cantstanya has quit [Remote host closed the connection]
erictapen has joined #nixos
<kalbasit>
gchristensen: that'd be great! I want to use it, and I'm not sure what to do with their shell based installation with embedded jar
<kalbasit>
gchristensen: thank you!
<manveru>
,libraries glenndavy
<{^_^}>
glenndavy: Don't install libraries through nix-env or systemPackages, use nix-shell instead. See https://nixos.wiki/wiki/FAQ/Libraries for details.
<manveru>
glenndavy: that said, you most likely want libffi
cantstanya has joined #nixos
<timclassic>
Does anyone know if there are any reasonable options if I want an aarch64 build system of my own with 8 (or gasp 16) GiB RAM?
janneke has quit [Ping timeout: 246 seconds]
<glenndavy>
manveru, ok thanks, reading up on libraries.
<glenndavy>
manveru, the reason im not looking at libffi, is theres a line near the top of the derviation that pretty much bails if we’re on darwin iirc.
<manveru>
hmm
<manveru>
no, that just disables tests on darwin
<glenndavy>
oh
<glenndavy>
oic
<glenndavy>
ok, great, thanks!
shibboleth has quit [Quit: shibboleth]
<glenndavy>
so to be clear, i can’t use a profile to ensure the libraries are available, i need to use shell?
<manveru>
i'd heavily recommend that :)
<manveru>
make a shell.nix with `with import <nixpkgs> {}; mkShell { buildInputs = [ libffi ]; }`
janneke has joined #nixos
<glenndavy>
yep, i do usually use shell, but was hoping to not have to for general cases
<manveru>
then you can run `nix-shell` and build your gems in it (just add whatever you want to buildInputs)
<glenndavy>
yep, understood
<glenndavy>
so, can i use a packageOverride as a means of achieving something similar?
<manveru>
for building gems or ruby apps, there's specifically bundix
<manveru>
which already includes the libffi dependency for the ffi gem
<ryantm>
jackdk: Yeah, anything we can do to make them easier to merge would be great. I've been slowly adding more and more checks to improve the PR quality. It looks like this particular one is bad because it causes 0 rebuilds.
<jackdk>
I don't have much nixos clout and no merge rights. Just want to get the issue on the table
<ryantm>
jackdk: If you have any ideas for streamlining the automated updates, that'd be great to hear.
vadim__ has quit [Remote host closed the connection]
<jackdk>
if there's enough hydra resources to do a darwin and linux build incl deps, and that passes, perhaps they should be botmerged? there's not much meaningful review possible
* DigitalKiwi
petitions that urls should always be quoted so that opening them works, alternatively a space between the end and the ;
<ryantm>
jackdk: There's lots of meaninful review possible, you can run the programs or link to the libraries, you can read the changelog, etc.
adetokunbo has joined #nixos
<jackdk>
true.
<DigitalKiwi>
(it often tries to include the ; as part of the url which is a 404; space or quotes makes it work)
<ryantm>
DigitalKiwi: I think there is some consensus around deprecating the URL primitive
reinhardt has joined #nixos
ninjin has quit [Remote host closed the connection]
<johnw>
is there a way to make Hydra rebuild a derivation with certain overrides, but only if it fails? My use case is that I have long-running builds that, if they fail, I want them to auto-rebuild with debugging on -- even though I expect that to also fail. This way, developers can look at the rebuilt derivation to see what the bug is.
ninjin has joined #nixos
kvda has joined #nixos
sgronblo has joined #nixos
dhkl has joined #nixos
Rusty1 has quit [Quit: Konversation terminated!]
sgronblo has quit [Ping timeout: 268 seconds]
xok has joined #nixos
<xok>
hello all...
<xok>
I've generated a perl module installation directives using the nix-generate-from-cpan
<xok>
and now this module is trying to use -lssl
<xok>
the error I'm getting is: *** Could not find OpenSSL
<xok>
I tried installing it with: nix-env -iA nixos.openssl
<xok>
it should be there but still can't install the plugin...
<xok>
can anyone help with this issue?..
<johnw>
please show the commands the causes the failure
bemeurer has quit [Quit: WeeChat 2.4]
<xok>
I can't show the commands...
<johnw>
why not?
<xok>
actually I can, but it wouldn't help..
<johnw>
not showing them helps less :)
<xok>
as I am trying to run nixops and install specific ( perl dancer ) environment...
<xok>
the command is as simple as this: nixops deploy -d MACHINE
<johnw>
is openssl a dependency of the derivation you're deploying?
<johnw>
even just telling me that nixops is involved helps
tertl3_ has quit [Quit: Connection closed for inactivity]
<xok>
well, it's not a dependency...
<xok>
but I've added that package...
<xok>
it still doesn't help...
<johnw>
that's not the same thing
<xok>
what do you mean?..
<johnw>
adding the package to the user environment is unrelated to satisfying a derivation's dependencies
<johnw>
Nix needs to know the graph of all dependencies of a project, for lots of different reason
<johnw>
nix-env -i foo doesn't establish a link between your Perl item and foo
kreisys has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
ddellacosta has quit [Ping timeout: 250 seconds]
mjrosenb has joined #nixos
orivej has joined #nixos
<immae>
Hello there! I have a derivaiton I would like to debug (because it fails), but I only have access to its drv file (the file in /nix/store/....drv ), since it is a gem built via bundix derivation. Do I have a way to open a debugging shell with just that drv file?
<xok>
johnw: thanks for your help anyway...
<xok>
johnw: ok, it seems the module that runs the test and fails is not Net::SSLeay
<xok>
but Net::HTTPS::Any instead..
<xok>
can you help me disable test for this now?..
<immae>
xok: usually doCheck = false;
<xok>
immae: where and how do I set it?..
<xok>
ok, never mind, I found it, trying now...
<immae>
ok :)
<immae>
it’s a derivation attribute, you should put it in the Net::HTTPS::Any attributes
<Arahael>
immae: I wonder if you could include it?
<immae>
Arahael: how?
<immae>
oh you mean directly in a derivation as build input?
<xok>
thank you guys...
<xok>
docheck was a great thing, it helped...
<immae>
cool
<xok>
I had to set it in several places and everything runs fine now...
<immae>
xok: sometimes check fail yes. It’s better when they can be enabled, but it’s not always possible
<xok>
what do you mean?..
<immae>
Well, there are a lot of doCheck = false in nixpkgs :p
<immae>
Arahael: someone reported a similar issue with another gem yesterday, and the solution works for me, so I’m good now, but I’m still interested in how to do the debugging myself
nD5Xjz has quit [Ping timeout: 245 seconds]
nD5Xjz has joined #nixos
<immae>
(poking manveru who solved the same issue yesterday with tzinfo ^, maybe you have some hints about how you debug those things?)
Havvy has joined #nixos
<immae>
I think it is a generic problem with gemspec in ruby 2.6.* and not just tzinfo btw
<jluttine>
i'm writing a script with pkgs.writeScript and this script uses nix-shell. what's the package that provides this command? i should call it as ${somepackage}/bin/nix-shell right?
<jluttine>
how do i find which package provides an executable?
<elvishjerricco>
jluttine: pkgs.nix is what you're looking for
<elvishjerricco>
,locate bin nix-shell
<{^_^}>
Found in packages: nix1, nixStable, nixUnstable
<elvishjerricco>
huh
<elvishjerricco>
anyway pkgs.nix is definitely it
<jluttine>
yeah, i found those three too so i was confused.. thanks!
<elvishjerricco>
bwe: I don't understand what that docker feature does. Does it just... spawn identical processes? That seems rather useless
zupo has quit [Client Quit]
<dminuoso>
elvishjerricco: What docker feature are you talking about?
<dminuoso>
Oh haha. The weird formatting of lines in weechat coupled with the window sizing hid that link from bwe.
<elvishjerricco>
dminuoso: `docker service scale`
Ariakenom has joined #nixos
<dminuoso>
That's just docker swarm
<elvishjerricco>
I do not know what that is :P
<elvishjerricco>
And the main page for it just tells you what advantages you can expect, not how it actually works :/ https://docs.docker.com/engine/swarm/
<dminuoso>
Pretty much the reason Docker is so useful, and pretty much the reason why Docker is not so useful for most people.
<dminuoso>
(At least the concept in itself)
<dminuoso>
The whole ecosystem of Docker/Swarm/Kubernetes/etc is to solve a problem most people never ever have.. :-)
<elvishjerricco>
It sounds like it can provision new instances somehow? Doesn't that require integration with your hosting provider?
<elvishjerricco>
I thought docker itself was independent of that
<dminuoso>
elvishjerricco: No. Think of it like...
<dminuoso>
elvishjerricco: A kind of (worker) thread pool management.. except it doesnt manage threads but docker containers.
<dminuoso>
And thing like docker service scale lets you tap into the pool size of containers.
<jluttine>
hmm.. nix-shell in writeScript script doesn't quite work: error: file 'nixpkgs' was not found in the Nix search path (add it using $NIX_PATH or -I)
<elvishjerricco>
dminuoso: So it just spawns or kills processes on one physical machine?
linarcx has joined #nixos
<jluttine>
how can i get the nixpkgs path when writing writeScript script?
adetokunbo has quit [Quit: This computer has gone to sleep]
<dminuoso>
elvishjerricco: Yes. Those processes themselves would be running docker containers.
<jluttine>
i tried using `-I ${pkgs}` but that didn't work
adetokunbo has joined #nixos
<elvishjerricco>
jluttine: How are you trying to use it? If you're trying to use Nix within a nix derivation's build process, you're going to have a bad time and shouldn't do that
<dminuoso>
elvishjerricco: (Actually I dont know whether it's actually *processes* and not *threads*, but the distinction is rather useless anyway)
<elvishjerricco>
Just write the program to be multithreaded
<dminuoso>
elvishjerricco: Again, its a solution to a problem most people don't have.
<jluttine>
elvishjerricco: I'm trying to write a script that would launch nix-shell and run a command inside that shell..
<jluttine>
elvishjerricco: this script is needed by one systemd service i'm writing
<dminuoso>
elvishjerricco: With docker you can have special networking between those containers (giving you stronger isolation), and you can scale across host machines
<elvishjerricco>
And what problem is that? Multithreading? There's tons of trivial ways to do that without a special docker feature
<elvishjerricco>
jluttine: Why does a systemd service need to get its programs from nix-shell? Why not just bake them in?
<jluttine>
elvishjerricco: the environment needs to vary run-time
<dminuoso>
elvishjerricco: Simply writing something "multithreaded" is hard actually.
jackdk has quit [Ping timeout: 268 seconds]
<elvishjerricco>
dminuoso: Sure
<dminuoso>
elvishjerricco: A single segfault will bring your whole process down for example.
<elvishjerricco>
But, you can use like gnu parallel to get the same effect as this docker scale thing
<jluttine>
elvishjerricco: whenever it is launched by the service, it should read the shell.nix file to determine what kind of environment is wanted
<jluttine>
elvishjerricco: so i cannot bake that in..
<elvishjerricco>
jluttine: Ah, ok. nixpkgs is found via an environment variable NIX_PATH
<dminuoso>
elvishjerricco: Docker gives you the network overlay, which is pretty useful
<dminuoso>
elvishjerricco: And builtin service discovery via DNS.
<elvishjerricco>
dminuoso: Hm, yea I guess that could be helpful.
<dminuoso>
(Well. Docker swarm to be precise)
<elvishjerricco>
jluttine: systemd services don't get hardly any environment variables by default; definitely not NIX_PATH
<jluttine>
elvishjerricco: yep.. that's the problem...
<dminuoso>
elvishjerricco: I mean if gnu parallel works for you, or just a single threaded application - then yeah docker swarm is just massive complexitivity overkill.
<elvishjerricco>
jluttine: Do you want `<nixpkgs>` to point to the nixpkgs from build time, or the current system's nixpkgs?
<jluttine>
elvishjerricco: maybe i need to pin the nixpkgs in shell.nix
erictapen has quit [Ping timeout: 245 seconds]
<elvishjerricco>
current channel's*
<jluttine>
elvishjerricco: not sure.. :/ maybe whichever is easier :)
<elvishjerricco>
jluttine: Easiest is build time
<dminuoso>
elvishjerricco: Another cool thing is that swarm automatically handles mutual TLS between the services - which greatly reduces your burden to handle PKI.
<jluttine>
elvishjerricco: or maybe i'll just pin nixpkgs in shell.nix..
<elvishjerricco>
`pkgs.path` refers to the path of nixpkgs. So set `NIX_PATH=nixpkgs=${pkgs.path}`
<jluttine>
elvishjerricco: thanks!!
<dminuoso>
(Which we all do far less than we pretend we do when we meet at conferences, hackathon or similar events)
<elvishjerricco>
Using TLS at all to communicate between services on the same machine seems completely unnecessary :P
<dminuoso>
elvishjerricco: Again. Multi-host. ;-)
<dminuoso>
elvishjerricco: Part of the reason of the network overlay (as opposed to using SHMEM, pipes or other techniques) is to be able to seamlessly scale across many machines.
<Ariakenom>
dminuoso: "_automatically_ handles _manual_ TLS" im sure this means something but I found it funny :)
<dminuoso>
Ariakenom: Oh heh
<dminuoso>
elvishjerricco: Kubernetes and Swarm are great features... if your problem begins "how do I deploy a service and make it scale across dozens/hundreds/thousands of servers painlessly, and set it up in such a way that the ops guys can look over the "swarm" of nodes running your service..
<dminuoso>
elvishjerricco: But like I said. Very few people actually have those problems.
<dminuoso>
For most others its just cargo cult tech
<elvishjerricco>
bwe: Anyway, you can use systemd's `@` syntax to define a service that can be spawned many times. Then you can just add `foo@1` `foo@2` etc to the wants of `multi-user.target`. They won't be containerized in anyway; look into nixos containers, both declarative and imperative in the NixOS manual if you need containerization.
<jfc>
Hey folks, is it possible to update my kernel without switching everything to unstable? I have a poorly supported integrated soundcard on my motherboard (codec CA0132), and it looks like there are some patches upstream which might address my sound issues.
Makaveli7 has joined #nixos
<johanot>
jfc: you can overlay linuxPackages
<jfc>
Awesome. Is there any documentation (or search terms) I can look into? I'm still unfortunately learning my way around nix
Makaveli7 has quit [Read error: Connection reset by peer]
Makaveli7 has joined #nixos
Makaveli7 has quit [Read error: Connection reset by peer]
Makaveli7 has joined #nixos
<johanot>
jfc: I would do: NIX_PATH=nixpkgs=/nix/var/nix/profiles/per-user/<user>/channels/<channel> nix search <foo>
<johanot>
but there's very likely an easier way :)
<jfc>
johanot: Thanks, definitely better than nothing! I dug through the github and found 4_19 in the unstable repo, which just finished compiling. Thanks again for the help, and wish me luck. :)
<johanot>
oh.. nix search has the -I flag.. so "nix search -I nixpkgs=/nix/var/nix/profiles/per-user/<user>/channels/<channel> <foo>" is as good as setting the NIX_PATH env var
sgronblo has quit [Ping timeout: 245 seconds]
<johanot>
jfc: I don't have channels on my system though.. I usually set nixpkgs to some git rev of a nixpkgs checkout, or a nixpkgs tarball
<immae>
manveru: yes, I actually got it in a few gems already. Do you have an idea of where it comes from, and more important, how did you figure it out the first time? I’m interested in being able to go into the build environment of a gem whose derivation comes from bundix, to help debugging, but the only thing I have is a drv file, which has all the information but cannot be directly used to spawn a shell.
xok has quit [Quit: Leaving.]
<manveru>
immae: i just read the error message :)
<manveru>
then i dug into rubygems to find the message and what causes it
<manveru>
and it turned out that it's a new check they introduced that makes sure all files mentioned in the gemspec are present, which for some reason isn't the case for some gems
<manveru>
but since the files array is actually only used to build a gem, we don't need it for actually running it, so i remove the entry in the gemspec to skip the check
<manveru>
which is a hack, but way faster than actually figuring out why the files don't match, so suitable for 3am :)
<{^_^}>
[nixpkgs] @dotlambda pushed 3 commits to release-19.03: https://git.io/fjIYt
<manveru>
my theory is that the gems were built with old rubygems versions, and there's a regression in the rubygems bundled with 2.6, but haven't had time to check it yet
<{^_^}>
[nixpkgs] @peti pushed commit from @cdepillabout to haskell-updates « Add option to disable library-for-ghci in the generic haskell builder. »: https://git.io/fjIYj
<{^_^}>
[nixpkgs] @peti pushed commit from @cdepillabout to haskell-updates « Add option to disable library-for-ghci in the generic haskell builder. »: https://git.io/fjIO0
<{^_^}>
[nixpkgs] @peti pushed commit from @cdepillabout to master « Add option to disable library-for-ghci in the generic haskell builder. »: https://git.io/fjIOE
<chisui>
Hey, I'm trying to set up NixOs on my new machine. After some time I got it running but unfortunately I have no networking manager in KDE. The whole application is missing. here is my configuration.nix https://pastebin.com/m8CPT9xN
<mikky>
I'll be happy if it was at least possible to install pyghmi to a virtualenv, but it fails with gcc compilation error due to incorrect includes: https://tinyurl.com/y2s8vh5m. I have absolutely no idea how to fix this on NixOS and it is, as far as I understand, a nixos-specific issue.
<Taneb>
Is there a way to use something like Hydra's RunCommand, when all the jobs in an evalutation of a jobset finish?
<pie__>
no it wont :( something is strictly evaluated somewhere, probabyl the recursiveupdate...really the problem is, when i run mkPluginInfrastructure, i get a {plugins = ...; withPlugins = ...}
<pie__>
and I want to be able to override the .plugins and have that influence the withPlugins
<pie__>
which lets me do something like: with import <nixpkgs> {}; let ghidra = callPackage ./default.nix {}; in ghidra.withPlugins (p: with p; [ ghidra-scala-loader ])
<pie__>
Profpatsch, do you know if // is strict?
<pie__>
hmm...I guess it wouldnt make sense if it was
<MasseR>
SSH_AUTH_SOCK is no longer set on 19.03, I can't find anything obvious on issues or `git diff origin/release-18.09..origin/release-19.03`
<clever>
genesis: just nix-build if you want to test things
<clever>
nix-env will keep every broken version around for months (to years), incase you want to perform rollbacks
<genesis>
oki, anyway i don't have space for rollback
<genesis>
i always gc .
<clever>
nix-env adds GC roots, so you cant gc those versions
<clever>
you have to delete gc roots via -d or --delete-older-than
<clever>
which can sometimes delete roots you did want
<genesis>
i do :)
<{^_^}>
[nixops] @PsyanticY opened pull request #1121 → ec2: Usinng boto3 in operations: stop/start/reboot/console-output → https://git.io/fjIZN
MasseR has quit [Quit: WeeChat 2.0]
<clever>
gchristensen: making progress on packaging electron, and damn does it ever take forever to even fetch the sources
<slabity>
If you think that takes forever you should try opening an electron app :P
azdle has quit [Read error: Connection reset by peer]
<clever>
slabity: i already do that on a daily basis
<clever>
slabity: i'm currently debugging a event loop bug in the core of electron
<averell>
in case anyone else is getting stack overflows on rebuild, i just had to set ulimit -s 1000000 because 100M wasn't enough.
Soo_Slow has quit [Remote host closed the connection]
<mikky>
I have installed a manually updated (i.e. from local nixpkgs tree) python lib. How do I make that lib "visible" to system-wide python environment?
<Orbstheorem>
Hi! Has anybody though of a command that finds which package provides a man page and runs `nix-shell -p package --run 'man dodo'`, in the same spirit as command-not-found?
rauno has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @peterhoeg opened pull request #58800 → openvpn: support for updating systemd-resolved with DNS servers → https://git.io/fjIcK
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
samrose_ has quit [Quit: Konversation terminated!]
<nefix>
betaboon: Thanks! That's what I was looking for! And sure, I'm going to let you know if it does finish with something decent and I don't drop it!
<jluttine>
i have systemd service with script defined. however, when i stop the service, the script isn't killed.. how to make systemd kill the script?
Fare has joined #nixos
<infinisil>
jluttine: It definitely should, what's the script?
i1nfusion has quit [Remote host closed the connection]
<jluttine>
well, i'm not 100% whether it kills the script but that jupyterhub process seems to remain and it's a blocking call so it should mean the script isn't killed
i1nfusion has quit [Remote host closed the connection]
drakonis has joined #nixos
linarcx has joined #nixos
i1nfusion has joined #nixos
illegalprime has quit [Remote host closed the connection]
Mr_Keyser_Soze has quit [Ping timeout: 245 seconds]
kleisli has joined #nixos
<pie__>
clever, so I think I want to write my own newScope function thats based on my plugin set?
linarcx has quit [Ping timeout: 245 seconds]
<pie__>
i mean, as opposed to pulling nix all of pkgs
kleisli has quit [Client Quit]
<pie__>
thats the easy part though
<clever>
pie__: you could use callPackageWith to create a new scope that contains nothing else
<clever>
pie__: but then its difficult to get non-plugin things like stdenv
<pie__>
wouldnt I just pass those as I've done so far, via the pkgs callPackage?
<pie__>
hm, not sure what that would look like actually
<pie__>
well, im not really interested in *specifically* callPackage
linarcx has joined #nixos
<pie__>
isnt it just a coincidence that callpackage and overrideScope are from the same thing?
<clever>
mostly
<pie__>
(besides the fact that makeScope happens to call newScope)
sgronblo has quit [Ping timeout: 250 seconds]
<pie__>
i mean, maybe i need callPackage, IDK, I just ?dont need? to be able to implicitly pass plugins to eachother
<pie__>
clever, am I wrong to think this way?
Thinkofname[m] has joined #nixos
aanderse-work has quit []
<clever>
pie__: i think you probably do want callPackage
<pie__>
though this still doesnt completely solve my problem, or im thinking about it wrong; I still dont see how this will let my withPlugins see overridden .plugins
<pie__>
or rephrased, how do I relate makeScope and mkPluginInfrastructure?
<pie__>
what question do i ask to allow me to answer this
<Taneb>
In the python library I'm packaging, running "python setup.py build" creates some "__init.py__" files that don't seem to be carried through in the nix version. Any idea what I could be missing?
<wolfman154>
10:42 wolfman154: https://hastebin.com/jayebaxoco.bashhttps://hastebin.com/odovacowip.rb I am using these instructions 17:03 gchristensen: your path to your config would be "${config.xsession.windowManager.xmonad.config}" and getting errors, the two links in this paste show the paste bin?
<wolfman154>
How would I change the syntax to get it to work?
fendor has quit [Remote host closed the connection]
slack1256 has quit [Remote host closed the connection]
fendor has joined #nixos
Soo_Slow has joined #nixos
<pie__>
clever, I pushed a new branch called overrides
<wolfman154>
10:44 wolfman154: 10:42 wolfman154: https://hastebin.com/jayebaxoco.bashhttps://hastebin.com/odovacowip.rb I am using these instructions 17:03 gchristensen: your path to your config would be "${config.xsession.windowManager.xmonad.config}" and getting errors, the two links in this paste show the paste bin?
<wolfman154>
What is the correct syntax? I’m new to nix?
<{^_^}>
[nixpkgs] @obadz pushed commit from @dtzWill to master « networkmanager-l2tp: 1.2.10 -> 1.2.12 »: https://git.io/fjI8j
<srhb>
wolfman154: I'm not sure what you're trying to do there.
<srhb>
wolfman154: Can you elaborate on what your goal is?
<pie__>
clever, I think I found out what I need newscope for, i need it for callpackage, and I need callpackage because of the way I made all-plugins.nix
<pie__>
i want to be able to pass ghidra and utilities
<pie__>
so i guess in need ghidra.callPackage so such a callpackage can be in scope for the user
<pie__>
but im still confused about how to construct a newScope; the newscope in pkgs seems to carry around all of pkgs, but then whats the point of the second makeScope parameter
<pie__>
that question seems bad but then I cant figure out what im confused about
<wolfman154>
srhb: I got xmonad to work two days ago by creating an xmonad.hs to link to now I am trying this: 17:01 wolfman154: gchristensen: https://rycee.gitlab.io/home-manager/options.html 17:03 wolfman154: gchristensen: scroll all the way to the bottom to xsession.windowManager.xmonad.config
sb0 has quit [Remote host closed the connection]
<wolfman154>
10:49 wolfman154: 10:44 wolfman154: 10:42 wolfman154: https://hastebin.com/jayebaxoco.bashhttps://hastebin.com/odovacowip.rb I am using these instructions 17:03 gchristensen: your path to your config would be "${config.xsession.windowManager.xmonad.config}" and getting errors, the two links in this paste show the paste bin?
<{^_^}>
[nixpkgs] @vcunat pushed 3 commits to release-19.03: https://git.io/fjI4m
leat3 has quit [Quit: WeeChat 2.3]
<srhb>
wolfman154: Though frankly managing the xmonad config inline that sounds really tedious, so I'd suggest putting it next to your home-manager config as xmonad.hs and having config = ./xmonad.hs;
i1nfusion has quit [Remote host closed the connection]
isidore has quit [Quit: WeeChat 2.2]
Fare has quit [Ping timeout: 264 seconds]
i1nfusion has joined #nixos
Glider_IRC has quit [Ping timeout: 245 seconds]
<wolfman154>
srhb: oh, so that’s how, thanks for the link, if I understand correctly I do the pkgs.writeText text into a a created file called xmonad.nix in the same directory as home.nix the call then change config path to that location and nix will then generate the xmonad.hs file for me, is this what you mean?
<wolfman154>
srhb: I’m typing on my phone so sorry for the typos
<srhb>
No, I literally mean an xmonad.hs that is just a regular haskell file.
Fare has joined #nixos
<srhb>
wolfman154: Upon building your configuration, nix will import that to the store.
<wolfman154>
srhb: so home.mix’s xmonad.config links to the xmonad.nix file created in the same directory, I just want to make sure I am doing this correctly?
<srhb>
wolfman154: There's no xmonad.nix file. There's home.nix and right beside it is xmonad.hs
rauno has joined #nixos
<swerlk>
Hi. I was hoping to get some help getting xpra to work using the xdummy driver. There is a more detailed description of the issue on the github issue tracker: https://github.com/NixOS/nixpkgs/issues/24652
<{^_^}>
#24652 (by calvertvl, 1 year ago, open): Cannot run xpra using Xdummy in parallel with desktop X environment
<wolfman154>
srhb: Oh ok, I see what you mean thanks for the tips, It’s good to know there’s two ways to do it, I think I’ll try to do it the nix way though
<wolfman154>
srhb++
<{^_^}>
srhb's karma got increased to 44
waleee has quit [Quit: WeeChat 2.4]
<srhb>
wolfman154: Okay. That just means you'll roundtrip through a nix string, which is sort of superfluous, but the end result will be the same.
<gchristensen>
it is just fine, and normal "nix way" to have it be a separate file like she suggests
ddellacosta has joined #nixos
<wolfman154>
Ok thanks for the tips, I struggle choosing between options.
Fare has quit [Ping timeout: 255 seconds]
<wolfman154>
srhb++
<{^_^}>
srhb's karma got increased to 45
<wolfman154>
gchristensen++
<{^_^}>
gchristensen's karma got increased to 98
freeman42x has joined #nixos
<srhb>
swerlk: I'm not sure I really understand the issue from that (I don't know xpra...) -- is this just a matter of making a driver available without changing the display manager?
<srhb>
(I'm guessing no, but perhaps more details will make the issue clearer)
wolfman154 has quit [Ping timeout: 246 seconds]
<ajs124>
has anyone here deployed bitwarden_rs on nixos?
<swerlk>
srhb: I am not too sure on the particulars of xpra itself, as I have only really been a user. But in essence it makes dummy X displays that can be pulled up and hidden, in a similar way to a screen miltuplexer. So usually the process is to start an xserver using `xpra start :DISPLAY_NUMBER` and then that sets up a screen that can be connected to on a particlar port
<swerlk>
I think part of the current problem is that there is no access to the dummy driver, which is causing this error when trying to run using start: `(EE) parse_vt_settings: Cannot open /dev/tty0 (Permission denied)`
<tobiasBora>
I tried to draw a stupid simple matplotlib drawing in python and I get the error:
<tobiasBora>
UserWarning: Matplotlib is currently using agg, which is a non-GUI backend, so cannot show the figure.
<tobiasBora>
why is nixos using agg, and how can I change that??
<swerlk>
tobiasBora: You can use `tkinter` with the withPackages function, and then manually set the backend to TkAgg. Or set the backend using matplotlibrc
<tobiasBora>
swerlk: I cannot make it default, rather than editing all the scripts in the world?
sgronblo has quit [Ping timeout: 250 seconds]
<tobiasBora>
oh, installing tkinter seems to be enough
<tobiasBora>
I'm wondering, is there any decent "default" install that installs all the deps?
<tobiasBora>
(the usual ones)
<swerlk>
tobiasBora: There is a weird way you might be able to do it using the `MPLBACKEND` environmental variable and setting that before you call the interpreter if you wanted to make it specific to a specific nix derivation
cfricke has joined #nixos
<srhb>
tobiasBora: For matplotlib? I don't think we have any sort of "collection" for that.
<srhb>
tobiasBora: Easy to make one, though.
<swerlk>
Collection?
<swerlk>
Oh nvm, sorry
<srhb>
swerlk: some sort of in-tree withPackages expression that has all the usual stuff.
<swerlk>
Ya
<swerlk>
That would be pretty helpful
<swerlk>
Actually my xpra question is from wanting to use python and matplotlib from a remote machine
<tobiasBora>
srhb: may be a good idea, it can be tedious to think to all the possible things to install. Like for KDE I keep forgetting deps, and I spent my life founding new issues because I forgot to install package XXX that usual KDE install have
<srhb>
tobiasBora: The other end of that spectrum is getting people to agree one some universal standard, though. :-P
<tobiasBora>
srhb: well a standard with too few packages is always better than no standard. And I think that KDE officially recommends some programs no? Like most of the people wants to visualize pictures, take screenshots, unzip/untar/tar/zip files.
<eacameron>
when I do `nix-shell -p clang` I get `clangd` on my path, but when I do `nix-build -E '(import <nixpkgs> {}).clang` I don't get anything with clangd. The difference is that the latter builds `clang-wrapped`. How can I get the standard clang?
<swerlk>
srhb: I'm very new to nix, so I am not quite sure how to add the xdummy to the list to drivers. Would you have a reference/know how to do that?
<sphalerite>
,locate clang
<{^_^}>
Found in packages: dpdk, psol, root, clang, swift, z88dk, llvm_4, castxml, clang_4, clang_6, clang_7, clang_8, llvm_35, llvm_37, llvm_38, llvm_39, clang_35, clang_37, clang_38, clang_39, linux.dev, clang-polly, clang_multi, jetbrains.clion, emscriptenfastcomp, linuxPackages.dpdk, vimPlugins.clighter8, linuxPackages_4_4.dpdk, linuxPackages_4_9.dpdk, linuxPackages_5_0.dpdk, vimPlugins.deoplete-go, linuxPackages_4_14.dpdk, and 19 more
<sphalerite>
,locate clangd
<{^_^}>
Found in packages: jetbrains.clion, llvmPackages.clang-unwrapped
<srhb>
swerlk: (Putting in just "dummy" will likely nuke your normal drivers from orbit unless the default list has the same option priority, which i don't think it does.)
<eacameron>
sphalerite: thank you. Where is the mapping that causes nix-shell -p to do something different?
<srhb>
eacameron: -p is essentially a mkDerivation with the -p package as a buildInput (roughly)
<sphalerite>
eacameron: it may be a setup hook, propagated build inputs, or something like that that takes effect when clang is used as a buildInput
mjacob has quit [Ping timeout: 252 seconds]
<eacameron>
oh I see.
<swerlk>
srhb: Hmm I added the driver and tried a rebuild switch but still have the permissions problem
<srhb>
swerlk: I'm not sure why that driver _would_ change that, but again, I don't know the software at hand. If this is unprivileged stuff, it sounds like your regular shouldn't really be able to touch that tty node anyway.
<srhb>
your regular user*
cfricke has quit [Quit: WeeChat 2.4]
<swerlk>
srhb: I used [ "nvidia" "dummy" ], which seemed to not take the whole system down. I should really be doing this locally but oh well
nD5Xjz has quit [Ping timeout: 245 seconds]
<srhb>
swerlk: Does the same issue persist as root?
<srhb>
swerlk: (which has write access, fwiw)
nD5Xjz has joined #nixos
<swerlk>
srhb: I am not sure how the dummy driver works but the description is to "allow a standard X11 server to be used by non-root users with the dummy video driver (xf86-video-dummy)." https://xpra.org/trac/wiki/Xdummy
<srhb>
swerlk: But good, that's at least a different error.
<srhb>
swerlk: I would suspect that the next step is feeding it an xserver config that has a screen defined which actually uses that dummy device.
<srhb>
swerlk: (massive disclaimer, I'm talking out of my arse here, in case it wasn't obvious)
<srhb>
swerlk: But as long as no one else has got any bright ideas, you can discard or use as you see fit ;-)
<swerlk>
srhb: Ah, well your arse is better than my brain at the moment since I haven't gotten anywhere
<swerlk>
swerlk: I'll try something like that when I get the time. I think I am just going to resign myself to remoting into a Mac at work at the moment (not totally fun)
<swerlk>
srhb: Well adding myself to the tty group changed the error a bit, so that seems like some part of the direction. The new error is `(EE) xf86OpenConsole: Cannot open virtual console 4 (Permission denied)`
<srhb>
swerlk: But maybe we're sidetracked if it's not supposed to try and use a console at all.
<srhb>
swerlk: In which case maybe feeding it an xorg config that only relies on dummy does help, but.. Guesswork. Anyway, building it currently..
<srhb>
swerlk: Maybe you can answer whether it should work like that? Like, does it take over a physical console normally?
<swerlk>
It does not
<srhb>
Then I guess the permissions and the tty stuff is a red herring.
<swerlk>
You get floating windows of any application running in the display. Similar to how the unified view in VMWare Fusion or Parallels works (if you have seen those before)
<srhb>
Ah, got it.
<srhb>
swerlk: Actually, can you tell me how to reproduce? What should I be running?
<swerlk>
srhb: The normal method to do this on another distro is to run `xpra start :10` to start a server and then that can be attached to using `xpra attach :10`
<pie_>
clever, I don't know why I had so much trouble with that...I'm not feeling fond of complicated fixpoint stuff. It's like I can barely read what I'm looking at
<srhb>
clever: I don't think that helps xpra point an Xorg process at a different file.
<pie_>
I read over it again and I think "how the heck did I miss that the first time around@
<srhb>
swerlk: So, I can do something like.. (bear with me, ... ) /nix/store/1s84kn47fpvhck84b042c5gbg4y60yy2-xpra-2.3.4/bin/xpra_Xdummy :10 -config /nix/store/1s84kn47fpvhck84b042c5gbg4y60yy2-xpra-2.3.4/etc/xpra/xorg.conf -configdir /nix/store/1s84kn47fpvhck84b042c5gbg4y60yy2-xpra-2.3.4/etc/X11 -modulepath /nix/store/fmf0lczh64a9sig0dbv50anyh8spxa9j-xf86-video-dummy-0.3.8
<srhb>
swerlk: It doesn't *work* but I'm getting closer, I think..
<swerlk>
srhb: I'm frightened
<srhb>
Yes, sorry about the paths, random hackery...
<srhb>
The point is that I'm pointing some backend part of it at the right config files and the dummy driver as well
<srhb>
swerlk: Now it finds the dummy driver and fails on missing the fb module (which is fair since I'm *only* pointing it at the dummy driver now)
<averell>
anyone have any tips on how to assimilate a 2GB vps (incl. zfs encrypted root)? I'm trying nixos-generators / kexec but i there's not enough ram to extract it to a ramdisk root. is there a way to maybe exclude linux unfree firmware (that's 200+ mb)?
sgronblo has quit [Ping timeout: 250 seconds]
<srhb>
swerlk: iow this is totally fixable. :-P
<tilpner>
averell: I have used the kexec tarball with a 2GB VPS
<averell>
how did you bootstrap that?
zupo has joined #nixos
<tilpner>
What do you mean? I copy my tarball and execute it
<swerlk>
srhb: Haha I trust this is true. So you are attempting to call the subcomponents directly
<averell>
let's say it's sysrescuecd, then i have a ramdisk root, and a HD. the HD is not mounted at / so i can't extract kexec.tar.gz there, and the / ramdisk is not big enough.
<srhb>
swerlk: Yes, just to sidestep some boring glue stuff and see if there's any fundamental blocker.
<averell>
just install some ubuntu or similar, and use that probably?
<srhb>
swerlk: To which I think the answer now is "no, there isn't"
<srhb>
swerlk: Ok, it runs now.. No idea if it _works_
pingiun has joined #nixos
<srhb>
swerlk: I needed to add the corresponding ModulePaths for xorg.xorgserver {lib/xorg/modules /lib/xorg/modules/extensions /lib/xorg/modules/drivers} and the one from xorg.xf86videodummy itself.
<srhb>
swerlk: My command line is this: result/bin/xpra start :10 --xvfb='Xorg -config /home/sarah/nixpkgs/xorg.conf'
<srhb>
swerlk: Where that xorg.conf is the one shipped with xpra, modified with the above modulepaths
<srhb>
swerlk: I think that's it for me tonight. Hope it helped :)
<wolfman154>
srhb: I removed gnome as my de and set xmonad as the default, when I reboot it works, but it shows xmonad as a tiny little box and the rest of my screen is black, I did not have this issue when when I installed xmonad in configuration.nix, the problem is with home-manager?
<srhb>
wolfman154: I'm off to read a book for the night, but perhaps if you pastebin all your config files, someone else may be able to help figure out what's wrong with it :)
<swerlk>
srhb: Thanks for all the help! I'll try to get that running on my system and let you know if it worked.
<swerlk>
srhb++
<{^_^}>
srhb's karma got increased to 47
<wolfman154>
srhb: ok later
<swerlk>
srhb: I assume karma is a thing that is valuable
<srhb>
swerlk: Oh yeah, 1-1 ratio to gold bars or so I hear.
<srhb>
:-)
<srhb>
'night
<swerlk>
night
<wolfman154>
I removed gnome as my de and set xmonad as the default, when I reboot it works, but it shows xmonad as a tiny little box and the rest of my screen is black, I did not have this issue when when I installed xmonad in configuration.nix, the problem is with home-manager?
<wolfman154>
Do you guys have any idea why this is happening?
<wolfman154>
I installed xmonad in home.nix?
<rycee>
wolfman154: Does it show anything in the user journal?
<wolfman154>
rycee: where’s that, sorry I’m still new to nixos?
obadz has quit [Ping timeout: 250 seconds]
ym555 has joined #nixos
trh_ has quit [Ping timeout: 245 seconds]
ng0 has quit [Remote host closed the connection]
<rycee>
You can run `journalctl --user -b` in a terminal to see the messages from the current boot. Have a look at the beginning to see if there are any obvious problems.
domogled1 has joined #nixos
ng0 has joined #nixos
<wolfman154>
rycee: ok I’ll take a look
<rycee>
wolfman12345: I've never seen a similar issue with xmonad.
<rycee>
wolfman154: Have you used the xmonad configuration previously?
<wolfman154>
samueldr: I’m pretty sure it’s xmonad opening the terminal,urxvt I think
<wolfman154>
I’m confused :(
obadz has joined #nixos
<rycee>
wolfman154: Do you have an ~/.xsession file?
<rycee>
Does it contain a line like `/nix/store/kqm963gvsldjzm1b80pz4va29i004j7j-xmonad-with-packages-8.4.4/bin/xmonad`?
wfranzini has quit [Remote host closed the connection]
<wolfman154>
rycee: let me check
fendor_ has joined #nixos
<wolfman154>
rycee: am I supposed to enable xmonad with both xserver and xsession?
civodul has joined #nixos
<Aleksejs>
Hi, recently I started to see this at the rebuild "Failed to start local-fs.target: Unit boot.mount is masked."
<Aleksejs>
I haven't changed a lot, only added one more virtual host (I have plenty of them)
<rycee>
wolfman154: You do not need to enable xmonad in the system configuration. It's pretty much enough to have `services.xserver.enable = true` in the system configuration.
wfranzini has joined #nixos
oida has joined #nixos
<wolfman154>
rycee: no there is no .xsession just .session-errors?
<rycee>
wolfman154: Do you have `xsession.enable = true` in your configuration?
<rycee>
In your Home Manager configuration, that is.
<wolfman154>
rycee: let me check, I just rebooted
drakonis_ has joined #nixos
<{^_^}>
[nixpkgs] @hedning pushed 175 commits to gnome-3.32: https://git.io/fjI0i
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<wolfman154>
rycee: don’t think so only xsession.windowManager.xmonad.enable = true;
jmgai has joined #nixos
<wolfman154>
rycee: so I need to add this to fix it 13:37 rycee: wolfman154: Do you have `xsession.enable = true` in your configuration?
<pingiun>
I mistakingly setup mutableUsers to false, but did not configure a password, is there any way to restore from this?
<rycee>
wolfman154: Yes, it must be added. And then relog.
<monokrome>
Does anyone know if there's a way to mount an "overlay" of an unencrypted device/mount so that files read/written from it are encrypted/decrypted on the fly? I am pretty sure there's a driver for this but forgot what it's called
<monokrome>
I'm experimenting w using IPFS to replace dropbox
<wolfman154>
rycee: I haven’t fixed hidpi so the words are microscopic, I can barely see, give me sec and I’ll let you know if it fixed it
<jmgai>
hi, I am trying to package multicorn, a postgres extension, i.e. a shared library, that embeds python. I can build the package and load it as a plugin into the postgres service of nixos. But when the shared library tries to load its python modules it fail, because it cannot find the modules. I think this should be fixable by setting PYTHONPATH, but how do I do that, since this should be done in the
<jmgai>
postgresql-and-plugins package of the service, which I would like to avoid touching.
<jmgai>
monokrome: do you mean encfs?
zupo has joined #nixos
<manveru>
jmgai: can't you add it to the env of the service?
<jmgai>
manveru: yes, but how. The service has an option to add plugins. I would like to give it the PYTHONPATH as soon as multicorn is added to the plugins.
<monokrome>
I wonder if I acn just store a luks device in there
<monokrome>
I think files are updated impartially complete each time in IPFS, so probably won't work O_o
<monokrome>
s/complete//
<manveru>
hmm
<jmgai>
I could add it via a systemd unit extensions, could I?
<manveru>
jmgai: a package can't change the config of a service, unfortunately
<jmgai>
fendor_ has quit [Ping timeout: 250 seconds]
<manveru>
adding to extraPlugins simply adds it to the paths
<manveru>
i thought there was a refactor of postgres at some point
<wolfman154>
rycee: it rebuilt but when I signed in it brought me to an empty gnome screen, how strange, I already removed gnome?
<{^_^}>
#38698 (by thoughtpolice, 51 weeks ago, open): Overhaul PostgreSQL packaging, native extensions, and NixOS support
<manveru>
jmgai: so, no, if you really want to add this to nixpkgs and work out of the box, you'd have to add an option to the postgres module i thinnk
<rycee>
wolfman154: Do you have a ~/.xsession file now?
<jmgai>
/byye
jmgai has quit [Quit: leaving]
jmgai has joined #nixos
<jmgai>
manveru: thanks for clarifying that.
<manveru>
i'm still hoping the refactor will make it, but probably not for 19.03 :(
<wolfman154>
rycee: after I sign in it gives me a blank screen and the mouse but it doesn’t let me do anything even if I switch to an older configuration?
<jmgai>
that refactoring looks interesting, I' still reading it
i1nfusion has quit [Remote host closed the connection]
<manveru>
yeah, it's pretty cool :D
<manveru>
extension handling is a mess atm
<wolfman154>
rycee: I don’t want to have to install nixos from scratch again but nothing is working
<rycee>
wolfman154: Did you check whether you have an ~/.xsession file?
<jmgai>
to be honest, the task isn't exactly trivial either
<jmgai>
because that multicorn is just a precondition for the actual postgres extension I am interested in
<wolfman154>
rycee: I can’t nothing works when I sign in?
<mabel>
I turned on the verbosity options on cmake and VERBOSE=1 in makeFlags, but to no avail... it still doesn't show what exact command failed and why
<mabel>
the last line before was "Leaving directory '/build/source/build'
<mabel>
and then above that, cmake -E cmake_simlink_library
<mabel>
*symlink
<clever>
mabel: can you pastebin the entire log?
<wolfman154>
rycee: so how do I turn on the graphical session back on?
<mabel>
hm... how do I snarf the log?
<mabel>
ah, 2>
<rycee>
wolfman154: Ok so you are in xmonad and it works for creating new terminals and such?
<clever>
make[2]: *** Waiting for unfinished jobs....
<mabel>
ohh......
<clever>
mabel: this is the main reason you had trouble finding it
<mabel>
I wonder... in the future, should I run nix-build inside of a neovim buffer or something for easy hunting?
<mabel>
or is there a way to grab the build log after the fact?
<clever>
mabel: simplest thing is to enableParallelBuilding = false;
<clever>
then it will stop immediately after failure
<mabel>
ahh
<clever>
`make -j8` will wait for the other 7 jobs to finish upon failure of one
<wolfman154>
rycee: now when I sign into xmonad, it shows the wallpaper of gnome (strange) and when I press alt shift enter to open terminal nothing happens?
<clever>
and if make is recursively calling make on more dirs, those 7 jobs could be very spammy
wfranzini has quit [Remote host closed the connection]
<wolfman154>
rycee: usually when I first sign into xmonad there is a blank screen, from there I create a terminal window?
wfranzini has joined #nixos
<{^_^}>
[nixpkgs] @malob opened pull request #58846 → vimPlugins: add plugins deoplete-fish, NeoSolarized, vim-choosewin, and vim-fish → https://git.io/fjIEX
<wolfman154>
rycee: super key is supposed to be swapped with alt but it still does not work?
<wolfman154>
rycee: frustration
<rycee>
wolfman154: I think you need to paste all your configurations somewhere. Otherwise it will be impossible to say what the problem is.
<wolfman154>
rycee: I can’t do that with the graphics disable, let me re enable gnome and then I’ll paste bin every thing
<mabel>
clever: that clue told me I needed to disable building tests, which totally fixed it! thank you!
<wolfman154>
My system crashed, I am going to have to reinstall, do you think you could help me tomorrow I could paste bin all the my configuration files?
sgronblo has quit [Ping timeout: 246 seconds]
<random_user>
Hey all! Raspberry Pi 3 Model B+ anyone? Can I get the boot menu with generations using uboot (like in grub or systemd-boot)? Why is the output on booting systemd distorted (heavily distorted, unreadable)? And why can't I run xorg (looks like a driver error)?
<samueldr>
hi! yeah, u-boot should provide the menu to select a boot entry
<wolfman154>
rycee: you made this,so you should be able to spot my errors
<samueldr>
though I haven't tested with a 3B+, with a 3B it will work
<samueldr>
random_user: except that with some screens u-boot's screen init might not work right and might require tweaks to config.txt to set the right type of HDMI (e.g. with DVI support) and resolution
bgamari_ has quit [Ping timeout: 246 seconds]
<rycee>
wolfman154: Yeah, I think it will be easier with the configurations.
<gchristensen>
wolfman154: why do you need to reinstall?
<samueldr>
random_user: as for the initial boot messages, might be related to that, maybe u-boot doesn't init the display right then the kernel continue with bad assumptions
<samueldr>
random_user: as far as xorg, not sure, I did run it with a 3B, but had to use linuxPackages_latest, IIRC, and there might have been fudging with vc4 and modesetting, but that was I think over a year ago
<wolfman154>
gchristensen: well I think I can fix nixos through tyy
<gchristensen>
what makes you think you need to reinstall?
<random_user>
samueldr: oh, thanks! I'm not very good with dev boards, but I followed the instructions on the wiki.
<gchristensen>
you've said in the past that you've had to reinstall several times. almost nobody needs to do this, so I suspect maybe you don't either
kvda has joined #nixos
<samueldr>
random_user: are you plugging to your screen via an HDMI to DVI solution? if so it might be something you need to tweak within config.txt
<wolfman154>
gchristensen: yeah you’re right, just a bad habit from ubuntu
kvda has quit [Client Quit]
<gchristensen>
what makes you think you need to reinstall?
<gchristensen>
like, what has taken place where you think you should
<random_user>
samueldr: I'm not sure, but I'll check it out!
<mabel>
is there a way I can detect the QT major version from being passed a qt derivation?
<random_user>
samueldr: thanks a lot! :)
ddellacosta has quit [Ping timeout: 268 seconds]
userp has joined #nixos
fenedor has quit [Ping timeout: 246 seconds]
<mabel>
there's a few packages that take in a qt derivation, but cmake has a setting that depends on the major version of the package
<wolfman154>
gchristensen: I’m brain dead I cannot even think right now
<sochan>
but the thing is, that doesn't work for `nix build`, only for `nix-shell`
<sochan>
because we are not trying to build our software with nix, we are just trying to prepare a build enviornment for it
<sochan>
so my question is: what is the proper way to define a build envioronement for a project without using mkDerivation which makes nix think that our project is a package we are trying to build
<sochan>
when we are not trying to build it
nD5Xjz has quit [Ping timeout: 250 seconds]
<Ralith>
sochan: what I do is I have a `package.nix` which uses nixpkgs conventions and is referenced by a `shell.nix`, which uses overrideAttrs to set `src = null;`, and a `default.nix` which defines a package set using pkgs.lib.makeScope, imports the proper nixpkgs revision, references any other custom packages you need, etc.
<sochan>
oh, that sounds like what we need
<sochan>
do you maybe have an example of that somewhere on github?
<sochan>
i'm not that well versed in nix, but maybe I can figure it out from your comment
<sochan>
or look for some examples of use of plg.lib.makeScope as you mentioned in some other projects on github
<Ralith>
yeah, or grep around in nixpkgs itself for uses of it
<sochan>
so you don't use mkDerivation at all in your default.nix?
i1nfusion has quit [Remote host closed the connection]
<sochan>
you use the makeScope method to pull in the dependencies you need
<sochan>
and you set src to null to avoid it being built? or somthing like that?
nD5Xjz_ has joined #nixos
nD5Xjz has quit [Ping timeout: 246 seconds]
<Ralith>
to avoid copying the source into the nix store, which is very slow when you have a build dir floating around in there for a large project and wastes space
<Ralith>
mkDerivation is used in the package.nix
<Ralith>
I misspoke slightly, also; the shell.nix should reference the default.nix which referencs the package.nix
<sochan>
hmmm, how is shell.nix referenced by default.nix? or are they separate?
<Arahael>
over7head: you are also missing puredarwin
<simpson>
You'd let NixOS manage GRUB, and use `boot.loader.grub.useOSProber = true;` I think.
<Arahael>
over7head: and reactos, also openbsd, netbsd...
<over7head>
thats why i dont want to allow NixOS to manage grub...i must staz with this one... it is configured to boot for example BSD custom keren and so..
<simpson>
There's probably a way to do it, but I don't know it.
<Arahael>
simply dont tell nix about how to set up grub.
<simpson>
Arahael: But then I'm not sure how one would get NixOS's boot entries into GRUB.
<Arahael>
dont give it a grub device, etc. and back up that grub for good measure anyway.
<Arahael>
simpson: excellent point.
Leo` has joined #nixos
<over7head>
Arahael, i didnt know theres puredarwin...
<over7head>
would give it a try
Thra11 has quit [Ping timeout: 250 seconds]
<Arahael>
over7head: and reactos. better than haiku in some ways. but haiku's networking is far better.
<over7head>
alos, i have plan for openbsd too...its different than freebsd
laalf has quit [Remote host closed the connection]
<over7head>
im some kind of OS hobist haha
<Arahael>
clearly. also freedos!
<Arahael>
your system probably already has minux (running cpu microcode)
laalf has joined #nixos
<Arahael>
there is seriously little reason to avoid virtualisation today.
<glenndavy>
hi there, wondering what i should use as my NIX_PATH to handle “error: file 'nixpkgs' was not found in the Nix search path (add it using $NIX_PATH or -I),” ?
<{^_^}>
[nixpkgs] @tadeokondrak opened pull request #58859 → vapoursynth: split python module out → https://git.io/fjIgL
zupo has joined #nixos
zupo has quit [Client Quit]
<glenndavy>
apparently NIX_PATH points at a directory of expressions, but afaik i have none out side the store