG_CONFIG_HOME but I don't know how to find the path to the file.
<srhb>
Either relative to the executable or with an absolute path from $out ?
<ldlework>
there's no $out in python
<srhb>
No, but in your nix expression there is, and you can use that as you like :)
garbas has joined #nixos
<ldlework>
I want the config file to work as normal if the package is being used by a non-Nix user
smolboye has quit [Read error: Connection reset by peer]
<emily>
srhb: thank you for all the help! I'll try and get that working tomorrow
<srhb>
emily: See you :)
smolboye has joined #nixos
balsoft has quit [Ping timeout: 244 seconds]
<ixxie>
srhb, tilpner - any tips on how to deduce the ipv6 gateway of a random VM from the inside?
<srhb>
ixxie: Um, why do you need it? :)
<ldlework>
srhb: FileNotFoundError: [Errno 2] No such file or directory: '/nix/store/nrj8vf0s0qc0c5s6l0vhk1k6pf2x8ggq-recap/lib/python3.6/site-packages/recap/data/config.toml'
<srhb>
ldlework: Is it there though?
<ldlework>
no why would it
<srhb>
I assumed you copied it there during installPhase or something?
<ldlework>
the file is in the $out of the recap package
<ldlework>
installPhase has nothing to do with it!
<ldlework>
the python package must work on non-nix OS lol
<srhb>
Okay, I guess I'm misunderstanding your question then.
<ldlework>
I have a python package, which has an entrypoint, IE, a bin script
thc202 has joined #nixos
<ldlework>
It has a command, such that I would like to copy a text file from *somewhere* to XDG_CONFIG_HOME/recap
<ldlework>
ls -la ./result/data ~/src/recap
<ldlework>
.r--r--r-- 372 root 31 Dec 1969 config.toml
<ldlework>
the python setup.py is putting config.toml into $out
<Aleksejs>
srhb: hi. Sorry, I just woke up and read your comment. I've checked your commits, there seems to be a small error in python-dotenv package. You use licenses.bsd, which is not in licenses file. It shows error "attribute bsd missing" during install
<srhb>
ldlework: And what's the question?
<srhb>
Aleksejs: Ah, thanks.
<ldlework>
srhb: how do I retrieve the file from within my python code
<ldlework>
such that it will find it on a nix and non-nix system
<ldlework>
i understand if I have to "patch" the file in the nix
<ldlework>
so that when the python app does the copy, it looks in the right place
<srhb>
ldlework: I think the right answer is the same: You must know the data file location relative to the module beforehand, and it seems that data_files should ensure that
<srhb>
ldlework: And then the path to the datafile is just relative to __file__ in the module
<srhb>
On Nix and non-Nix
<ldlework>
it isn't relative to the module lol
<srhb>
Why isn't it?
<ldlework>
maybe you're thinking of package_data?
<ldlework>
where the data files literally end up inside the package/ ?
<ldlework>
data_files install things relative to sys.prefix but nix that isn't actually the case, the files end up in the python library's nix package
<ldlework>
in $out
<ldlework>
and absolute paths don't work at all
<ldlework>
like if you try to say this data file should be installed into /usr/share or /etc or something
<ldlework>
it just doesn't end up anywhere from what I can tell
<ldlework>
so you use a relative path and it ends up in the library's $out
<ldlework>
the python package, the code files, end up in the python interpreter's package
<srhb>
ldlework: Perhaps a small example package would help, I may be reading the python documentation wrong.
<ldlework>
so if you use __file__ you're inside the python package..
<srhb>
And you may be right that it's package_data and pkgutil.get_data instead
<ldlework>
it builds for me but not you or infinisil
<ldlework>
i've pinned nixpkgs
<ldlework>
what's with that?
<srhb>
ldlework: Do you have sandboxing turned off?
<srhb>
nixos-option nix.useSandbox
mayhewluke has quit [Ping timeout: 272 seconds]
<ldlework>
srhb: that just exit without output
<srhb>
ldlework: You're on NixOS, right?
<srhb>
ldlework: Anyway, I suspect you have it off and that your build (or the build of a dependency) is impure and using the network, breaking it for sandboxed builds.
mayhewluke has joined #nixos
<ldlework>
srhb: yeah
<ldlework>
i am on nixos
<srhb>
set nix.useSandbox = true
<ldlework>
this is going to like rebuild my whole system isn't it
<srhb>
Nope.
<ldlework>
oh that was fast
<srhb>
It only affects new builds.
<ldlework>
well uh
<srhb>
And it really ought to be on by default.
<ldlework>
the package still built for me
<srhb>
That's probably because you already built a package that should fail
<ldlework>
srhb: I got it to work with package_data
jpab has joined #nixos
<srhb>
Great :)
<srhb>
Aleksejs: Fixed. Feel free to upstream if you're interested in maintaining them. :)
<Aleksejs>
srhb: thanks!
ThatDocsLady has quit [Quit: Leaving]
<srhb>
Why is it that sometimes bsd-3 clause is written with bullet points, and sometimes with numbered bullets >>
<Aleksejs>
ampy itself works now but it's commands don't work. I'm trying to figure out if it's a problem with ampy itself or with esp32 firmware
<srhb>
Aleksejs: Well, I doubt I have the right tools to really test it.
<srhb>
But I figured this would be a starting point. :)
<Aleksejs>
yeah, I will test it today
Ariakenom has joined #nixos
logzet has joined #nixos
revtintin has quit [Quit: WeeChat 1.9.1]
orivej has quit [Ping timeout: 244 seconds]
hellrazor has joined #nixos
alex`` has joined #nixos
<Aleksejs>
srhb: I've figured out where the problem was - it was with my setup. ampy works fine
Asmadeus has quit [Ping timeout: 240 seconds]
Asmadeus has joined #nixos
orzo has quit [Ping timeout: 268 seconds]
<srhb>
Aleksejs: Nice :)
das_j has joined #nixos
<das_j>
Hey, little question about nix 2. I'm currently deploying my Debian infrastructure with Ansible, which means I have a default.nix that returns a derivation with all required packages (Ansible and some python libs) as build inputs. When launching nix-shell in this directory, I get a new shell with all required packages. When using `nix run` however, it complains that there is no valid `src` or `srcs`. What do I
<das_j>
have to do differently here?
jtojnar has quit [Ping timeout: 252 seconds]
<das_j>
Setting unpackPhase to "" doesn't work either
jasongrossman has left #nixos ["ERC (IRC client for Emacs 26.1)"]
<Thra11>
If I find a mistake in my pull request after it gets merged, should I create a new pull request to fix it, or can I add new commits to the original pull request?
<LnL>
you can't reopen it if it was merged, but you can create a new one with the same branch
<inquisitiv3>
Does somebody know if there is a finished derivation for Haskell IDE Engine in Nix? I can't find it when I search, bit I got a lot of matches so it's possible that I missed it.
<Enzime>
how are the dependencies meant to be specified
<infinisil>
Enzime: add the argument to the list at the top and to propagatedBuildInputs
<Enzime>
infinisil: what's the difference between propagatedBuildInputs and buildInputs?
<infinisil>
No time or will to explain right now, python just needs it
<inquisitiv3>
It seems that I can't sign in to Github using my Yubikey on NixOS. When I insert the key the error message "Something went really wrong". I'm using Chromium, and there isn't any problems with an similarly configured Chromium instance on my Fedora machine. I've checked open issues on Github in the Nixpkgs repo, but it seems that there isn't any reported issues with it.
<Enzime>
alrighty
<infinisil>
I hope to eventually find (or write) a good and complete explanation of propagatedBuildInputs myself
<bart->
hi
<inquisitiv3>
Does it make sense to try to get HIE running in a nix-shell environment, or do I only make it more complicated than what it needs to be?
<inquisitiv3>
infinisil: Are you saying that it *should* be installed in a nix-shell environment?
<infinisil>
inquisitiv3: I'm doing it at least, to have the right ghc with the nix packages the projects needs
<{^_^}>
[nixpkgs] @jluttine opened pull request #45644 → pythonPackages.rlp: 1.0.1 -> 1.0.2 (with some new deps) → https://git.io/fAmV7
bruno2 has joined #nixos
<inquisitiv3>
infinisil: So I've installed Cachix and authenticated my machine in the regular environment. Should I navigate to the project directory, start a nix-shell with `nix-shell -p ghc` (I only need a very basic environment) and when run 1) `cachix use hie-nix` 2) `nix-env -iA hies -f https://github.com/domenkozar/hie-nix/tarball/master` and lastly 3) `hie-8.2 --help`?
<inquisitiv3>
I'm just guessing right now. Don't have any directly relevant knowledge how to set this up.
<infinisil>
Nah, installation can't be per nix-shell
<inquisitiv3>
When how should I do?
<infinisil>
1) then 2)
<infinisil>
Then get some working cabal on PATH (this can be through installing cabal-install or through nix-shell)
<inquisitiv3>
Doesn't that install a specific version of HIE system-wide? Or is HIE built against several different versions of GHC installed at the same time?
<inquisitiv3>
Little confused :S
<infinisil>
hies contains all versions (8.0, 8.,2 8.4) as binaries hie-8.0, hie-8.2, hie-8.4
<infinisil>
While hie-nix's hie80 attribute only contains version 8.0, as a binary "hie", same for the others
philippD has joined #nixos
<inquisitiv3>
Aha, thanks!
<infinisil>
You need the correct hie for your ghc version
<Mic92>
something that has a man page would be nice
<infinisil>
inquisitiv3: Unfortunately this update can't be used yet, but if it eventually can, there will be a hie-wrapper binary to automatically select the correct binary
<infinisil>
Mic92: you mean because it got removed?
<Mic92>
infinisil: yes.
<monotux>
so I copied this file (https://goo.gl/kkPLpF) and imported it into my system wide configuration.nix, and setup a backup job. I have some settings in /root/.ssh/config that has to be used for the backup job to work, but it's not being applied which makes the backup job fail. how do I make sure that my root ssh config is applied when running the serv
<monotux>
ice?
<inquisitiv3>
infinisil: You say that I should get a working Cabal installation, but you mention Stack previously. Should I still get Cabal? I only need GHC as of now (just beginning to learn Haskell).
<inquisitiv3>
Mic92: `nix-repl` has been replaced with `nix repl`.
<Mic92>
inquisitiv3: I know. I want to update the tutorial therefore
FRidh has quit [Quit: Konversation terminated!]
<infinisil>
inquisitiv3: hie needs either stack or cabal to work. It will be easier to use stack
<inquisitiv3>
infinisil: Sorry if I'm slow to understand, but do you mean that I shouldn't use a different development environment? That I should simply install stack in my regular environment with `nix-env`?
<infinisil>
Yeah
<inquisitiv3>
Why with `nix-env` and not system-wide?
<infinisil>
Stack itself handles different dev environments, the stack version itself doesn't have any influence on what
<infinisil>
that*
<infinisil>
inquisitiv3: ah yeah you can do that too
<inquisitiv3>
So you mean that I should use Stack's environments instead of nix-shell?
<inquisitiv3>
Yeah, that's is obvious even for me...
<infinisil>
Just install stack globally
<infinisil>
`stack build` automatically enters a nix-shell with the correct things
<infinisil>
And hie used that command to build your project
<inquisitiv3>
So now I've installed Stack globally.
mayhewluke has quit [Ping timeout: 272 seconds]
mayhewluke has joined #nixos
Mateon3 has joined #nixos
aarvar has quit [Ping timeout: 272 seconds]
Mateon1 has quit [Ping timeout: 252 seconds]
Mateon3 is now known as Mateon1
<inquisitiv3>
So now I only need to navigate to the parent directory for directory containing all the *.hs files I've and 1) run `stack new [...]` and 2) run `stack build`?
ixxie has joined #nixos
rardiol has joined #nixos
das_j has joined #nixos
<infinisil>
Something like that, stack new doesn't work with an existing dir though
<infinisil>
inquisitiv3: i'd just do a new `stack new' and copy the files to it, look up online how files have to be layed out and how the .cabal file works
<inquisitiv3>
Why did you describe giving `stack new` two arguments? I'm looking trough stack's documentation and it only seems to take one argument?
<das_j>
Hm, is there any way to add my own options to the generated option list? (options.json)
<infinisil>
inquisitiv3: maybe ask in #haskell instead
<das_j>
I essentially need to add my own modules to the `nix-build --no-out-link '<nixpkgs/nixos/release.nix>' -I nixpkgs=$(NIXPKGS) -A options` command
<inquisitiv3>
infinisil: True.
<inquisitiv3>
There isn't anything left for me to do to get HIE running on my machine? When I run `stack build` the correct version of HIE will start?
orivej has quit [Ping timeout: 252 seconds]
orivej has joined #nixos
aanderse has quit [Ping timeout: 276 seconds]
<infinisil>
inquisitiv3: stack won't do anything for the correct hie version, it doesn't add any hie itself
<ixxie>
anybody ever considering adding `nixops install` command that uses kexec to install NixOS on some cloud VM?
<infinisil>
inquisitiv3: in fact it's the other way around: hie starts stack. But you need to select the correct hie version yourself (until that hie-wrapper is there)
<infinisil>
The stack.yaml file shows the stackage lts version, which determines the ghc version and therefore the hie version you need
<infinisil>
Right now the default should be 8.4
<tilpner>
ixxie - I don't really see how that could be done well without any user config or a ton of templates
b has joined #nixos
<mudri[m]>
Often when I run `nix-collect-garbage --delete-older-than 14d`, a lot of symlinks go missing from /var/run/current-system/sw/bin, and I have to reinstall everything via nix-env. Why could this be happening?
<inquisitiv3>
infinisil: Aha. So I should run `hie-8.4` to get hie and stack running?
<ixxie>
tilpner: it would be implemented on a per-backend basis I figured
<infinisil>
inquisitiv3: yea
<tilpner>
ixxie - Maybe if you limited support to VMs with a single disk, that could work
<ixxie>
tilpner: that is what I mean, a way to produce pretty basic NixOS machines
<inquisitiv3>
infinisil: Do I have to think about which version of GHC Nix supplies?
<infinisil>
inquisitiv3: nope, it's selected by stack (but provided by nix), and nix provides all common ghc version (and ones hie can use)
<ixxie>
tilpner: this worked for Hetzner, it doesn't include the steps of creating the machine or building Clever's kexec yet, but it could easily be added
<bitonic>
and i'm not entirely sure how to debug it
<tilpner>
infinisil - That's not any shorter though
<tilpner>
(Not that it matters)
<mudri[m]>
bitonic: I'm sure I have, but I can't remember why.
<ixxie>
tilpner: neat, do you have that in a repo somewhere?
<infinisil>
tilpner: yea
<bitonic>
mudri[m]: do you have a clue on how to debug this?
<mudri[m]>
bitonic: What was the command?
<jluttine>
i'm packaging a python package that has a dot in its name: backports.shutil-which. how should i name it in nixpkgs?
<inquisitiv3>
infinisil: You said that I need to start HIE and Stack by running `hie-8.*` manually in the command line until the `hie-wrapper` script is finished and available. Do I still need to following instructions to get it working with VS Code?
<bitonic>
otherwise i guess i'm going to start having to strace this :|
<infinisil>
inquisitiv3: you don't if you don't intend to use different ghc versions
<bitonic>
infinisil: that also doesn't work, but the error is a bit different
<infinisil>
inquisitiv3: just tell vscode to use the "hie-8.4" binary (or only install hie-nix' hie84 attribue which will then make you have the standard "hie" binary which is the version 8.4)
<ixxie>
tilpner: in the end its the same but shifts the code into Nix, which is better of course. But my point stands I think: this could be a handy util to nixops, especially if wrapped in cloud provider specific APIs/CLIs to also create the machine and pass the IP to it.
<tilpner>
ixxie - Maybe. I don't use nixops. IIRC there was interest in removing provider-specific stuff from nixops and replacing it with terraform
<infinisil>
bitonic: not entirely sure, both aren't working for me i think. I actually never got ssh substituters to work with nix 2.0 :/ maybe check if there's an issue for it. sphalerit also had the same broken terminal problem at some point i think
<inquisitiv3>
infinisil: Sorry if I'm very slow and don't understand, but what do you mean with the text in the parenthesis?
<bitonic>
infinisil: that is quite sucky
<bitonic>
thanks for the info, i'll google
<infinisil>
inquisitiv3: the hies attribute installs all hie versions in binaries called hie-8.*, but if you only use 8.4, you can install the hie84 attribute directly, which will then give you the "hie" binary. By default editors look for a "hie" binary, so probably it will just work out of the box
<ixxie>
tilpner: interesting, do you know who to ask about this?
rprije has quit [Ping timeout: 264 seconds]
<tilpner>
ixxie - I remember gchristensen talking about it, don't know about actual activity/progress though
<gchristensen>
I already do it, but it is a bit janky
<inquisitiv3>
infinisil: I installed hies according the instructions on https://github.com/domenkozar/hie-nix. I've `hie-8.0`, `hie-8.2` and `hie-8.4` available in my path. Should I have done it any other way?
<ixxie>
gchristensen: so what does terraform do exactly?
<gchristensen>
creates the servers and writes a .nix file
<gchristensen>
ixxie: but my recommendation is don't bother with converting to NixOS and instead use/support providers which already support NixOS in a nice way.
<gchristensen>
hetzner cloud allows uploading a disk image iirc, aws already has nixos, packet.net... I'm happy to work with other providers to get them supporting NixOS, too
<inquisitiv3>
infinisil: Is it the second part?
<inquisitiv3>
"Or for just a specific GHC and get hie:" and below
<ixxie>
gchristensen: oh cool, didn't know that about Hetzner
* gchristensen
looks it up
<ixxie>
gchristensen: I recently switched to Hetzner so
<inquisitiv3>
infinisil: Thanks for all the help! I wouldn't be able to do that myself. I really appreciate you taking your time!
<gchristensen>
you can do it!
hamishmack has quit [Ping timeout: 268 seconds]
<tilpner>
gchristensen - borg build ^ ?
<inquisitiv3>
Why should it be so complicated to only get a language extension to work on a machine? :(
silver has joined #nixos
<infinisil>
inquisitiv3: ;)
<genesis>
infinisil : is it oki for xkbd finally ?
<tilpner>
gchristensen++
<{^_^}>
gchristensen's karma got increased to 23
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fAmKX
<infinisil>
genesis still no sorry. Why do you keep bash and perl arguments?
<genesis>
for the 2 scripts in share/xkbd
<infinisil>
genesis: they aren't used in the code and should therefore be removed.
<genesis>
i will fix them later
<infinisil>
They don't have any effect
<genesis>
the perl one would be usefull to fix i think
<infinisil>
You seem to have a misunderstanding on how package declarations work
<genesis>
but i'd prefer look at the Makefile.am later, and add the proper regex if i see a need.
<inquisitiv3>
infinisil++
<{^_^}>
infinisil's karma got increased to 20
<infinisil>
genesis: when you have a function like { stdenv, bash }: ... in nix and you're not using the bash argument anywhere, it can be removed, because it has no effect, the nix evaluation doesn't use that argument
<genesis>
even to patchshebangs them or try to generate locale later in my todolist ?
<genesis>
using xkbd-config
<ixxie>
tilpner: I don't see any such option and I have a server set up; just the snapshots option
<infinisil>
genesis: this has nothing to do with it
<tilpner>
ixxie - Click on your server, then there's a list of menu options on the left-ish side
<infinisil>
genesis: it's like creating a function `int foo(int x, int y) { return x; }` in C, you can remove the y argument because it's never referenced
<ixxie>
tilpner++
<{^_^}>
tilpner's karma got increased to 8
<ixxie>
indeed
<ixxie>
and NixOS is there
<infinisil>
genesis: your bash and perl arguments there are never referenced from the nix function implementation, so they can't have any effect
<{^_^}>
#45623 (by bignaux, 1 day ago, open): xkbd: init at 0.8.18
<genesis>
yeap, i thought i need them for the scripts, my bad, i remove and let it go.
<infinisil>
genesis: i mean, i'm assuming that you have verified the program working in its current state
<genesis>
it is
<genesis>
but i can't generate fr xkbd so i've a qwerty one
<genesis>
that's better than have the xkbd error in navit
<infinisil>
Not sure what this has to do with it
<tilpner>
gchristensen - You could always ask support, if this is something you care about: support@hetzner.com . Would you just tell them that there's demand, or do the work for them (as far as possible)?
<genesis>
xkdb-config can generate a xkdb profile according to your local, but it is more work i don't want to do right now to fix that upstream
<infinisil>
inquisitiv3_: the message says the error is on line 38
jluttine has joined #nixos
<inquisitiv3_>
infinisil: Line 38 is the one there "in {" is on. Do you want the whole config_
<inquisitiv3_>
?
<infinisil>
inquisitiv3_: Yea
<inquisitiv3_>
infinisil: Updated the gist
BlessJah_ has quit [Ping timeout: 272 seconds]
<infinisil>
inquisitiv3_: Ah yes, put lines 36-38 to line 2 instead
lejonet has quit [Ping timeout: 252 seconds]
<gchristensen>
inquisitiv3_: it looks like you've sort of gotten in to a mode where infinisil is doing the thinking and you're doing the typing. we're all volunteers here, and I worry you risk over-asking. if you're going to use Nix, you're going to need to learn Nix, and that means experimenting and trying stuff and reading error messages to build up your own ability to debug problems. We're still here to help, but we need
<gchristensen>
you to be holding up your own end of bargain, by trying to learn and solve your own problems before coming for help. If you need help with how to do this better, https://www.mikeash.com/getting_answers.html has a nice bit on good IRC etiquette for how to solve problems, and failing that find help.
<infinisil>
Thanks gchristensen, well put :)
<gchristensen>
:)
ArahaelPi has joined #nixos
init_6 has quit [Ping timeout: 244 seconds]
<inquisitiv3_>
gchristensen, infinisil: I'm sorry if I'm overasking. I guess I simply is tired of this problem and want to get this working ASAP. Sorry again.
<gchristensen>
it is okay! I get in to the same mode sometimes, especially with something new to me.
aszlig has quit [Quit: Kerneling down for reboot NOW.]
aszlig has joined #nixos
netixx has joined #nixos
<netixx>
Hi all, I have been working/rebasing a PR about openvswitch for a while, it's been reviewed by a couple of people. Is there any chances to have this merged into master (hopefully before 18.09 fork) ? https://github.com/NixOS/nixpkgs/pull/35127
<genesis>
i wonder, i put xkdb in my buildinputs, but navit can't find it, if i installed xkdb, it could
Izorkin has quit [Disconnected by services]
Izorkin has joined #nixos
<genesis>
so navit wants xkdb available in $PATH, how to do that ?
<genesis>
same with espeak, i need it cmdline, is that the right case for propagatedbuildinputs ?
<symphorien>
Just wrap the binary
<genesis>
with prefix path ?
<symphorien>
Yes
<genesis>
oki thanks
<symphorien>
Or patch the source to use an absolute path of you prefer
<genesis>
i don't since it's in user config, make it unreadable with nix path :)
edwtjo has joined #nixos
sigmundv__ has quit [Ping timeout: 268 seconds]
kenshinCH has joined #nixos
kenshinCH has quit [Client Quit]
<philippD>
I want to use a single file as src when calling mkDerivation but it doesn't work how I want to. I have set { src = ./test.md; unpackCmd = ":"; ... } but I get an error: "unpacker appears to have produced no directories" And my buildPhase script isn't called.
orivej has quit [Ping timeout: 244 seconds]
michaelpj has quit [Quit: ZNC 1.6.5 - http://znc.in]
<genesis>
yes
<infinisil>
philippD: unpackPhase = "cp $src .";
worldofpeace has joined #nixos
ixxie has joined #nixos
<philippD>
ah so the variable had the wrong name
<philippD>
unpackPhase = ":"; works for me
<infinisil>
philippD: builds get run in a directory which unpackPhase is supposed to populate
<infinisil>
":" just does nothing, so this won't get populated, which is fine when you're referring to the source by using the `$src` env var
<philippD>
I use ${src} which points to the nix store so that works too and I don't have to copy another file
<{^_^}>
[nixpkgs] @dotlambda pushed commit from @fgaz to master « endgame-singularity: init at 0.30c (#45057) »: https://git.io/fAm5l
<{^_^}>
[nixpkgs] @vkleen opened pull request #45655 → fetchpatch: should use the buildPackages patchutils → https://git.io/fAm5a
goibhniu has joined #nixos
r2d2leboss has joined #nixos
<r2d2leboss>
Hello. Is someone (a person, a team or a company) working on adding support to Nix (the package manager) on Windows without using WSL ? I would like to help but I would like to know if there has been previous works.
<{^_^}>
[nixpkgs] @vkleen closed pull request #45655 → fetchpatch: should use the buildPackages patchutils → https://git.io/fAm5a
simukis has quit [Read error: Connection reset by peer]
simukis has joined #nixos
<sevcsik>
Hi, could somebody point me to
<sevcsik>
Hi, could someone point me to a nix expression which builds an electron app from source? I'm having trouble with a lot of binary NPM packages electron depends on
<keith_analog>
Hi, I would like to find a way to configure my ipfs service in nixos to use the version of ipfs from the unstable channel (but continue to use the default channel otherwise). Could someone help me find the configuration I need? Thanks!
<mdash>
keith_analog: let unstable = import <nixos-unstable> {}; in {... services.ipfs = { package = unstable.ipfs; ... }; }
<pikajude>
is there an effort right now to implement incremental compilation of haskell packages in nixpkgs?
<blankhart>
i am having trouble building a haskell tensorflow project. nix-shell fails complaining that it is missing executable protoc (google binary for protocol buffers). what is my next step?
<LnL>
add protobuf?
<jonge>
blankhart: LnL: i think the compiler is in "pkgs.protobufc" while the lib stuff is in "pkgs.protobuf"
<LnL>
,locate /bin/protoc
<{^_^}>
Found in packages: protobuf, haskellPackages.protobuf-simple
<blankhart>
it should already be there
<blankhart>
oh
IRCsum has quit [Remote host closed the connection]
<jonge>
LnL: oh i just learned that protobufc has the c bindings. i have a project where both is needed as dependency and accidentally interchanged their meanings, oops. :'-O
<clever>
blankhart: pkgs.protobufc also contains a protoc symlink
<scode>
Apologies in advance for being dense, but I'm waiting for the fix to https://github.com/NixOS/nixpkgs/issues/45500 to land in the unstable channel. However, I think I'm misreading what's what in Hydra, because as far as I can tell the issue is present in unstable (as per nix-channel --update nixpkgs). But following links to hydra it appears to have relatively new builds. What am I missing?
<{^_^}>
#45500 (by mdorman, 3 days ago, closed): firefox+browserpass "is not allowed to refer to the following paths:
<r2d2leboss>
sevcsik: yes however it only works on Windows 10 ant not on Windows 7
mayhewluke has quit [Ping timeout: 252 seconds]
keith_analog has quit [Quit: Konversation terminated!]
mayhewluke has joined #nixos
<sevcsik>
Ah, sorry, you said *without* WSL...
<blankhart>
more generally how do i sequence the build so that nix installs a binary like protoc before building other dependencies?
<blankhart>
i tried putting protobuf in libraryToolDepends and setupHaskellDepends, but that doesn't seem to work
<clever>
blankhart: does the haskell package contain a library?
<pikajude>
snack is promising, but I've never gotten it to compile anything before
<clever>
pikajude: ive heard that snack doesnt support multiple cabal projects at once
<clever>
pikajude: and the main thing i want to use it on, is ~20 cabal projects depending on eachother
<blankhart>
clever: yes
<clever>
blankhart: can you pastebin your nix files?
<pikajude>
oh snack isn't meant to be used as just a development tool?
<pikajude>
huh
<pikajude>
interesting
<pi3r_>
Another way to ask this: is there any benefit to wrap the derivation inside pkgs.haskellPackages if you don't need to tweak your deps.
<clever>
pi3r_: yeah, i think those are the same
periklis has quit [Ping timeout: 276 seconds]
<clever>
pi3r_: i try to put it all in an overrides, because it makes it simpler to access the deps if you need them, and when the project grows, more things can depend on that package
<blankhart>
nix-shell
<clever>
blankhart: did you re-launch the shell? what does `type protoc` return?
<{^_^}>
[cabal2nix] @peti pushed to master « yaml: work around incomplete dependency specification in upstream Cabal file »: https://git.io/fAmAX
<blankhart>
clever, thanks! it complains that there is no libraryToolDepends attribute where you used it in shell.nix, but stops complaining if that is changed to setupHaskellDepends
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fAmAF
<daveo>
I am trying to install duplicity 0.7.18 from nixpkgs. I have cloned the repo and when I do a nix-build -A duplicity it copies it to the store but does not set my path. How do I have it also set the path
<clever>
daveo: nix ignores whatever you have currently installed while building
<clever>
so that wont make any difference
<sphaugh>
is packageOverrides evaluated before any overlays?
<daveo>
clever: shucks I was hoping that would fix my problem... error message is `BackendException: B2 backend requires B2 Python APIs (pip install b2)`
<clever>
sphaugh: not sure, i try to avoid mixing them
<clever>
daveo: the nix expression for the package will need to be fixed to include B2
<daveo>
clever: ok looking at that now
snajpa has joined #nixos
jperras has quit [Quit: WeeChat 2.2]
<{^_^}>
[nixpkgs] @volth opened pull request #45661 → perlPackages.SVNSimple: fix build with non-default perl → https://git.io/fAmpu
<bigvalen>
I'd like to turn my Nixos box into a windows domain controller. It's nearly 20 years since I messed with domain controllers, so I've no idea what I'm doing. I had a look at https://wiki.samba.org/index.php/Setting_up_Samba_as_an_Active_Directory_Domain_Controller and it seems to recommend using 'samba-tool' which ... might conflict with how NixOS does things.
<clever>
daveo: youll also need to add it to another file
<tilpner>
One problem might be (not familiar with python packaging) that backblaze-b2 is using buildPythonApplication. Perhaps it should use a library equivalent?
<daveo>
looking
<daveo>
I see that in nixpkgs/pkgs/development/tools/backblaze-b2/default.nix:5
<tilpner>
The directory structure seems fine though
<tilpner>
Is there an error?
<daveo>
Not sure if other people want/need b2 support for duplicity in the nixpkgs. If so I can put in a pull request
<daveo>
no error it seems to work fine
<daveo>
it is now backing up all my stuff to b2
__monty__ has quit [Read error: Connection reset by peer]
<tilpner>
It could be an off-by-default option
pi3r_ has quit [Ping timeout: 268 seconds]
<daveo>
good idea. that is beyond my knowledge at this point :)
<daveo>
I've seen what the options look like before but I don
<daveo>
't know exactly what the syntax is
<Thra11>
Wow. nixos hydra's aarch64 build server can build qtwebengine in just over 2 hours! That's fast!
<daveo>
Thra11: I don't know regular benchmarks for the qtwebengine build... what are they?
<Thra11>
daveo: All I know is my little machine takes over a day.
<Thra11>
It doesn't have a massive amount of RAM though, so it ends up swapping a lot in the later stages of the build
<daveo>
Thra11: trying to find the build now. unless you got a link
<bgamari>
I have tried an overlay with `mkDerivation = args: haskell.lib.enabledDWARFDebugging (mkDerivation args)` but this strange with the curious error
<bgamari>
anonymous function at /nix/store/797yi4rh4mzs1vfx6im9yznc9l4sj7bb-source/pkgs/development/haskell-modules/generic-builder.nix:13:1 called with unexpected argument 'mkDerivation', at /nix/store/797yi4rh4mzs1vfx6im9yznc9l4sj7bb-source/lib/customisation.nix:69:12
<clever>
bgamari: mkDerivation cant be wrapped like that
<bgamari>
hmm, alright then
<clever>
bgamari: let me check what i did...
<bigvalen>
Hmm. Any attempt to run nixos-rebuild is failing for me, because postfix was installed incorrectly.
<bigvalen>
[root@nas:/nix/store/r0z0z67cav0yffbpdpcz2f12b3vd9dwj-postfix-3.0.3/sbin]# ls -l |head -3
<bigvalen>
Any idea why it'd have prefixed the destination symlink with 'installdir' ?
<bgamari>
clever, sure, that will do
<bgamari>
clever, thanks!
<bigvalen>
Is there any way to mutate something in /nix/store to fix this ?
<clever>
bgamari: you want to fix the postfix derivation and then nixos-rebuild again
<daveo>
I think somehow that my duplicity/b2 backup scheme is inefficient. I was also looking at borg, but it looks like I have to make the incremental backups on a local drive and then use rclone to send them. I just want to send the backups to b2. Anybody use borg and just send the backup off without storing it locally?
<bgamari>
clever, do I? ;)
<bgamari>
bigvalen, ^
<clever>
bgamari: oops, yeah
<bigvalen>
clever, derivation ? I just have: packages.postfix.enable=true;
<tilpner>
daveo - Maybe slow, but AFAIK duplicity uses less space than borg
<clever>
bigvalen: you need to add an override to fix postfix
<daveo>
tilpner: ok I also just read it can do incremental as well
<clever>
bigvalen: version 3.3.1 doesnt have this issue, so it may already be fixed, what does `nix-channel --list` report?
<daveo_>
tilpner: backblaze seems to reccomend restic over duplicity
<daveo_>
tilpner: using restic right now. worked out of the box and it is much better. at least in terms of the info it puts to stdout :)
pi3r_ has quit [Quit: Leaving]
<adamantium>
clever: do we need legacy mounts? I was considering not using legacy, tested it with non-root datasets so far and it seems to work
graphene has joined #nixos
seanparsons has quit [Quit: ZNC 1.6.5 - http://znc.in]
seanparsons has joined #nixos
logzet has quit [Ping timeout: 264 seconds]
chris| has joined #nixos
<daveo_>
clever: just saw your message on vnstat and iftop. I have a KDE widget always up that shows how much I have ingoing and outgoing on the network
<clever>
daveo_: ah, that also does the same job as vnstat
<clever>
daveo_: though iftop can show more details, and tell you which domain is doing it all
lawlesseel has quit [Ping timeout: 250 seconds]
jperras has joined #nixos
aswanson has quit [Quit: WeeChat 2.0]
hamishmack has joined #nixos
<dhess>
Hi all, I've started using dhallToNix for bits of configuration. It's working pretty well, but I just ran into an issue that I hadn't considered.
<dhess>
dhallToNix depends on dhall-nix, which is a Haskell program. If you try to use it with config that is generated for, say, an aarch64-linux, then it fails, because GHC isn't available for that platform.
<dhess>
Now, there's really no reason why the config needs to be *generated* on that arch. Is there a way to tell Nix to build the derivation on a different system than the target system's arch? (Without needing to resort to a full cross-compile)
<clever>
daveo_: dhallToNix would need to be modified to accept 2 system params, then import nixpkgs twice, with those 2 values
<clever>
dhess: oops, ^^^
<daveo_>
clever: loving iftop but I only see ips. do you need something for it to resolve dns and show names??
<dhess>
clever: hmm, bummer.
<dhess>
well maybe it's time for me to figure out cross-compiles
lawlesseel has joined #nixos
Thra11 has quit [Ping timeout: 252 seconds]
<daveo_>
clever: nevermind now I see more than just ip addresses. Fascinating to see what all is interacting with this box
<dhess>
It would be nice if Nixpkgs supported an "all" pseudo-architecture like Debian does, where it doesn't matter where the derivation is built.
<clever>
daveo_: yep
<clever>
dhess: re-importing nixpkgs with builtins.currentSystem can get that effect
<clever>
dhess: but then it introduces an impurity
<dhess>
clever: Got any examples of that in your nix-tests repo?
<clever>
dhess: import <nixpkgs> { system = builtins.currentSystem; }
<clever>
that will always be for the current system, then you need to grab the dhall-nix from it
<daveo_>
or try to package it up and deploy it on kubernetes
<clever>
daveo_: iftop is just a much simpler form of the same thing, it uses pcap to gather the data
<infinisil>
Seems like a pain to package
<daveo_>
maybe. by package I meant make a Dockerfile and maybe a helm chart for it
<dhess>
clever: ok if I did that, would it be a problem for Hydra to build that?
<dhess>
I'm not really clear on when imports are Hydra are kosher
<clever>
dhess: hydra would eval it on an x86-64, so currentSystem will be x86-64
<dhess>
oh ok that's fine then
<clever>
dhess: but you wont import <nixpkgs> but instead `import pkgs.path { system = builtins.currentSysstem; }`
<clever>
dhess: and maybe also do something with config and overlays
<dhess>
clever: I have a fixed Nixpkgs so I'll use that
<dhess>
And maybe I'll use system = x86_64-linux so that it's consistent. That should remove the impurities I think?
<clever>
dhess: but that can cause problems if you try to build it on a real aarch64, it will demand an x86-64 build slave
<daveo_>
I just so wonder how safe this rpot thing is. Makes me want to recreate my own version of it to check all the components instead of using all the binaries in tatsu-i's repo. Call me paranoid
<dhess>
clever: I always deploy from either macOS or x86-64
<dhess>
and the macOS systems always have x86-64 remote builders
ted_ has joined #nixos
binaryphile has quit [Ping timeout: 268 seconds]
<dhess>
clever: that worked like a charm. Thanks!
<dhess>
I'll have to remember that trick.
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
makefu has quit [Ping timeout: 272 seconds]
<daveo_>
trying rpot in a vm now
Dedalo has joined #nixos
rprije has joined #nixos
orivej has quit [Ping timeout: 252 seconds]
kyren has quit [Ping timeout: 264 seconds]
thc202 has quit [Ping timeout: 244 seconds]
kyren has joined #nixos
ericsagnes has quit [Ping timeout: 264 seconds]
goibhniu has quit [Ping timeout: 268 seconds]
lfam has joined #nixos
lfam has quit [Quit: Leaving]
lawlesseel has quit [Remote host closed the connection]