<lopsided98>
Has anyone noticed that large binary cache downloads will timeout on slow connections, even if they are making steady (but slow) progress?
<lopsided98>
I can also reproduce it with channel updates
Synthetica has quit [Quit: Connection closed for inactivity]
<jlv>
I'm trying to add a new package and am running into a bit of a snag. It is a qt app, and I am getting the error "Can not use "WebEngine" module which has not yet been found.". I have added qtwebengine to buildInputs, so I am not sure why it can't be found.
<jlv>
The full error is CMake Error at /nix/store/hy0z060pmjwrdrbb0qrii7hnwif8cz1y-qtbase-5.12.0-dev/lib/cmake/Qt5Core/Qt5CoreMacros.cmake:372 (message): Can not use "WebEngine" module which has not yet been found.Call Stack (most recent call first): CMakeLists.txt:602 (QT5_USE_MODULES)
<gchristensen>
worldofpeace: there is no error checked in. what error did you find?
<gchristensen>
worldofpeace: ... d'oh!
<gchristensen>
wrong branch! checking again.
<jlv>
Looks like the error is while building qtbase?
selfsymmetric-mu has left #nixos ["gone to the land of dead hiccups and extinguished light bulbs"]
enh^ has quit []
<gchristensen>
"The following elements occur in listitem:..." and "These elements contain listitem..." sections are really useful
<worldofpeace>
it seems that I was looking at the wrong docs :P
<gchristensen>
I wonder what you were looking at?
<worldofpeace>
So I guess I should wrap all of that in a <listitem> gchristensen ?
<worldofpeace>
I think it was an older document
<gchristensen>
ah
<gchristensen>
let me look at the document, one sec
<Cale>
Anyone have a hint about how one can enter a shell that has a specified ghc version *and* Cabal library version?
<gchristensen>
yeah that would make sense for that particular list (however, I sort of think the structure of that particular document is a bit strange :))
eo^ has joined #nixos
magnetophon has quit [Ping timeout: 272 seconds]
lambda-11235 has quit [Quit: Bye]
<Cale>
There's a Cabal_2_4_1_0 package, but just installing it doesn't seem to be enough somehow -- we still get errors from cabal-install of other packages which seem to indicate that Cabal 2.0.1.0 is in use
kvda has joined #nixos
<gchristensen>
worldofpeace: so yeah, just wrap that up in a <varlistentry><term>Meson</term><listitem>...
<jlv>
Is there any way to force a nix-build to rebuild instead of using the cache?
<gchristensen>
--substituters ''
judson has joined #nixos
<gchristensen>
or, if you've already built (or fetched) it and want to force another build, you can do --check, , which checks to see if the build was binary reproducible
rouma has quit [Read error: Connection reset by peer]
detran has quit [Ping timeout: 245 seconds]
rouma has joined #nixos
<jtojnar>
worldofpeace: I want to enable doCheck on enchant
<jtojnar>
do not want to step on your toes
<gchristensen>
worldofpeace: I don't really like the idea of using pname for the repo name, as it isn't really the parameterized like that -- it is grahamc/xmloscopy... I dunno. it is a crazy fine point. would you like me to change it to repo = pname?
<gchristensen>
I will if you'd like me to :)
<worldofpeace>
That's what I mean't actually `repo = pname`
<worldofpeace>
only nitpicking that because I went a little crazy with expression style today #66136
<worldofpeace>
jtojnar: I'd think with multiple people it's still like a write lock, and those people have access. That would be a good analogy to be in CONTRIBUTING.md.
<{^_^}>
[nixpkgs] @teto pushed commit from @regellosigkeitsaxiom to master « stlink: 1.3.0 -> 1.5.1 »: https://git.io/fjQqh
<jackdk>
kalbasit: my pi3b+ is not happy right now, but I think if you set hardware.enableRedistributableFirmware = true; you get those without having to ask for them specifically
magnetophon has quit [Remote host closed the connection]
___laika has quit [Ping timeout: 258 seconds]
cfricke has quit [Quit: WeeChat 2.5]
cfricke has joined #nixos
<andi->
gchristensen: update on the slack workspace workflow: Just go to https://slack.com/ssb/add and after a while a text box appears where you are supposed to copy the link. Just copy that and switch to the slack window. It will magically do things then.
astrofog has quit [Remote host closed the connection]
<manveru>
WilliamHamilton: usually, yeah
linarcx has quit [Read error: Connection reset by peer]
linarcx has joined #nixos
<manveru>
check out appimage-run
<WilliamHamilton>
manveru is there some trick I should know? The AppImage website only says "make this executable and run it as if it were a script", but I get:
<WilliamHamilton>
$ ./ledger-live-desktop-1.12.0-linux-x86_64.AppImage zsh: no such file or directory: ./ledger-live-desktop-1.12.0-linux-x86_64.AppImage
<manveru>
yeah, that's what appimage-run is for :)
<WilliamHamilton>
which seems to me a problem with `ld`
<WilliamHamilton>
aha! :D
johanot has quit [Ping timeout: 245 seconds]
<manveru>
you can't run them directly since for some reason appimages aren't linked statically
<{^_^}>
[nixpkgs] @teto pushed commit from @jonringer to master « cmatrix: 1.2a -> 2.0 »: https://git.io/fjQOj
pietranera has joined #nixos
<alexarice[m]>
Hi everyone, has the method of turning off suspend on lid close changed? I have `services.logind.extraConfig = "HandleLidSwitch=ignore"` in my `configuration.nix` and this used to work but it doesn't seem to be working any more
<teto>
alexarice[m]: there are some settings specifically to control this, check man configuration.nix
WilliamHamilton has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @doronbehar opened pull request #66192 → <!-- Nixpkgs has a lot of new incoming Pull Requests, but not enough people to review this constant stream. Even if you … → https://git.io/fjQ3X
xkapastel has quit [Quit: Connection closed for inactivity]
<{^_^}>
[nixpkgs] @doronbehar closed pull request #66192 → <!-- Nixpkgs has a lot of new incoming Pull Requests, but not enough people to review this constant stream. Even if you … → https://git.io/fjQ3X
o1lo01ol1o has quit [Remote host closed the connection]
<nh2>
matthewbauer Ericson2314: in C/C++ projects, what are good ways to propagate the information that some library depends on some other library when static linking is used? let's say we have `myProject` depends on `libA` which depends on `libB`. Is there any mechanism that ensures that `libB.a` will be passed to `myProject`'s link?
<andi->
nh2: usually pkg-config has a way to propagate that information
<nh2>
andi-: what about libs that don't use pkg-config, do we have a convention in nixpkgs? Or should I rather make a .pc file for that software?
<andi->
nh2: no idea, you either have to add that information manually or submit a patch that adds pkgconfig support upstream IMO
__monty__ has joined #nixos
MmeQuignon has joined #nixos
noudle has joined #nixos
ajirx has quit [Ping timeout: 268 seconds]
jtojnar has quit [Remote host closed the connection]
nadley has joined #nixos
<Mateon1>
Hi, so I recently switched to using overlays for installing some packages (which seems to have uninstalled most of my existing packages for some reason... ugh), anyway, I have a conflict because two packages in the same overlay provide the same binary (clang and binutils both provide `as`). How do I set package priorities from an overlay?
<clever>
,libraries Mateon1
<{^_^}>
Mateon1: Don't install libraries through nix-env or systemPackages, use nix-shell instead. See https://nixos.wiki/wiki/FAQ/Libraries for details.
<clever>
that includes compilers as well
<Mateon1>
Ughh, nix-shell is painful to use in practice when you just want to write a quick source file in /tmp and compile it
<Mateon1>
It's fine when you actually start writing a project, but for anything smaller than that it just gets in the way
<clever>
Mateon1: gcc wont work when used outside of nix-shell
<{^_^}>
[nixpkgs] @primeos pushed to master « dav1d: Add withTools but disable all tools by default »: https://git.io/fjQsI
<clever>
and you dont need to write a nix file, just `nix-shell -p foo`
<Mateon1>
It won't? I'm pretty sure I used a global gcc before
<{^_^}>
[nixpkgs] @kcalvinalvin opened pull request #66195 → Add support for cmt (Chromium Multitouch drivers), custom drivers for chromebooks. → https://git.io/fjQsL
<clever>
Mateon1: it wont be able to find any other library youve installed
<Mateon1>
Right, I don't care about libraries when not writing a project, though
<Mateon1>
I literally only want to compile a self-contained binary that prints something to stdout
<clever>
one min
<Mateon1>
Alright, gcc installs fine globally, in that case I don't need clang (and will probably just alias it to gcc for now)
<Abednego>
let derive2 = derive { snapshot = "2019-07-01"; };
<Abednego>
<Abednego>
Any thoughts on how to debug this? Or where I should report it?
abbiya has quit [Quit: abbiya]
<balsoft[m]1>
Well, there has to be at least a `derivation` call somewhere in there
<balsoft[m]1>
Or, likely, even `stdenv.mkDerivation`
<balsoft[m]1>
I don't know anything about R, so I haven't looked into this part of nixpkgs, but if you need to debug, you can look for where the call for one of those functions is and the go from there
justan0theruser has joined #nixos
justanotheruser has quit [Ping timeout: 246 seconds]
<Abednego>
I don't think one needs to know anything about R to do this - sadly I don't know enough about nix
__monty__ has joined #nixos
vmandela has quit [Quit: Leaving]
mexisme has joined #nixos
<simpson>
It depends on how your co-workers consume nixpkgs. They could patch a local copy of nixpkgs, or configure an overlay, or likely there's some sort of nasty nix-shell one-liner.
<Abednego>
Probably easiest to point them at my nixpkgs repo
<Abednego>
simpson: thanks
<exarkun>
Do module imports support URLs? eg can I import straight from github?
___laika has joined #nixos
mexisme has quit [Ping timeout: 252 seconds]
Mister_Magister has quit [Excess Flood]
Mister_Magister has joined #nixos
<Miyu-chan>
exarkun: "Yes"
<Miyu-chan>
You can't import straight from GitHub, but you can import from derivation.
<Miyu-chan>
IIRC, some /nix/store generating primops are not considered derivations, so there's also that.
<sphalerite>
Abednego: didn't you author a bunch of supertux levels??
___laika has quit [Ping timeout: 248 seconds]
<Abednego>
Not me - that must be a different Abednego
<manveru>
sphalerite: it's a biblical name, so probably pretty common :)
<Miyu-chan>
I'd expect a builtins to generate a path and not a string.
<manveru>
Miyu-chan: paths are strings
<sphalerite>
Abednego: anyway, the shortest way is probably to have them pass -I nixpkgs=https://github.com/abednego/nixpkgs/archive/branch-name.tar.gz to whatever command they're using
<Miyu-chan>
manveru: Paths are mostly strings.
<manveru>
well, with a special context
<Miyu-chan>
Yeah.
ericsagnes has quit [Ping timeout: 264 seconds]
<manveru>
but copyPathToStore must be a derivation, right?
<Miyu-chan>
TL;DR, I think NIX_PATH tries to find the most specific search path, and uses the remaining paths relative to that.
<Miyu-chan>
Oh yeah, nixpkgs doesn't.
npmccallum has joined #nixos
linarcx has joined #nixos
<Miyu-chan>
I guess a more actual use-case would be nix tools that are outputs of nixpkgs.
selfsymmetric-pa has quit [Remote host closed the connection]
<Miyu-chan>
nixops for example doesn't have facilities for configuring the path of ec2-ami.nix
<{^_^}>
[nixpkgs] @marsam opened pull request #66203 → terracognita: init at 0.1.6 → https://git.io/fjQne
<Miyu-chan>
Also, kinda curious. How long do you think it'd take for flakes RFC to be implemented if accepted? NIX_PATH being disabled by default seems a bit huge.
<yorick>
Miyu-chan: the implementation is there already
<Miyu-chan>
Also, kinda curious. Maybe there should be facilities for multi-flake repos? If it allows circular dependencies, then haskellPackages and pythonPackages can be in the same repo(maintaining the monorepo approach), while keeping them in separate flakes(for what purpose tho, I guess?)
jtojnar has joined #nixos
layus[m] has joined #nixos
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
ddellacosta has quit [Ping timeout: 244 seconds]
<{^_^}>
[nixpkgs] @worldofpeace pushed commit from @emmanuelrosa to master « openshot: use wrapQtAppsHook »: https://git.io/fjQnO
<detran>
quick question - I've got jdk11 installed, and I've got clojure installed. However, when I run (System/getProperty "java.version") inside of clojure it reports "1.8.0_212"
<detran>
java -version returns openjdk version "11.0.3-internal" 2019-04-16
<detran>
where in the world is that jdk 8 coming from? Or how can I find out?
mmlb3 has quit [Ping timeout: 258 seconds]
Abednego has joined #nixos
<clever>
detran: run `nix-store -qR` on the path of the closure binary
waleee-cl has joined #nixos
<detran>
ah, it looks like theres an openjdk-8u212 in the output list
<BoipiSigre>
I installed a JDBC driver, now i must put the path on archive file in parameter file. How find the path ?
o1lo01ol1o has quit [Remote host closed the connection]
<WhittlesJr>
I have two python packages to submit, but I do not have an entry for myself in maintainer-list.nix. What's the proper protocol for submitting PRs in this case? Two PRs, each with the new maintainer entry? One PR? Submit one and wait till it's pulled then submit the other?
ndi^ has joined #nixos
fendor has quit [Ping timeout: 244 seconds]
<sphalerite>
WhittlesJr: if they're related, I'd make one PR for all three changes
<WhittlesJr>
sphalerite, what if they're not strictly related to each other?
<gchristensen>
sounds tricky :)
o1lo01ol1o has joined #nixos
__monty__ has quit [Ping timeout: 245 seconds]
drewr has joined #nixos
<sphalerite>
WhittlesJr: I'd say probably make one PR if they're both straightforward packages
drewr has quit [Remote host closed the connection]
<WhittlesJr>
sphalerite, makes sense. One PR with 3 commits then?
<gchristensen>
moving to nixos-unstable has made my computer much less usable: firefox very regularly just ... uh ... exits. anyone seeing behavior like this?
hyper_ch2 has quit [Remote host closed the connection]
hyper_ch2 has joined #nixos
zupo has quit [Client Quit]
hyper_ch2 has quit [Remote host closed the connection]
__monty__ has joined #nixos
WhittlesJr has quit [Remote host closed the connection]
zupo has joined #nixos
hyper_ch2 has joined #nixos
hyper_ch2 has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @averelld opened pull request #66219 → python.pkgs.tensorflow: fix hash for gpu build → https://git.io/fjQla
bakakuna has joined #nixos
sigmundv has quit [Ping timeout: 268 seconds]
<manveru>
emilsp: they both have direnv plugins, they basically check what's the nearest .envrc to your file/project and modify the environment of the editor to match what it generates
<emilsp>
aha, fair. I think I could survive without that.
<emilsp>
is there a way I can package up a deriviation into a tarball?
<manveru>
well, i use the emacs direnv-mode all the time, it's super helpful :)
hyper_ch2 has quit [Remote host closed the connection]
<lambda-11235>
Although, my timeline doesn't appear on X11 with missing breeze theme, which is why I had to add it.
<gchristensen>
huh
damesca has joined #nixos
<lambda-11235>
No segfault though
byteflam1 has quit [Ping timeout: 244 seconds]
Abednego has joined #nixos
byteflam1 has joined #nixos
Abednego` has joined #nixos
Abednego has quit [Ping timeout: 272 seconds]
Abednego` has quit [Ping timeout: 268 seconds]
<gchristensen>
lambda-11235: I wonder if xdenlive should have ffmpegFull in its path too
<lambda-11235>
Ah right, it probably should. It needs ffplay
<gchristensen>
btw I have no idea what I'm doing. I've never edited video. I've only made a meager attempt at it once this weekend :0
nDuff has joined #nixos
mexisme has joined #nixos
byteflam1 has quit [Ping timeout: 246 seconds]
<{^_^}>
[nixpkgs] @worldofpeace opened pull request #66222 → virtualbox: use wrapQtAppsHook → https://git.io/fjQ8O
Abednego` has joined #nixos
<lambda-11235>
I made a meager attempt last night. An attempt that had me spending hours figuring out these missing deps, followed by 5 mins of editing.
<emilsp>
is there a way I can get the ugly android ndk toolchain into my PATH ?
ee1943 has quit [Read error: Connection reset by peer]
aiverson has joined #nixos
<petercommand>
I am trying to build an environment with libgobject-2.0.so.0 in LD_LIBRARY_PATH
<petercommand>
what package should I sepcify as buildInput?
<petercommand>
I tried glib, but glib doesn't work
ee1943 has joined #nixos
<aiverson>
I'm having some trouble getting LUKS set up on 19.03 on a laptop. It gives me the "Waiting for device" error message then "$DEVICE is unavailable" error message. This happens with volumes autodetected by nixos-generate-config and manually configured as best as I can figure out how. Does anyone have troubleshooting advice?
<aiverson>
err... "waiting for device" status log, not error message
gagbo has quit [Remote host closed the connection]
gagbo has joined #nixos
<emilsp>
can I somehow change the pkgs.stdenv.targetPlatform.config to not be my host in the repl ?
palo has quit [Quit: WeeChat 2.4]
andreas303 has quit [Remote host closed the connection]
byteflam1 has joined #nixos
palo has joined #nixos
_kwstas has joined #nixos
_kwstas has quit [Remote host closed the connection]
_kwstas has joined #nixos
<iqubic>
is it possible to specify a set of run-time depedencies in a nixpkg derivation? Like say program 'foo' calls executable 'bar' as part of normal run-time operations. Can I make it so that installing 'foo' also pulls 'bar' into path?
<{^_^}>
[nixpkgs] @kalbasit pushed commit from @marsam to master « terracognita: init at 0.1.6 (#66203) »: https://git.io/fjQ8A
<zeta_0>
hello guys quick question, in what directory do the haskellPackages get installed because i want to use this haskell cheatsheet after installing it in home.nix? https://nixos.org/nixos/packages.html#haskellpackages.cheat
<__monty__>
clever: Maybe you know. I'm trying to use buildEnv's `pathsToLink` but it causes evaluation to run out of memory. What's going on?
<clever>
__monty__: can you gist your current nix expr?
<__monty__>
clever: Sure, it's just an empty buildEnv added to my home.nix currently though.
<zeta_0>
so where do i access the cheat sheet pdf after installing `haskellPackages.CHeatSheet`? after the rebuild will it tell me where it is located?
byteflam1 has joined #nixos
<clever>
zeta_0: the $out has a bin/cheatsheet so you could probably run that binary
<clever>
zeta_0: $data has the pdf in it, but there is no easy way to access it, but that path will be in the Paths_CheatSheet.datadir :: String, which the code may make use of
<__monty__>
clever: Yes, didn't ask over there because this was a home-manager config, not nix-darwin, though I do run home-manager from nix-darwin.
<emilsp>
clever: and what arg do I use to feed a path to a zlib.so for patchelf ?
<__monty__>
clever++
<{^_^}>
clever's karma got increased to 164
<simpson>
zeta_0: There's a formal study of security properties and program effects (confidentiality, integrity, etc.) but I don't know what that subfield is called.
<aiverson>
rnhmjoj: I've got a laptop with an nvme drive. The first partition is an unencrypted EFI partition. The second is the LUKS partition. The LUKS partition has an LVM system in it containing a swap volume and an ext4 volume for the fs root.
jkachmar has joined #nixos
<marek>
under which conditions it is okay for a package derivation to contain no files?
<mdash>
marek: what would it contain then
<marek>
mdash: probably only subpackages, but the main derivation itself is empty
<marek>
just an empty dir in nix store
<marek>
I'm little bit unsure, as it produces no output what so ever, probably a regression
<mdash>
strange!
<zeta_0>
clever: what do you mean by `$out` and `$data`?
coderobe has quit [Read error: Connection reset by peer]
<adisbladis>
marek: Which package?
<clever>
zeta_0: split output packages
<adisbladis>
buildGoPackage has an empty $out for example
<clever>
zeta_0: any package in nix can have multiple outputs, which can be GC'd independently
bakakuna has joined #nixos
<adisbladis>
marek: hmmm. Weird
<{^_^}>
[nixpkgs] @danielfullmer opened pull request #66223 → kdeconnect: Use qt's mkDerivation → https://git.io/fjQ4r
_kwstas has quit [Quit: _kwstas]
<zeta_0>
clever: oh
byteflam1 has joined #nixos
nDuff has joined #nixos
<__monty__>
Hmm, any way to make cache substitution work better on a very slow connection? I'm getting tons of timeouts and it looks like previously fetched paths aren't remembered.
<petercommand>
I got this error when trying to execute Zettlr.."Settings schema 'org.gtk.Settings.FileChooser' is not installed"..
<petercommand>
not sure how to install a settings schema..
<kalbasit>
is there a function that can return NIX_PATH with nixpkgs modified given a NIX_PATH?
zeta_0 has quit [Remote host closed the connection]
<Yaniel>
has anyone got hardware encoding to work in obs-studio
<Yaniel>
?
lambda-11235 has quit [Ping timeout: 246 seconds]
Shoubit has quit [Ping timeout: 246 seconds]
<averell>
is there a way to tell a remote builder to fetch dependencies itself? i'm basically just nixo-rebuilding with --builders ssh://bla, and it's uploading tons of stuff
<__monty__>
averell: Only in two steps, instantiate things on the remote then nix-copy-closure them over, or substitute, which should work at that point.
<__monty__>
I agree this is a painpoint.
<clever>
averell: one sec...
<__monty__>
Figures clever has something for this.
<clever>
213 nix = {
<clever>
250 builders-use-substitutes = true
<clever>
246 extraOptions = ''
<clever>
that allows the build slaves to use their own cache config
_kwstas has joined #nixos
<jlv>
What is an the easiest way to replace a file before building?
<jlv>
Still trying to get seadrive-gui working. I have a theory that I would like to quickly test.
<petercommand>
I was trying to install Zettlr..not sure how to get yarn to not fetch from remote servers when using yarn2nix..
<clever>
petercommand: yarn --offline
<petercommand>
yes
chloekek has quit [Ping timeout: 248 seconds]
<petercommand>
didn't work
Wizek has quit [Quit: Connection closed for inactivity]
<balsoft[m]1>
Hey everyone! Is anybody here using the `androidenv.emulateApp` function for VM's? If so, how do I go about finding out the correct package and activity id's?
<clever>
arianvp: android doesnt obey the FHS, and all the libraries are in /system/lib i believe
<arianvp>
I see. but I need to cross compile to android. so perhaps I need to use that attribute?
waleee-cl has quit [Quit: Connection closed for inactivity]
chloekek has joined #nixos
<clever>
arianvp: and anything built normally with nix expects a /nix/store/, so you either need to do some funny stuff to make the libraries co-operate with /system/lib/ or just go full static
<arianvp>
clever: ah I see
<arianvp>
clever: next problem. Play Store requires me to publish 64 bit libraries as well
<clever>
arianvp: ahh, thats probably a bug in nixpkgs then
<clever>
i'm not that familiar with the internals of the cross-compile stuff yet
<exarkun>
is there an idiom for, hmm, forking? a nixos module. there is a service module in nixos that I want to customize for my own purposes... but I can't copy, edit, and load the copy because all the options conflict with the in-tree options...
ajirx_ has joined #nixos
<exarkun>
do I just have to give it a different name?
<exarkun>
"disabledModule" looks like it's probably what I want... but I'll explain what I'm trying to do, too
<exarkun>
services.tahoe allows for some tahoe configuration to be defined but not all of it
<exarkun>
so broadly, I'm making a version of that module that lets me control all the configuration I want
<simpson>
exarkun: FWIW you have my blessing to make a less-shitty version of that module. I am not offended.
<clever>
exarkun: sounds like a good use for disabledModule, and then file a PR once its working nicely
<aiverson>
And now I have a kernel panic when I try to make it let me start a recovery shell.
<exarkun>
clever: cool, that sounds about right
<exarkun>
simpson: it doesn't seem shitty, just incomplete :) also some of the configuration I want to control doesn't even exist in tahoe master yet ...
<pie_>
qyliss: i have some questions about your config
<pie_>
qyliss: you have a default.nix at the top level that imports the overlays, but you also have nixpkgs.overlays set in the nix module, why?
phreedom has quit [Ping timeout: 260 seconds]
psyanticy has quit [Quit: Connection closed for inactivity]
<qyliss>
pie_: all good questions that I do not know the answers to right now but might when I am more awake
<Hedgework>
Hey, Nix newb here... is there an easy way to cough up the list of things I've installed on the fly via "nix-env -iA whatever.package" so that I can figure out what I want to put in my declarative config for the next machine?
<clever>
Hedgework: nix-env -q
<Hedgework>
clever: Thank you!
jlv has quit [Ping timeout: 260 seconds]
<pie_>
qyliss: ty, trying to make a good config here >.>
<pie_>
qyliss: idk if you know about NIXOS_EXTRA_MODULE_PATH, i added that to `activate` to have it automatically import a common set of modules i want, as opposed to importing it via imports = [ ... ], do you know any reason that might be bad? (line 23 https://bpaste.net/show/sg_A )
linarcx has quit [Ping timeout: 248 seconds]
<clever>
pie_: i would just make my own foo.nix, that has both common-modules and configuration.nix under imports
<aiverson>
Can anyone here help me figure out getting an install under LUKS running?
<pie_>
clever: yeah its a difference of one line i guess
<clever>
aiverson: what problem are you having?
<clever>
pie_: and more likely to work via nixos-rebuild
<aiverson>
When I try to boot it, it can't find the partition that LUKS is on.
<pie_>
clever: ah hm
<pie_>
clever: yeah ok
<clever>
aiverson: can you pastebin your configuration.nix ?
<clever>
aiverson: you can boot the install ISO again, luksOpen, and mount everything back under /mnt/
<pie_>
qyliss: clever: not sure if this config is compatible with rebuild to begin with now that i look at it
<aiverson>
I can do that. I can luksOpen and mount everything.
<aiverson>
I can read and write the filesystem, and even nix-enter it.
<aiverson>
The boot loader registers the nixos generations written to /boot and the nixos initrd gets loaded and enters stage 1 boot. And then it complains about /dev/whatever-I-try not existing. I've used the autogenerated hardware-configuration.nix; I've written my own lines to use alternate names.
<clever>
aiverson: if you pastebin both the configuration.nix and hardware-configuration.nix, i can take a look at them
<clever>
aiverson: gist helps when dealing with multiple files at once
fendor has quit [Remote host closed the connection]
fendor has joined #nixos
<aiverson>
Ok, I got it fixed. nixos-generate-config wasn't detecting the luks volume and adding the luks device line correctly, and when I tried to replace the hardware-configuration.nix entirely, I forgot to include the nvme kernel module in the initrd.
<aiverson>
Now I just need to figure out why i3 isn't actually starting when it should when I log in.
<{^_^}>
[nixpkgs] @xvello opened pull request #66226 → Add bluez-qt as an explicit dependency of plasma5 → https://git.io/fjQRU
<clever>
aiverson: what FS is ontop of the luks?
<{^_^}>
[nixpkgs] @xvello closed pull request #65440 → bluedevil: make bluez-qt a propagatedBuildInputs → https://git.io/fjy42
<aiverson>
The LUKS contains an LVM containing a swap volume and an ext4 volume.
<clever>
aiverson: ah, yeah, nixos-generate-config has trouble finding the path from fs->lvm->luks
<clever>
so it doesnt know the luks is required
zupo has joined #nixos
<aiverson>
Is there something I can do to fix that in the future and make a PR against nixos-generate-config? Is that an existing issue with someone working on it?
{^_^} has quit [Remote host closed the connection]
__monty__ has quit [Quit: rebootying]
{^_^} has joined #nixos
jlv has joined #nixos
gchristensen has joined #nixos
<gchristensen>
,botsnack
<{^_^}>
Oh thanks, have a cookie yourself
<gchristensen>
oops :/
<jlv>
I'm getting an error "/nix/store/m93jjbdb9nvc1f38fxbf14rs4r1g013a-curl-7.65.3/lib/libcurl.so.4: no version information available" when trying to run a package I built. Not sure what to do about that.
<johanot>
:D you are spoiling it
<johanot>
gchristensen:
<damesca>
This feels like a dumb question - what's the difference between nixpkgs-19.03-darwin and nixos-19.03?
<damesca>
So if I'm using nix on a mac, all my packages are coming from nixpkgs-19.03-darwin?
<arcnmx>
different set of tests gate the channel releases I think, because darwin channel doesn't care if any nixos/linux specific tests don't pass?
<arcnmx>
and can update sooner because yeah kernel and things don't need to wait to be built? not sure how it works out in practice...
<damesca>
I'm looking to pin packages in my git repo that contains my config, and I have a mac for a laptop and a linux homeserver - so I should pin nixos-19.03 and nixpkgs-19.03-darwin and use those as my channels for linux/mac respectively?
<damesca>
I feel like my thinking isn't right and I'm missing something
<damesca>
/multiple things
<pie_>
qyliss: ok so the top level default.nix is to load the overlays as soon as possible i think, but im not sure how that gets called
<simpson>
That doesn't sound wrong. The same configuration might build slightly different packages on a lower level, but that shouldn't matter.
<pie_>
qyliss: which is to say, my problem is, i dont have that top level default.nix and when im defining my modules, they dont see my overlays, because i dont set them anywhere
<damesca>
simpson: good to know, thanks
<{^_^}>
[nixpkgs] @domenkozar merged pull request #64286 → haskellPackages.inline-c-cpp: fix build on darwin → https://git.io/fj6WF
<qyliss>
pie_: top-level default.nix is what nixpkgs points to
<pie_>
hm....right
<pie_>
qyliss: not in activate?
<qyliss>
in nix.nixPath
<pie_>
ok so i guess i might have a bootstrapping problem
<pie_>
i expect overlays to already work on the first load with my current setup
<pie_>
s/load/rebuild/
<pie_>
hm maybe -I will wokr
<Thra11>
I'm really loving NixOS this evening. Kept misconfiguring my system, but NixOS just lets me switch back to a working generation (or worst case power-cycle and boot into a working generation)
<zeta_0>
is the package `ihaskell` broken for all ghc versions? i really want to use this package?
marek has quit [Ping timeout: 248 seconds]
justan0theruser is now known as justanotheruser
wfranzini has quit [Remote host closed the connection]
<zeta_0>
and the package `haskell-overridze` as well?
marek has joined #nixos
o1lo01ol1o has quit [Ping timeout: 258 seconds]
ixxie has quit [Ping timeout: 268 seconds]
<pie_>
qyliss: that doesnt make sense...if thats true how does the following from `activate` work?: `nix build -f '<nixpkgs/nixos>' --out-link "$link" "$@" system` work? wouldnt that need <nixpkgs/nixpkgs/nixos>?
<pie_>
to be fair NIX_PATH in activate is export NIX_PATH=nixos-config=$d/config/nodes/$sys/configuration.nix:$d
gagbo has quit [Ping timeout: 246 seconds]
<pie_>
yeah ok, with $d that makes sense because it looks for $d/nixpkgs/nixos
<qyliss>
pie_: no, because the root default.nix re-exports nixpkgs/default.nix
gagbo has joined #nixos
<zeta_0>
has anyone here tried `ihaskell`?
WhittlesJr has quit [Remote host closed the connection]
<pie_>
qyliss: i dont really get how that applies here...
<pie_>
qyliss: the top level default.nix and importing <nixpkg/nixos> from $d=the_root_of_the_tree both make sense to me in isolation, but passing -I nixpkgs=the_root doesnt seem to change anything (and im surprised it works at all given that it should be pointing <nixpkgs/nixos> to a nonexistent location?
<pie_>
that should be defined in nixpkgs-overlays/mapDir.nix
<pie_>
qyliss: if i import the top level in the repl, it works
<lejonet>
What would be the easiest way to override a feature of a kernel? (I want to enable the explicitly disabled ia32emulation, that is done for the hardened kernels)
<lejonet>
I can see that in the function hardenedLinuxPackagesFor (which is used to generate the hardened kernel packages) that features.ia32Emulation = false; Would I have to make my own hardenedLinuxPackagesFor and hardened kernel packages in an overlay or is there an easier way?
<pie_>
aaah what the heck NIX_PATH=nixos-config=./config/nodes/nixos/configuration.nix:. nix repl also works in the repl...
<lejonet>
symphorien: that one only goes through commandline and modules, not features
<symphorien>
There is a section "custom configuration"
<lejonet>
I have overriden the features thing with the kernelPatches so I do have ia32 emulation turned on for my kernel, but I need the actual feature to be set to true, to satisfy installing steam
<symphorien>
Ah I see
<lejonet>
Should've added why I needed to do the feature overriding, sorry bout that
<symphorien>
No idea then, sorry
_kwstas has quit [Quit: _kwstas]
__monty__ has quit [Quit: leaving]
<lejonet>
Had I "only" needed the feature on the kernel, the custom configuration in that wikipage is enough :) (that is where I started)
<lejonet>
symphorien: I'm just gonna make my own package, with my own function that doesn't set features.ia32emulation = false; (generic.nix sets it to true by default)
<lejonet>
and put it in an overlay
<lejonet>
and use that for kernelPackages
Edes has quit [Quit: leaving]
jsgrant_ has joined #nixos
<pie_>
qyliss: ok 1) i think i messed up something in the overlay override of lib causing the infrec 2) isnt {lib, ...} supposed to be pkgs.lib? dont see why that broke the way it did...
damesca has quit [Remote host closed the connection]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<{^_^}>
[nixpkgs] @yurrriq opened pull request #66234 → pythonPackages.bugwarrior: depend on jira, comment out test deps → https://git.io/fjQEm
zupo has joined #nixos
<pie_>
or not...im hella lost.
jlv has joined #nixos
liberiga has joined #nixos
hyperfekt has joined #nixos
ramses_ has quit [Remote host closed the connection]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<hyperfekt>
y'all i'm so happy. i managed to pin NixOS with no checkout :D i did have to chase a bug for two days but it turns out i had simply forgot an import statement '=D
doyougnu has quit [Ping timeout: 246 seconds]
lambda-11235 has quit [Quit: Bye]
<pie_>
apparently accessing pkgs.lib in a module causes infinite recursion? wtf?
<hyperfekt>
not usually. could it be that you're using it to import more modules?
<pie_>
ok actually, guess i should make a repro or something, ugh
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/fjQE2
ericsagnes has joined #nixos
<infinisil>
I've gotten pretty good at stopping infrec causes in modules :)
<hyperfekt>
yeah, apparently the module args depend on the closure of imports somehow
<Guillaum>
what binutils should I override if I want to override all binutils involved in pkgCross.aarch64-multiplatform. I tried to override pkgsCross.aarch64-multiplatform.buildPackages.binutils-unwrapped, but gcc for aarch64 is still using the old binutils.
<pie_>
so, by passing overlays to the top level nixpkgs funciton
<hyperfekt>
mckeankylej: That's for the updater, if you just want a specific version it'd probably be easiest to just override the source of one of chromium packages (stable/dev/nightly).
evanjs has joined #nixos
<pie_>
hyperfekt: ok but in that case how the heck do i get my overlay stuff to show up in the lib argument
<infinisil>
pie_: What's ./nixpkgs?
<mckeankylej>
hyperfekt: how do I do that I tried `pkgs.chromium.override { src = .. }` but src wasnt an option
<pie_>
infinisil: checkout of nixpkgs
<infinisil>
Hm..
<evanjs>
Question re adding to maintainers. Is there a point in time when I am expected to add myself to maintainers for a package? Like one PR, multiple, etc?
<{^_^}>
[nixpkgs] @xaverdh opened pull request #66235 → kernel/modules-closure.sh: fix the modules closure → https://git.io/fjQEw
<hyperfekt>
mckeankylej: override is for the arguments passed to the nix file, you want overrideArgs for the arguments passed to mkDerivation
<pie_>
evanjs: i dont think theres any official policy
byteflam1 has quit [Ping timeout: 246 seconds]
<pie_>
hyperfekt: override...Args?
Thra11 has quit [Ping timeout: 258 seconds]
<pie_>
huh
<infinisil>
pie_: What are you evaluating exactly to cause inf rec?
<evanjs>
pie_: Alright. Figured as much, just didn't want to make assumptions.
<infinisil>
evanjs: Feel free to add yourself as maintainer even if you haven't done any work on it. Even being interested in helping it keep running is better than no maintainer
<hyperfekt>
pie_: You should be able to pass your overlay to the nixpkgs you're importing (like I suggested) in the arguments, like (import <nixpkgs> { overlays = ... })
<infinisil>
pie_: Ah there's your problem. You can't use pkgs in imports
<pie_>
hyperfekt: oh right. thats crappy though :(
<pie_>
infinisil: yeah thats what hyperfekt said
<infinisil>
Ah didn't see
<pie_>
infinisil: is this documented? :P
<hyperfekt>
haha, i'm not the one for beautiful solutions, i'm glad if my own stuff works at all :b
<pie_>
x( thanks anyway1
<infinisil>
pie_: I haven't seen anything, soo probably not
erasmas has quit [Quit: leaving]
ng0 has quit [Quit: Alexa, when is the end of world?]
<hyperfekt>
Hm, we should fix that. But maybe with input from someone who knows /why/ this is actually happening.
* pie_
points to nbp? :P
mexisme has quit [Ping timeout: 250 seconds]
<mckeankylej>
hyperfekt can you show me an example of overrideArgs
<infinisil>
hyperfekt: pie_: It's happening because the list of imports depends on pkgs, but pkgs can change depending on overlays that are applied, and you can only figure out all the overlays applied if you evaluate all imports, so there's the infinite recursion
<evanjs>
infinisil: alright, that essentially answers my question. Thanks!
<infinisil>
(because there's the nixpkgs.overlays option you can set in potentially all NixOS modules)
<pie_>
infinisil: why does the list of imports depend on pkgs?
<pie_>
i mean im pretty smol brain, but to me it looks like my use of pkgs here is orthogonal to the set of overlays
<infinisil>
pie_: fixed points don't magically allow you to make stuff depend on itself. They only work when there's no cycles
troydm has joined #nixos
<infinisil>
> fix (x: x)
<{^_^}>
infinite recursion encountered, at (string):256:9
<pie_>
obviously theres a cycle somewhere but i feel like there doesnt need to be..
<pie_>
let me reread what you said like 5 more times :I
<hyperfekt>
infinisil: That can't be it, I get recursion with any module argument (which have no relation to overlays), not just nixpkgs ones.
<pie_>
one should probably look at the traceback
<infinisil>
pie_: So a bit more clearly: To know what pkgs.lib is, you need to evaluate *all* overlays, because all of them could change .lib. To know all overlays, you need to know all NixOS modules. To know all NixOS modules, you need to traverse the `imports` list of all of them. But the only way you can traverse it for that module of yours is to evaluate the expression that uses `pkgs`
<hyperfekt>
pie_: lmk when that one makes sense to you :'D
___laika has quit [Ping timeout: 245 seconds]
<infinisil>
hyperfekt: Not sure what you mean by that. If you have a concrete case of infinite recursion I can probably tell you where the cycle is
<pie_>
infinisil: ok that sounds a bit more plausible
<infinisil>
hyperfekt: Pretty much same reason: To know all imports it needs to evaluate test, to know test it needs to evaluate all imports
orivej has joined #nixos
<infinisil>
The set of values that can be a dependency of `imports` is limited to ones that don't go through the module-loop, such as `lib`
<{^_^}>
[nixpkgs] @worldofpeace opened pull request #66236 → Reorganize GNOME tests, re-enable LightDM for release-combined → https://git.io/fjQuT
<hyperfekt>
Yeah, I was just saying it can't have anything to do with the overlays. So in the pkgs case it's not about knowing about all overlays, but about not knowing pkgs, right?
<hyperfekt>
Cause iirc even if you don't set pkgs with _module.args it'll give you a recursion. Or is the only thing taken to evaluate pkgs from the configuration the overlays?
<infinisil>
Ah yeah it's not overlay-specific. I think in the pkgs case it's only about overlays and config, (because I think only nixpkgs.overlays and nixpkgs.config can change pkgs)
<infinisil>
If those two options didn't exist, infinite recursion probably wouldn't occur if you used `pkgs` in `imports`
<infinisil>
Ah right
<infinisil>
You're right, it would still be a problem because pkgs is set through _module.args
jlpeters has joined #nixos
<hyperfekt>
So we should document that you can't use pkgs or self-defined module arguments to determine imports, but lib is fine.
<hyperfekt>
(or config)
lord| has quit [Quit: WeeChat 2.5]
<hyperfekt>
It would be interesting to have an index of errors and causes in the manual
gagbo has quit [Ping timeout: 248 seconds]
gyroninja__ has joined #nixos
mckeankylej has quit [Remote host closed the connection]
gagbo has joined #nixos
<hyperfekt>
Is _module.args even documented anywhere?
<infinisil>
config also won't work
<hyperfekt>
yeah, i added that in the next line down :b
<asymmetric>
what's the advantage? if it is that i can create directories and control permissions in a systemd service, then i can do the same inside /etc right?
<infinisil>
/etc is not being set up with systemd, it's some etc.sh script in nixpkgs
<growpotkin>
asymmetric: It depends on the file, but I generally try to avoid manually controlling any of the files in /etc because so many of them are generated. It's not necessarily wrong but it might create conflicts
<infinisil>
systemd with systemd.tmpfiles would provide a lot more control
evanjs has quit [Quit: WeeChat 2.5]
<asymmetric>
thx, i'll look into using /var/lib
<asymmetric>
any particular module i can use as inspiration?
<asymmetric>
i've been looking at tt-rss
<infinisil>
Searching for `systemd.tmpfiles` in nixpkgs should give you lots of examples
<growpotkin>
asymmetric: I missed the beginning of the conversation but I've written a handful of Nix services. What exactly were you trying to build?
<asymmetric>
growpotkin: pi-hole
<pie_>
infinisil: did I do something wrong here? https://bpaste.net/show/j9Nw im enabling the options but none of my traces run...
<growpotkin>
oh nice
<pie_>
infinisil: it is getting imported because syntax errors show up...
___laika has quit [Ping timeout: 272 seconds]
<infinisil>
what do you mean none of your traces run?
<infinisil>
Oh the trace up there
<pie_>
err, yeah just one
<infinisil>
pie_: An option won't be evaluated unless you use it :)
<infinisil>
Lazyness and such
<pie_>
the real problem is its complaining about not setting a boot device but this is copied straight out of my config, and re: im enabling the stuff
<pie_>
infinisil: yeah but i set mine.grub.encryptedZfs = true;
<infinisil>
Oh
<infinisil>
Hm...
<pie_>
qyliss's stuff looks good but damn what am i doing :I
<pie_>
im breaking everything
<qyliss>
pie_: if you ask me tomorrow I'll help you out
<pie_>
qyliss: thanks, im hoping i can get some stuff working till then xD
<qyliss>
glhf :)
<infinisil>
pie_: Oh, you can't use mkIf like that
<infinisil>
pie_: `mkIf foo { ... } // mkIf bar { ... }` doesn't work
<pie_>
...
<pie_>
ffs lol
<infinisil>
You need `mkMerge [ (mkIf foo { ... }) (mkIf bar { ... }) ]`
<pie_>
[incoherent screaming]
gagbo has quit [Ping timeout: 248 seconds]
<pie_>
that looks fine but geez, i guess i need to read the list of all the types in the manual
<infinisil>
I guess it's kind of bad that this doesn't give an error
<pie_>
infinisil: so this isny even nix basically :P
<pie_>
im learning an EDSL
gagbo has joined #nixos
<infinisil>
Yeah pretty much :)
<pie_>
gd it
<infinisil>
Actually one thing that inspired nixlisp is how you can't really do proper dsl's in nix, because the standard language features get in the way
<pie_>
infinisil: not @ nixlisp; ugh the rabbithole just keeps getting deeper :P
<growpotkin>
Oh damn have you been working on LISP tools for Nix?
<pie_>
0_0 whats going on here
<infinisil>
:P
* jackdk
wonders if it's a good idea to generate nix from m4 macros and IFD them...
<pie_>
infinisil: did you embed a lisp
<growpotkin>
I just had a major breakthrough the other day for making shells
<pie_>
growpotkin: im listening
<growpotkin>
I even got it to work on Mac lol
<growpotkin>
here one sec let me make a gist
<growpotkin>
basically i prop up a localized asdf and sbcl config in the project as a part of the `shellHook`.
<growpotkin>
It's certainly not perfectly "pure" as we normally like in Nix but it's miles better than nothing, and it doesn't assume that the caller has any LISP configurations/libs available on their system to start with.