matthewbauer has quit [Remote host closed the connection]
<nh2[m]>
dtz: at some point we probably want to configure the ghc to be itself linked at least almost-statically, right now in the dynamic link it has 150 ms startup time
Lev50 has quit [Quit: This computer has gone to sleep]
Mr_Keyser_Soze has quit [Ping timeout: 265 seconds]
zearen has joined #nixos
oltoAltn has joined #nixos
Supersonic has quit [Ping timeout: 260 seconds]
Supersonic has joined #nixos
olto has quit [Ping timeout: 265 seconds]
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Have-Quick has quit [Quit: Have-Quick]
<rprije>
Hi. How can I get a list of attributes in nixpkgs? ie finding sibling attributes to haskellPackages
Mr_Keyser_Soze has joined #nixos
jperras has quit [Ping timeout: 248 seconds]
<rprije>
Also, is there any way of listing everything on the same hierarchy as nixpkgs itself?
<rprije>
*same hierarchy level
<rprije>
Concretely, I'm trying to find where haskellPackages_ghcjs went and in the process trying to learn the tools that would allow me to figure it out on my own.
jperras has joined #nixos
<infinisil>
rprije: Hmm, I often just use rg in a nixpkgs checkout to find stuff
<infinisil>
One way to list packages in an attrset is to use `nix-env -qaA haskellPackages`
<infinisil>
rprije: But to find this stuff out, I'd recommend to look at the definition of things. In this case, how haskellPackages is defined
xcmw has joined #nixos
<infinisil>
And (almost) all top level attributes are defined in nixpkgs/pkgs/top-level/all-packages.nix
<rprije>
infinisil, Thanks. That's basically what I've been reaching for but it'd be nice if there were a simple query for it.
MP2E has joined #nixos
<infinisil>
Well you could evaluate it with nix directly and process it however you want, e.g. like this: nix-instantiate --eval -E 'builtins.attrNames (import <nixpkgs> {}).haskellPackages' --json | jq '.[]' -r
<infinisil>
^ outputs a list of all attributes under haskellPackages
<rprije>
Ooh, that's helpfully general. Thanks!
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
andreabedini has joined #nixos
Mr_Keyser_Soze has quit [Ping timeout: 245 seconds]
ryantrinkle has joined #nixos
contrapumpkin has joined #nixos
halfbit has quit [Ping timeout: 245 seconds]
alphor has quit [Ping timeout: 256 seconds]
cstrahan_ has quit [Ping timeout: 256 seconds]
cstrahan_ has joined #nixos
alphor has joined #nixos
truthadjustr has quit [Quit: leaving]
oida has quit [Remote host closed the connection]
oida has joined #nixos
mizu_no_oto has joined #nixos
igo95862 has joined #nixos
spear2 has joined #nixos
Mr_Keyser_Soze has joined #nixos
nckx has quit [Quit: Updating my GNU GuixSD server — gnu.org/s/guix]
Mr_Keyser_Soze has quit [Ping timeout: 276 seconds]
andreabedini has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<hoshineko>
wait, nevermind, i was being stupid
<Havvy>
Upgrading from 17.09 to 18.03, KDE5 is only half functional. There's no desktop or application bar but yakuake still works, although with lots of graphical issues. X.X
<{^_^}>
[nixpkgs] @aneeshusa opened pull request #41976 → salt: be Python version agnostic → https://git.io/vh6D8
Ariakenom has joined #nixos
andreabedini has joined #nixos
asuryawanshi has joined #nixos
<{^_^}>
[nixpkgs] @aneeshusa opened pull request #41977 → Salt minion improvements → https://git.io/vh6Do
knupfer has joined #nixos
reinzelmann has joined #nixos
asuryawanshi has quit [Remote host closed the connection]
<tfc[m]>
hey there. i wrote a default.nix expression for building a c++ project. it contains `buildInputs = [ ... gmp.dev ... ];` and about this thing i have a question: in order to make things within `nix-build` see GMP headers, it has to be `gmp.dev` and not just `gmp`. from reading the docs i thought this would be done automatically, is that correct?
<Guanin_>
I'm running nixOS on a somewhat old laptop right now and have trouble with libGL/GPU drivers. The GPU is an AMD/ATI Mobility Radeon HD 5650. I tried amdgpu and ati as video drivers, but with both firefox and nextcloud are segfaulting (firefox takes the x server down with it :/)
<charlesh`>
Hi, I do not understand something with fetchFromGitHub
Guanin_ has quit [Remote host closed the connection]
<srhb>
charlesh`: Yes?
<charlesh`>
especially, the sha256 for fetchFromGitHub seems not to be the object Sha ?
<srhb>
charlesh`: Correct, it's a hash of the files as downloaded in your store
<charlesh`>
OK, what's the correct way to get that sha ?
<srhb>
charlesh`: rev is the regular git sha
<srhb>
charlesh`: nix-prefetch-url or just put in a wrong sha and watch the error message.
<srhb>
!tofu
<charlesh`>
yes, for rev I got it, it may be a release
<{^_^}>
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.
<charlesh`>
ok
<srhb>
charlesh`: Nowadays nix will move the download into place under the right sha, so it should only download once anyway.
<charlesh`>
thanks for your answers !
<srhb>
You are welcome. :)
<Guanin_>
sorry if I missed something, X got killed again
<Guanin_>
so it seem to happen some time after libGL failed (with r600_dri.so missing and similar messages)
Turion has quit [Ping timeout: 264 seconds]
<Guanin_>
ok, the error is "libGL error: unable to load driver: r600_dri.so"
<pie_>
i tried to do this but when i try to import glfw in python it complains that it cant find the shared library: pypi2nix -v -V "3.6" -e imgui[glfw] -e glfw -E glfw
<pie_>
raise ImportError("Failed to load GLFW3 shared library.")
selaux has joined #nixos
<joepie91>
trying to install a markdown viewer that's written in haskell; apparently, I first need to install the universe :)
<gchristensen>
this is because of mutable user support
<gchristensen>
and the ability to create a user without specifying a uid
<gchristensen>
two things I don't love
<Myrl-saki>
Is there a "dump" attrset in config? An attrset which accepts anything, basically.
hoshineko has quit [Ping timeout: 240 seconds]
<infinisil>
types.unspecified
<Myrl-saki>
infinisil: I mean in config itself.
<infinisil>
Needs an option
<Myrl-saki>
Right. Technically, you could just make `options.dump = mkOption ..` irght?
<infinisil>
Yeah
hoshineko has joined #nixos
schoppenhauer has joined #nixos
<Myrl-saki>
Oh right. I accidentally made this mistake which didn't cause an error. I'm thinking it's not an error on purpose.
<Myrl-saki>
Can you make an option attr, not an option? Basically, a constant.
<infinisil>
readOnly = true
<Myrl-saki>
I'm thinking it's useful for something like
<Myrl-saki>
oh
johanot2 has joined #nixos
<Myrl-saki>
infinisil: default + readOnly?
<infinisil>
Yup
<Myrl-saki>
Thanks.
freeman42x]NixOS has joined #nixos
freeman42x]NixOS has quit [Remote host closed the connection]
freeman42x]NixOS has joined #nixos
<Myrl-saki>
Oh yeah.
Neo--- has quit [Read error: Connection reset by peer]
Neo-- has quit [Read error: Connection reset by peer]
<Myrl-saki>
What's the technical difference between { x = null; } and { }? I'm guessing it's the set's type?
Rusty1_ has joined #nixos
Neo--- has joined #nixos
Neo-- has joined #nixos
<srhb>
Myrl-saki: Not really type. One has an attribute called x, the other does not.
<Myrl-saki>
srhb: Right. I'm wondering why it wasn't encoded that {} <-> { foo = null; }
<srhb>
Ah, so you're asking why nix does not consider them equal?
<Myrl-saki>
srhb: Pretty much.
<srhb>
That sounds like the kind of quesitons only eelco can answer. Language design choice.
<srhb>
I can only conclude that they are semantically different in the language. :P
<gchristensen>
well x is defined in the first one and the second one doesn't
<Myrl-saki>
gchristensen: My idea is basically every unassigned attr is null.
<gchristensen>
but it isn't
<gchristensen>
nix-repl> { }.foo
<gchristensen>
error: attribute 'foo' missing, at (string):1:1
<gchristensen>
nix-repl> { foo = null; }.foo
<gchristensen>
null
<Myrl-saki>
gchristensen: I mean, why is it not designed that way?
<__monty__>
Myrl-saki: Why would you want it to be? Null is muy annoying.
<Myrl-saki>
I can see a few cases, one is that it's sometimes better to error than just null.
<Myrl-saki>
Wait, BRB, dinner.
<gchristensen>
yeah, the major point to me is pretending unset attrs is null lets you be super sloppy
xy2_ has quit [Ping timeout: 245 seconds]
xy2_ has joined #nixos
MP2E has quit [Remote host closed the connection]
<charlesh`>
Do you have a language exemple where this kind of set relation holds ?
<charlesh`>
{ } == { x = null; }
jtojnar has quit [Quit: jtojnar]
<__monty__>
charlesh`: I'm thinking maybe perl because of autovivification?
jtojnar_ has joined #nixos
jtojnar_ is now known as jtojnar
camsbury has joined #nixos
<tilpner>
Myrl-saki - Early failure is good. If ({ }).foo was null, every typo could silently break things
<tilpner>
(And usually not in the place you'd expect)
<charlesh`>
__monty__: indeed, didn't know about that perl design
<__monty__>
charlesh`: I'm actually not sure perl's comparison would accept such different values but it's conceptually similar because something that "is" empty can "contain" any attribute. (Was trying to clarify pretty sure I failed.)
<charlesh`>
Yup, i'm not sure either perl accept the equality checks. Clojure has a similar semantic where {}.x == {x = null;}.x but {} isn't equal to {x = null;}
<__monty__>
Yeah, that's the behavior I'd expect from perl too.
<gchristensen>
"In the Perl programming language, autovivification is the automatic creation of new arrays and hashes as required every time an undefined value is dereferenced."
<hodapp>
that hurts me
<{^_^}>
[nixpkgs] @nlewo opened pull request #42010 → skydive: init at 0.17.0 → https://git.io/vhiBg
<__monty__>
hodapp: It's something some people rave about but that sounds like lunacy to us reasonable folk.
<LoneTech>
so it's like applying the defaultdict behaviour to the existence of the dictionary itself?
<hodapp>
though, I guess Go even sort of does this for certain data types (e.g. nil in Go is basically a 0-length list)
<__monty__>
LoneTech: Yes.
<infinisil>
Myrl-saki: There's something related though:
<infinisil>
> { ${null} = 10; }
<{^_^}>
{ }
pie_ has joined #nixos
<charlesh`>
> ${null}
<{^_^}>
error: syntax error, unexpected DOLLAR_CURLY, at (string):146:1
<charlesh`>
${null}
<charlesh`>
what does ${null} evaluates to ?
<gchristensen>
{ ${null} = ...; } is a dirty trick to support some dirty code
<infinisil>
You can use the ${..} syntax (or "${..}") in attributes to set a dynamic attribute name
<gchristensen>
and probably doesn't need to exist :)
<infinisil>
And null is a special case which doesn't set it
<__monty__>
But that still doesn't give you {}.foo == null. That's more like {}.null == somethingNotNull.
<infinisil>
Yeah, that's why it's just something related
<__monty__>
What I'm wondering rn: Who could possibly put this to use?
<infinisil>
The ${null} = ..?
<charlesh`>
that one, and also when do you want { foo = null; } == { } ?
<gchristensen>
nixpkgs uses a ${null} once
<gchristensen>
except its ${avarwhichmaybenull}
<__monty__>
charlesh`: I think what Myrl-saki wants is not { x = null; } == {} but autovivification.
<infinisil>
Yeah, some attribute which should only be set on certain conditions
<kiloreux>
I am trying to use NIX_CFLAGS_COMPILE inside my nix expression. But the values aren't being reflected on the compilation. Any idea how I can set gcc compiler flags /
<infinisil>
__monty__: Usually one would just do this:
<gchristensen>
__monty__, charlesh`, Myrl-saki, oh then they could use the library function for something like it
<infinisil>
> { }.someattributewhichmayexist or null
<{^_^}>
null
<infinisil>
> { foo = 10; }.foo or null
<{^_^}>
10
<__monty__>
gchristensen, infinisil: Why can't that be inverted? { attr=null unless it should exist; }?
<gchristensen>
huh?
<gchristensen>
yeah, so I don't know why nixpkgs doesn't use the more sane option
<gchristensen>
being "cute"
<infinisil>
Hmm, I guess it could, I don't see anything that wouldn't work with that right now
<__monty__>
Ah, being clever for clever's sake.
trevorriles has joined #nixos
<infinisil>
clever++
<{^_^}>
clever's karma got increased to 5
<{^_^}>
clever's karma got increased to 1
<infinisil>
Oh, sry
<infinisil>
(2 bots running)
<__monty__>
O.o two bots with the same nick?
<__monty__>
What is this voodoo?
<gchristensen>
{^_^} is an IRC <-> RabbitMQ gateway
<gchristensen>
all the messages get broadcast to multiple workers, which can choose to reply or not
<infinisil>
And i happen to run my backend along with the same backend but an experimental version of it right now
<gchristensen>
this is so that you don't see a {^_^} leave and join every time a teeny tiny code change has to happen
<tilpner>
__monty__ - ZNC could do the same thing
knupfer has joined #nixos
deepfire` has joined #nixos
<gchristensen>
the IRC <-> RabbitMQ gateway is like 50loc, so not a lot to mess up or fix. the last thing I fixed was detecting if infinisil's karma bot replied with "Did you mean infinity? Increasing infinity doesn't do anything.." and substituting it with a different message.
<infinisil>
I'm confused about platform/architecture/system myself though
johanot2 has quit [Ping timeout: 248 seconds]
<__monty__>
fendor: If you mean why is 8.2.2 the most recent, I think that's because it's the most recent stackage LTS.
<fendor>
__monty__, oh, that makes sense
<fendor>
infinisil, how can i see the haskell compilers? nix search haskell.compilers does not work :(
<__monty__>
fendor: Maybe haskell.compiler
* pie_
tries disabling his driver line
<infinisil>
fendor: Hmm yeah nix search doesn't list them, but they are there. You can use `nix-build '<nixpkgs>' -A haskell.compiler.<tab>` to see the possible values
<phry>
I'm having a packge build that calls `python .. +launch $(which sphinx-build)`. The problem is that sphinx-build is a wrapped python script, so python tries to interpret shell as python. Anyone familiar with a good workaround for that one?
shah^ has joined #nixos
<pie_>
ok switching to intel driver fixed
<mupf>
Hello guys, I'm having trouble patching a binary. Also I'm quite new to nixos. I tried sticking to the manual but 'ldd' still complains:
<mupf>
I used nix-locate to find the package required to fulfill these requirements.
<mupf>
still.
<phry>
did you patch the loader itself?
<mupf>
I did
<mupf>
I also fixed a few other dependencies using patchelf
<mupf>
but till now, I always had a few left.
<phry>
can you post a pastebin of what you patched so far?
<tilpner>
And are you sure you want to patch it? There's easier ways to just run it
<infinisil>
phry: Can you just change the build phase to call the wrapped binary directly?
<__monty__>
infinisil: What's the best way to find out the default nixpkgs ghc? The one that haskellPackages is compatible with?
<infinisil>
__monty__: Looking at the definiton of haskellPackages
<Myrl-saki>
infinisil: __monty__: gchristensen: Hey, I'm back now!
<Myrl-saki>
__monty__: What's autovivification?
<mupf>
phry: I will, but busy right now. I mention you later. tilpner easier ways?
<phry>
infinisil: yes, but then of course it has missing dependencies. I tried patching the PATH of sphinx-build into the Makefile calling python that then calls.sphinx-build-unwrapped.
<tilpner>
mupf - Try running it with steam-run. nix-shell -p steam-run --run 'steam-run ./your-game'
<phry>
but that introduces a whole new problems :/
<mupf>
phry: I basically patched the rpath with "with pkgs; lib.makeLibraryPath [ openssl openal xorg_sys_opengl stdenv.cc.cc.lib ]
<phry>
Myrl-saki: the script is interpreted as if it were python altough it is bash, actually wrapping python
<phry>
and I don't know enough about python to fix it from that angle -.-
<Myrl-saki>
phry: Right. What I mean is that why do you need to call the script(with its dependencies)?
<infinisil>
Myrl-saki: 15:26:04 <gchristensen> "In the Perl programming language, autovivification is the automatic creation of new arrays and hashes as required every time an undefined value is dereferenced."
<phry>
Myrl-saki: that's part of the build process of kitty
<__monty__>
phry: But kitty's already packaged, no?
<Myrl-saki>
infinisil: Thanks.
<phry>
__monty__: not v0.11.0, which seems to have some changes around building the documentation
<phry>
(about 250 commits to be exact)
<Myrl-saki>
phry: I don't understand it either, but I'd rather patch the build script instead.
<__monty__>
phry: And you can't call the script instead of python...script.sh?
shah^ has quit [Ping timeout: 248 seconds]
<phry>
__monty__: nope, then it complains about the python package "kitty" not being vailable
<Myrl-saki>
phry: Wrapped scripts accept arguments. Have you tried those ?
<Myrl-saki>
phry: I also don't have much experience in Python, so yeah.
<phry>
hrm... :/ had a bit of a breakthrough, but now it tries to write into /homeless-shelter... it's never easy *sigh*
johanot has quit [Quit: Leaving.]
sorixelle has quit [Ping timeout: 265 seconds]
Maxdamantus has quit [Ping timeout: 260 seconds]
Maxdamantus has joined #nixos
<pie_>
is there an easy way to install nixos on a different drive if im already in a nixos system?
thetet has joined #nixos
__monty__ has quit [Quit: leaving]
lonokhov has joined #nixos
<phry>
pie_: just like you installed nixos in the first place: nixos-generate-config (or copy it over) and nixos-install
<pie_>
i installed from a livecd or something
<pie_>
its been a while :D
<phry>
which actually was a fill nixos installation, just on a cd ;)
<phry>
so everything from the manual on installing nixos applies to you, except the "insert usb stick" part ;)
<pie_>
sure, i figured, i just dont remember much
<pie_>
ok cool
<pie_>
can i install an up to date system with an old installer?
<phry>
I believe it will install the version that the host system is running at. so if you isntall it from your running system, the target will be the same state
<ryantm>
hodapp: looks like it's working for me. I'll paste some stuff in a sec.
<hodapp>
ryantm: it's definitely not checking out submodules for me. CMake complains of it, and if I run with nix-shell -K I see the directories in question are empty
<ryantm>
hodapp: when you specify the hash of a derivation it's called a fixed-output derivation, and it tries to get it from the store.
<ryantm>
hodapp: If you alter a fixed output derivation, you need to make the hash something nonsensical, or use nix-prefetch-url to get the new hash
igo95862 has joined #nixos
<ryantm>
hodapp: I usually just replace a few characters of the hash with q
Ralith_ has quit [Ping timeout: 264 seconds]
<hodapp>
raaaawr... why do so many CMake builds have undeclared dependencies
Ralith_ has joined #nixos
<{^_^}>
[hydra] @rbvermaa pushed to master « Make /api/push-github public for private Hydra instances. This makes it possible to use webhooks to trigger evaluations … »: https://git.io/vhiPp
<sphalerite>
hodapp: because non-nix systems let you get away with that sort of thing? :)
<infinisil>
nh2[m]: I didn't know there was extend for the compilers though
<sphalerite>
hodapp: but it does keep track of all the sources, including the build scripts themselves, and rebuilds everything that needs to be rebuilt when changes occur
<sphalerite>
hodapp: of course that's not helpful in the nix context, but for incremental compilation it's very valuable
Ariakenom has quit [Ping timeout: 245 seconds]
<sphalerite>
(and of course a system like bazel's, at least as far as I understand it, is still better)
MercurialAlchemi has quit [Ping timeout: 265 seconds]
winem_ has quit [Ping timeout: 268 seconds]
jperras has quit [Quit: WeeChat 2.1]
<tilpner>
Anyone good at Rust packaging here? I'm trying to use https://github.com/spotifyd/spotifyd , but I can't do it with either buildRustPackage or carnix :(
<infinisil>
Not an expert either, but what's the problem with carnix?
<tilpner>
I think carnix is buggy, uses the wrong URL
<tilpner>
But I got past that by removing that [replace]
justanotheruser has quit [Ping timeout: 264 seconds]
<tfc[m]>
hey there, i have a `release.nix` that returns a set with a lot of different keys that have derivations as values. what command can i tell a colleague that would quickly list what keys that set has on the shell?
<tfc[m]>
`nix-instantiate` just gives out the hashed paths that don't say much
<tilpner>
infinisil - And then carnix fails on error[E0463]: can't find crate for `base64`
<infinisil>
tilpner: Using latest version of carnix? (0.8.5)?
<tilpner>
Huh
<tilpner>
Why is master only at 0.7.2
<infinisil>
Oh it might not be in nixpkgs yet I guess, I just checked crates.io, and it seems to often have new versions
<tilpner>
I'll try that one
<infinisil>
tfc[m]: `nix-instantiate --eval -E 'builtins.attrNames (import ./release.nix)'` should work
<infinisil>
Or `nix eval '(builtins.attrNames (import ./release.nix))'`
<tfc[m]>
infinisil: yes that works, thanks! i did however hope for something shorter. nevermind, this works for me. :)
<infinisil>
Hmm, I think there's a way with nix-env
trevorriles has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<tfc[m]>
infinisil: it prints the same package name all the time... the thing is that i am not altering the `name` field of the derivation because i am just building it in different variants (compiler, with and without sanitizers, etc.), so i would not like to rename the actual package name.
migy_ has quit [Quit: migy_]
trevorriles has joined #nixos
<tfc[m]>
but the set contains `packagename-variant1` `packagename-variant2` etc.
<tfc[m]>
so the nix-instantiate commandline that you just posted before lists exactly those
<infinisil>
Yeah, the nix-env variant lists derivation names, not attribute names
<tilpner>
error: attribute 'rand' at /cfg/pkgs/spotifyd/spotifyd/Cargo.nix:1053:5 already defined at /cfg/pkgs/spotifyd/spotifyd/Cargo.nix:1043:5
<infinisil>
tilpner: Hmm, I'd patch that Cargo.toml file, but not sure how to do that
<tilpner>
infinisil - Manual editing is acceptable
<tilpner>
(But not of the Cargo.nix)
Lev50 has joined #nixos
<tilpner>
What would you change?
<infinisil>
tfc[m]: Auto-completion should work too: `nix-build release.nix -A <tab>`, should list all attributes
<infinisil>
tilpner: What's Cargo.nix?
<tfc[m]>
yes you're right! this is perfect. thanks man!
<infinisil>
:)
matthewbauer has quit [Remote host closed the connection]
<tilpner>
infinisil - Even after manually fixing Cargo.nix duplicates, it takes an argument "buildRustCrateHelpers", which I don't know how to provide
<infinisil>
Phew, out of my league
<tilpner>
pemeunier - Are you here by any chance? (I know it says AFK for 11 days, but...)
<nh2[m]>
domenkozar: you need to tune the cachix upload parallelism or server settings a bit, my gigabit internet is very bored at 5 Mbit/s :) What protocol/approach is it currently using for uploading?
hotfuzz_ has joined #nixos
<cransom>
is there a #cachix? if not, maybe there should be.
<pie_>
uh how do i upgrade nix before up do --upgrade for channels
<pie_>
i got some warnings about "unknown setting signed-binary-caches" but it hasnt failed yet
<{^_^}>
[nixpkgs] @matthewbauer merged pull request #42023 → ghcjs: add version to haskellCompilerName → https://git.io/vhiMZ
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/vhi71
<norfumpit>
im doing an upgrade to unstable channel, and i see nix is compiling libreoffice from source. is this normal? shouldn't it fetch from cache?
<pie_>
norfumpit, might not exist in the binary cache then yet
<norfumpit>
pie_: is that likely?
<pie_>
well since its unstable, i guess yeah
<pie_>
i mean im not sure :D
<{^_^}>
[nix] @matthewbauer opened pull request #2230 → Fix optimise-store.cc to skip more paths on macOS → https://git.io/vhi77
<norfumpit>
can i somehow opt out of compiling packages from source? like, don't upgrade unless theres a binary
silver_hook has quit [Ping timeout: 256 seconds]
slyfox has quit [Quit: :(]
avn has quit [Ping timeout: 240 seconds]
matthewbauer has joined #nixos
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
<vorot93[m]>
after upgrading to unstable my desktop just keeps suspending when screen is off. Why is that? Tweaking logind via configuration.nix didn't help
<vorot93[m]>
I have `services.logind = { lidSwitch = "ignore"; extraConfig = "IdleAction=ignore" };` in configuration.nix.
<roconnor>
Um, is there a package that will give me access to a 32-bit gcc on a 64-bit machine, if this question even makes sense?
matthewbauer has joined #nixos
<roconnor>
ah stdenv_32bit!
astrofog has quit [Remote host closed the connection]
fendor has quit [Ping timeout: 276 seconds]
Guest21370 has quit [Ping timeout: 245 seconds]
knupfer has quit [Ping timeout: 255 seconds]
Sonarpulse has quit [Ping timeout: 276 seconds]
keegans has joined #nixos
<Unode>
is there any switch to nix-env that allows me to know a priori how much space is a --dry-run going to take? (best estimate assuming no package will be built from source)
<Unode>
domenkozar: Using cachix for the first time after https://github.com/NixOS/nixpkgs/pull/41964 . Derivation is pulling quite the list of dependencies and a good chunk of them is going to be compiled locally. If I understand the purpose of cachix, shouldn't it cache itself and its dependencies?
<Lev50>
Running pip install in nix-shell is giving me an error like 'error: could not create '/nix/store/<hash>-python3-3.6.5/lib/python3.6/site-packages/multidict': Permission denied'
<Lev50>
It looks like multidict needs to be built as part of the install process, and it can't write to /nix/store/....
sir_guy_carleton has joined #nixos
<Lev50>
Is the solution to this to put python36Packages.multidict in default.nix?
trevorriles has quit [Read error: Connection reset by peer]
<avn>
lev50: is one of options. Or use any of tools making .nix from requirements.txt
johanot has quit [Remote host closed the connection]
justanotheruser has joined #nixos
<judson>
I'm returning to an old issue I've had (on my own): my system gets confused about which chromium to use. I've started to figure out some of the xdg-settings stuff...
<judson>
But it seems like after update, the new chrome executable doesn't recognize the running one.
<judson>
I've imagined in the past it was because the nix-store path to the exe changes... but folks here report that they have no such issue.
work_ has joined #nixos
<{^_^}>
[cabal2nix] @peti pushed to master « update-stackage.sh: ignore Stackage constraints on json-autotype »: https://git.io/vhPfb
<judson>
Dimly related: zoom-us from current unstable an master of the repo: sh: /nix/store/2kcrj1ksd2a14bm5sky182fv2xwfhfap-glibc-2.26-131/lib/libc.so.6: version `GLIBC_2.27' not found (required by /nix/store/8h26kzkrx1mpm2lxkky6yvd764985phh-v4l-utils-1.14.2/lib/libv4l2.so.0)
<judson>
Oh, dammit. It's that weird OpenGL hack thing.
alexteves_ has quit [Remote host closed the connection]
camsbury has quit [Ping timeout: 276 seconds]
camsbury has joined #nixos
mkoenig_ has joined #nixos
mkoenig has quit [Ping timeout: 248 seconds]
xy2_ has quit [Ping timeout: 256 seconds]
sanscoeu_ has joined #nixos
nschoe has joined #nixos
sanscoeur has quit [Ping timeout: 260 seconds]
<catern>
how fast will python37 appear in nixpkgs? should I just go ahead and make a pull request to add python37Packages with the rc?
<catern>
I want to take a hard dep on python37 for some new software I'm writing :)
<Mic92>
catern: we had a pull request for that open right away with the first alpha or so
<Mic92>
should not take that long.
<catern>
ah okay, great
<catern>
then I will wait in excitement :)
<Mic92>
it may be not become the default 3er version though
<catern>
that's fine
<Mic92>
what is the interesting news of that version? data classes?
<{^_^}>
[nixpkgs] @Ericson2314 opened pull request #42034 → darwin bootstrapping: Avoid overriding aliases for LLVM → https://git.io/vhPkN
<catern>
a number of things, dataclasses, contextvars, 'lazily' evaluated type annotations
<catern>
it's nothing super exciting but it's stuff I'd like to have
asymmetric has joined #nixos
knupfer has joined #nixos
<Mic92>
I am not sure why there is an HTMLCalender in the standard library
Olgierd has quit [Quit: WeeChat 1.9.1]
roberth has quit [Ping timeout: 245 seconds]
xy2_ has joined #nixos
Lev50 has quit [Ping timeout: 268 seconds]
woodson_ has joined #nixos
asymmetric has quit [Read error: Connection reset by peer]
<dtz>
grahamc: your latest comment on the Nix issue got formatted in a silly way haha
<rycee>
Anybody know whether it is at all possible to convince Firefox to look for add-ons as XPI files somewhere inside `~/.nix-profile`?
sigmundv has quit [Ping timeout: 240 seconds]
wrl has joined #nixos
stites has quit [Quit: WeeChat 1.6]
<rycee>
I was looking, e.g., at http://kb.mozillazine.org/Installing_extensions but it only mentions the profile directory and the Firefox installation directory, which is not very promising.
<colemickens>
So, how do I get nix-copy-closure to use a nonstandard ssh port?
sanscoeu_ has quit [Remote host closed the connection]
<colemickens>
I'm very surprised this is so impossible to discover.
<makefu>
colemickens: maybe use .ssh/config ? not sure though
<makefu>
if you find how to do it, put it in the wiki :D
<colemickens>
I don't want to do that. I need a 'portable' command I can invoke.
<colemickens>
Will do. I need to start contributing back to it more. I miss Arch Wiki.
<colemickens>
(btw, I think your suggestion works, it very well may be the only supported method. I'm partially frustrated because I swear I went through this same question/discovery process three weeks ago and can't remember what I figured out.)
<makefu>
https://nixos.wiki is there for you. the content is not as advanced as the arch wiki but we are doing our best
<makefu>
NIX_SSHOPTS
<makefu>
according to `man nix-copy-closure`
<makefu>
NIX_SSHOPTS='-p 2222' or something
<{^_^}>
[nixpkgs] @Ericson2314 merged pull request #42034 → darwin bootstrapping: Avoid overriding aliases for LLVM → https://git.io/vhPkN
<{^_^}>
[nixpkgs] @matthewbauer opened pull request #42037 → Fixes for GMP + prebuilt mobile → https://git.io/vhPqf
<{^_^}>
[nixpkgs] @Ericson2314 pushed 44 commits to staging: https://git.io/vhPqm
<colemickens>
and now I feel bad, it was in the man. :/ thank you @makefu, I'll go right now and try to find a good wiki spot to put that with some keywords.
<colemickens>
are there good places in the wiki for "scenarios"
<colemickens>
like, "building your laptop configs on your beefy server and then copying the closures around".
<colemickens>
There are random blog posts around, but it'd be nice to document the "scenario" end-to-end, if nothing else, for the next time I need to do this and have forgotten how.
<makefu>
colemickens: i think there is nothing like this yet
jmiven has quit [Quit: co'o]
<makefu>
but you could just create a page with your scenario
<judson>
I think the opengl mesa-noglu drivers are linking against glibc 2.6 somehow.
<judson>
I mean, ldd says so, but I can't find the why...
jmiven has joined #nixos
<samueldr>
colemickens: I like using `nixos-build ... --build-host $remote` for this :)
<colemickens>
I like to build the config with `nix build` first, which doesn't seem to have that option, very confusingly.
<samueldr>
any reason for nix build instead of nixos-rebuild build?
<colemickens>
I'm a bit unclear on why `nix build` and `nix-build` both exist, which one is the way forward, if `nix build` is going to grow the remote build options, etc.
<colemickens>
(it does appear it special case 'localhost' though, I had to use '127.0.0.1' to get it to actually use ssh and build on remote), and now I'm off to figure out how to do remote builds w/o root ssh access.
alex`` has quit [Ping timeout: 245 seconds]
<dtz>
fwiw I do remember disabling libgmp in the ghc builds for some reason, might be same reason for what you're seeing... or might be cause of what you're seeing :)
rihards has quit [Quit: rihards]
alex`` has joined #nixos
<infinisil>
Nix quiz: What does the following evaluate to: let x = "foo"; in ({ x ? x }: x) {}
<judson>
Can anyone help me? I cannot get zoom working on NixOS. It and v4l_utils (its dep) build against glibc-2.27, but the mesa-noglu OpenGL driver (also a dep) build against Glibc-2.26. It crashes at runtime. But I can't see glibc as a buildInput anywhere.
<judson>
Maybe I need to enable nvidia on this machine...
myshoe has quit [Quit: Lost terminal]
guest-35gcfv has joined #nixos
<guest-35gcfv>
hello everyone
simukis has quit [Ping timeout: 256 seconds]
Ariakenom has quit [Quit: Leaving]
<guest-35gcfv>
i’ve been looking in the documentation, and I can’t seem to find an option to boot into multi-user instead of graphical target. anyone who can tell me where to look or have a quick fix? :)
<guest-35gcfv>
well, set the option in configuration.nix that is.
<dtz>
re:zoom-us might be easiest to just install zoom from same channel providing your drivers
<dtz>
silly libraries
<infinisil>
guest-35gcfv: what do you mean by 'multi-user'?
<srhb>
Yes, zoom-us works with master on master(/unstable) or 18.03 on 18.03
chrisaw has joined #nixos
<guest-35gcfv>
infinisil: boot in to console mode instead of the default display manager.
<srhb>
It's an unfortunate side effect of not taking into consideration the differences caused by libglvnd between the versions. My bad.
<avn>
dtz: I better say "silly electron", which have 100500 versions and all of them shipped as compiled binaries
<chrisaw>
How long does it usually take for a change which happened in nixpkgs repo to make it through to unstable channel?
<infinisil>
guest-35gcfv: Ah, so you just don't want it to start X at the beginning?
robstr has quit [Quit: WeeChat 1.9.1]
<dtz>
lmao I did not know that
<srhb>
chrisaw: It depends. Last time it was about 11 days. You can check the graphs, let me find the link...
<infinisil>
guest-35gcfv: X is only started when you have `services.xserver.enable = true;`
<guest-35gcfv>
infinisil: exactly. maybe i should just have started out with that hehe
organixpear has joined #nixos
<organixpear>
Hello. I need some help.
<guest-35gcfv>
infinisil: so if i install a window manager, i'm still able to start it?
alex`` has quit [Quit: WeeChat 2.1]
<chrisaw>
srhb: Thanks! Likely a daft question but is there a channel which builds directly from nixpkgs more quickly? Need a fix from nixpkgs for a couple of packages. I submitted issues in GitHub and they're both fixed now but waiting for that fix to trickle through the channel I'm using.
<srhb>
chrisaw: nixpkgs-unstable-small if there's such a thing. or cherry pick. Or pull in just a few packages from straight up master.
<organixpear>
I am trying to run a 3D windows game using wine on the latest x86_64 nixos. When I try to run the game I get: libGL error: unable to load driver: i965_dri.so. I have mesa-noglu installed
<guest-35gcfv>
infinisil: might have read the option description a bit wrong then...
<infinisil>
guest-35gcfv: Ah, with that option i think everything X is disabled. So you want it to have everything from X, but just not start when booting up?
<guest-35gcfv>
infinisil: yeah
<infinisil>
Ah hold on
knupfer has quit [Remote host closed the connection]
<elvishjerricco>
srhb: Unfortunately, I believe the right column is the timestamp of the commit, not the time the channel advanced
<guest-35gcfv>
infinisil: oh, thats make sense. gonna give try :) thank you!
<srhb>
elvishjerricco: Really? That makes it rather not useful. I thought it reflected the data in the graph (which by itself also does not identify the delta)
<srhb>
Oh well
<elvishjerricco>
srhb: I could be wrong. I think I remember checking the commits and finding the exact same time stamp.
Intensity has quit [Ping timeout: 260 seconds]
<gchristensen>
would it be terrible if I changed the meaning of the second column?
<{^_^}>
[nixpkgs] @matthewbauer opened pull request #42050 → ghc8.4: build statically on prebuilt iOS → https://git.io/vhPl2
<elvishjerricco>
gchristensen: Depends on how many people use it. I'd love it for that column to be the advance time
<gchristensen>
I haven't fixed it because I don't want to mess people up
<{^_^}>
[nixpkgs] @yegortimoshenko pushed 2 commits to master: https://git.io/vhPlK
<elvishjerricco>
gchristensen: Maybe you should offer a parallel history.json file, so that you can always add fields without destroying any reasonable workflow?
<gchristensen>
that would be reasonable
<gchristensen>
but would probably involve mmaking the tracker script better than a shell script
<elvishjerricco>
Hah, you don't like writing code generators in bash? :P
<infinisil>
gchristensen: My backend stays connected fwiw
<chrisaw>
infinisil: That's what I'm doing - I think to be honest I'm just matching to the group group.
aarvar has joined #nixos
<samueldr>
ah good, must have skimmed over the paragraph, sorry
MP2E has joined #nixos
<gchristensen>
samueldr: it is new as of after your question :)
<samueldr>
oh
<gchristensen>
srhb: ^
<gchristensen>
no longer useless!
matthewbauer has joined #nixos
solaire has quit [Quit: Leaving]
<chrisaw>
infinisil: Figured it out - I needed to use python3 instead of just python and I also needed to use overridePythonAttrs instead of just overrideAttrs. Thanks for the hint!
<infinisil>
Nice :)
<chrisaw>
infinisil: Oh... doh. Spoke too soon. :(
thetet has quit [Read error: Connection reset by peer]
<colemickens>
lol no matter what I do, `autossh [anything]` just prints help.
<colemickens>
wow, how do I keep running into commands today that don't parse args properly.
<colemickens>
It doesn't handle -M0 (-M 0) and nixos-rebuild doesn't accept --long-flag=value (--long-flag value)
matthewbauer has quit [Ping timeout: 245 seconds]
<infinisil>
colemickens: autossh is pretty much useless, ssh itself can do it better
pie_ has joined #nixos
* colemickens
is lazy to look up flags
<colemickens>
you're right though, I'll drudge through my dotfiles and figure out where my alias is from last time I did this.
<infinisil>
I have my config ready with this, hold on
<colemickens>
samueldr: how do you deal with signing when doing nixos-rebuild on a remote --build-host? I keep getting signing complaints, even after provisioning keys and distributing the public key to my box where I'm starting the build.
<colemickens>
it's not so bad that I can't figure it out ;)
<colemickens>
thanks!
phdoerfler has quit [Ping timeout: 240 seconds]
<infinisil>
That's something slightly different. You only need ServerAliveInterval, ClientAliveInterval (on the server), ExitOnForwardFailure, the Control* ones, and pack that into a restarting systemd service
<infinisil>
:)
hiratara has joined #nixos
tzemanovic has joined #nixos
<chrisaw>
infinisil: It doesn't seem to matter which package I override - the patch I'm trying to add never gets applied during a rebuild. It's very strange!
<pie_>
is there a consensus on the best way to currently manage user dotfiles declaratively?
<chrisaw>
Tried python, python3, python36 - none of them seem to make a difference to the actual build. :| Is there a simple way to check what "python group" (I guess?) a dependency belongs to?
<infinisil>
chrisaw: The `rec` is not needed, python -> python2 -> python27, so you probably want to override python27 instead, it should be patches = oldAttrs.patches ++ ..., are you sure it's overrideAttrs instead of overridePythonAttrs?
<infinisil>
Whether the overlay gets used depends on whether something needs that package, do you have warrant in your configuration.nix somewhere?
<infinisil>
Ah and you mentioned it's a dependency? If so, you need to do this override within the package that it depends on
<pie_>
ok so nix search is a thing but is there a nix search for config options
<pie_>
because i keep needing to use the nixos.org page
tzemanovic has quit [Remote host closed the connection]
<infinisil>
pie_: There's a man page: man configuration.nix
<chrisaw>
infinisil: Ah that could be the problem. So it's being pulled in from home-assistant - so I would need to tell that package to use my custom version?
camsbury has quit [Ping timeout: 248 seconds]
<infinisil>
chrisaw: Ye
tzemanovic has joined #nixos
<chrisaw>
infinisil: FYI - without the rec { neither python nor python36 (it's actually using python 3.6 so must need python3)
<pie_>
infinisil, oh god lol
<chrisaw>
*exists - missed the vital part. :|
<infinisil>
chrisaw: Huh? What's the error when you remove rec?
<pie_>
how do i fix man pages globally
<chrisaw>
infinisil: undefined variable 'python36', same with 'python3' and 'python' if I switch to those.
<infinisil>
chrisaw: gist that exact file?
<chrisaw>
Full file yeah
<infinisil>
In the gist you showed me, the rec really shouldn't have any influence on anything
<chrisaw>
infinisil: Then called in using nigpkgs.overlays = [ (import ./pkgs/warrant) ]; - nothing too fancy there.
<chrisaw>
*nixpkgs
<pie_>
for example i dont have a man page for resolve.conf
<pie_>
*resolv
<manveru>
i think that's in the manpages package
<pie_>
how....does that work?
<manveru>
hm, or not
<infinisil>
chrisaw: What rec does is this
<infinisil>
> { foo = 10; bar = foo; }
<{^_^}>
{ bar = <CODE>; foo = 10; }
<infinisil>
> { foo = 10; bar = foo; }.bar
<{^_^}>
{ value = { value = <CYCLE>; }; }
<infinisil>
Oh right, um
<infinisil>
> { fooo = 10; barr = fooo; }.barr
<{^_^}>
undefined variable 'fooo' at (string):149:21
<infinisil>
chrisaw: But the only attribute you are defining in that gist is "python", but you never reference this in the rest of the code, so the rec shouldn't have an effect..
<chrisaw>
infinisil: Ah I see! In the context of a pkg overlay though doesn't it basically expand out the pkgs group?
<samueldr>
committed 11 months ago, says "allows nix-copy-closure to work through trusted users"
<chrisaw>
infinisil: As in - with it - "python =" works, without it "pkgs.python" would probably achieve the same thing. I may be way off the mark here - still learning!
<colemickens>
@samueldr: oh interesting, that seems easier (if not questionably lazier) than provisioning and copying around keys, and skirts me around needing to deal with building on multiple machines.
<infinisil>
chrisaw: I can't reproduce, can you gist the file again that produces the "undefined variable 'python36'" error?
<pie_>
manveru, the description seems a bit misleading then, it talks about "developer" manpages in the description
<pie_>
not that its something particularly important to get hung up on
sanscoeur has quit [Ping timeout: 264 seconds]
<manveru>
i think those are just the manpages that aren't directly associated with a package otherwise
<manveru>
never really looked into it
<pie_>
what if i get something with nix-shell -p?
xcmw has joined #nixos
<manveru>
you should have the manpage in your nix-shell as well
<pie_>
ok so for -p fim, i guess the fim package just doesnt implement the man pages output or something?
xy2_ has quit [Ping timeout: 245 seconds]
<colemickens>
@samueldr I hadn't actually considered that I could stop cloning nixpkgs on all of my machines and manually syncing it. That's a pretty nice simplification to my workflow as well.
<manveru>
tbh i also never use -p, i write shell.nix files ^^;
<colemickens>
Ultimately, I'd like to get something that tries to rebase my custom commits on top of whatever channels/nixos-unstable-branch I'm tracking, and then rebuilds my nixos configs for all my machines, and then I can push/pull at will with whatever ssh/http strategy I want.
<manveru>
it's so short, and then i can just use direnv with it: `with import <nixpkgs> {}; mkShell { buildInputs = [ nixops crystal shards dbus libxml2 ]; }`
<colemickens>
I assume there might be a way to coax hydra into doing that for me
<samueldr>
colemickens: my manual sync involves a private remote clone with a branch
<infinisil>
chrisaw: You can remove that line, what you're intending to do there is like this already
jperras has joined #nixos
<samueldr>
as for custom commits, I have none. I may be lucky, never had trouble with simply using overlays
<colemickens>
Hm, I don't want to maintain an overlay just because I want some commits that haven't landed in my channel yet. Or maybe I'm overcomplicating things.
<colemickens>
Like I bumped up libinput and sent a PR, but I don't want to switch to 'master'.
<colemickens>
Is there an easy way to pull that PR effectively via an overlay? Or am I better off just maintaining those cherry-picks until nixos-unstable advances far enough to drop them?
<infinisil>
colemickens: You could define your nixpkgs to be the channels nixpkgs with the pull requests patch applied
chrisaw has quit [Ping timeout: 240 seconds]
<colemickens>
Hm, without maintaining my own branch and needing to rebase whenever the channel advances? Sorry to be needy, but you don't have an example do you? Or one I could extrapolate from?
<catern>
does anyone have a current, running example of using pypi2nix to deploy Sentry?
<chrisaw>
infinisil: Oh sorry, you were talking to colemickens.
<{^_^}>
[nixpkgs] @Ericson2314 opened pull request #42056 → llvm 6: Fix libcxxabi impurity and darwin sanitizers → https://git.io/vhPze
<chrisaw>
infinisil: I removed the line in question, solved the error. What I was asking though is - how can I actually ensure home-assistant uses my overridden package now? Or better yet - can I force it to be used globally?
<{^_^}>
[nixpkgs] @matthewbauer pushed to master « cc-wrapper: fix eval error on mac »: https://git.io/vhPzH
<colemickens>
Hm, is there a historical record of what nixos-unstable has pointed at?
Acou_Bass has quit [Quit: byeeeeeeeeeeeeeee]
Acou_Bass has joined #nixos
<colemickens>
infinisil: thank you very much by the way, this is definitely interesting to me.
seanparsons has quit [Read error: Connection reset by peer]
TonyTheL1on has joined #nixos
<infinisil>
colemickens: How you could use this as your normal nixpkgs: add a new channel (e.g named unpatched), use `import <unpatched> {}` at the top instead, then set in NIX_PATH nixpkgs=/path/to/my/gist.nix (nix.nixPath on NixOS)
TonyTheLion has quit [Ping timeout: 264 seconds]
<{^_^}>
[nixpkgs] @Ericson2314 merged pull request #42056 → llvm 6: Fix libcxxabi impurity and darwin sanitizers → https://git.io/vhPze