<kalbasit_>
worldofpeace: where did you put the rev at exactly?
<kalbasit_>
(noob here, sorry)
<worldofpeace>
You'll have to override that as well.
<worldofpeace>
And at line 29
<kalbasit_>
oh
lassulus_ has joined #nixos
<samueldr>
worldofpeace: don't want to seem harsh, but you sure this will build 0.25.2? `version` is already 0.25.2, and while locally I get a 0.25.2 URL from github, it seems to still not result in 0.25...
<worldofpeace>
samueldr: No :P This piece of software has a lot going on I think.
<worldofpeace>
samueldr: You're not harsh
<samueldr>
I used `${builtins.trace version version}` and it does look as expected
<samueldr>
which makes me think there's maybe something with how go packages are built
* samueldr
tries something dumb
<worldofpeace>
samueldr: Totally
<worldofpeace>
deps?
lassulus has quit [Ping timeout: 264 seconds]
lassulus_ is now known as lassulus
<samueldr>
okay, so using *something else* than minikube as `src` doesn't build (as I said, dumb) so it's somehow still using the minikube checkout at the right tag (presumably)
<worldofpeace>
when you say 'right tag' you mean the one we want or the other?
<samueldr>
seemingly downloads 0.25.2
<Myrl-saki>
samueldr: Ping. I have a nixpkgs-18.03 which might work for 3.0 kernels.
<samueldr>
Myrl-saki: interesting, with cross-builds there's not much caching, so it may be right to have an additional overlay for this
<samueldr>
I don't have a samsung device on-hand, and they use a custom bootloader, so I can maybe help you port my project to it, but there's a lot of unknowns
<samueldr>
though the good news is that armv7 is tested to boot
<samueldr>
(as in, android boot.img generates properly and boots)
<worldofpeace>
kalbasit_: I think what I've told you works aside that you need to also override the localkube.
<Myrl-saki>
samueldr: Nice. I also got pmOS to boot. I think I've said this yesterday?
<kalbasit_>
worldofpeace: yea I just done that, it still reports the 0.26.0 version though.
<kalbasit_>
I'll just have to test it out locally and see what version of kubernetes will I get with it
<samueldr>
Myrl-saki: yes, and that's an important point to have done as you have a known working state
<Myrl-saki>
samueldr: Yep.
<worldofpeace>
kalbasit_: who and what is reporting 0.26.0?
<Myrl-saki>
samueldr: I'll message you once I have more free time.
<samueldr>
Myrl-saki: so it's possible to build the kernel and transplant it into the working pmOS thing
<samueldr>
and vice versa, test a nix-built initrd with a known working kernel
<{^_^}>
[nixpkgs] @ldesgoui opened pull request #42860 → murmur service: prevent failure launch by waiting until network is available → https://git.io/fignh
jb55 has quit [Ping timeout: 265 seconds]
<samueldr>
I'm not 100% positive I understand, but it looks like it's mostly a way to allow overriding parts of the gnome set?
Drakonis has quit [Remote host closed the connection]
<worldofpeace>
Let's say I wanted to add another package set to nixpkgs. Like how? :D
<cheshircat>
Hey, is there any documentation for getting ocaml packages packaged with nix, or should I just fumble around with stuff that exists in nixpkgs? The package I'm trying to add is patoline, which isn't on opam and build with a makefile, but has ocaml dependencies
andymandias has quit [Ping timeout: 260 seconds]
<cheshircat>
I've managed to compile it statefully with opam, but I'm having trouble getting it to work with the nixified ocaml dependencies
<kalbasit_>
to me it is lol. Been too wrapped up in Arch world
<samueldr>
though it's not new, it doesn't have the mindshare some feel it should deserve :)
<kalbasit_>
there's quite the learning curve to it
<kalbasit_>
but I think it's worth the time and effort
<samueldr>
I'm not sure there's anyone that really put a moderate amount of effort that's really gone back out
<kalbasit_>
the worst thing for me is when I get to work and " Good morning, oh shit my system is broken again, I'll be hiding in a conf room for a few days :@"
<kalbasit_>
samueldr: true
<samueldr>
I just loved it how I was able to get back working in under an hours when my SSD failed last year
pikajude has quit [Remote host closed the connection]
<samueldr>
(not catastrophically, but enough to need to use my spare laptop)
<samueldr>
clone config, get right project files, continue working
pikajude has joined #nixos
andreabe_ has joined #nixos
andreabedini has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @Infinisil opened pull request #42861 → [WIP] Idris packages clean ups and updates → https://git.io/fPJl4
<Lisanna>
ugh, hydra still doesn't support git submodules?
<Lisanna>
oh well, was going to need to hack on hydra myself anyways
jperras has quit [Ping timeout: 240 seconds]
<elvishjerricco>
Lisanna: I have a project using git submodules that hydra builds just fine
<Lisanna>
elvishjerricco strange, I don't see any commits in hydra that would have fixed this issue
<Lisanna>
elvishjerricco "/nix/store/k0704k0sbd6a0vigd5f487ksb73plmyr-hydra-2017-11-21/bin/.nix-prefetch-git-wrapped: line 160: cd: too many arguments"
graphene has quit [Remote host closed the connection]
graphene has joined #nixos
knupfer has quit [Ping timeout: 256 seconds]
sigmundv has joined #nixos
joehh has quit [Ping timeout: 265 seconds]
kiloreux has joined #nixos
tzemanovic has quit [Remote host closed the connection]
<kiloreux>
I am getting "error: while evaluating the attribute ‘pkgs’ of the derivation ‘group_packages’"
<kiloreux>
"cannot convert a function to JSON"
<kiloreux>
Does anybody know what this mean ?
tzemanovic has joined #nixos
patrl has joined #nixos
tzemanovic has quit [Ping timeout: 268 seconds]
patrl has quit [Client Quit]
<Lisanna>
is there a reason why there aren't any uses of releaseTools.sourceTarball that don't use the nixpkgs built-in distribution phase (they all define their own)?
<srhb>
kiloreux: Seems fairly clear to me. You're trying to convert a function to JSON somewhere.
robstr has quit [Read error: Connection reset by peer]
palo_ has joined #nixos
palo_ has quit [Changing host]
palo_ has joined #nixos
palo has quit [Ping timeout: 276 seconds]
<kiloreux>
Not really for me. I have juist copied the nix expressions from the nixpkgs repo.
<srhb>
kiloreux: Perhaps you could pastebin what you're doing
<srhb>
kiloreux: And php56 and php56Packages aren't paths that are generally linkable.
<srhb>
kiloreux: Are you trying to get a filesystem tree of every php package and php as well?
<kiloreux>
Yes, that's what I am trying to do.
<srhb>
kiloreux: It's gonna be a fair bit more complicated than what you're doing. Essentially you'd want to use php and phpPackages from your imported nixpkgs, not some out of tree inclusion.
<kiloreux>
srhb, Do you have any example on how I could achieve that ?
<srhb>
kiloreux: But then, you'd want paths = [ php ] ++ phpPackages. However, phpPackages is an attrset, not a list of paths. So you'd need to convert it to a list of paths first. But then quite a few of those attributes don't even evaluate (at least not with php5) so you're going to have to filter them out first.
<srhb>
kiloreux: No, it's quite a bit of work
<srhb>
But that's some of the steps
<kiloreux>
I see. Thank you srhb <3
<srhb>
kiloreux: Honestly, there's not that many php packages, you might be better off including the ones you want explicitly.
genesis has quit [Remote host closed the connection]
<srhb>
Lisanna: No, that will lose you the other attributes. Do what clever suggested with // :)
<Lisanna>
srhb oh right, 'cause that'll make vmTools and rpmDistros an otherwise empty attrset
<Lisanna>
so I have to merge them manaully
<srhb>
Lisanna: Right :)
<Lisanna>
Ok, so a bit more complicated, but doable!
<Lisanna>
and much better than maintaining my own patches on top of nixpkgs
<srhb>
For sure.
genesis has joined #nixos
<Mic92>
What channel does travis' nix image use by default?
<booglewoogle>
hey guys, still trying to build https://github.com/FasterMelee/Ishiiruka with "nix-build -E '(import <nixpkgs> {}).callPackage ./fastermelee.nix {}'" on this nix-expression https://pastebin.com/cgCSRx6A, getting this error (at 100%! :( ). Another wall I've hit and can't seem to find useful ideas in the net. anyone got a hunch?
asuryawanshi has quit [Remote host closed the connection]
<srhb>
theo: https://nixos.org/nixpkgs/manual/#chap-overlays -- essentially overlays allow you to add/modify/remove packages in your nixpkgs without including them in the nixpkgs proper tree.
<srhb>
theo: They are lists of functions that modify nixpkgs.
<theo>
Mic92: it's still very basic so I'm not looking to share it just yet
<theo>
srhb: thanks, I'll take a look. I did much reading of the manuals but didn't find that page in particular
<srhb>
theo: trivially, ... self: super: { myHello = super.hello }; -- is an overlay that adds a myHello package that is an exact duplicate of the hello package
<srhb>
theo: Definitely worth a read :)
<theo>
srhb: my goal at the moment is to have windowManger.<my window manager name>.enable = true; in my /etc/nixos/configuration.nix and have X use my window manager
<theo>
will overlays be able to achieve this?
tzemanovic has quit [Remote host closed the connection]
<clever>
theo: overlays cant set nixos options, only make changes to packages in nixpkgs
<Mic92>
theo: then you need to define your own module, which is also possible.
<tilpner>
Mic92 - How does the locking work? Does it require a PR for every update, or does it update periodically? (Or does it not use the locks at all?)
tzemanovic has joined #nixos
<theo>
Mic92: I've adapted 2bwm's 2bwm.nix module for my own wm, but I don't know how to include my module in my system
<Mic92>
tilpner: at the moment I use travis's cron feature, that update all repositories that evaluate properly
ij^ has joined #nixos
<Mic92>
However I am looking for a push-based approach, that updates faster
<tilpner>
Mic92 - Ah, I see. You'd still run into the fetchurl ttl, but I guess 1h is fine-ish
ericsagnes has quit [Ping timeout: 240 seconds]
newhoggy has quit [Remote host closed the connection]
<clever>
theo: add it to the imports list in configuration.nix
<ij^>
I can't use the ruby's "curb" gem installed with a "bundle install" inside a "nix-shell -p curl", because of a "symbol __resolv_context_get_preinit, version GLIBC_PRIVATE not defined in file libc.so.6 with link time reference" when I run the rails server. Full error: https://clbin.com/VkxKJ
<clever>
theo: everything you put into imports is a nixos module
<tilpner>
Mic92 - What's the current interval for that travis cron?
<Mic92>
tilpner: unfortunaly a day, that's why I want to migrate to different option. But I first wanted to get something working.
<theo>
clever: ah thanks for the tip, I'll try that now. I read what you said about overlays on the wiki as well, thanks for the clarification
<Mic92>
this is the smallest interval supported
<Mic92>
I could probably use the API to increase this
tzemanovic has quit [Ping timeout: 248 seconds]
<tilpner>
Mic92 - Why does it use fetchGit?
<{^_^}>
[nixpkgs] @alyssais opened pull request #42866 → Fix buildPythonPackage example → https://git.io/f1Dnd
<tilpner>
(As opposed to fetchTarball or fetchFromGitHub)
<Mic92>
tilpner: with locking did you mean the individual repositories or the NUR repo itself?
<theo>
clever: I remember the issue now - I have manager to package and install my window manager using nix-env so my user has access to it
<theo>
clever: but it is not accessible globally and my configuration.nix doesn't know about it
<theo>
clever: any ideas how to let the package be known globally?
<tilpner>
Mic92 - The individual repos. I was curious about the time that passes between me pushing something to my repo, and being able to talk about nur.repos.tilpner.* on IRC
<theo>
clever: I tried cloning nixpkgs, adding my package and using nixos-rebuild switch -I nixpkgs=/path/to/my/nixpkgs but that threw up all sorts of errors and I don't want to maintain a local copy of nixpkgs
<srhb>
theo: package is solved by overlay, module with imports.
xy2_ has quit [Ping timeout: 256 seconds]
<tilpner>
Mic92 - Yes, please don't lock it into GitHub any more than necessary. It should support non-GH URs
<tilpner>
(higher latency by default is okay)
<tilpner>
Mic92 - Why does the packageOverrides snippet use fetchGit? (As opposed to fetchTarball or fetchFromGitHub)
<theo>
srhb: I read the man page you linked but I'm not sure I've quite got my head around it
<theo>
srhb: I can't see how this would help me package the wm
<Mic92>
tilpner: I thought it updates faster, also I did not benchmarked it.
<ij^>
I maybe figured out why — ruby's installed with an older glibc into $HOME/.rbenv/…, but I've updated nixos, so bundler now takes a fresher one from system.
<tilpner>
Mic92 - Unless it keeps a cache of bare git repos somewhere, I don't see how. I'd expect fetchFromGitHub to be faster
<Mic92>
tilpner: builtins.fetchGit does have a cache afaik.
<LnL>
yes, but it's not ideal for multiple repos
pie_ has quit [Ping timeout: 240 seconds]
newhoggy has joined #nixos
<Mic92>
NUR only promotes fetchGit for NUR itself, not the including repositories
<tilpner>
Mic92 - Still, the .tar.gz of master is 12K currently, the caching wouldn't help much
<tilpner>
(But I have not timed it, so who knows)
<tilpner>
(In a quick test, fetchTarball feels faster)
<tilpner>
And I feel like UX could greatly benefit from allowing for fetchTarball within repos.json as well
<Guanin>
Hi, I've got an application that requires files in a relative directory to the binary. Would it make sense to just create a symlink for that files out of the nix storage? (So that the derivation contains the binary and a symlink `config`, which points e.g. to /var/lib/foo_config)
freeman42x]NixOS has quit [Ping timeout: 256 seconds]
<Mic92>
tilpner: I would prefer to have some kind of version control on the downstream side.
<tilpner>
Mic92 - What's the advantage of that?
<Mic92>
transparency
<booglewoogle>
hey, trying another route now with a nix-shell to be able to granulate the build steps. now i'll get this error https://pastebin.com/rsX4WSed (Xrandr not found -- Checking for module 'xi>=1.5.0' -- No package 'xi' found) with this default nix https://pastebin.com/wR4kfpYX , which includes xorg.libXrandr. anything wrong about the default.nix so it can't find the library?
<Mic92>
and going back in time
<tilpner>
Mic92 - Going back in time can be prevented with git repos too
<Mic92>
I cannot prevent people from force pushing, but I don't expect many people from doing that.
<tilpner>
Unless you make that part of the verification process, then I see the benefit
<tilpner>
E.g. "new revision must be child of old revision"
<Mic92>
`git bisect` should be possible in most cases
<Mic92>
This is not about security, but safety
<tilpner>
I don't expect many people to use the tarball route for untransparency either
<tilpner>
There should always be a version-controlled repo on the other side (but not always git), I just don't see the advantage of fetching that repositories when I could just fetch a checkout
<Mic92>
NUR fetches a checkout.
<tilpner>
By fetching the repository, no?
<Mic92>
for github it fetches the zip archive.
<tilpner>
Oh
Guest92966 is now known as fgaz
newhoggy has quit [Remote host closed the connection]
<Mic92>
But there is no generic way to resolve version control systems to archive urls.
<tilpner>
That's why I wondered about allowing users to provide archive urls directly
<tilpner>
But I didn't see your github.com optimisation
ThatDocsLady has quit [Remote host closed the connection]
<LnL>
not just zip files
ThatDocsLady has joined #nixos
knupfer has joined #nixos
<Mic92>
I accept support for other methods as pull requests, but I will probably not implement them myself. There are other items higher in the priority list.
lstanley has quit [Ping timeout: 248 seconds]
Neo-- has joined #nixos
zimbatm_ has joined #nixos
zimbatm_ has quit [Client Quit]
joehh has joined #nixos
<Lisanna>
with the releaseTools.rpmBuild function & friends, is there a way to deal with building an RPM that depends on another RPM that you have built with the same mechanism?
__Sander__ has joined #nixos
<Lisanna>
diskImageFun takes an extraPackages argument, but it's a list of strings that refer only to the distro's normal package distribution
ekleog has quit [Quit: WeeChat 2.0]
lstanley has joined #nixos
<Lisanna>
Oh, looks like I can just wrap the base distro disk image with fillDiskWithRPMs
<Lisanna>
:D
<Lisanna>
this is too easy...
kiloreux has joined #nixos
init_6 has joined #nixos
graphene has quit [Remote host closed the connection]
newhoggy has joined #nixos
graphene has joined #nixos
graphene has quit [Read error: Connection reset by peer]
andymandias has quit [Ping timeout: 264 seconds]
graphene has joined #nixos
<ij^>
Why is it that after having no gcc, "nix-shell -p hello --run 'gcc --version'" shows a gcc present?
<clever>
ij^: nix-shell uses stdenv.mkDerivation, which provides a gcc
graphene has quit [Remote host closed the connection]
<ij^>
Ah. Well, I need to pass gcc6 to ruby-build scripts, but -p gcc6 doesn't work.
andymandias has joined #nixos
graphene has joined #nixos
<clever>
ij^: it may help to look at how gcc7Stdenv is defined
<clever>
the gcc6Stdenv is missing for some reason, but its trivial enough to just make it on the fly
<booglewoogle>
the nixos beauty is starting to unfold for me
ericsagnes has joined #nixos
philippD has joined #nixos
Crackville has joined #nixos
Crackville has left #nixos [#nixos]
<booglewoogle>
trying my luck again: does anybody have an idea as to how this error https://pastebin.com/rsX4WSed can be possible when building from this nix-shell https://pastebin.com/wR4kfpYX (xorg.libXrandr included, from how I understand the nix-shell system so far)?
liuyb has quit [Ping timeout: 260 seconds]
<clever>
booglewoogle: try adding pkgconfig to the buildInputs
<tilpner>
Mic92 - You mention sharing modules. Do you have an example of using them?
nschoe has joined #nixos
<tilpner>
Mic92 - Would nur.repos.*.modules be an attrset of paths or an attrset of functions? (Only paths seem to work, but your example imports them)
jperras has joined #nixos
<tilpner>
(I see the example in your readme, but I wonder if it works)
wgas has quit [Quit: Leaving]
b has joined #nixos
jperras has quit [Ping timeout: 248 seconds]
pie_ has joined #nixos
ekleog has joined #nixos
<Mic92>
tilpner: it works
<tilpner>
Huh. I get infinite recursion even for an empty module
<booglewoogle>
phew. now getting this error https://pastebin.com/4dzJ6J2M (gtk libs not found) - still when building in a nix-shell with this default.nix https://pastebin.com/hzYTSe1n , which includes all sorts of gtk libs I could find in nixpkgs. :(
<tilpner>
infinisil - With what value though? If we use callPackage, doesn't nur depend on pkgs too?
<theo>
well now that I've accidentally pasted it I may as well ask: I've seen this as syntax for adding a custom packages with overlays, but how would this be included in a configuration.nix?
<infinisil>
Yeah specialArgs is the only thing that would work probably
<tilpner>
infinisil - The alternative is imports = [ ((import <nixpkgs> {}).callPackage ../nur {}).modules.dummy ];, but that's ugly and fragile
<infinisil>
And it doesn't use your overlays
blahdodo has joined #nixos
<tilpner>
It does here
<tilpner>
But yes, probably not portable
<infinisil>
the ones in nixpkgs.overlays I mean, which however is the reason this whole thing gives inf rec to begin with
<kiloreux>
If I have a custom installPhase the postInstall step is never executed. Is this by design ?
<tilpner>
infinisil - I don't set nixpkgs.overlays, so that's okay
<infinisil>
tilpner: It still needs to traverse all modules to notice that though, but it can't when a module depends on pkgs itself
<tilpner>
Right
<infinisil>
pkgs needs evaluated nixpkgs.overlays, which needs to traverse all modules to collect all overlays, which would need to evaluate pkgs to get the NUR module, etc.
<infinisil>
It's unfortunate..
<tilpner>
kiloreux - Add "runHook postInstall" to the end
<kiloreux>
Thank you tilpner
<tilpner>
It's not just overlays, also nixpkgs.localSystem
<tilpner>
(And config of course)
newhoggy has joined #nixos
<{^_^}>
[nixpkgs] @Ma27 opened pull request #42871 → ocserv: init at 0.12.1 → https://git.io/fDeHb
<infinisil>
Ah right
ThatDocsLady has quit [Remote host closed the connection]
ThatDocsLady has joined #nixos
theo has quit [Quit: Lost terminal]
<Mic92>
tilpner: packagesOverides
<aminechikhaoui>
hm how to tell `nix build -f build.nix --builders <foo>` to try to substitute stuff using the target builder ?
<Mic92>
tilpner: It only worked for me by accident because I was using let nur = (import (builtins.fetchGit { url = "https://github.com/nix-community/NUR"; })) {}; in
<Mic92>
that in turn does triggers an import in the default.nix: { pkgs ? import <nixpkgs> {} }:
<Mic92>
I am not sure if there is a way that does not require that import
mariatsji has quit [Ping timeout: 265 seconds]
pie_ has quit [Ping timeout: 265 seconds]
mariatsji has joined #nixos
<nschoe>
acowley, hi, just wanted to thansk for your help on setting ROS with nixOS. I have been running it inside a docker container, as you suggested, shared the X11 socket. This seems to works perfectly. At least for learning :)
<nschoe>
acowley, also I was trying to find a Haskell package and I saw your name on `roshask` so I might bother you again in the future 'bout that =)
<Mic92>
I have removed the incorrect documentation for nixos modules for the moment.
<booglewoogle>
can someone explain to me the difference between building this nix-expression https://pastebin.com/BtY5x44v with "nix-build -E '(import <nixpkgs> {}).callPackage ./fastermelee.nix {}'" and building the same release by hand in this nix-shell https://pastebin.com/5gmDAxgd ? from what I understand, they do the same thing, they fail at very different steps though:
<booglewoogle>
- the former reaches 100% and then can't find "/tmp/nix-build-FasterMelee.drv-0/source/include" (is there a way I can edit the CMakeLists.txt before the build starts when using nix-build?)
<booglewoogle>
- the latter can't find the gtk-libraries, using the same inputs as the former though?
adamt has joined #nixos
<symphorien>
nix-build runs in a sandbox, usually
<adamt>
Hiya. Is it possible to just calculate the output path, without actually building anything?
<symphorien>
to edit files, use a patch or sed the files in postPatch for example
<symphorien>
adamt: nix-build "<nixpkgs>" -A hello --dry-run
sellout- has joined #nixos
<booglewoogle>
hum, okay. thanks!
andymandias has quit [Ping timeout: 240 seconds]
<tilpner>
Mic92 - It might be possible to provide a path to a modules attrset using just builtins and copied functions from lib
<shlevy>
Any rust users around?
<tilpner>
Yes-ish
<srhb>
adamt: or eval the outPath attribute
<adamt>
symphorien: What about a more nix 2.x way? :P (also, that does seem to work for hello, but when building a local nix expr it doesn't print the output path when using --dry-run for some reason. Hmm.)
<adamt>
srhb: Aha.
<Mic92>
tilpner: if you have a better solution, I would appretiate a pull request. At the moment we can still break things.
<shlevy>
Running into a problem building 'ring' with carnix on macOS, some kind of unsupported asm directives, and I'm pretty lost on what the issue is... upstream docs suggest macOS is supported
<adamt>
srhb: Is it a toplevel attr of the result?
sellout- has quit [Ping timeout: 248 seconds]
<srhb>
adamt: It's a top level attribute of the derivation, usually
<Mic92>
shlevy: we also had problems with some sse tests in rustc.
<tilpner>
Mic92 - It would be a special-case for nixpkgs-less things, and it adds some complexity to implementation and usage. Answering the question of whether the usecase justifies that, is up to you. I can try locally though
<Mic92>
LnL: I did not had a further look into it.
<LnL>
no, but it failed because it's only allowed in nightly builds
<srk>
nschoe: *would
<Mic92>
If you can make progress on it, just push on my branch. I probably will not make any progress with it this week.
jperras has quit [Ping timeout: 264 seconds]
<efx[m]>
What command must I execute to install a package exported by nix copy --to tmp/mycache? I need to do this manually as the servers do not have network access to each other.
winem_ has joined #nixos
<efx[m]>
additionally, does nix copy --to /tmp/mycache create a local binary cache?
<LnL>
Mic92: I'll probably be busy with darwin stuff, but if that bootstrap issue is fixed I'll can try building without the simd patch
<tilpner>
Mic92 - Dependency on nixpkgs is solveable, but installation as pkgs.nur or nur is not, at least not without adding to specialArgs. And I don't know a way to do that, that could reasonably be added to your README
<tilpner>
(Installation + usage imports)
<infinisil>
If I ever become ruler of the world I'm going to ban infinite recursion
<infinisil>
You'll have to write all programs in a total programming language and prove that there can't be infinite recursion
<Mic92>
tilpner: the dirty trick could be (with import <nixpkgs> {}; pkgs.callPackages (...))
<LnL>
efx[m]: there's a difference between --to /tmp/foo and --to file:///tmp/foo, the first is a store while the second is a binary cache
<tilpner>
Mic92 - My first thought after seeing NUR was to share hacks/not-quite-good-enough-for-nixpkgs modules
<efx[m]>
ooh, iLnL: ah, thanks for the
<efx[m]>
clariclarification
<Mic92>
tilpner: my intention was to reduce the maintainance efforts that has to be centralized in nixpkgs.
<Mic92>
and to give experiments better discoverability.
<tilpner>
Sure, that goes for the packages side, and it will fulfill that purpose
<tilpner>
But it might not be easy to use for sharing modules
shiver32 is now known as shiver
ihar has joined #nixos
pie_ has joined #nixos
<tilpner>
Even tx0.co/1l inf-recs. Someone please correct me, but I think that'd be required for imports = [ nur.* ]; to work
<shlevy>
Anyone know if P-E-Meunier on github is anyone here? :)
<goibhniu>
pemeunier looks promising :D
<shlevy>
:) thanks
<tilpner>
Note that [pemeunier] idle: 9 days, 07 hours 12 minutes 16 seconds, signon at: Sat, 23 Jun 2018 09:32:20
<LnL>
pretty sure I've talked to him in here
<shlevy>
pemeunier: Looks like CARGO_CFG_TARGET_OS is getting set wrong on macos (it gets set to "darwin" instead of "macos"). Looking into the docs for that var now, just wanted to let you know and see fi you have any thoughts on the right way to figure it out for all platforms
<{^_^}>
[nixpkgs] @dotlambda pushed commit from @timokau to master « pythonPackages.backports_shutil_get_terminal_size: 1.0 -> 2016-02-21 (#42840) »: https://git.io/fy1dS
mroutis has joined #nixos
<clever>
aminechikhaoui: the goal is to get 128gig free, and keep anything that remains, so it can still act as a cache
<clever>
aminechikhaoui: but the new min-free and max-free in nix.conf work better, doing a larger gc less often, rather then a small gc every day at the same time
<mroutis>
anyone using nixpkg with ruby/ruby on rails? I'd love to use nix to manage my ruby versions, but I'm afraid to face a lot of issues with bundix (never tried it, tho), any recommendations?
<LnL>
clever: have you used those
<aminechikhaoui>
clever: ok thanks
andymandias has joined #nixos
<clever>
LnL: i have them enabled on 2 or 3 of my machines
<clever>
LnL: import-from-derivation sometimes triggers a GC and eats stuff the build needs, but hasnt computed the path to yet
<clever>
LnL: it has also caused builds to fail randomly, by deleting something thats in use by a building derivation
<clever>
aminechikhaoui: oh, min-free also lets you go a lot lower, because it will auto-gc at the start of any action on the store, so you dont need to keep enough free to survive the next 24 hours
<clever>
aminechikhaoui: but you still need enough free to complete max-jobs in parallel
<aminechikhaoui>
clever: ok I'll probably experiment with that
jensens has quit [Ping timeout: 276 seconds]
<manveru>
mroutis: while i'm admittedly biased, why don't you try bundix? :)
<{^_^}>
[nixpkgs] @marsam opened pull request #42876 → recoll: Add support to build without QT → https://git.io/fyFDw
crmlt has quit [Ping timeout: 248 seconds]
<mroutis>
manveru: <3 didn't expected to find a contributor; being honest, I don't feel prepared to "solve"/"debug" any possible errors during the transition (bundler -> bundix) phase
wchresta has quit [Ping timeout: 264 seconds]
<mroutis>
also, I haven't heard of any member of the rails community using bundix, so I felt kind of pushing to far outside the classic "convention over configuration"
<mroutis>
I'll give it a try, now that I don't feel it like an isolated effort :P
<manveru>
well, there are members of the nix community using rails... if that counts :)
<nschoe>
Hi all, just to confirm: the package 'autoconf' should not be installed in the user's profile right? When it's needed for dev, it should be put into a shell.nix and sourced, am I right?
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
<LnL>
yes, imho
<cocreature>
Hey, I am trying to use callCabal2nix in combination with fetchFromGitHub but it still seems to apply revisions from hackage. is there a way to disable this?
Mrmaxmeier has quit [Remote host closed the connection]
Mrmaxmeier has joined #nixos
eren has joined #nixos
<Lisanna>
Ugh, need some help with overlays
<Lisanna>
I'm trying to overlay an additional distro on vmTools.rpmDistros so that I can get it to show up in diskImageFuns (diskImageFuns is an attrset produced from the contents of rpmDistros). Unfortunately I have my overlay structured like this:
<Lisanna>
which means that I'm more or less just overlaying a new rpmDistros on top of the old one inside the old contents of vmTools, so diskImageFuns is always going to be the old one!
<Lisanna>
and if I try to grab diskImageFuns from self.vmTools I get an infinite recursion ):
<Lisanna>
diskImageFuns has a direct dependency on the value of rpmDistros
<Lisanna>
wait, do I need to explicitly override stuff in overlays for that to work? i.e., is it always going to just pick up the old values for all attributes?
<Lisanna>
re-reading the overlays example and that's looking like what's going on here
<Lisanna>
...oh, overlays are probably only applied one layer deep
<Lisanna>
so if I override something at the top-level (vmTools) it will appear as changed in self, but overriding stuff below the top level won't propagate the effects the same
<{^_^}>
[nixops] @AmineChikhaoui merged pull request #921 → Ignore evaluation errors in destroy and delete operations, fixes #917 → https://git.io/vxbtQ
<{^_^}>
[nixops] @AmineChikhaoui pushed 2 commits to master: https://git.io/fSORr
<mpickering>
Right so nixpkgs.vmTools.diskImageFuns is the old version
<mpickering>
Because you don't modify it at all
<mpickering>
why are you not trying to use nixpkgs.diskImageFuns?
<Lisanna>
mpickering what do you mean?
<mpickering>
You don't modify `vmTools.diskImageFuns` at all with your overlay
<Lisanna>
mpickering oh, I mean, I am considering doing that, but I'd have to more or less copy and paste the existing implementation
<Lisanna>
and swap out the references
<Lisanna>
I want to add something to rpmDistros and have the change automatically propagate itself to vmTools.diskimageFuns, like it would if I wasn't trying to manipulate something two layers deep
<Lisanna>
since this would work fine for package dependencies at the top layer of nixpkgs
<Lisanna>
so, trying to think of possible solution that don't involve copying and pasting the existing code for vmTools.diskImageFuns and tweaking it slightly
<mpickering>
The problem here is how the vm/default.nix is designed
<Lisanna>
yeah, I guess there's no really easy answer here
<Lisanna>
since it's a recursive attrset
<Lisanna>
there's not really a way to "overlay" stuff on a basic dumb recursive attrset
<mpickering>
If instead the reference was self.rpmDistros
<Lisanna>
and have that change the evaluation of other stuff in that attrset automatically
<Lisanna>
yeah
<mpickering>
and there was some kind of fixpoint then what you are doing would work
<mpickering>
I don't know how to set that up immediatley
<Lisanna>
so vm/default.nix would need its own overlays system more or less
<Lisanna>
in order for me to be able to get what I'm trying to achieve here
<ldlework>
Are those really kernel-managed containers though?
matthewbauer has joined #nixos
<Lisanna>
ldlework they're virtual machines
<Lisanna>
built on top of the magical vmTools.runInLinuxVM derivation
<Lisanna>
oh well, thankfully it's only a few lines of copying and pasting
<ldlework>
oh crap
<ldlework>
I was scrolled up... way WAY up.
<ldlework>
lol
<Lisanna>
^^
<mpickering>
perhaps if you add a "self" parameter
<Lisanna>
mpickering is there a better way that doesn't involve directly modifying nixpkgs src?
<mpickering>
Don't think so
booglewoogle has quit [Ping timeout: 245 seconds]
<mpickering>
Looks to me that recursive attribute sets should be avoided really
aarvar has joined #nixos
<Lisanna>
mpickering hey, it's better than if half of this stuff was in a let statement
<Lisanna>
then I'd be completely screwed
<Lisanna>
cool, it worked!!!
jperras has joined #nixos
<mpickering>
Lisanna: I feel like the nix language is too low level, it should help people more do the right thing
<mpickering>
good it works for you
matrium has joined #nixos
<Lisanna>
yeah, it might be nice if recursive attrsets implicitly came with some abilities to override their contents in a way that is meaningful for affecting its self-references
<ldlework>
Nixlang is pretty intense as far as learning goes.
<mpickering>
You mean like an object orientated language Lisanna :) ?
<Lisanna>
lol
<Lisanna>
I'm sure guix has a solution to this problem...
<mpickering>
I already knew haskell well so the fact that it's functional and relies on fixpoints actually made it easier for me
<Lisanna>
nix is honestly my first functional language... I learned about fix points and let statements from their implementations and usage in nix
jperras has quit [Ping timeout: 255 seconds]
<mpickering>
I'm serious.. the denotational semantics of object orientated languages end up looking a lot like how nix is designed
Sonarpulse has joined #nixos
<Lisanna>
yeah
<Lisanna>
attrsets become sort of these ad-hoc immutable objects with methods
<Lisanna>
stuff like python3.withPackages
<matrium>
hi, it seems like ghcjs is broken when compiling packages with primitive-0.6.4.0. Hence I'm trying to override the the primitive package to 0.6.4.0, following the instructions from Gabriel Gonzalez' nix-haskell tutorial. But I'm just getting the error "builder for '/nix/store/jjab8qnhfams3ws0f00i7kzb5xq624xd-primitive-0.6.3.0.drv' failed with exit code 1" and can't trace down the root
<mpickering>
matrium: paste what you have
<matrium>
*trying to override it to 0.6.3.0
<mpickering>
What is the error?
<mpickering>
It's possible it just doesn't work with 0.6.3.0?
<ldesgoui[m]>
hey, I pushed a PR but the eval bot complains about access to a gitlab .git URI being forbiddt during access mode, I'm using fetchGit with just the Url and the rev. Anything I can do to fix?
<mpickering>
the error is your project failing to build
<Nissim>
i am new to nix os ans would like some help please.
<tilpner>
With what?
<gchristensen>
go ahead and ask :) we're here to help!
<Nissim>
I am 16 and new to nixos, got this project to do for uni.
<Nissim>
how do i find the 256Sha for a repo on gitlab
<tilpner>
Try nix-prefetch-git
<tilpner>
,tofu
<{^_^}>
To get a sha256 hash of a new source, you can use the Trust On First Use model: use probably-wrong hash (for example: 0000000000000000000000000000000000000000000000000000) then replace it with the correct hash Nix expected.
<Nissim>
can we not do that from bash? i mean like doing this: shasum -a 256 code.tar.gz ?
<tilpner>
No, .tar.gz is not deterministic
<tilpner>
(But nix-prefetch-git can be used from CLI)
silver has joined #nixos
<Nissim>
how do i use nix-prefetch-git i cannot find the download link, you mind sharing please.
ajs124 has left #nixos ["Machine going to sleep"]
ryanartecona has joined #nixos
<tilpner>
nix-shell -p nix-prefetch-git
graphene has joined #nixos
<Nissim>
okie downloaded nix-prefetch-git, and doing building in a moment.. thanks Mr tilpner
griff_ has joined #nixos
leat has quit [Remote host closed the connection]
<gchristensen>
not safe to assume Mr around these parts :) best to just use nicknames and skip mr / mrs / ms
<mkaito>
worldofpeace: it's on staging waiting for a suitable world rebuild time.
<mkaito>
rebuild takes over a day on my beast workstation. Hydra is gonna have fun.
xy3_ has joined #nixos
contrapumpkin has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<tnks>
can't find builtins.hasContext in the docs.
xy2_ has quit [Ping timeout: 248 seconds]
<worldofpeace>
mkaito: Thanks, It looked like it was still waiting for changes to be made
<tnks>
what exactly is "context"?
<mkaito>
worldofpeace: no, FRidh just forgot to review his change request. probably busy or something.
<worldofpeace>
mkaito: definitely
wchresta_ has joined #nixos
<mkaito>
feel free to bump the PR or something. I've started to run my system off a fork where I cherry pick from PRs I like, just so I don't have to wait for nixpkgs.
contrapumpkin has joined #nixos
leat has quit [Remote host closed the connection]
wangoe has quit [Quit: WeeChat 1.9.1]
leat has joined #nixos
griff_ has quit [Quit: griff_]
<mkaito>
which is really only an option if you don't mind recompiling everything yourself, which I don't.
tzemanovic has joined #nixos
<infinisil>
tnks: I think I figured it out
<infinisil>
tnks: Strings can have a context, which is a set of derivations representing the dependencies of that string. Nix uses it to determine dependencies
chessai has joined #nixos
<infinisil>
tnks: So e.g. when I declare a string "cp ${pkgs.hello} $out", Nix will add the hello derivation to the strings context to know that if this string ever gets used, hello shall be a dependency of it
tzemanovic has quit [Ping timeout: 268 seconds]
<infinisil>
And you can't use strings with context in certain scenarios, e.g. in a derivation name
<{^_^}>
the string '188avy0j39h7iiw3y7fazgh7wk43diz1-hello-2.10' is not allowed to refer to a store path (such as '!out!/nix/store/c271aryv5jzm0aq640lb7gw0102id5f5-hello-2.10.drv'), at /nix/store/cc5d0bfpbw5r...
<infinisil>
But there's a builtin to discard the context:
<Guanin>
(asked it earlier already) I've got an application (phantombot) that requires files in a directory relative to the binary (not changeable without patching a lot). What is the best approach to build a package for this? put a symlink to a directory in /var/lib/ in that place?
<infinisil>
,context = Nix strings can have a context representing a set of derivations this string depends on. E.g. "foobar ${pkgs.hello}" has the hello derivation in its context, "foobar" doesn't have any context.
<{^_^}>
context defined
<worldofpeace>
mkaito: I've tried forking nixpkgs and having your own patch set on top of it. Always ended up having to compile something which I don't have time for.
ryanartecona has quit [Quit: ryanartecona]
<mkaito>
that's why it hasn't been merged
<mkaito>
half the world depends on mesa
<mkaito>
a mesa update causes a world rebuild, which takes a very long time
<mkaito>
the basic idea is to group up world-rebuild-causing patches so Hydra doesn't have to do the same monster rebuild more than once.
asuryawanshi has quit [Read error: Connection reset by peer]
asuryawa_ has joined #nixos
leat has quit [Remote host closed the connection]
leat has joined #nixos
redj has joined #nixos
griff_ has joined #nixos
ma27 has quit [Ping timeout: 240 seconds]
<infinisil>
Guanin: Yeah that's the best way I know of
Ross has quit [Ping timeout: 248 seconds]
nallar has joined #nixos
nallar is now known as Ross
<nh2[m]>
is there a command like `nix build nixpkgs.libsodium` that instead of creating a `result` symlink, it just prints the store path?
<symphorien>
Guanin: if it annoys you you can use libredirect but this is not the most beautiful way.
<nh2[m]>
actually what I want is to just get whatever `nix-env -iA` would install for a given package, its store path
leat has quit [Remote host closed the connection]
<Guanin>
Thanks infinisil and symphorien
leat has joined #nixos
<Guanin>
Of course, the best way would be to make that directory changeable, but that would probably be way to much work :/
<infinisil>
nh2[m]: --no-link
ryanartecona has joined #nixos
asuryawa_ has quit [Remote host closed the connection]
<infinisil>
Guanin: You could just use a defaulted argument to your package file
<Guanin>
Yes, that would probably be the way to go right now :)
<Guanin>
Thank you
* infinisil
checks if other packages do this, because he thought there were some
<nh2[m]>
infinisil: actually `nix build` doesn't seem to create `result/`, some other command I ran before must have created it
<nh2[m]>
infinisil: --no-link also doesn't print any store path
<nh2[m]>
in other way, is there a shorter way of doing this: nix-store --realise $(nix-instantiate --expr '(import <nixpkgs> {}).libsodium.dev')
<infinisil>
nh2[m]: nix-build --no-out-link '<nixpkgs>' -A hello
chreekat has quit [Ping timeout: 256 seconds]
<infinisil>
Yet another reason not to use the new nix command :P
<nh2[m]>
infinisil: perfect!
<infinisil>
Guanin: There is nix itself which takes `stateDir ? "/nix/var"`
<infinisil>
But I can't find any others
robstr has quit [Quit: WeeChat 1.9.1]
<Guanin>
I'd guess that the reason for this is not comparable to my problem, but it should probably work with that approach. I'll try it in the next few hours :) Thanks so far
sir_guy_carleton has quit [Quit: WeeChat 2.0]
<infinisil>
:)
ckauhaus has quit [Quit: Leaving.]
leat has quit [Remote host closed the connection]
leat has joined #nixos
graphene has quit [Remote host closed the connection]
<Baughn>
If I leave nix.conf' max-jobs low, then builds take very long indeed. Build parallelism is sorely needed; most builds aren't themselves parallelised.
<Baughn>
But if I leave cores (as in `make -j $CORES`) low, then the system goes underutilized any time it's waiting for a single, parallelizable build.
<Baughn>
If I turn them both up, however, then I run out of memory.
<Baughn>
(Also nixos-generate-config leaves both at 32, which seems unfortunate.)
<Baughn>
In short, I suppose I want jobs to be scheduled in a smarter manner, maximizing average throughput without OOMing my system.
<Baughn>
Hmm~
<Baughn>
I've built a scheduler to do that once before.
<Baughn>
But this would be deep in the guts of Nix, right?
<infinisil>
Hmm I mean how would you know what derivations take a lot of memory?
<infinisil>
Or do you mean it should just temporarily cancel other derivations to make the one requiring lots of memory work?
<infinisil>
That might work
<Baughn>
If enableParallelBuild is set, then it takes lots of memory. :V
<Baughn>
But that's not quite what I mean. I guess my ideal state would be:
knupfer has joined #nixos
<Baughn>
While setting up a build, we scan the derivations and build a dependency tree. This happens already, presumably.
leat has quit [Remote host closed the connection]
<Baughn>
This can be modelled as a DAG. What we want to do is optimize a particular graph problem I don't know a name for, where we have a system that's N 'slots' wide (N being min(#ofcores, ram/2GB) or something), and we want to feed that entire DAG in order through said system. It's a scheduling problem, in other words.
chreekat has quit [Ping timeout: 245 seconds]
leat has joined #nixos
<Baughn>
Mapping the graph to a 2D grid, we want to minimize the *height* while making the width always <= N.
<Baughn>
Or, as a heuristic, we want to maximize the average fill of the system.
Nissim has quit [Quit: Page closed]
<Baughn>
What makes the problem complicated is that *some* nodes have variable width. Anything that has enableParallelBuild set can be reshaped between being tall and being wide...
<Baughn>
...
<Baughn>
This is not a good model. :P
<Baughn>
The optimization criteria is clear enough. I want builds to finish as quickly as possible, without OOMing, which means not running too many instances of GCC simultaneously.
<Baughn>
That seems like the right branch of math, thanks. :D
<Baughn>
I was forced to reinvent half of that from scratch. Nobody could point me in the direction.
<hodapp>
if you topo-sort the DAG and monitor how many slots are left, you can sort of step through the graph one node at a time and only proceed if you have enough slots for what's next
<infinisil>
A lot of stuff can be modeled with flow graphs
Nissim has joined #nixos
<Baughn>
One fun hack, though...
<Baughn>
We could patch the kernel to, for the nix-build namespace, leave processes suspended after exec until explicitly started by the scheduler.
<Nissim>
hello all, looking for some help with flask HTTPBasicAuth on nixos
<Nissim>
my app is a python flask app but it will not run as it is looking for flask and HTTPBasicAuth. would any one know please how to install these packages on nixos.
<LnL>
have you looked at the python section in the nixpkgs manual?
<hodapp>
but if said DAG is topo-sorted, there are multiple valid ways to still topo-sort, and at any given point you have a set of nodes that could be run next, and you should be able to optimally choose that based on the slots it will occupy
<infinisil>
hodapp: the "optimally" is the problem
simukis has quit [Ping timeout: 256 seconds]
<Baughn>
infinisil: I'm not sure the theorem is directly applicable, but it's at least in the right direction. Thanks.
<hodapp>
though, may need a bit of restructuring to be a flow network. in the form they give there, you'd be doing something closer to having N 'machine' nodes that you're trying to decide the assignments of, I think
asuryawanshi has quit [Ping timeout: 265 seconds]
disasm has quit [Ping timeout: 255 seconds]
<hodapp>
if I get a chance later, will dig out my algorithms textbook and look at some of the examples they had there...
<Baughn>
I'll go with the simple heuristic at first, I think. That's something that should be implementable without too much ado -- if any of this is.
<infinisil>
Pretty sure this problem might be NP-complete
<infinisil>
A lot of problems are
leat has quit [Remote host closed the connection]
leat has joined #nixos
<hodapp>
toposort can get you pretty far still for a heuristif
<hodapp>
heuristic*
<infinisil>
I mean it doesn't really optimize for anything though, it just calculates an ordering of dependencies
<Baughn>
A greedy algorithm did the work quite well last time.
<hodapp>
I'm not calling toposort an optimization algorithm, I'm saying that it makes a lot of quick-and-dirty optimization very easy
* Baughn
notes that the tests currently *fail* for Nix.
myshoe has quit [Quit: Lost terminal]
leat has quit [Remote host closed the connection]
leat has joined #nixos
<nh2[m]>
can I disable the "impure path ... used in link" check? I want to build a nix package on Ubuntu against Ubuntu's libc
<Baughn>
nh2[m]: You can include it as an explicit input.
lhart has joined #nixos
<Baughn>
Of course that'll copy it.
<nh2[m]>
Baughn: how would I go about that? Right now I just do `export CC=/usr/bin/gcc` in `preConfigure`
<nh2[m]>
Baughn: because otherwise my program doesn't run. I want to build against libsodium and use nix to build me one. I can't use nixpkgs's libsodium because that is built a newer glibc than my Ubuntu has, so it uses a symbol that my Ubuntu's glibc doesn't have. I also can't link everything against Nix's glibc because for that I would have to convince my entire build tooling to use nix's libraries
<nh2[m]>
so my idea is: Use nix's convenient way of declaring sources and build steps, to build an .so file that works with my Ubuntu's glibc
<Baughn>
Mm. Not sure you can do that.
<Baughn>
There's a lot of machinery in Nix that exists explicitly to prevent you.
<Baughn>
Nix generally works better if you go all-in, rather than try to mix. I usually say that with the intent of pointing people at NixOS, but compiling everything against nixpkgs is a good start.
leat has quit [Remote host closed the connection]
leat has joined #nixos
<nh2[m]>
Baughn: I mean NixOS is fine and well, but nix has the explicit goal on working on non-NixOS as well, and technically there doesn't see to be any reason to prevent people to opt out of purity it to use it as a glorified EBUILD system
<nh2[m]>
though I'm cetainly aware nobody may have cared yet
<nh2[m]>
but sometimes it's just that undocumented flag that you've never heard of that solves it ;)
wchresta_ has quit [Quit: Yaaic - Yet another Android IRC client - http://www.yaaic.org]
<symphorien>
there is a nativeStdenv, maybe try using it insteaf of stdenv (wild guess)
andymandias has joined #nixos
<nh2[m]>
symphorien: where is that? I cannot find that with grep
<nh2[m]>
I'm not even sure why `impure path ... used in link` is appearing at all. It's in ld-wrapper.sh and I'm explicitly setting `LD=/usr/bin/ld`
<nh2[m]>
symphorien: `export NIX_ENFORCE_PURITY=` gets me past the purity check (yay!) but I get another error `./conftest: relocation error: /nix/store/2kcrj1ksd2a14bm5sky182fv2xwfhfap-glibc-2.26-131/lib/libc.so.6: symbol __tunable_get_val, version GLIBC_PRIVATE not defined in file ld-linux-x86-
<nh2[m]>
64.so.2 with link time reference` in libsodium's autoconf
<nh2[m]>
do you know of any instructions how to use that native stdenv?
<nh2[m]>
like am I supposed to just replace `pkgs.stdenv` by something that contains the native stdenv?
joehh has quit [Ping timeout: 245 seconds]
<nh2[m]>
if yes, where would that value sit in the nixpkgs attrset?
<symphorien>
no sorry
knupfer has quit [Remote host closed the connection]
logzet has quit [Remote host closed the connection]
Lisanna has quit [Quit: Lisanna]
leat has quit [Remote host closed the connection]
leat has joined #nixos
joehh has joined #nixos
<PolarIntersect>
How do people handle external USB disks on NixOS? Just mount manually into their home directories?
<samueldr>
I'm a bit lazy in both ways, and use gnome-disks to mount them
<samueldr>
lazy since I didn't setup a neater way to mount
<nh2[m]>
I think the current issue happens because nix passes it `"-L/nix/store/2kcrj1ksd2a14bm5sky182fv2xwfhfap-glibc-2.26-131/lib", "-rpath", "/nix/store/2kcrj1ksd2a14bm5sky182fv2xwfhfap-glibc-2.26-131/lib",` so that makes it pick nix's glibc for the ./conftest tool
<samueldr>
and lazy since I don't use udisksctl on the CLI
<PolarIntersect>
Ah. I thought maybe people did that, but I don't use gnome because it requires running gnome-desktop now
<samueldr>
I don't know (haven't verified) how much gnome-disks brings in from gnome
<PolarIntersect>
Would be lovely if I could run my WM in gnome but that doesn't work w gnome3
<PolarIntersect>
samueldr: I think it just reiies on their being a gnome session
<samueldr>
ah, I'm not talking about their service for mounting, I'm talking about a specific tool
<samueldr>
`nix-shell -p gnome3.gnome-disk-utility` if you want to check
cnidario has quit [Remote host closed the connection]
<samueldr>
it's then called gnome-disks
<samueldr>
`nix-shell -p gnome3.gnome-disk-utility --run gnome-disks` even
<Myrl-saki>
I'll have to move the Nix expression forward.
<Myrl-saki>
Err
<Myrl-saki>
The Nixpkgs
<Myrl-saki>
now I have to do a full system rebuild.
<Myrl-saki>
; ^ ;
<Myrl-saki>
Welps. I'm disabling the checks lol
<Myrl-saki>
Also, is `nix-build` accepting a /nix/store/.drv a Nix 2 thing?
<Myrl-saki>
For Nix 1, I had to use nix-store --realise
worldofpeace has joined #nixos
matthewbauer has quit [Read error: Connection reset by peer]
leat has quit [Remote host closed the connection]
leat has joined #nixos
<infinisil>
Myrl-saki: $(nix-build --no-out-link '<nixpkgs>' -A nix1)/bin/nix-build $(nix-instantiate '<nixpkgs>' -A hello)
matthewbauer has joined #nixos
<infinisil>
Given an error for me, but works with normal nix-build, so I guess it's a Nix 2.0 thing :P
<Myrl-saki>
infinisil: That's actually what I did.
<Myrl-saki>
Oh okay. :P
<Myrl-saki>
infinisil: About dying due to not a symlink?
worldofpeace has quit [Client Quit]
<infinisil>
The fact that you can just run such a command is such a good selling point of nix
<infinisil>
Myrl-saki: Yeah not sure what the means
worldofpeace has joined #nixos
<LnL>
nix-store -r $(nix-instantiate ...)
<Myrl-saki>
infinisil: I didn't know it was a Nix 2.0 thing, so I did `ln -s /nix/store/foo.drv /home/myrl/foo.drv; nix-build /home/myrl/foo.drv` and it works lmao
<infinisil>
LnL: Yeah, but apparently Nix 2.0's nix-build can do nix-store -r 's job
<LnL>
yep
<Myrl-saki>
I 100% thought my Nix was broken. :P
<Myrl-saki>
infinisil: LnL: THanks.
<infinisil>
I have yet to see Nix being broken even once
<Myrl-saki>
infinisil: Nix, the binary, or Nix the "ecosystem"?
tzemanovic has quit []
<LnL>
I have, was using my branch to 'fix' distributed builds
<infinisil>
I did discover a weird semantic in the Nix evaluator I think should be changed: nix#2109
<Myrl-saki>
infinisil: I had my /nix/store broken once because the power plug is on the same level as my chair, and my chair has wheels. And this probably happened like once a day.