<jophish>
how can I stop the cmake hook being run?
cpennington has quit [(Remote host closed the connection)]
frankpf has quit [(Ping timeout: 252 seconds)]
<jophish>
ah dontUseCmakeConfigure
eacameron has joined #nixos
radvendii has joined #nixos
eacameron has quit [(Ping timeout: 240 seconds)]
nekroze has joined #nixos
<jophish>
oh shit, I compiled the wrong arch for ncurses into my cross compiled ghc
<nekroze>
How can I get go2nix to use vendored deps?
<NixOS_GitHub>
[nixpkgs] copumpkin closed pull request #24964: make-disk-image: change to be less VM-centric (master...make-disk-image-refactor) https://git.io/vSdmF
radvendii has quit [(Client Quit)]
eacameron has joined #nixos
vaibhavsagar_ has joined #nixos
vaibhavsagar has quit [(Remote host closed the connection)]
eacameron has quit [(Ping timeout: 252 seconds)]
eacameron has joined #nixos
<NixOS_GitHub>
[nixpkgs] copumpkin opened pull request #25165: google-compute-image: switch to use the common make-disk-image.nix (master...google-image-common) https://git.io/v9kKg
eacamero_ has joined #nixos
eacameron has quit [(Ping timeout: 268 seconds)]
griff_ has quit [(Quit: griff_)]
griff_ has joined #nixos
griff_ has quit [(Client Quit)]
griff_ has joined #nixos
griff_ has quit [(Client Quit)]
eacamero_ has quit [(Ping timeout: 260 seconds)]
griff_ has joined #nixos
griff_ has quit [(Client Quit)]
systemfault has joined #nixos
hexagoxel has quit [(Ping timeout: 260 seconds)]
griff_ has joined #nixos
griff_ has quit [(Client Quit)]
griff_ has joined #nixos
griff_ has quit [(Client Quit)]
griff_ has joined #nixos
griff_ has quit [(Client Quit)]
griff_ has joined #nixos
eacameron has joined #nixos
griff_ has quit [(Client Quit)]
griff_ has joined #nixos
griff_ has quit [(Client Quit)]
griff_ has joined #nixos
griff_ has quit [(Client Quit)]
griff_ has joined #nixos
griff_ has quit [(Client Quit)]
<nekroze>
\q
nekroze has quit [(Quit: leaving)]
griff_ has joined #nixos
hexagoxel has joined #nixos
griff_ has quit [(Client Quit)]
griff_ has joined #nixos
griff_ has quit [(Client Quit)]
dbmikus_ has joined #nixos
eacameron has quit [(Ping timeout: 260 seconds)]
griff_ has joined #nixos
griff_ has quit [(Client Quit)]
griff_ has joined #nixos
griff_ has quit [(Client Quit)]
griff_ has joined #nixos
griff_ has quit [(Client Quit)]
eacameron has joined #nixos
eacameron has quit [(Ping timeout: 240 seconds)]
Tipping_Fedora has joined #nixos
eacameron has joined #nixos
Wizek has quit [(Quit: Leaving)]
dbmikus_ has quit [(Quit: WeeChat 1.7)]
dbmikus_ has joined #nixos
dbmikus_ has quit [(Client Quit)]
ryanartecona has joined #nixos
eacameron has quit [(Ping timeout: 252 seconds)]
hamishmack has quit [(Ping timeout: 258 seconds)]
<NixOS_GitHub>
[nixpkgs] copumpkin opened pull request #25166: brightbox-image: switch to use the common make-disk-image.nix (master...brightbox-image-comon) https://git.io/v9k6P
jsgrant__ has quit [(Quit: Peace Peeps. o/ If you need me asap, message me at msg[at]jsgrant.io & I'll try to get back to you within 24 hours.)]
schoppenhauer has quit [(Ping timeout: 268 seconds)]
schoppenhauer has joined #nixos
hamishmack has joined #nixos
<Tipping_Fedora>
Hi guys, how would I replace wpa_sullpicant with network manager at install?
eacameron has joined #nixos
justan0theruser has joined #nixos
justanotheruser has quit [(Ping timeout: 240 seconds)]
Piece_Maker has joined #nixos
Acou_Bass has quit [(Ping timeout: 240 seconds)]
Piece_Maker is now known as Acou_Bass
alx741 has quit [(Quit: alx741)]
endformationage has quit [(Quit: WeeChat 1.7)]
markus1219 has quit [(Remote host closed the connection)]
markus1189 has joined #nixos
markus1209 has quit [(Remote host closed the connection)]
<joko>
The only workaround I could propose is to copy manually the built closures to the Hydra server
justbeingglad has joined #nixos
marsel has quit [(Ping timeout: 258 seconds)]
<joko>
Hopefully more people will start having this problem and it will be noticed
<lukec>
joko: Thanks for the link. I guess I'll dive into the source when I get a moment.
davidak has quit [(Quit: Leaving.)]
freusque has quit [(Ping timeout: 255 seconds)]
justbeingglad has left #nixos []
<NixOS_GitHub>
[nixpkgs] LnL7 pushed 34 new commits to master: https://git.io/v9kHd
<NixOS_GitHub>
nixpkgs/master 8d26be2 Matthew Bauer: antiword: supports darwin
<NixOS_GitHub>
nixpkgs/master 32d95b1 Matthew Bauer: dvdauthor: supports darwin
<NixOS_GitHub>
nixpkgs/master e75fde6 Matthew Bauer: dconf: supports darwin
jmsb has joined #nixos
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
MercurialAlchemi has quit [(Ping timeout: 260 seconds)]
eacameron has quit [(Remote host closed the connection)]
dmi3y has joined #nixos
tommyangelo has joined #nixos
MercurialAlchemi has joined #nixos
dmi3y has quit [(Quit: dmi3y)]
dmi3y has joined #nixos
<lukec>
joko: I have gotten past my problem (by setting max_output_size which I didn't know about)
nslqqq has quit [(Ping timeout: 240 seconds)]
<lukec>
joko: I do notice in your issue that the number you set it to is one larger than what will fit in 32 bits.
<lukec>
joko: try setting it to something smaller.
justbeingglad has joined #nixos
freusque has joined #nixos
Sonarpulse has quit [(Ping timeout: 258 seconds)]
justbeingglad has left #nixos []
moet has quit [(Ping timeout: 240 seconds)]
jmsb has quit [(Remote host closed the connection)]
nslqqq has joined #nixos
Itkovian has joined #nixos
blahdodo has quit [(Remote host closed the connection)]
justelex_ has quit [(Excess Flood)]
justelex_ has joined #nixos
blahdodo has joined #nixos
marsel has joined #nixos
kampfschlaefer has joined #nixos
freusque has quit [(Ping timeout: 268 seconds)]
lukego has joined #nixos
mounty has quit [(Ping timeout: 252 seconds)]
mounty has joined #nixos
ambro718 has joined #nixos
<lukego>
Hoi! I am looking for a way to copy a closure from the nix store into a file, to be transferred to another machine and imported. I have a store path, and this leads to a directory that includes symlinks to other store paths that would need to be included. How? :-)
<lukego>
spacefrogg: I want to copy to a file, as with tar
<lukego>
spacefrogg: manpage for '--export' says that it does not copy a closure
<spacefrogg>
lukego: Read on.
nicknovi1 has joined #nixos
<spacefrogg>
lukego: It gives you an example of doing exactly what you want to do.
nicknovitski has quit [(Ping timeout: 260 seconds)]
<lukego>
spacefrogg: but the example seems to imply that 'nix-store -qR $path' will print the store paths for everything in the closure, right? and that's not what I am seeing in my gist
<spacefrogg>
lukego: Well, that is what -qR is supposed to do. What makes you think otherwise?
<lukego>
spacefrogg: Have you seen my gist? is that the expected behavior? I run 'nix-store -qR $path' on a path that contains symlinks to 10 other paths, and none of those other paths are included in the output. I expect to see all of them, as well as their own dependencies.
<spacefrogg>
lukego: How does your root path come to be?
<lukego>
nix-store --import
<lukego>
--add, rather
<lukego>
I suppose that by constructing the symlinks directly and adding them with 'nix-store --add', nix can't track the dependencies and doesn't know about them?
<spacefrogg>
lukego: You seem to suppose right.
<lukego>
OK. Thanks. I had imagined the GC was "scavenging" the contents to discover the links automatically, like a conservative garbage collector.
nschoe has joined #nixos
alex-v` has joined #nixos
<spacefrogg>
lukego: GC and Dependencies are treated on different levels.
DutchWolfie has joined #nixos
DutchWolfie has quit [(Changing host)]
DutchWolfie has joined #nixos
<spacefrogg>
lukego: dependencies are stored in a database.
<joko>
lukec: thanks for mentioning this, but it fails still :(
<spacefrogg>
lukego: If you nix-store --add'ed your content, why don't you do this on the other system as well?
eacameron has joined #nixos
<lukego>
Is there any easy way to "fix" this? e.g. to scavenge one store path to find its references to the store, and then construct a new store path with the same contents but declaring those dependencies? (otherwise I need to rewrite this code that imports data into the store.)
<spacefrogg>
You have to rewrite the code.
<lukego>
I'm using the nix store as a database of sorts. I import a bunch of data, do a bunch of transformations on it (with a shell script that calls nix-store/nix-build, similar to styx), and the scenario now is that I want to extract an intermediate transformation step to another machine to inspect and debug it
<spacefrogg>
In the nixpkgs manual you will find pointer to declare explicit dependencies on other store paths
alex-v has quit [(Ping timeout: 268 seconds)]
hyphon81 has quit [(Ping timeout: 252 seconds)]
<spacefrogg>
It resides in a nix-support directory.
<lukego>
I suppose the "proper" way would be to construct my store paths using nix-build instead of 'nix-store --add' so that the dependencies are tracked by nix from the beginning
dmi3y has quit [(Quit: dmi3y)]
mounty has quit [(Read error: Connection reset by peer)]
<spacefrogg>
lukego: That sounds better, yes. Have to leave, sorry.
<lukego>
spacefrogg: Thanks for your help!
<spacefrogg>
yw
<lukego>
Seems like I have done this wrong and have to start over :-|.
eacameron has quit [(Ping timeout: 260 seconds)]
lukec is now known as themk
MichaelRaskin has quit [(Quit: MichaelRaskin)]
<lukego>
New question :-). Is there some "gist" like way to host sources for nix expressions that are potentially large (e.g. virtual machine image)? The problem I have is that I write nix expressions to process data, potentially many gigabytes, and I want to be able to run this code on any machine -- not just the one where I already have the data in local files
derjohn_mob has quit [(Ping timeout: 258 seconds)]
mounty has joined #nixos
<lukego>
I suppose that I could import data into the nix store, use the store path as a unique ID (quasi-url), and move it around with nix-copy-closure. but this seems to imply that all relevant machines have ssh access with read/write to a common server, which may be a barrier
spacefrogg has quit [(Ping timeout: 255 seconds)]
xadi has joined #nixos
thc202 has joined #nixos
<lukego>
I was thinking that e.g. an ipfs url could be more practical than a nix store path, but I have never worked with ipfs so maybe that is just opening a new can of worms
mounty has quit [(Ping timeout: 260 seconds)]
simukis_ has joined #nixos
mounty has joined #nixos
ambro718 has quit [(Ping timeout: 240 seconds)]
goibhniu has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 1 new commit to master: https://git.io/v9kN4
<NixOS_GitHub>
[nixpkgs] peterhoeg pushed 1 new commit to master: https://git.io/v9kNV
<NixOS_GitHub>
nixpkgs/master 229e338 Peter Hoeg: mpv: 0.24.0 -> 0.25.0
<domenkozar>
lukego: iirc for runtime dependencies it scans contents of files in store path, but not symlinks
newhoggy_ has joined #nixos
<domenkozar>
which is arguably wrong
<domenkozar>
niksnut: is there a good reason Nix doesn't consider symlinks for runtime closure?
<lukego>
Just now it seems like store paths are not so useful as names if they are going to bypass the dependency tracking mechanisms. So would need to always refer to objects using the derivations that build them. So have to think of a way to make that concise like a store path.
<domenkozar>
lukego: you can just write them to a file in store path
newhoggy has quit [(Ping timeout: 260 seconds)]
<domenkozar>
echo /nix/store/path >> $out/deps
dmi3y has joined #nixos
<lukego>
domenkozar: is this just an isolated restriction though, or am I "fighting the system"?
newhoggy has joined #nixos
<domenkozar>
lukego: it's currently the only way to declare runtime dependencies
<domenkozar>
many parts of our infrastructure do it
python476 has joined #nixos
<domenkozar>
in fact, if you specify propagatedBuildInputs, what it does under the hood is, write those build inputs to $out/nix-support/propagated-build-inputs
<lukego>
domenkozar: thanks. So if this is basically just working around an obscure bug in nix, not treating symlinks as content, that it sounds okay.
newhoggy_ has quit [(Ping timeout: 245 seconds)]
mounty has quit [(Remote host closed the connection)]
<themk>
lukego: just to clarify (because I think there may be some miscommunication happening here). You are not creating this directory of symlinks with a nix expression?
<lukego>
themk: right. I'm creating it with mkdir/ln and then importing it with nix-store --add
newhoggy has quit [(Ping timeout: 245 seconds)]
dmi3y_ has joined #nixos
mounty has joined #nixos
<lukego>
This is perhaps analogous to doing pointer arithmetic in a garbage collected language. No worries if you have a "conservative" GC, but totally hosed if you have "precise" GC
dmi3y has quit [(Ping timeout: 255 seconds)]
<themk>
So, I'm pretty sure that runtime dependencies are discovered by looking at the build inputs, and scanning all the things for those hashes.
<lukego>
I've been assuming that nix is conservative having seen references to discovering "runtime" dependencies by scavenging file contents
<themk>
But by doing what you are doing you are bypassing all that.
<lukego>
Hm. Yes, and the trick of writing the store paths into a file is not helping
<lukego>
themk: The behavior I expected was that the runtime dependencies would be every store path found in the contents, but it sounds like you are saying it will be the _subset of the declared build inputs_ found in the contents
freusque has joined #nixos
<themk>
Yes, it has to be the build inputs.
<lukego>
and doesn't look like 'nix-store --add' lets you specify build inputs as store paths
<lukego>
So the message I am getting is that "pointer arithmetic is not okay" i.e. creating store paths manually and including them in builds
<the-kenny>
Anyone having issues with redshift in a recent NixOS master? It can't connect to my x server, supposedly because it can't access ~/.Xauthority
<domenkozar>
lukego: are those store paths added to nix store beforehand?
<lukego>
domenkozar: yep
mudri has joined #nixos
<domenkozar>
lukego: I'd say this qualifies for a bug report with a reproducible test case
<lukego>
themk: The problem I have now is that the ultimate input to my nix expression is messy, like a 1GB directory containing lots of subdirectories, and the first thing that I want to do is preprocess that to make a set of compressed tarballs of each interesting component. However, if I can't reference those tarballs via their nix stores afterwards, it seems like I need to always carry around the gigantic messy directory as the source of m
spacefrogg has joined #nixos
<themk>
lukego: could you use fetchurl?
Isorkin has quit [(Ping timeout: 258 seconds)]
dmi3y_ has quit [(Quit: dmi3y_)]
Isorkin has joined #nixos
nicknovi1 has quit [(Ping timeout: 240 seconds)]
mounty has quit [(Ping timeout: 258 seconds)]
<lukego>
themk: Maybe. Question is, how do I write a script that preprocesses messy input outputs URL(s) where it can be fetched?
<lukego>
domenkozar: but other people are saying it is the expected behavior...
<lukego>
domenkozar: and I suppose that I can understand that perspective. It's a bit weird if the name of a derivation becomes a dependency _only_ if it happens to not yet have been GCd from the store.
chakerbenhamed has joined #nixos
<lukego>
s/derivation/store path/
<domenkozar>
ah, yes, you have to add it to the gc root.
dmi3y has joined #nixos
<lukego>
I would really like to have a command 'file2url <filename>' that takes a while, hosts it _somewhere_, and prints a url where it can be accessed. Is there an easy way to do this now in 2017?
<lukego>
If the file were small then 'gist' would do exactly that, but the file can be large.
<domenkozar>
amazon s3 cli does that
<lukego>
Could be the answer
<lukego>
but, then I suppose it is uploaded to the account of the person running the program, so it implies that everybody needs to expose S3 objects. I'm not sure people want to do that -- I'd worry about a DDoS hitting my credit card.
<NixOS_GitHub>
[nixpkgs] aszlig closed pull request #14053: Handle submodules for type "either" (master...either-with-submodules) https://git.io/vaXLj
newhoggy has joined #nixos
<lukego>
IPFS seems more appealing, in that it could just run on e.g. a few hetzner servers, but that might be too much bother to setup
<themk>
lukego: where is your messy input coming from? Is that a tarball hosted somewhere already?
simp__ has joined #nixos
<simp__>
Hey. Is Nix compatible with the majority of linux software?
<simp__>
And is it stable enough to use as a main os?
thinkingeye has joined #nixos
<themk>
simp__: yes, and yes :) What software in particular are you concerned about?
<lukego>
themk: It's diagnostic information from an application. Basically coredumps. I want to be able to import a coredump into the store, and I want to be able to aggregate them together e.g. "collection of 100 coredumps from version A". Then I want to be able to use these objects as inputs to diagnostic tools (gdb-like)
newhoggy has quit [(Ping timeout: 240 seconds)]
<the-kenny>
Any ideas on how to find a "ghost" service? systemd keeps trying to start redshift (as a user service) even though I can't find the unit anywhere (neither the file nor via systemctl list-units)
<simp__>
themk: Are there any negatives about Nix? Or maybe not negatives but potential negatives, things one should know about before using it?
<goibhniu>
simp__: you should be aware that you can't run binaries (which haven't been packaged) on NixOS without some extra work
<simp__>
Ah, and how hard is it to do that?
<lukego>
themk: Suppose that I run a big application on my machine and it crashes producing 20 coredumps from related applications. If you want to help me diagnose the problem I wanna just give you a URL that you can paste into your debugging to see all the data. I'd like for you to get preprocessed/compressed reprsentations of the data, and skip any particular files that you already have
<goibhniu>
simp__: I believe you can set up a FHS like environment which you can use (this is how Steam works on NixOS) ... or you can package the binary
<simp__>
Any good articles I should read about Nix?
<goibhniu>
simp__: you may also want to check if the software you use is already in nixpkgs or not
<thinkingeye>
simp__: I did it successfully with some binaries from my university
<lukego>
themk: So the code I am debugging now is like "import --group test1 /run/coredumps-1/* --group test2 /run/coredumps-2/*" which should produce an ID/URL/... that references two groups of coredump files in such a way that a debugger can load them
<niksnut>
how was "studio-product-process-set" created? where did its derivation get those symlinked store paths from?
<niksnut>
for comparison: nix-store -qR /run/current-system/sw
<goibhniu>
simp__: also, you can use nix on top of another distro before switching to nixos
<lukego>
niksnut: I created those symlinks manually with basically 'ln -s $(nix-store --add path) foo'
<domenkozar>
it would still adhere to restriction?
<niksnut>
since this is nix-store --add, not a derivation
<domenkozar>
niksnut: so runtime dependencies need to be part of build-time deps?
<niksnut>
yes
<goibhniu>
makefu: nice!
dmi3y has quit [(Quit: dmi3y)]
<simp__>
I don't get it exactly though. I'm looking at i3-gaps' documentation and on Nix they simply say to add a few lines to configuration.nix. Is there no installation? Is there some software checking that file and taking the necessary steps to apply the confiuration?
<niksnut>
in a derivation, you can use "builtins.storePath /nix/store/18br0r75ccg4ckbzlvzzm60s2fg5wxgf-studio-product" though
<niksnut>
so you can have a little derivation that creates symlinks to a list of builtins.storePath calls
thinkingeye has quit [(Quit: Konversation terminated!)]
<domenkozar>
lukego: so tldr: you'll have to use nix to generate a runtime closure
<themk>
niksnut: Nice, I didn't know about builtins.storePath. lukego does that solve your problem then?
<niksnut>
but stuff added via nix-store --add will never have references
<goibhniu>
simp__: you use the command `nixos-rebuild` to evaluate the configuration file, create a new system generation and switch to it
<lukego>
So the notion I am getting is that a script for importing data into the nix store should probably do this: 1) Preprocess the source data. 2) Generate a nix expression that references the sources e.g. by filesystem paths. 3) Run nix-build.
<lukego>
The basic problem I have is that I am doing this piecemeal, breaking the input into parts and adding each one with 'nix-store --add', but this makes them hard to reference
<themk>
You can have multiple outputs in nix.
<simp__>
That sounds really interesting, goibhniu. I'm definitely giving nix a try.
<lukego>
Or I suppose that I could take more care to link things together in a good way, but I am probably fighting the system, and prone to e.g. race conditions if there is a nix store garbage collection during import that deletes from of the components before I have a chance to reference them
<simp__>
I assume the big software all works fine in Nix without that packaging business, right? (Vim, Chromium or some equivalente browser, compilers and interpreters of popular languages?=
<LnL>
niksnut: did you see the resolve-system-dependencies.pl segfaults on hydra, could that be related to the perl upgrade on staging or is it a hydra issue?
<lukego>
themk: builtins.storePath probably solves the immediate problem but I suspect it just kicks the can down the road.
<goibhniu>
simp__: yep
<lukego>
themk: Since my store paths are probably not reachable from GC roots between the time that I create them and the time that I reference them
<LnL>
ah, maybe some of the new machines don't have
<LnL>
the pre-build-hook = in their nix.conf?
<simp__>
Also, goibhniu. Are there any issues with this kind of packaging or similar processes? Like, bugs, performance issues, etc.. (think linux ports of windows software)
ericsagnes has quit [(Ping timeout: 240 seconds)]
<lukego>
(random question: does Hydra serve up the nix store? Can you e.g. construct a URL that will lead to a closure for a given store path?)
spacefrogg has quit [(Quit: WeeChat 1.7)]
<lukego>
Just trying to think about whether I need something like S3 for hosting data or whether a nix store would do the job more directly.
<LnL>
niksnut: looks like it might be mac2, I think that one had been offline for a while
<domenkozar>
lukego: you can serve it with http or ssh with a simple daemon
<domenkozar>
depending on your use case
<lukego>
Could perhaps use nix-copy-closure instead with some pseudo-identity e.g. a disclosed public key that only allows read-only access to the store
<themk>
lukego: checkout nix-serve
<lukego>
domenkozar: I suppose the use case is basically 'nix-copy-closure' but for unauthenticated clients
<goibhniu>
simp__: I guess the time it takes to get security patches available as binaries is an open issue
<lukego>
These store-serving programs look good. Seems like it is reasonable to have a "well known server" that is serving up its nix store, and then to reference files their via their store paths
<themk>
lukego: Just make sure you add them as gc-roots
<lukego>
Though... Have to remember that referencing things by their store paths directly is what got me into a mess just now.
<NixOS_GitHub>
[nix] edolstra closed pull request #1347: Fix small grammar issue about page (master...sm-grammar-fix) https://git.io/vSpgP
<lukego>
themk: Yeah. Still... I am wondering whether it is a better design to do a series of operations on the nix store (e.g. --add, etc) from a shell script, or instead to generate a nix expression that will do everything from within the nix domain
<themk>
Do a nix expression.
yee has joined #nixos
<themk>
Feed the output of nix expressions to other nix expressions
<lukego>
I suppose that if I write builtins.storePath "/nix/store/..." then that could be downloaded from a binary cache?
<themk>
the intermediate results will be cached.
<themk>
If you are serving your nix store, then send the pipeline of expressions to someone else, they will download your itermediate result rather than recalculating everything.
<simp__>
Is NixOS supposed to be used as a main OS atm? In the presentations of people talking about it they're usually using mac
<themk>
I use it as my main OS
<clever>
i have nixos on my laptop, desktop, router, nas, 2 netbooks, and 2 servers
<lukego>
themk: Thing is that sending a pipeline of expressions is not as catchy as sending a URL or a store path
<simp__>
For how long have you been using, themk and clever? And do you see yourself continuing to use it?
<lukego>
themk: but yeah I feel like I need to design everything with nix expressions, then make the shell interface a very simple bit of syntactic sugar.
<clever>
simp__: about 2 years, and i dont see myself ever using another distro
<themk>
simp__: nearly 3 years.
<simp__>
Damn.
* goibhniu
has been using it for ~7 years and can't imagine using a distro without such features
<lukego>
simp__: I'm running nixos on servers and I wouldn't consider going back to ubuntu, fwiw.
<simp__>
How does it compare to something like Arch Linux?
<makefu>
lukego: i feel the same way. everything else just seems backwards in regards to configuration management and package consistency
<simp__>
I liked the simplicity and elegance of Arch, am looking to a similar thing.
<simp__>
for*
<themk>
simp__: in some ways NixOS is simpler than Arch.
<themk>
I used Arch before NixOS as my main.
<simp__>
Should I just install it right away or maybe play around with it in a virtual machine?
<simp__>
Also, it's nice to see that the community here is active
<goibhniu>
install it, you're going to love it :D
<simp__>
Some distros are just so dead
<themk>
simp__: It's quite different to your regular Linux distro.
<simp__>
I think I'll install it right away
<themk>
And as such, many resources on the web aren't immediately useful to NixOS users.
<simp__>
Any hardware concerns I should have?
<themk>
So, it can be a bit daunting if something goes wrong.
<simp__>
Is there some NixOS wiki?
<themk>
But you can always rollback on NixOS :)
<simp__>
Yes, that sounds so useful
<themk>
simp__: NixOS wiki is old and outdated unfortunately. Use the NixOS manual.
<simp__>
Any particular reason as to why they don't have a wiki?
lukego has quit [(Quit: Page closed)]
<clever>
simp__: spammers defacing everything
<simp__>
Yeah, well there must be some way of controlling for that
goibhniu has quit [(Ping timeout: 245 seconds)]
Lisanna_ has quit [(Quit: Lisanna_)]
goibhniu has joined #nixos
Isorkin has quit [(Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org)]
helpme has joined #nixos
helpme is now known as Guest64534
gilligan_ has joined #nixos
reinzelmann has quit [(Ping timeout: 260 seconds)]
<manveru>
simp__: there's the github wiki
spacefrogg has joined #nixos
whenov has quit [(Quit: Leaving.)]
ravloony has joined #nixos
tommyangelo has quit [(Ping timeout: 240 seconds)]
FRidh has joined #nixos
derjohn_mob has joined #nixos
<simp__>
Is setting up dual booting with windows straightforward?
obadz has quit [(Ping timeout: 260 seconds)]
obadz has joined #nixos
<Guest64534>
hey guys, maybe someone have time to help me with building portfolio performance (https://github.com/buchen/portfolio). I get the following error if i run nix-build "cannot auto-call a function that has an argument withoout a defaul value ('stdenv')". The default.nix looks like this http://textuploader.com/drh5w
<goibhniu>
simp__: I think it's just a matter of configuring the chain loader in grub
<NixOS_GitHub>
[nixpkgs] emanueleperuffo opened pull request #25170: gnomeExtensions.dash-to-dock: init at version v57 (master...dash-to-dock) https://git.io/v9ITM
vaibhavsagar_ has quit [(Ping timeout: 252 seconds)]
vaibhavsagar__ has quit [(Client Quit)]
vaibhavsagar has joined #nixos
<simp__>
Installing NixOS :)
<clever>
:D
<clever>
there is also a nixos manual and a rogue game on alt+f8 and alt+f9
<clever>
more easily accessed if your on the non-graphical iso
<simp__>
Well I downloaded the graphical image
<clever>
graphical is usualy better, gparted is just so much simpler then cli
<clever>
and you can use a proper browser to look things up
ravloony has quit [(Ping timeout: 260 seconds)]
freusque has quit [(Quit: WeeChat 1.7)]
newhoggy has joined #nixos
nslqqq has joined #nixos
<simp__>
Could UEFI be causing this? "An error occured in stage 1 of the boot process, which must mount the root filesystem on '/mnt-root' and then start tage 2."
<c0c0n3>
goibhniu: yip, I did that yesterday to no avail, but I'm going to try it again now on a fresh container just in case I did something wrong, will get back to you in 10, thx!
<goibhniu>
c0c0n3: I presume you can log into the container with `nixos-container root-login`, right?
<c0c0n3>
yes, thats right
<c0c0n3>
can do that no prob, running `ip addr` from inside the container gives:
<simp__>
clever: usb stick
<c0c0n3>
eth0@if4: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000 link/ether a6:b9:4c:04:6b:39 brd ff:ff:ff:ff:ff:ff link-netnsid 0 inet 10.233.1.2/32 scope global eth0 valid_lft forever preferred_lft forever inet6 fe80::a4b9:4cff:fe04:6b39/64 scope link valid_lft forever preferred_lft forever
<clever>
simp__: did you use dd or some other tool?
<simp__>
dd
<simp__>
I've managed to boot now
<clever>
simp__: ah
<goibhniu>
c0c0n3: is apache listening on that address?
<c0c0n3>
going to check now...
<simp__>
I chose nomodeset and it booted, clever. But I actually don't know what nomodeset is.
<c0c0n3>
goibhniu: yes, apache is up and running, and listening to the ip address (10.233.1.2) bound to the virtual eth. If I run curl http://10.233.1.2/ from inside the container, I get apache's index page
<clever>
simp__: that controls the video modes
<NixOS_GitHub>
[nix] edolstra pushed 1 new commit to master: https://git.io/v9Iq8
<NixOS_GitHub>
nix/master 9b63bb8 Eelco Dolstra: nix-shell -p: Use runCommandCC...
<goibhniu>
c0c0n3: I guess it's a firewall thing then
<simp__>
Any reason as to why one would work but not the other, clever?
<clever>
c0c0n3: ah yeah, and the container has its own firewall, you would have to allow port 80
<clever>
simp__: cant think of one
nslqqq has quit [(Ping timeout: 240 seconds)]
<c0c0n3>
clever: goibhniu: thx guys, going to check! didn't think of that cos it wasn't in the manual, so i assumed the example in the manual should have worked out of the box, but yah, a firewall issue makes perfect sense, thx!!!
vaibhavsagar has quit [(Ping timeout: 260 seconds)]
<goibhniu>
c0c0n3: that's confusing alright, if you have time, maybe you could open an issue for that?
newhoggy has quit [(Remote host closed the connection)]
xadi has joined #nixos
<NixOS_GitHub>
[nixpkgs] nagisa opened pull request #25173: fish: Do not overwrite the inherited environment (master...fish-env-overwrite) https://git.io/v9Im8
<c0c0n3>
goibhniu: yah, I'm still debugging it tho, disabling the firewall doesn't help w/ trying to access the port 80 from the host, packets are still dropped (using tcpdump to monitor) but now I can ping the container from the host, which I couldn't do earlier....
<clever>
c0c0n3: and keep in mind there are 2 firewalls, try iptables-save both inside and outside the container
<clever>
thats a easy way to just list the current fw config
nslqqq has joined #nixos
gilligan_ has quit [(Ping timeout: 255 seconds)]
c0c0n3 has quit [(Ping timeout: 260 seconds)]
<simp__>
"After having mounted your installation partition to /mnt, you must mount the boot partition to /mnt/boot." having trouble with the latter. "mount LABEL=BOOT /mnt/boot" just says mount point /mnt/boot does not exist
<c0c0n3>
there's still something funny w/ the network config but then again, it could be me doing something wrong
<lukego>
Just thinking about my "nix store as database of diagnostic info" design. I reckon I will split it into a shell frontend (dealing with messy inputs) and a nix backend (dealing with clean inputs). You can run 'frontend ./messydir' and it will grovel through the directory to put the inputs into canonical formats and run 'nix-build -E backend [ (coredump url1) (coredump url2) ... ]'
<c0c0n3>
if someone has the time i can explain
<lukego>
Then since everything has come into the nix universe from within 'nix-build' I can hopefully use the store paths to copy closures around, etc.
<avn>
domenkozar: something strange, I also got empty file ;/
<clever>
dcz: that will modify the stdenv to just use gcc5 by default
<NixOS_GitHub>
[nixpkgs] NeQuissimus pushed 1 new commit to release-17.03: https://git.io/v9IGz
<NixOS_GitHub>
nixpkgs/release-17.03 93870d9 Tim Steinbach: ammonite: 0.8.2 -> 0.8.3...
freusque has joined #nixos
<dcz>
clever: thank you! it worked
<simp__>
Seems adding the boot flag hasn't worked
<clever>
simp__: let me double-check this end
<simp__>
When booting from the disk I installed Nix on I get "This is not a bootable disk. Please insert a bootable floppy and press any key to try again ..."
eacameron has joined #nixos
<clever>
simp__: can you boot from the usb again?
<clever>
simp__: and then pastebin the output of running "efibootmgr"
<simp__>
Well I did, but it's still not showing up on uefi utility
<clever>
dcz: the gcc in nix is modified to never look in /usr/include/
<clever>
you must put a package with those headers into the buildInputs
ZoomZoomZoom has joined #nixos
<NixOS_GitHub>
[nixpkgs] nagisa closed pull request #25173: fish: Do not overwrite the inherited environment (master...fish-env-overwrite) https://git.io/v9Im8
<dcz>
what I mean is the package does not name its files correctly in comparison to the system one
<clever>
i think you also need to use #include <cimg/CImg.h>
<dcz>
clever: yes, I think that's what should not be required
<clever>
ah
<clever>
id say its a bug with the cimg package then, one min
<dcz>
(don't take m my word, I'm just trying to compile what someone else made)
<simp__>
clever: "EFI variables are not supported on this system"
<simp__>
Is the output
<clever>
simp__: you need to boot the usb via efi if you want to be able to install an efi os
<clever>
simp__: efi is weird like that
<simp__>
yeah but I did
<ZoomZoomZoom>
Hi everyone! Trying to write my first module. How can I easily convert a list of sets of two ints to a string to pass as a parameter, looking like "000-111,222-333,444-555" ?
<clever>
simp__: and you ran that as root?
<simp__>
efibootmgr? yeah
<clever>
simp__: can you pastebin the output of "mount" ?
<simp__>
Oh, I meant to say when I installed
<simp__>
Not right now, it's not running via efi
griff_ has joined #nixos
<clever>
ah
<clever>
you also need to boot via efi to even see the efi config
<simp__>
Because that requires me to go in the UEFI utility which isnt detecting the usb
<clever>
simp__: what filesystem did you use for / and how big is /boot?
<simp__>
ext4 and boot is 1GB I think
<simp__>
minimum 500mb
<clever>
simp__: it would be a bit of a waste of space, but you can change /boot into a bios boot partition (the bios_grub flag in gparted), and then switch to non-efi booting
<Guest64534>
are there some examples how packaged somthing that builds with maven?
<clever>
simp__: after you change the type, mount root to /mnt, /boot is no longer required, re-run nixos-generate-config --root /mnt ; nixos-install
<clever>
simp__: oh, and set boot.loader.grub.device = "/dev/sda"; (using the path for the hdd)
<simp__>
But before all that, do I need to boot via efi?
<clever>
dcz: ah, i see why i cant override cimg, the author didnt make the package very well
<clever>
simp__: i feel efi is optional, and i have never gotten it to work on any machine
<clever>
dcz: yeah, this cimg package is a bit of a mess
<simp__>
Alright, I'll try what you suggested.
<simp__>
Do you think maybe it's windows causing the issue?
<simp__>
I could wipe out the drive where windows is installed
<clever>
simp__: simpler/faster to just unplug that hdd
<simp__>
Is it possible that it's windows though?
<clever>
simp__: i did also have issues where the bios 100% ignored the boot order i set, and always booted the efi windows
<clever>
i had to disable efi entirely to make it stop
pi3r has joined #nixos
<dcz>
where can I find X11/xlib.h ?
<simp__>
I don't see any grub flags on gparted though
<clever>
simp__: it should be called bios_grub
<simp__>
There isn't anything like that
<clever>
simp__: can you screenshot the page your looking at?
<simp__>
is there any screenshotting program in this?
<clever>
simp__: what happens when you hit the printscreen key?
<simp__>
clever: I don't have the grub settings in my configuration file, only systemd-boot. Should I just comment those out and write the ones relating to grub?
<clever>
simp__: yeah, boot.loader.grub.enable = true; and the one i gave before
justbeingglad has joined #nixos
<clever>
systemd-boot only works on gpt
<clever>
grub works on both
<dcz>
can you when installing packages inside nix-shell, are thay actually installed anywhere?
<dcz>
*-can you
<clever>
dcz: when you open a nix-shell, it will always put everything in /nix/store/, and it will modify some env variables so tools can find it
<clever>
dcz: things with a bin/ folder go into $PATH, and everything goes into the gcc search path
<dcz>
oh, I see
<clever>
and since its only changing env variables, it wont persist after you exit the shell
<clever>
and you could use a conflicting version of something in a different terminal by running nix-shell on a different default.nix
justbeingglad has left #nixos []
<avn>
clever: it put auto-gc-root also ;)
<clever>
so you could have one terminal with gcc4, and another with gcc5
<clever>
avn: nix will scan /proc/*/env and a few others, and wont gc anything that can be in-use
<clever>
avn: so just putting it into the $PATH roots it
<dcz>
I was hoping to update dependencies without editing default.nix every time
<jophish>
woo, I've cross compiled ghc
<clever>
dcz: you can also point things to a local source directory
<clever>
jophish: its an aarch64 cpu right?, does it have backwards compat? which also has to be enabled in the kernel
<jophish>
clever: yeah, I can execute armv7 binaries on this machine
<jophish>
clever: It's worth noting that the interpreter for the ghc wrapper is using an x86_64 bash
<jophish>
(which fails)
<jophish>
so my build system is clearly not quite right
<clever>
jophish: ah, that will need to be fixed then
<jophish>
however trying to execute the file underneath does give that illegal instruction
<clever>
jophish: one trick, you can copy that shell script to $HOME, then just edit the right bash into the copy, and try running it
<clever>
ah
<clever>
jophish: one min
<jophish>
ghc made it tricky to cross compile
<jophish>
I'm compiling it from buildPackages (which doesn't fit into the standard nix idioms)
<jophish>
I'll rewrite my expression so that it compiles in pkgs
<clever>
root@raspberrypi:~# /nix/store/rbd67ylkmx2fpci7k4d4kaf6lnwrikyv-binutils-2.28/bin/objdump -d /nix/store/rbd67ylkmx2fpci7k4d4kaf6lnwrikyv-binutils-2.28/bin/objdump | less
<clever>
jophish: try running an arm objdump against the ghc with -d
<jophish>
ok
<clever>
root@raspberrypi:~# /nix/store/rbd67ylkmx2fpci7k4d4kaf6lnwrikyv-binutils-2.28/bin/readelf -A /nix/store/rbd67ylkmx2fpci7k4d4kaf6lnwrikyv-binutils-2.28/bin/readelf
<Dezgeg>
maybe 'readelf -a' too
<clever>
Tag_CPU_name: "7-A"
<clever>
and -A as well, which shows sub-arch level stuff (v6 vs v7)
<Dezgeg>
or gdb it, once it hits SIGILL do x/5i $pc to see what instruction it crashed on
<clever>
though if you have a v7 cpu, it should just work on v6 and v7
jophish_ has joined #nixos
c0c0n3 has quit [(Quit: Page closed)]
jophish_ has quit [(Client Quit)]
<jophish>
wow, objdump over ssh killed gnome-terminal
<jophish>
how is that even possible
<dcz>
clever: you mentioned you used nix-repl to look for packages, how to do it?
<clever>
dcz: nix-repl '<nixpkgs>' to load up nixpkgs
<clever>
dcz: then you can just tab-complete any package, or :b hello to build one
<clever>
jophish: i'm not seeing neon in my rpi2 builds, you may want to double-check that the cpu can support everything listed
<clever>
Tag_FP_arch: VFPv3-D16
<jophish>
hmm, I just compiled with armv7l multiplatform from nixpkgs
<clever>
i also have a different FP_arch
mudri has joined #nixos
<Dezgeg>
in the worst case the build system is looking in /proc/cpuinfo and adding gcc flags based on that....
<Dezgeg>
(I think gmp used to do that)
<lukego>
Is there an idiom for a shell script to loop over the elements of a nix list of derivations? like if 'srcs = [ d1 d2 d3 ]' and I want to write shell code in a builder like 'for s in ${srcs}; do ... end'
<clever>
lukego: if srcs= is on the derivation, nix will turn it into a $srcs env variable automaticaly, then you can use a normal bash for loop
<lukego>
clever: does that work with runCommand?
<clever>
lukego: or you could also concatMap in nix to unroll the loop
<clever>
jophish: 2017-04-24 09:26:18 < Dezgeg> or gdb it, once it hits SIGILL do x/5i $pc to see what instruction it crashed on
<Dezgeg>
run, then once it crashes, bt & x/20i $pc
<lukego>
clever: I think I understand why that doesn't work, but I am not sure what I should do instread.
<clever>
lukego: just $srcs, ${srcs} tries to reference a nix variable
<jophish>
ah, sorry!
<lukego>
clever: oh, thanks !
<lukego>
clever: I was thinking that the 'env' parameter is not working, but of course it is defining the shell script environment, not making nix variable bindings.
<jophish>
mcr 15, 0, r6, cr7, cr10, {5}
<clever>
lukego: and if you still want to protect the variable using the bash ${foo}, it needs escaping: "echo \${srcs}"
<Dezgeg>
now that's a fun one to google for :-D
<dcz>
clever: what's the formula to switch to gcc48?
<clever>
dcz: put gcc48 into the .override of stdenv, where gcc5 currently is
<Dezgeg>
is it coming from the binary you compiled and not from something like gmp?
<jophish>
Dezgeg: looks like it's haskell: │0x036fe598 in stg_putMVarzh$def ()
<Dezgeg>
okay
<jophish>
I'm going to correct my nix expression for ghc to make sure that bash is sourced from the correct package set
<clever>
Dezgeg: coprocessor 15 handles performance counters, and from your last link, i think it also does some atomic stuff
<Dezgeg>
cp15 does a million things
<clever>
heh
<clever>
with 16 of the, id expect it to be more spread out
<Dezgeg>
I suppose the other 15 are for silicon implementers to use for their proprietary stuff
JagaJaga has joined #nixos
justelex_ has quit [(Quit: Konversation terminated!)]
dcz has quit [(Ping timeout: 240 seconds)]
<simp__>
anyone here using i3 with nixos?
justelex_ has joined #nixos
<clever>
simp__: xorg crashing after you login?
justelex_ has quit [(Read error: Connection reset by peer)]
<simp__>
No, I just can't do anything
<clever>
ah
joepie91 has quit [(Quit: Konversation terminated!)]
<clever>
are you able to ssh in from another box?
<simp__>
mod + d or mod + <enter> or mod + 2 none do anything
<simp__>
but its not frozen because I can see the time below changing
<simp__>
No clever
joepie91___ has joined #nixos
joepie91___ has quit [(Changing host)]
joepie91___ has joined #nixos
<simp__>
Do rollbacks only rollback configuration or also software?
joepie91___ is now known as joepie91
<simp__>
installed packages I mean
<clever>
simp__: it also rolls back the software installed system wide
<simp__>
ah
<clever>
but not the stuff done with nix-env (that has its own --rollback option)
<simp__>
Oh alright
ryantrinkle has quit [(Ping timeout: 268 seconds)]
<simp__>
also it says "Session: none + i3"? Not sure what that means.
<clever>
i just stick to plain old xfce
<clever>
so i havent had time to learn what wierd things i3 can doo
<simp__>
yeah me neither Ive only started using it recently
<simp__>
but found it really nice
c0c0n3 has joined #nixos
contrapumpkin has quit [(Ping timeout: 240 seconds)]
<c0c0n3>
hi there, having trouble with imperative container networking, can anyone help?
<goibhniu>
simp__: are you using slim?
dridus has joined #nixos
<goibhniu>
(as your display manager)
<goibhniu>
hi c0c0n3, please ask your question, maybe someone can help
<c0c0n3>
cool, here goes
<c0c0n3>
if i create a container and then immediately start it with: nixos-container create foo; nixos-container start foo
<c0c0n3>
then i can ping it with: ping -c 1 $(nixos-container show-ip foo)
<jophish>
clever: so on the aarch64 build I'm not able to find libgcc_s.so.1
eacameron has joined #nixos
<jophish>
I'm putting this down to the same issue as the incorrect bash
<c0c0n3>
but after a few seconds the virtual network interface on the host loses its ip address
<clever>
jophish: i think it should be rpath'd into the elf at compile time
<c0c0n3>
so you can't reach the container anymore, i.e. ping -c 1 $(nixos-container show-ip foo) fails
<c0c0n3>
any idea why this is happening?
<goibhniu>
c0c0n3: ah, you need to tell network manager to not try to manage that interface
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<c0c0n3>
ha! will try that, thx!!!!
<goibhniu>
there is a short section about that in the manual
<goibhniu>
you're welcome
simukis_ has quit [(Quit: simukis_)]
<c0c0n3>
goibhniu: yip, you're right, i even recall reading it, doh! sorry
taeradan has quit [(Read error: Connection reset by peer)]
<goibhniu>
no worries!
contrapumpkin has joined #nixos
<simp__>
Hm, I don't think so goibhniu. How do I check? I haven't changed much
<lukego>
So I reckon that I cooked up a decent backend for my "use nix store as database of application diagnostic information." There are API functions (import.nix) to construct each kind of debug info and then I hand-wrote an example of what the frontend should produce from a big messy dir (test.nix). https://gist.github.com/fb00d7d562be09fc442bba7c30734d68
<goibhniu>
simp__: it's the default, but I believe it's a bit buggy, it might be worth testing something else e.g. lightdm or sddm
<lukego>
The dependencies look good in 'nix-store -qR' now
<simp__>
goibhniu: I assume I change that in configuration.nix
newhoggy has quit [(Remote host closed the connection)]
<bennofs>
simp__: can you switch to a tty and see if the i3 process is running?
byteflame has joined #nixos
<bennofs>
simp__: also take a look at ~/.xsession-errors
<simp__>
well I do get in i3
ericsagnes has joined #nixos
<simp__>
i3status is works
nslqqq has quit [(Ping timeout: 260 seconds)]
nicknovi1 has joined #nixos
Guest75137 has joined #nixos
newhoggy has joined #nixos
MercurialAlchemi has quit [(Ping timeout: 255 seconds)]
Itkovian has joined #nixos
gilligan_ has quit [(Quit: Leaving)]
newhoggy has quit [(Ping timeout: 260 seconds)]
newhoggy has joined #nixos
<jophish>
I wish nix would tell me if I try to perform a build which has already been attempted unsuccessfully
<jophish>
I suppose I could always `nix-store --read-log ...drv`
nslqqq has joined #nixos
<gchristensen>
whoa jophish, that'd be cool
<jophish>
cached failures
iyzsong has quit [(Quit: bye)]
<jophish>
at the moment I'm compiling with various option permutations. it's be nice to warm the cache by attempting them all. come back tomorrow and find the one which worked very quickly
<jophish>
ah, build-cache-failures
<bennofs>
jophish: isn't that what cached failures does for nix? :o
<bennofs>
jophish: ah, I see you found it
<bennofs>
jophish: it would be nice though if it was possible to sync that from hydra.nixos.org sometimes
<jophish>
I wonder why it's off by default
<clever>
jophish: oh, there is another thing of interest
<gchristensen>
maybe it manifests itself as a "mysterious failure"
<clever>
jophish: if you set succeedOnFailure=true; in a build, then any fatal error code will update $out/nix-support/failed and "succeed"
<clever>
then nix will consider it as finished, not failed
byteflame has quit [(Ping timeout: 258 seconds)]
<clever>
and will cache whatever it left in $out
<jophish>
clever: oh wow, that's great
<jophish>
just what I want
acarrico has joined #nixos
<gchristensen>
so many wonderful secrets in nix
<clever>
i think hydra also counts that as failed, but lets you download the $out
<clever>
so you could log details on the failure to $out
spinus has joined #nixos
<clever>
you can also set failureHook to a snippet of bash, that will do such logging
<clever>
for example, save config.log to $out, so you can debug it without having to reproduce it locally
<gchristensen>
:o
<clever>
but you would need a general idea on what could fail in the future, and make it save things before it fails
estewei has joined #nixos
<gchristensen>
yeah
<gchristensen>
it'd be neat if there could be an escape hatch in the failure hook that lets a user get a shell in the build
<clever>
may be of more use in standardized languages like haskell or rust, where there is a standard package format already
erasmas has joined #nixos
<clever>
and the tooling has known log files that it hides info in
<clever>
gchristensen: for that, i build with -K, then nix-shell and cd
bennofs has quit [(Ping timeout: 245 seconds)]
<gchristensen>
yeah, it is a bit annoying still though :)
Itkovian has quit [(Read error: Connection reset by peer)]
byteflame has joined #nixos
<clever>
and some things like QT need a preconfigure hook ran to make qmake even function
<gchristensen>
yeah
<clever>
and nix-shell'ing creates a __nix_qt5__ in the pwd
<clever>
and if that already exists, nix-shell fails
<clever>
so you must damage the build state to even open a shell in it
<clever>
which reminds me of a bug i saw in qt on osx
<clever>
one of the install phases was trying to install things to a dependency
<clever>
and because i setup nix-daemon on the mac, it lacked write to its inputs
<clever>
(as it should!)
<clever>
but a normal nix-shell installPhase fails, because it also lacks write to $out
<clever>
and root gives it write to the inputs and doesnt reproduce the failure
<clever>
so i had to manualy make $out, and give myself write, to simulate a nix sandbox
vandenoever has joined #nixos
<clever>
gchristensen: oh, that gives me an idea
<clever>
gchristensen: another nix-shell mode, beyond --pure, where it just treats nix-daemon as an ssh server, and gives you a shell inside a real sandbox
<clever>
and it could just throw away $out when your done, impurities are not allowed in the real /nix/store!
ThatDocsLady has quit [(Quit: Arma-geddin-outta-here!)]
ThatDocsLady has joined #nixos
Sonarpulse has joined #nixos
<gchristensen>
:o
<gchristensen>
sounds great :o
bennofs has joined #nixos
<clever>
gchristensen: also, i need to look into that ssh agent forwarding idea with nix
<clever>
cant remember who has push access to accept it
byteflame has quit [(Read error: Connection reset by peer)]
<jophish>
clever: if we write alll the derivations like that then you're going to be sure to never make any mistakes for fear of aborting the build on playback!
<clever>
jophish: you wouldnt replicate the commands exactly, but it would remind you about weird things you may have had to edit
<clever>
gchristensen: oh, and i think this part of the manual needs work, using a builder like that can often completely break the ability to override things
<clever>
gchristensen: i think some of these are also un-needed
<gchristensen>
yeah
<clever>
for example, i cant remove nano, ever
<gchristensen>
I know
<clever>
and i should have the choice to remove curl, rsync, strace, tar, diff
<NixOS_GitHub>
[nixpkgs] bennofs force-pushed haskell-fix-override from f207ff3 to 84e7bb1: https://git.io/vKAP7
<NixOS_GitHub>
nixpkgs/haskell-fix-override 84e7bb1 Benno Fünfstück: haskell: preserve overrideScope on override...
<gchristensen>
environment.noRequiredPackages
ixxie has joined #nixos
gsora has quit [(Ping timeout: 240 seconds)]
endformationage has joined #nixos
gal_bolle has quit [(Ping timeout: 258 seconds)]
Guest75137 has quit [(Ping timeout: 258 seconds)]
<jophish>
It would be nice if on a cached build failure, nix-store --read-log was called
<jophish>
not too hard to write a script for though
<ZoomZoomZoom>
I'm writing a service unit and I need to create a directory before starting a service (in a root-owned dir). I put an mkdir into preStart part of the service config, but it errs on launch, saying "permission denied". Shouldn't it be run as root?
Mateon1 has joined #nixos
<bennofs>
ZoomZoomZoom: is the dir on the nix store? the nix store is mounted read-only
<NixOS_GitHub>
[nixpkgs] domenkozar pushed 1 new commit to master: https://git.io/v9I5B
<NixOS_GitHub>
nixpkgs/master 9f11d17 Domen Kožar: add papertrail at 0.10.1...
<LnL>
clever: lol
<ZoomZoomZoom>
nope, it's in /var/lib
<NixOS_GitHub>
[nixpkgs] domenkozar pushed 1 new commit to release-17.03: https://git.io/v9I52
<NixOS_GitHub>
nixpkgs/release-17.03 77016a0 Domen Kožar: add papertrail at 0.10.1...
<clever>
ZoomZoomZoom: is a user set on the service?
<jophish>
Also, is anyone else persistently weirded out by all the nix tools paging their output by default?
<dash>
jophish: doesn't everything do that these days
<dash>
git etc
Hoolootwo has joined #nixos
<ZoomZoomZoom>
clever, yes, serviceConfig.User is set. Is this a problem? I thought it just sets the user under which the actual service is started
<domenkozar>
jophish: PAGER=
<clever>
ZoomZoomZoom: try running the "id" command in prestart, before the mkdir
<domenkozar>
I find it convenient
aneeshusa has joined #nixos
<symphorien>
ZoomZoomZoom: use
<ZoomZoomZoom>
clever, indeed, preStart runs under same user
<symphorien>
PermissionsStartOnly in man systemd.service
<symphorien>
sorry for the spurious line
<clever>
ZoomZoomZoom: which i think would be put into nix as serviceConfig.PermissionsStartOnly = true;
freusque has quit [(Quit: WeeChat 1.7)]
<ZoomZoomZoom>
clever, symphorien, TIL! Thanks a lot.
newhoggy has quit [(Remote host closed the connection)]
<symphorien>
besides I think there is a special system to declare that a directory should be created by systemd
<clever>
yeah, ive heard of that as well
<symphorien>
but I can't remember its name
MichaelRaskin has joined #nixos
Itkovian has joined #nixos
<bennofs>
symphorien, clever, ZoomZoomZoom: systemd-tmpfiles can create various kinds of things
<ben>
what does callPackage = super.lib.callPackageWith super; do, intuitively? :(
<jophish>
domenkozar: ah, good thinking. thanks!
chakerbenhamed has quit [(Ping timeout: 268 seconds)]
<clever>
ben: it creates an instance of callPackage, that will search inside super for the default argument values
<ben>
Isn't that what super.callPackage would already do?
<bennofs>
ben: super.callPackage takes the dependencies from self
<unlmtd[m]>
could nixos FS be interfering with vim-man plugin (to read man pages in vim) - I get 'No manual page for X' - and theres no such issue in the repo
<unlmtd[m]>
clever: I swear youre first on my list of 'ppl to fund when Im rich'
<clever>
heh
newhoggy has quit [(Ping timeout: 260 seconds)]
<CMCDragonkai>
what woudl clever be doing with all that money?
<unlmtd[m]>
ya sounds like that problem
<unlmtd[m]>
hes clever hell know what to do with it
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to staging: https://git.io/v9Ijs
<NixOS_GitHub>
nixpkgs/staging 10af6b4 Vladimír Čunát: perl*, statsmodels: fix evaluation problems...
<CMCDragonkai>
:)
<clever>
ive memorized large chunks of nix and nixpkgs code, and have ideas on how to improve a number of htings
<taktoa>
unlmtd[m]: sometimes I feel like the best decision I've ever made was convincing clever to get into nixos
<qknight>
manveru: it has the reproduce locally
<unlmtd[m]>
taktoa: woha thanks for doing that
<unlmtd[m]>
taktoa: now convince everyone to get into notos
<taktoa>
:-)
<qknight>
manveru: i want to build 'tests.leaps' on hydra
<qknight>
manveru: but i don't know how to write the build.nix file which would build it
<clever>
unlmtd[m]: a list of random things i have plans to finish, a gui installer for nixos, ssh agent forwarding in nix-build, a haskell display manager for nixos, fuse mounting for /nix/store
<manveru>
oh... i have no experience with that
CMCDragonkai has quit [(Read error: Connection reset by peer)]
newhoggy has joined #nixos
<unlmtd[m]>
I like the idea of the minimal nixos
<unlmtd[m]>
clever: maybe you can write it up
<taktoa>
also, unlmtd[m], I once had the thought that if I ever start a software company, I'll hire clever to sit in a slack channel (#clever) to debug everyone else's problems
<unlmtd[m]>
hold off on the ssh thing until ipfs matures, that could be a big game changer
<gchristensen>
taktoa: obviously
<unlmtd[m]>
well, fuck slack
<taktoa>
sure, whatever, IRC is okay too
<clever>
unlmtd[m]: me and taktoa also have plans for an entire chat program, text+voice+video
<pi3r2>
Could someone with `qt` experience give us a hand ;-) ?
<unlmtd[m]>
thats ok Im planning to unlicense all the code one day, soon I guess
ryanartecona has quit [(Quit: ryanartecona)]
<unlmtd[m]>
Ill offer a way to strip every form of IP on everything
<clever>
unlmtd[m]: we also have #arcane-chat on freenode already
fnlkj has quit [(Quit: ?)]
<simpson>
unlmtd[m]: You may find that folks can no longer use your systems.
<taktoa>
I think literally one line of text has ever been sent on that channel though
cfricke has joined #nixos
jmsb has joined #nixos
ilyaigpetrov has joined #nixos
<sphalerite>
taktoa: AGPL! :p
ZoomZoomZoom has quit [(Quit: Leaving)]
<taktoa>
unlmtd[m]: Just to set the record straight: you realize that because arcane-chat links into libtoxcore, which is licensed GPLv3, we *have* have arcane-chat be licensed under GPLv3. If at some point we stop depending on libtoxcore, then we can have a conversation about adopting more a permissive license. We chose to use libtoxcore because it was a library that clever is familiar with and it solves many of the problems we were facing (authentication, hole-
<taktoa>
punching, encryption, etc.).
aneeshusa has quit [(Ping timeout: 260 seconds)]
<MichaelRaskin>
sphalerite: do you understand AGPL? Be careful: if you say yes, I will ask questions about it (not that I can force you to answer them, of course).
<taktoa>
I *think* I understand AGPL but would like to know if there are more subtleties to it than my current understanding of it.
<taktoa>
MichaelRaskin: ^
newhoggy has joined #nixos
<MichaelRaskin>
Well, basically I try to imagine it being applied to some systems I have written and fail to understand the details. The main question is what _config_ you are supposed to be distributing with the system.
<MichaelRaskin>
For many ends we do need to treat any interesting enough config as code… on the other hand, you need to store access secrets outside what you publish… then it becomes complex enout to count as a derivative body of code
<NixOS_GitHub>
[nixpkgs] indiscipline opened pull request #25181: Implement aria2 service for controlling a daemon via rpc. (master...aria2.service) https://git.io/v9LJf
newhoggy has quit [(Ping timeout: 245 seconds)]
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to staging: https://git.io/v9LJE
<NixOS_GitHub>
nixpkgs/staging 9f0de95 Vladimír Čunát: net-snmp: fix build by using older perl for now...
<unlmtd[m]>
taktoa: you should really look at matrix
acertain has joined #nixos
<qknight>
my hydra reports always: GC Warning: Out of Memory! Heap size: 3047 MiB. Returning NULL!
<taktoa>
unlmtd[m]: yeah I know a decent amount about matrix and it's on my list of things to potentially migrate to to avoid tox
<qknight>
but i want it to build hello
<dash>
Matrix _seems_ really good. I need to learn what's bad about it :)
<clever>
qknight: 32 or 64bit machine?
newhoggy has joined #nixos
<unlmtd[m]>
it uses http as connector
<unlmtd[m]>
which is a bit heavy
<unlmtd[m]>
but very convenient for now
<taktoa>
my current list of alternatives is: Matrix, WebRTC, Signal, something custom
<simpson>
unlmtd[m]: Hm, "heavy" how? I'll certainly agree that we should all be considering Capn Proto.
<qknight>
clever: 64 bit machine
<qknight>
clever: i don't get how to create a hydra project, it probably wants to evaluate nixpkgs or something
<simp__>
Does xorg come bundled with nixos?
<clever>
qknight: do you want to build a custom project, or all of nixpkgs?
<ToxicFrog>
Hah.
<ToxicFrog>
I found the reason my certificates keep getting their permissions screwed up.
<sphalerite>
MichaelRaskin: I have a vague idea of it
<ToxicFrog>
Every ACME renewal timer has a pre-run script
<sphalerite>
Probably not enough to answer any questions about it
<qknight>
peti: https://ibb.co/ftYeO5 <- i've added this values but i get no jobs ever: There are no matching jobs.
byteflame has joined #nixos
<qknight>
peti: i tried to evaluate it manually several times
<ToxicFrog>
Looks like this was introduced in pr/24532
<qknight>
clever: i try petis docs ATM
<Tipping_Fedora>
hi guys, I was wondering about the current support of Nixos on the raspberrpi, can anyone point me in the direction to look? (or just tell me)
<clever>
qknight: there is no hydra-tutorial sub-dir in the hydra-tutorial repo, so the nix expression in that screenshot is invalid
simendsjo has joined #nixos
<clever>
qknight: you want just jobset-nixpkgs.nix
<qknight>
clever: oh yes, i tried that later. without that subdir it doesn't work either. let me try gain
<clever>
qknight: the $NIX_PATH is also clear, and you didnt tell it where to find nixpkgs
<NixOS_GitHub>
[nixpkgs] ahmedtd opened pull request #25184: python/wrapper.nix: Don't try to wrap pyc files (master...fix-python-wrapper) https://git.io/v9LI0
<sphalerite>
taktoa: what's wrong with the security on tox?
<jophish>
clever: do you know where I should start looking to determine why the wrong bash is being put in the `ghc` script and why the RPATH on ghc is incorrect
<jophish>
I assume it's because of an incorrect RPATH that ghc can't find libgcc_s.so.1
cfricke has quit [(Quit: WeeChat 1.7)]
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to master: https://git.io/v9LIM
<NixOS_GitHub>
nixpkgs/master cf3800b Tuomas Tynkkynen: Merge remote-tracking branch 'upstream/staging' into master
<taktoa>
sphalerite: it's mostly pretty good, but there were two incidents that give me pause.
<taktoa>
sphalerite: first incident was a bug clever found in which they were using libsodium in a way that resulted in memory containing private keys not being properly scrubbed.
<ToxicFrog>
(Depending on your configuration, it is a possible security issue, since it can make certificates readable by groups you did not expect to have access)
<qknight>
clever: awesome, it works! thanks so much for your help!
<Guest13938>
Hello, hope someone could help me a bit with the installation from live usb. I'm having trouble mounting the usb to the /mnt directory has the volume is already mounted in read-only to /iso (nixos manual chapter 2.9)
<kuznero>
I am try to specify unstable when querying for fsharp41... but still it is not there. Probably missing something
<kuznero>
Please help
derjohn_mob has quit [(Ping timeout: 260 seconds)]
<clever>
jophish: the cross-compile stuff has been redone since i last used it, so id need to look into it again
<taktoa>
sphalerite: oh also, unrelated to security, but tox has a habit of opening so many UDP connections that it will DoS some peoples' routers. so for that reason alone we may need to migrate away from it (or open a PR).
<sphalerite>
ah ok
<goibhniu>
hi Guest13938 you want to mount your root filesystem on /mnt, not the usb drive
<jophish>
if you search for gnu/bin/ghc you'll see the script generation
marsel has quit [(Ping timeout: 255 seconds)]
digitus has joined #nixos
<qknight>
clever: on more question, can hydra handle private git repos? if so, is it documented how to do so somewhere?
<qknight>
s/on/one
<ToxicFrog>
I want both rspamd and exim to run as group 'mail' so that exim can write to the rspamd socket.
<clever>
qknight: yeah, ive done that as well
<clever>
qknight: first, do "sudo -u hydra -i" and run "ssh-keygen" to make a keypair, no passphrase
<ToxicFrog>
But each one, if you tell it to run as a specific group, implicitly creates that group using users.extraGroups, and you can't define the same group twice.
<Sonarpulse>
I've been fixing travis errors and beefing up commit messages
Guest13938 has quit [(Ping timeout: 260 seconds)]
newhoggy has joined #nixos
<rsa>
anyone have any experience with overriding version dependencies for haskell packages? (in my case xmonad-wallpapers requires xmonad ==0.12.* and current is 0.13)
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to master: https://git.io/v9LYV
<NixOS_GitHub>
nixpkgs/master 0537320 Tuomas Tynkkynen: networkmanager: Temporarily disable -Wformat-security until next release...
byteflame has joined #nixos
newhoggy has quit [(Ping timeout: 255 seconds)]
davidak has joined #nixos
davidak has quit [(Client Quit)]
goibhniu has quit [(Ping timeout: 260 seconds)]
peacememories has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
peacememories has joined #nixos
betaboon has joined #nixos
betaboon has quit [(Changing host)]
betaboon has joined #nixos
<disasm>
savanni: hey, saw in IRC history you were working on an icinga2 package for nixpkgs. How far did you get on that? If you wouldn't mind sharing what you've done so far, I'd be happy to give it a go.
ntruessel has joined #nixos
<savanni>
I honestly didn't get very far at all. And it ended up not being relevant. I had wanted it for a local experiment, which I abandoned with my colleague just put it into a production environment.
<savanni>
I recall it being quite a mess, but probably manageable now that I know nix packaging a lot better.
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 4 new commits to master: https://git.io/v9LGa
<NixOS_GitHub>
nixpkgs/master 75608e1 Jason A. Donenfeld: wireguard: 0.0.20170409 -> 0.0.20170421
<NixOS_GitHub>
nixpkgs/master 600a2f2 Jörg Thalheim: wireguard: make myself maintainer
<NixOS_GitHub>
nixpkgs/master 8deb0e1 Jörg Thalheim: wireguard: do not fail on warning for older kernel
ryanartecona has quit [(Quit: ryanartecona)]
mudri has quit [(Ping timeout: 240 seconds)]
ambro718 has quit [(Ping timeout: 240 seconds)]
tokudan has joined #nixos
<copumpkin>
Sonarpulse: can't do it today, but probably sometime this week
<copumpkin>
Sonarpulse: especially if smaller than it was and well explained :)
<Sonarpulse>
right
<Sonarpulse>
I'm still thinking about the best way to split
<Sonarpulse>
but I'm adding more explanations in the meantime
<Sonarpulse>
it's sort of easier to describe what things become than how they changed
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to master: https://git.io/v9LZ3
<NixOS_GitHub>
nixpkgs/master 9437c36 Tuomas Tynkkynen: perlPackages.ModuleBuild: 0.4214 -> 0.4222
acowley is now known as acowley_away
thinkingeye has quit [(Quit: Konversation terminated!)]
betaboon has quit [(Quit: Leaving)]
m3tti has quit [(Ping timeout: 260 seconds)]
newhoggy has joined #nixos
<jophish>
When cross compiling, which package supplies the target's libgcc_s.so?
<jophish>
I can see it in the store in gccCrossStageFinal
<jophish>
however I'm not sure of the best way to access that
<jophish>
(basically my problem is that the RPATHs of the resulting binaries are not being patched correctly, in fact they are empty!)
<Sonarpulse>
jophish i thought libgcc was always statically linked
<Sonarpulse>
as compliler-rt is
byteflame has quit [(Ping timeout: 268 seconds)]
<jophish>
Sonarpulse: perhaps I'm solving the wrong problem here then
sigmundv__ has joined #nixos
newhoggy has quit [(Ping timeout: 255 seconds)]
<Sonarpulse>
jophish: what's the symptom?
ryanartecona has joined #nixos
byteflame has joined #nixos
mbirkis has quit [(Quit: WeeChat 1.4)]
<jophish>
Sonarpulse: when I run the binary on the target machine I get: ./ghc: error while loading shared libraries: libgcc_s.so.1: cannot open shared object file: No such file or directory
<Sonarpulse>
woah
<Sonarpulse>
what does the DT_NEEDED of the normal native one look like?
<Sonarpulse>
or the libs
<Sonarpulse>
err scratch libs if you're sure this is GHC itself failing to start, not some weird linking/th/etc of ghc being used
thinkingeye has joined #nixos
<jophish>
I'm not sure
<jophish>
well, I'm not doing a compile or anything weird
<jophish>
and LD_DEBUG shows it failing very quickly
<jophish>
LD_DEBUG=all shows less than a page!
nschoe has joined #nixos
<nschoe>
Hi all
<nschoe>
I'm trying to install `nixos.yed`, and it fails with 'builder for ‘/nix/store/263g58gc6pk2ylgh8x7mpkm1af2zk4x4-yEd-3.16.1.zip.drv’ failed to produce output path ‘/nix/store/0vdkf7gzr1mni3w7sr14rclsf0igzq26-yEd-3.16.1.zip’'. Can you explain what is going on?
<jophish>
nschoe: can you gist the whole log?
<jophish>
it sounds like that file couldn't be downloaded
<nschoe>
jophish, yes indeed. So it sorts of explains how to download it manually, but that's odd, right?
<jophish>
It's not typical, some derivations require it though
<nschoe>
jophish, but I don't understand why it could not download the file automatically? All other derivations do it automatically, right?
<jophish>
nschoe: it might require navigating a website to get it
<nschoe>
jophish, how, I see.
<clever>
nschoe: licensing reasons
<nschoe>
Thanks for your help ^^
<nschoe>
clever, this is weird, it used to be free...
<nschoe>
But okay, I understand. Will try something else, then :)
ThatDocsLady has quit [(Quit: Arma-geddin-outta-here!)]
pbogdan has joined #nixos
newhoggy has joined #nixos
<NixOS_GitHub>
[nixpkgs] Ericson2314 opened pull request #25186: top-level: common subexpression elimination for some cc-wrapper (master...ccwrapper-cse) https://git.io/v9LCx
<ben>
Hey, I wanna write an overlay that bumps the version of a nixpkgs package, but only if it's lower than what I was gonna bump it to, but I get infinite recursion issues :'(
<ben>
what's wrong with self: super: if super.blah.version < newVersion then { blah = super.blah.overrideAttrs { ... }; } else { blah = super.blah; }?
ixxie has quit [(Ping timeout: 255 seconds)]
dridus has quit [(Quit: znc? wassat?)]
thinkingeye has quit [(Quit: Konversation terminated!)]
Itkovian has joined #nixos
sigmundv__ has quit [(Remote host closed the connection)]
dridus has joined #nixos
<Tipping_Fedora>
hi, how can I add a user to the sudoers file? evertime i reboot the user is lost from there( as it stands I manually add that user into the file)
newhoggy has joined #nixos
<ben>
probably with security.sudo.extraConfig in your configuration.nix
<ben>
possibly you can just put the user in the wheel group instead
<ben>
if that's what you want
<dash>
Tipping_Fedora: easy way is put 'extraGroups = [ "wheel" ];' in your user definition in configuration.nix
<Tipping_Fedora>
ok thanks
<clever>
Tipping_Fedora: you can also add yourself to the wheel group
<gchristensen>
ben: are you pj?
<ben>
no
<gchristensen>
ok
mbirkis has joined #nixos
dridus has quit [(Quit: znc? wassat?)]
newhoggy has quit [(Ping timeout: 260 seconds)]
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<Tipping_Fedora>
thanks guys for the help
dridus has joined #nixos
<sphalerite>
is gccCrossStageStatic the appropriate package if I don't need a libc?
<simpson>
Stupid question: Does anybody use rvm/bundler-style Ruby stuff in Nix? What's your setup look like?
roconnor has joined #nixos
newhoggy has joined #nixos
zarel has joined #nixos
<ben>
i answered my overlay question from earlier: i need to do the conditional inside the overlay's { }, not outside, or laziness won't save me
<zarel>
hi hi
<copumpkin>
when does ikwildrpepper return?
nschoe has quit [(Ping timeout: 258 seconds)]
<c74d>
ben: I don't think comparing versions works like that; at least, there are functions for comparing versions that don't just do that (search for "version" in lib/strings.nix)
<ben>
oh, cheers
thinkingeye has joined #nixos
<simpson>
Wow, I just finally understood. It's not "Wild R. Pepper", but "I Know Why I Love Dr. Pepper", isn't it!?
newhoggy has quit [(Ping timeout: 255 seconds)]
<copumpkin>
"I want Dr. Pepper" in Dutch, as I understand it
<copumpkin>
ik wil dr. pepper
sigmundv has joined #nixos
<ntruessel>
Ok, found a patch to gn from arch that seems to work better
ogkloo has joined #nixos
<simpson>
That's a pretty good nick. Haskellers have really good nicks. Or good names, like "Philip" or "Shachaf".
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to master: https://git.io/v9LBE
<NixOS_GitHub>
nixpkgs/master c90998d Tuomas Tynkkynen: Revert "rustc: don't build on i686"...
<copumpkin>
simpson: but yours probably works too :)
<steveeJ>
how can I get a self reference in a recursive derivation?
<copumpkin>
steveeJ: (rec { }) is just a construct for writing an attrset that can refer to its own keys in its values
<copumpkin>
steveeJ: derivation knows nothing about that
<steveeJ>
copumpkin: so my question would be, how can I introduce the derivation to itself?
<steveeJ>
LnL: I'll read up on it and see if I can apply that
<copumpkin>
steveeJ: I'd step back and tell us what you actually want to do :)
<LnL>
what are you trying to do exactly?
<copumpkin>
it has a reference to its output inside the builder as $out
<steveeJ>
copumpkin: LnL: I want a nix variable out what is $out in the installPhase
<copumpkin>
why does it need to be on the nix side?
<copumpkin>
nix never knows (inside a derivation) the hash of the output
<copumpkin>
after the derivation is built, you can know its output path
<LnL>
you can't do that, it would result in an evaluation cycle
<steveeJ>
copumpkin: something inside me wants not to believe this :-D
<copumpkin>
1.12 introduces the notion of a placeholder
<LnL>
why can't u use $out?
<steveeJ>
LnL: which I have produced many of before I came here to ask the question ;-)
<copumpkin>
steveeJ: you can't operate on the hash before the hash is computed on something that includes the hash
<steveeJ>
LnL: $out seems problematic beccause if it's used in another derivation's installPhase it will be evaluated there
<copumpkin>
steveeJ: unless you break sha256 to figure out fixed points of it :P
<copumpkin>
the 1.12 placeholder notion lets you talk about the output in Nix before you have it, but that's about it
python476 has joined #nixos
mudri has joined #nixos
<steveeJ>
copumpkin: that would be enough, because I would actually not use this placeholder inside the derivation, just declare it there
<copumpkin>
it still won't let you use it
<LnL>
can you give an example?
<copumpkin>
the usual use for placeholders is for configureFlags and such
<steveeJ>
the way I have it now is in drvA i have relPath, and I write ${drvA}/${drvA.relPath}
<steveeJ>
in drvB
<copumpkin>
oh, you want the derivation to be able to advertise some things about itself to other nix stuff?
<copumpkin>
for that I'd use passthru
<steveeJ>
drvA.relPath is used within drvA"s installPhase like so: $out/${relPath}
<copumpkin>
which is evaluated separately
newhoggy has joined #nixos
<LnL>
yeah, you can probably use passtru for this
<steveeJ>
I just want to not repeat drvA :-D
<sphalerite>
Can I use nixpkgs to get a cross-compiler without a libc?
<copumpkin>
let relPath = "foo/bar/baz"; drv = stdenv.mkDerivation { name = "haha"; passthru.fullPath = "${drv}/${relPath}"; } in drv
<copumpkin>
that would work fine
<copumpkin>
steveeJ: then with the snippet I wrote above, you'd write ${mydrv.fullPath}
<copumpkin>
and it would evaluate to the composed path
DutchWolfie has quit [(Quit: Konversation terminated!)]
<copumpkin>
or you can pass through the relPath alone if you need to use it in isolation
<LnL>
how I understand it is that mkDerivation ignores passtru allowing you to add attributes to the drv that won't affect the drv
gilligan_ has joined #nixos
<jophish>
Sonarpulse: do you know how to get the path the libgcc_s.so reliably, for example with a cross compiler it's at /nix/store/...-gcc-cross-stage-final/aarch64-linux-gnu/lib64/libgcc_s.so
<jophish>
the aarch64-linux-gnu/lib64 seems a little tricky
<Sonarpulse>
jophish: I don't know off hand
<Sonarpulse>
I'm not sure I'd bother
newhoggy has quit [(Ping timeout: 245 seconds)]
<Sonarpulse>
one off hack to get you that native build!
<jophish>
true, although I'd quite like to leave this build in nixpkgs, and to have it work for other archer
<jophish>
arches*
<NixOS_GitHub>
[nixpkgs] edwtjo pushed 1 new commit to master: https://git.io/v9LzZ
<NixOS_GitHub>
nixpkgs/master 0277345 Edward Tjörnhammar: nixos, i2pd: remove, no longer needed, extip hack
<disasm>
savanni: thanks! I'll play around with it from scratch :)
<jophish>
Although even better would be to get gcc statically linking everything!
<steveeJ>
copumpkin: this is really, really useful! thanks!
<steveeJ>
sphalerite: I have an expression somewhere that does exactly that because I can't figure out how to get one *with* libc :-D do you have one?
<LnL>
steveeJ: a drv is basically just an attribute set with { type = "derivation"; }, so you can add attributes after the builder is evaluated that won't affect the output
<m3tti>
hey there how could i build a dev environment with happstack
thinkingeye has quit [(Quit: Konversation terminated!)]
<sphalerite>
steveeJ: SuperH version 3, an architecture originally designed by Hitachi and now maintained by Renesas
<sphalerite>
well I think it's on version 4 now
<sphalerite>
but my calculator has an SH3 processor
alx741 has joined #nixos
peacememories has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<gilligan_>
sphalerite, i never tried - but for starters this should just get fixed for everyone using broadcom-sta right now, which should be most ppl running nixos on a macbook I guess
newhoggy has quit [(Ping timeout: 240 seconds)]
<sphalerite>
I'm on a MacBookPro8,2 with B4331 wifi, been using b43 as long as I've used nixos
<NixOS_GitHub>
[nixpkgs] Ericson2314 pushed 2 new commits to master: https://git.io/v9LgQ
<NixOS_GitHub>
nixpkgs/master bef5278 John Ericson: top-level: Common subexpression elimination for cc-wrapper usage...
<NixOS_GitHub>
nixpkgs/master ee2d787 John Ericson: Merge pull request #25186 from obsidiansystems/ccwrapper-cse...
<steveeJ>
sphalerite: how can I access the cross compiler with your expressions?
<gilligan_>
sphalerite, macbook air here
<sphalerite>
steveeJ: well I use it to cross-compile existing packages. e.g. I do nix-build kobo.nix -A rsync to get rsync for it
mkoenig has joined #nixos
<steveeJ>
LnL: oh I see, but it's potentially very dangerous to do so
<steveeJ>
sphalerite: my use case is I want a nix-shell that has a cross-compiler for the target
<sphalerite>
steveeJ: probably nix-shell expr.nix -A stdenv.cc
<steveeJ>
sphalerite: interesting, I'll give it a try
<sphalerite>
steveeJ: wait no, maybe stdenv.ccCross. But stdenv.cc seems to target the host arch
<LnL>
steveeJ: what do you mean? it has no impact on the build
newhoggy has joined #nixos
peacememories has joined #nixos
peacememories has quit [(Client Quit)]
<steveeJ>
LnL: oh, so this is only valid for the attribute "passthru", everything else *is* considered for the hash
<LnL>
for stdenv.mkDerivation yes, the reason why other attributes change the hash is because they are exposed as environment variables in the builder
<steveeJ>
LnL: I think I'm tripping over the time aspect, and I somehow thought there is a window between evalauating and building the derivation where I could modify it
<steveeJ>
LnL: I'm not well informed when hashing takes place and what is part of the input
<LnL>
passtru is ignored by the builder so you can't access those at build time
<sphalerite>
steveeJ: derp, it's not nix-shell expr.nix -A stdenv.cc because that yields an environment in which to build stdenv.cc. Using -A hello works, or probably (not tried it) nix-shell -p 'with import ./expr.nix; stdenv.ccCross' will too
<sphalerite>
steveeJ: so nix-shell expr.nix -A hello, and $CC in the resulting shell points to the appropriate compiler
byteflame has quit [(Ping timeout: 260 seconds)]
thinkingeye has quit [(Quit: Konversation terminated!)]
newhoggy has quit [(Ping timeout: 260 seconds)]
byteflame has joined #nixos
<sphalerite>
yay, that was the same problem I was having with my SH3 cross-compiler. Just invoking nix-shell wrong!
reinzelmann has quit [(Quit: Leaving)]
<sphalerite>
steveeJ: thank you for helping us help you help us all.
<copumpkin>
steveeJ: yeah, passthru is "magic" in that consumers see its attributes on the derivation return value directly, but it doesn't affect the hash
<disasm>
quick question for folks using i3... if the config is specified in a nix declaration, a restart doesn't update the config because well the window manager is launched with the old config in the store as a param... So other than logging out and losing my window config are there any other options to update my config?
<NixOS_GitHub>
[nixpkgs] fpletz pushed 6 new commits to master: https://git.io/v9La7
<NixOS_GitHub>
nixpkgs/master afcf15b Franz Pletz: pythonPackages.pygit2: 0.24.0 -> 0.25.0...
<NixOS_GitHub>
nixpkgs/master e74ea42 Franz Pletz: avahi service: add reflector option
<NixOS_GitHub>
nixpkgs/master e68c3f0 Franz Pletz: atlassian-confluence: 6.0.3 -> 6.1.1
<clever>
disasm: i think the display manager will typicaly exit when you logout
<clever>
disasm: then systemd has to re-launch it, with the new config
<clever>
but some display managers may try to optimize that away
AllanEspinosa has joined #nixos
<disasm>
clever: yeah, logging out is what I was trying to avoid, lol :) I guess I either just move the config to my home dir and don't use nix to define it or deal with logging out once a month or so when I make a change to the config file :)
Filystyn has quit [(Quit: Konversation terminated!)]
<Sonarpulse>
trigger finger is itching on that dead code removal
bennofs has joined #nixos
<krux02>
Sonarpulse: I am new to nixos, and I try to build my project written in Nim in nixos
<krux02>
libX11.so: cannot open shared object file
sellout- has quit [(Quit: Leaving.)]
newhoggy has joined #nixos
<Sonarpulse>
krux02: ummmm can you run file on that?
<symphorien>
krux02: do it in nix-shell -p gcc xorg.libX11
<Sonarpulse>
I know nothing about nim
<Sonarpulse>
ahhh
<Sonarpulse>
missing not corrupt file
<Sonarpulse>
gotta rule out simple hings first
<krux02>
ok I could solve it by setting LD_LIBRARY_PATH in default.nix
<Sonarpulse>
shlevy: do you have time to review a tiny PR for your iOS stuff in a few minutes?
<krux02>
not sure if it is the correct way to do it
ntruessel has quit [(Quit: Page closed)]
<Sonarpulse>
krux02: ah you have default.nix for your project? good
<symphorien>
there is buildInputs for that
<Sonarpulse>
yeah
<symphorien>
buildInputs = [ all the libraries you need ]
<Sonarpulse>
*you need at runtime :D
<krux02>
buildInputs has already SDL2 set, but putting X11 in there does fix the dependency
<Sonarpulse>
great!
newhoggy has quit [(Ping timeout: 260 seconds)]
<krux02>
is this correct? buildInputs = [ pkgs.SDL2 pkgs.xlibs ]
ambro718 has joined #nixos
<NixOS_GitHub>
[nixpkgs] Ericson2314 opened pull request #25190: crossStdenv on iphone: Just get info from `targetPlatform` (master...ios-stdenv-tiny-refactor) https://git.io/v9LXw
newhoggy has joined #nixos
m3tti has quit [(Ping timeout: 258 seconds)]
aneeshusa has joined #nixos
aneeshusa has quit [(Client Quit)]
<krux02>
adding SDL to the build inputs does not resolve X11
<krux02>
which buildInputs contains the libX11.so?
newhoggy has quit [(Ping timeout: 255 seconds)]
acowley_away is now known as acowley
marsel has quit [(Ping timeout: 252 seconds)]
thetet has joined #nixos
<krux02>
adding xorg.libX11 does not make the program find X11
newhoggy has joined #nixos
<symphorien>
that is strange : its output contains libX11.so
<krux02>
I don't know if it is important, but ldd on the program does not list libX11
<symphorien>
on which program ?
<krux02>
well it is my own SDL2 application written in nim
<krux02>
it basically jusd creates an opengl context and renders some screensaver stuff
<krux02>
nothing big, except that it is written in Nim
newhoggy has quit [(Ping timeout: 255 seconds)]
<krux02>
so I don't know if nim interferes there somehow to load the libX11 library
<Profpatsch>
seanparsons: It doesn’t give you what you want, though.
<krux02>
how do I find out what part of a programs tries to load libX11.so
<krux02>
?
<Profpatsch>
That is configureFlags
<Profpatsch>
krux02: What do you mean by “part”?
<hodapp>
Profpatsch, bennofs: yes, there are, and I tried this first, and it's not working.
<bennofs>
hodapp: oh, what is the package in question
<hodapp>
ezdxf
<steveeJ>
does anyone have experince using alternatives to travis-ci with nix?
ryanartecona has joined #nixos
newhoggy has quit [(Ping timeout: 260 seconds)]
sellout- has joined #nixos
k2s has joined #nixos
k2s has quit [(Client Quit)]
nschoe has joined #nixos
gilligan_ has quit [(Ping timeout: 240 seconds)]
agjacome has quit [(Remote host closed the connection)]
<Profpatsch>
seanparsons: Wait.
<bennofs>
hodapp: hmm usually it works like this
<bennofs>
hodapp: idk what is broken in this case
<Profpatsch>
I just noticed that the configureFlags are always output when haskell packages are built x
<Tipping_Fedora>
hi, how well does nixos duel boot with another Linux os?
<dash>
Tipping_Fedora: just fine
<hodapp>
bennofs: figured it out, it's .zip rather than .tar.gz
<seanparsons>
Profpatsch: I don't think that includes the cabal flags though does it?
<Profpatsch>
seanparsons: Why, yes, those are the cabal flags
<seanparsons>
Looking at the cabal file for xmonad-contrib it doesn't define a default for use_xft, which I assume means it's false by default.
<dash>
Tipping_Fedora: just add it the other one in the grub config
<Tipping_Fedora>
ok
<Tipping_Fedora>
thanks
<NixOS_GitHub>
[nixpkgs] Ericson2314 pushed 2 new commits to master: https://git.io/v9L7G
<NixOS_GitHub>
nixpkgs/master db5a921 John Ericson: crossStdenv on iphone: Just get info from `targetPlatform`
<NixOS_GitHub>
nixpkgs/master af6e4c5 John Ericson: Merge pull request #25190 from obsidiansystems/ios-stdenv-tiny-refactor...
<7JTAA1KGK>
[nixpkgs] Ericson2314 closed pull request #25190: crossStdenv on iphone: Just get info from `targetPlatform` (master...ios-stdenv-tiny-refactor) https://git.io/v9LXw
ertesx has joined #nixos
<Sonarpulse>
niksnut: would it be acceptable to have some cross tests as part of nixos- or nixpkgs-unstable?
<Sonarpulse>
now that this stuff is less broken, I want to hold it to a higher standard
<eacameron>
Is there an easy way to have a whitelist of directories in a source filter?
<eacameron>
This seems like it would be a common use case but it's annoyingly hard to write by hand.
mkoenig has quit [(Remote host closed the connection)]
newhoggy has joined #nixos
<copumpkin>
sounds more than acceptable to me :)
<copumpkin>
if anything, it should be unacceptable to do otherwise
<Profpatsch>
ericnoan: You have a very strange version then.
<bennofs>
ericnoan: oh I didn't notice nixos.com vs nixor.org
<bennofs>
s/nixor/nixos
<Profpatsch>
toplel
<Profpatsch>
ericnoan: Hm, it’s some french site for me.
<ericnoan>
it was rush hour as well
<bennofs>
ericnoan: btw, there is also nixos.org/nixpkgs/manual and nixos.org/nixos/manual if you weren't aware
newhoggy has joined #nixos
mudri has joined #nixos
<NixOS_GitHub>
[nixpkgs] Profpatsch closed pull request #25070: haskell: add a ghcDefault pkgset, pointing to current default (master...ghcDefault) https://git.io/v9vh0
<ericnoan>
yes i know, i initially only read the nixos/manual, and saw there were manuals for nix and nixpkgs, going to read them as well
<ericnoan>
thats probably why i felt something was missing
<ericnoan>
Profpatsch: yeah i get a french pron site
<hodapp>
is the site French, or the porn?
<Profpatsch>
ericnoan: My spam filter saved me then. :)
<ericnoan>
hodapp: im to busy reading nix docs to find out
newhoggy has quit [(Ping timeout: 240 seconds)]
fresheyeball has joined #nixos
<fresheyeball>
hey so I managed to get an illegal configuration line in my nix.conf
bennofs has quit [(Ping timeout: 252 seconds)]
<fresheyeball>
because I did something dumb in configuration.nix
<fresheyeball>
but now I can't rebuild the os!
<fresheyeball>
what should I do?
wak-work has quit [(Ping timeout: 240 seconds)]
<symphorien>
reboot and select a previous configuration in grub
<fresheyeball>
symphorien: really?
<symphorien>
there should be a nix command for that but I don't know it :°
<fresheyeball>
if I make a mistake in configuration.nix I need to reboot to get my nix.conf fixed!?
newhoggy has joined #nixos
<ericnoan>
fresheyeball: when you boot, you can select the second option. nixos keeps old configurations
<fresheyeball>
ericnoan: I know
<fresheyeball>
its just seems dumb
<fresheyeball>
I put a list of strings in binaryCachePublicKeys
<fresheyeball>
instead of concating them to one string
<fresheyeball>
now I need to reboot
<fresheyeball>
that doesn't seem right
<symphorien>
I see something called nixos-rebuild --rollback
<ericnoan>
it is nixos-rebuild switch --rollback
<symphorien>
does it work ?
roconnor has joined #nixos
<fresheyeball>
no even that throws errors because my my nix.conf
<symphorien>
you might also override nix conf with an environment variable
<ericnoan>
there is also "nixos-rebuild switch --repair"
<symphorien>
in last resort, you can remout -o rw /nix/store but I strongly discourage you to do so because nixos-rebuild does'nt configure grub correctly after that
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 2 new commits to master: https://git.io/v9LbC
<NixOS_GitHub>
nixpkgs/master 18de0c3 panaeon: skypeforlinux: init at 5.1.0.1; added myself as maintainer
<NixOS_GitHub>
[nixpkgs] copumpkin closed pull request #25165: google-compute-image: switch to use the common make-disk-image.nix (master...google-image-common) https://git.io/v9kKg
hiratara has joined #nixos
newhoggy has quit [(Ping timeout: 240 seconds)]
Wizek_ has joined #nixos
nicknovi1 has quit [(Ping timeout: 252 seconds)]
<copumpkin>
phreedom: you around?
newhoggy has joined #nixos
mizu_no_oto has quit [(Ping timeout: 245 seconds)]
newhoggy has quit [(Ping timeout: 240 seconds)]
nicknovi1 has joined #nixos
krux02 has quit [(Remote host closed the connection)]
<Sonarpulse>
though let me know if I should elaborate it
kampfschlaefer has quit [(Ping timeout: 252 seconds)]
<copumpkin>
all I'm really looking for is a super high level "I am John and I am rewriting cross compilation in ways X, Y, Z, because the old way was busted, and made no sense, and the new way makes a lot more sense. I hereby define host to be A, target to be B, and use that consistently across all my commits. The approach is different by passing around stdenvs P, Q instead of the old approach which did W, V, which sucked because D, E"
<copumpkin>
I just have no context
<copumpkin>
I know you're doing massive changes for cross-compilation all over the place
<copumpkin>
I just have no coherent high-level story for why you're doing it, what your thought process was, etc.
<Sonarpulse>
ok
<Sonarpulse>
copumpkin: would that belong on this PR or the parent PR?
fnlkj has joined #nixos
<copumpkin>
I can review things on a small level and say "indeed you did rename all variables properly with no typos" but not having the high-level context makes my review less useful to you because I'm only doing basic syntactic checks and looking for obvious stupid stuff
contrapumpkin has quit [(Quit: My MacBook Pro has gone to sleep. ZZZzzz…)]
<copumpkin>
Sonarpulse: I don't really know or care, as long as it's somewhere :)
<Sonarpulse>
ok
<copumpkin>
and findable from the various PRs
<Sonarpulse>
fair enough
<copumpkin>
(this is true for all reviews, FWIW)
<copumpkin>
just becomes more necessary for larger ones
sigmundv has quit [(Quit: Leaving)]
<Sonarpulse>
let me beef up commit message, and also double check cross compilation docs
<Sonarpulse>
copumpkin: hah oops the doc note on deprecated "stdenv.cross" already assumes this change
justelex has quit [(Ping timeout: 240 seconds)]
sigmundv has joined #nixos
ambro718 has quit [(Ping timeout: 255 seconds)]
contrapumpkin has joined #nixos
acertain has quit [(Remote host closed the connection)]
<NixOS_GitHub>
[nixpkgs] Ericson2314 pushed 4 new commits to master: https://git.io/v9tqq
<NixOS_GitHub>
nixpkgs/master a7d8913 John Ericson: top-level: `stdenv.cross` is now only defined with host != build...
<NixOS_GitHub>
nixpkgs/master a7068ac John Ericson: Rewrite a few `stdenv.cross` uses that *should* be `targetPlatform`...
<NixOS_GitHub>
nixpkgs/master 49c99b7 John Ericson: cross-stdenv: Only prune most overrides in the final stage...
<Sonarpulse>
copumpkin: updated the commit message and merged
newhoggy has joined #nixos
<Sonarpulse>
insofar that i'm kind of the sole person this deep in the weeds, I guess that makes sense
<copumpkin>
hmm fair enough :)
digitus has quit [(Quit: digitus)]
<copumpkin>
where's the "my name is John" blurb? :) :) :)
<copumpkin>
regardless, I think such a thing should exist
<copumpkin>
even if you do go merging things without getting other people to look at them
griff_ has quit [(Quit: griff_)]
bennofs has quit [(Ping timeout: 240 seconds)]
drasich has joined #nixos
newhoggy has quit [(Ping timeout: 260 seconds)]
cpennington has quit [(Remote host closed the connection)]
<Sonarpulse>
copumpkin: err now I'm confused again
JagaJaga has quit [(Ping timeout: 258 seconds)]
<Sonarpulse>
copumpkin: i get that part of your drift is no one but me knows what the hell this madness is I'm doing :)
<copumpkin>
I'm saying that your goal isn't just to get sensible cross-compilation into nixpkgs; it's to make sure the rest of us who maintain all sorts of bits and pieces of nixpkgs affected by your changes are able to maintain it without breaking your stuff
<copumpkin>
if you just merge it all in
<copumpkin>
we'll either break it because we don't understand it
newhoggy has joined #nixos
<copumpkin>
or people will get frustrated and work around it
dbmikus_ has joined #nixos
<Sonarpulse>
copumpkin: ah
<copumpkin>
don't just throw it over the wall, is all I'm saying
<Sonarpulse>
i guess part of my idea is once i do all my changes
<copumpkin>
you (should) want us to understand it
<Sonarpulse>
it will be less ambiguous what one is suppose to do
<Sonarpulse>
not because it is easy to understand
<Sonarpulse>
but because it is easy to cargo cult
<Sonarpulse>
whereas before i found it hard to cargo cult
<Sonarpulse>
so i do try to document everything in that manual section
<Sonarpulse>
but I'm also optomistic that the more i merge the *less* I'll have to worry about breakage
<copumpkin>
ok :)
sigmundv has quit [(Quit: Leaving)]
newhoggy has quit [(Ping timeout: 260 seconds)]
puffnfresh has quit [(Ping timeout: 258 seconds)]
sigmundv has joined #nixos
kanzure has quit [(Ping timeout: 268 seconds)]
hhao has joined #nixos
oahong has quit [(Ping timeout: 240 seconds)]
puffnfresh has joined #nixos
newhoggy has joined #nixos
sigmundv has quit [(Client Quit)]
sellout- has quit [(Quit: Leaving.)]
sigmundv has joined #nixos
erasmas has quit [(Quit: leaving)]
newhoggy has quit [(Ping timeout: 268 seconds)]
JagaJaga has joined #nixos
ryanartecona has joined #nixos
simp__ has joined #nixos
byteflame has joined #nixos
newhoggy has joined #nixos
kanzure has joined #nixos
sigmundv has quit [(Read error: Connection reset by peer)]
newhoggy has quit [(Ping timeout: 260 seconds)]
drasich has quit [(Ping timeout: 260 seconds)]
griff_ has joined #nixos
newhoggy has joined #nixos
sigmundv has joined #nixos
kanzure has quit [(Ping timeout: 260 seconds)]
sdll has joined #nixos
newhoggy has quit [(Ping timeout: 260 seconds)]
taktoa has quit [(Remote host closed the connection)]
<sdll>
Team, could you please tell me how I can add a file to the package directory in the /nix by pointing a path to the file in configuration.nix? For example, I would like to add a new encryption template to wicd. To achieve that, I need to add a file to the directory /nix/store/xqyqgwcs3y26srcnwrcp4wlss5wawhav-wicd-1.7.2.4/etc/encryption/templates . Could yo please tell me how this can be achieved?
<sdll>
maiklein has joined #nixos
<sdll>
What is the best way to reference to the install directory of a package in configuration.nix?
<simpson>
sdll: Would it be okay to add the file to /etc/encryption/templates instead? /etc can be managed somewhat from configuration.nix.
<simpson>
maiklein: You might find it easier to write a Nix expression for the package that you are trying to build.
<simpson>
And then using that expression for development, deployment, etc.
<sdll>
simpson: Thank you very much for your reply! I did not know that I could do that -- could you please tell me how I can add a file to /etc then from configuration.nix?
<maiklein>
I also need xcb for my own libraries, I guess I could also build them with nixos but I am not sure if I want to do that yet.
<maiklein>
This is my first day with nixos
<simpson>
maiklein: It's okay, you'll get used to it.
<simpson>
sdll: Looking for docs.
markus1209 has joined #nixos
AllanEspinosa has quit [(Ping timeout: 260 seconds)]
JagaJaga has quit [(Ping timeout: 240 seconds)]
markus1219 has joined #nixos
newhoggy has quit [(Ping timeout: 260 seconds)]
kanzure has joined #nixos
<sdll>
simpson: Thank you for your time! Let's suppose, for convenience sake, that I have a new encryption template in ~/.templates/new_template
<sdll>
I wonder whether there is any way to execute shell script creating the necessary substructure in /etc and then doing ln -s from ~/.templates/new_template to the correct place in /etc
<simpson>
$(man configuration.nix) is that same information FWIW.