Synthetica has quit [Quit: Connection closed for inactivity]
<samueldr>
the main biggest limitation is that it will not print arguments of composed functions, but that's something that can't be done without executing the functions as some e.g. fetchFromGitHub are composed in different ways!
<drakonis>
samueldr, nice.
<samueldr>
but *calling* them with the composed arguments should work since it, in the end, it sloppily creates a nix expression
<samueldr>
I mean, I'm not sure of what I said previously, maybe I'm wrong, but I thought tests for 18.09 still were working like they should
<samueldr>
though this doesn't look like they broke, this looks like... nothing I've seen before
<worldofpeace>
yeah 404 wasn't the value I was expecting
<samueldr>
gchristensen: any idea about that? looks like the tests are marked as a success, but the log output is nginx's 404
<samueldr>
ah
<samueldr>
well, /undefined is not a valid URL
<samueldr>
so that explains the 404
<worldofpeace>
where's `/undefined` ?
<samueldr>
ah, looking at the network log
<samueldr>
I made the log viewer, if you didn't know
<worldofpeace>
excellent, I'm talking to the correct person for sure then
<samueldr>
ah, so the attempt doesn't have a log_url key :/
<mabel>
is there a reason to prefer nix-shell over nix run -f shell.nix?
<mabel>
(for using an isolated build environment for development)
<samueldr>
mabel: probably none, as long as it's for your use and not for scripting; `nix *` commands have not been declared as stable for scripting use
<mabel>
I think I want "nix run", otherwise the buildEnv derivation I declared won't actually be built.
<mabel>
unless there's a way around that with nix-shell... I guess I could make a dummy derivation that inherits its parent?
<mabel>
(so that the buildEnv comes in as a resolved dependency)
ottidmes has quit [Ping timeout: 240 seconds]
<samueldr>
worldofpeace: looks like the viewer itself isn't at fault, there seems to really be no log on the server :/
<samueldr>
it *looks* like the success state is good though, but cannot say 100% for sure, but something like 99% sure it's really succeeded
<siraben>
Any users of XFCE on HiPDI displays? How do I get it to scale up at the login screen?
<worldofpeace>
samueldr: Yeah it got a success state, and you're saying just because the log is missing doesn't mean it didn't do the thing.
<samueldr>
worldofpeace: yeah, there's a small backing API, and from a thousand foot view, it looks like it might bave misplaced the log, but the result should be good AFAICT
Purple-mx has quit [Quit: removing from IRC because user idle on matrix for 30+ days]
<jackdk_>
I am trying to add additional tools to the shell for a package I'm nixifying: https://pastebin.com/GbTgzJPJ nix-build works, so I think default.nix/mudcore.nix is okay, but nix-shell complains about coercing a set to a string
<worldofpeace>
samueldr: Now I'm on the watch if an misplacements occur again. Thanks for the reassurance :)
<samueldr>
worldofpeace: it might be tests related
<worldofpeace>
samueldr: That thought went into my head simultaneously
koschei[m] has quit [Quit: removing from IRC because user idle on matrix for 30+ days]
<mabel>
(alternatively, is there an easy way to fetch a pypi hash?)
<mabel>
,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.
<yl[m]>
mabel: the first path `nixpkgs in `<nixpkgs/applications/default.nix> ` is a channel name yes
<mabel>
so in that repo, they reference <nix-fetchers/...>
<mabel>
but that github repo does not appear to be a channel, is it?
<yl[m]>
it might help you avoid doing the make-extra-fetchers dance
<yl[m]>
shlevy: might be able to help you with this
cyounkin_ has joined #nixos
cyounkins has quit [Ping timeout: 246 seconds]
<jackdk_>
is there a way to get the date of a git commit through nix? I'm writing a default.nix which is checked into the repo it builds, and I don't want to update it by hand all the time.
<mabel>
hmm.... I think I should probably stick to tofu. this stuff is a few monads too deep for me. interesting concept though
<jackdk_>
I am aware of commitIdFromGitRepo, which seems to take me halfway to what I want (which is to have builds from git automatically get the -unstable-YY-MM-DD in their version
cyounkin_ has quit [Ping timeout: 240 seconds]
pie_ has quit [Remote host closed the connection]
pie_ has joined #nixos
lassulus_ has joined #nixos
<mabel>
argh, now I'm having fetchPypi give me a 404! but I'm able to download using pip install (in virtualenvwrapper, of course)
<mabel>
is there a log somewhere?
<mabel>
ohhh, I think it's because fetchPypi doesn't fetch wheels
lassulus has quit [Ping timeout: 250 seconds]
lassulus_ is now known as lassulus
<worldofpeace>
yl[m]: I responded on your gist :)
<mabel>
is there a way to get the log of a failed builder?
<mabel>
nix log doesn't show if there's only a drv
justanotheruser has joined #nixos
<worldofpeace>
mabel: fetchPypi can fetch wheels
<Church->
Thinking in goroutines hurts mah think meats
<worldofpeace>
mabel: give fetchpypi `format = "wheel";`
<{^_^}>
[nixpkgs] @tohl opened pull request #53434 → sbcl updated, tested on nixos x86_64 → https://git.io/fh3ZA
<worldofpeace>
justanotheruser: That executable is in `bluez`. Though if you're in nixos you should enable `hardware.bluetooth.enable` in configuration.nix
<mabel>
I sorta wonder if maybe we should just use pip for doing the fetching
<simpson>
Sadly pip is not cooperative with Nix. Maybe someday, but given the current direction of CPython I doubt it.
<worldofpeace>
^
<gchristensen>
pssst for ZFS on laptops: services.zfs.autoScrub.enable = true; services.zfs.autoScrub.interval = "daily"; systemd.services.zfs-scrub.ConditionACPower = true;
<mabel>
hm... what if we just replicated its URL fetching method?
<mabel>
as far as I could see, pip427 doesn't specify a URL convention, only a file name convention
<mabel>
notice that in my gist, the broken link and the working link both have the correct file name
<worldofpeace>
mabel: Ooh I didn't see your gist
<worldofpeace>
mabel: The filename for the working url uses a _ to seperate and not a -
<worldofpeace>
pname that is
<mabel>
ah! I was dumb. that works!
<mabel>
of course, it doesn't know how to unpack the wheel for some reason
<mabel>
but it did fetch it!
<worldofpeace>
mabel: Though I think you need a special condition for whether it's py2 or py3
<mabel>
yeah. right now I've hardcoded py3, but I'll fix it when I send a PR
<worldofpeace>
mabel: I'll be a reviewer for sure :)
<mabel>
"not know how to unpack source archive /nix/store/dfi3hqmph7k5k3kwa4fndqzgghya5a3z-komodo_pydbgp-10.2.2.89895-py3-none-any.whl"
<mabel>
odd... I specified format="wheel"
<worldofpeace>
mabel: move that out of `fetchPypi` and inherit it like you did with version
<worldofpeace>
`buildPythonPackage` takes `format` also
<mabel>
d'oh! noowww it works
<mabel>
thanks for the help! <3
<worldofpeace>
no problem :)
<jackdk_>
so I figure I can do something evil with builtins.runCommand but that's not really ideal. any thoughts?
<jackdk_>
er, it's not a builtin, but you know what I mean
endformationage has quit [Quit: WeeChat 2.3]
<fusion809>
I've rebuilt from nixos-unstable, with nixos-rebuild boot, after NIX_PATH points to nixos-unstable, and while the kernel and all software is at those in nixos-unstable, for some reason /etc/os-release still reads 18.09 as the version. Is there a way to change this? Make it read NixOS Unstable or 19.03? Besides manually editing it, which would only change it until system reboot.
<gchristensen>
jackdk_: to clarify, is this a security report?
Rusty1 has joined #nixos
sigmundv_ has quit [Ping timeout: 244 seconds]
<jackdk_>
gchristensen: no.
<gchristensen>
ah, ok *phew* what's up?
<jackdk_>
it's in scrollback, but to sumamrise: I'm hacking on a personal project, and the project comes with nix expressions to build it. I think I want to automatically generate versions like unstable-YYYY-MM-DD, and I've made that work using runCommand
<jackdk_>
this feels a bit evil, because it means that the build process pulls in git just to pick out a date, and it feels a bit IFDey
<gchristensen>
ah
<jackdk_>
when packaging external software in nixpkgs or whatever, it's easy because you can set the version and date when you pick a ref to build against
<jackdk_>
(the end goal is to have a personal overlay for my projects, because they don't need to go into nixpkgs, but that means doing dodgy things with git etc doesn't feel great)
<mdash>
jackdk_: mmm. take the date as a parameter then?
<jackdk_>
that was my initial thought too, but I'm not sure that's the best if I want a .nix file I can easily point at, call callPackage on, etc
<jackdk_>
I feel like I want a packagename.nix that stays like something that you can callPackage easily, but that's not a hard requirement (X-Y problem?)
<jackdk_>
then in default.nix I could do git-fu and an overrideAttrs to plug in the correct version
<jackdk_>
but if I do set up an overlay and point it to packagename.nix then I still don't know what version to call _that_. Maybe the correct thing is to curry the function?
<jackdk_>
date: { dep1, dep2, dep3, ..., depN }: let foo bar baz in stdenv.mkDerivation { etc } ?
<samueldr>
LOGO is recent nough (commit date 2018/10/11) so it's not like we missed a nixos release :)
<samueldr>
recent enough*
<worldofpeace>
nough said :P
<ldlework>
logo was a fun language
<worldofpeace>
Turtles are great
<dtz>
i like turtles
<dtz>
:)
<samueldr>
da turtle zealot
<worldofpeace>
I was waiting for that
<ldlework>
mi nelci lo tartaruga
<jackdk_>
is there a convenience function for running a shell command and capturing its output to a string (currently, I'm doing builtins.readFile (runCommand "name" ...))
lord| has quit [Ping timeout: 268 seconds]
<worldofpeace>
samueldr: We don't have that commit right? Since it's in v240
<samueldr>
nah, but isn't that only documentation?
<gchristensen>
jackdk_: since it isn't allowed in nixpkgs, no
<worldofpeace>
samueldr: Oh yeah should be good I think
<worldofpeace>
Fedora's had it for a while too
<samueldr>
worldofpeace: if something breaks due to that, AFAIUI they're not os-release compliant :)
<mdash>
jackdk_: the whole idea of a nixos config is that it's a static description of a single configuration, it should always produce the same output
<samueldr>
AFAIUI nothing in there is limited to those fields, and any field could be missing
<mdash>
mdash: so if you want a new build every day you'll hav eto do that outside the overlay
lord| has joined #nixos
<worldofpeace>
samueldr: Think so too, going for it. Probably need a release note for nixos.
<samueldr>
worldofpeace: why would it need a release note?
<samueldr>
unless I misunderstood something
<worldofpeace>
Look how cool we are we have a logo :)
<jackdk_>
mdash gchristensen: ok, thanks. then am I asking the wrong questions? is what I'm trying to do with nix forcing a square peg into a round hole?
<worldofpeace>
Such a prudent highlight
<simpson>
jackdk_: What *are* you trying to do with Nix?
<{^_^}>
[nixpkgs] @worldofpeace opened pull request #53435 → nixos/version: add LOGO to /etc/os-release → https://git.io/fh3cU
<mdash>
jackdk_: if you want a "build the latest thing" command, i'd just write a shell script that invokes the git command you have in mind, then passes its output as an arg to nix-build
<jackdk_>
simpson: I have a project at https://git.sr.ht/%7Ejack/mudcore/tree/nix-foolery . mudcore.nix is meant to be as close to a normal package specification as possible. I want nix-build to give me a build with the correct version number, so default.nix is doing builtins.readFile (runCommand ...) stuff
<jackdk_>
I also want to eventually put my software into an overlay, in which case I want to also use the project's mudcore.nix, but in _that_ case I would pass in an explicit src and version (probably builtins.fetchGit and a string literal, respectively)
<simpson>
jackdk_: Mm. To take a step back, why is it that you want Nix to build the version number? I suppose that your build system may require it somehow?
<jackdk_>
good question. to comply with the rules in https://nixos.org/nixpkgs/manual/#sec-package-naming I want the date of the current commit, so if you check it out and `nix-env -f . -i` it's installed with a familiar style of version number
<mdash>
jackdk_: that's only important for upstream submissions to nixpkgs
<{^_^}>
[nixpkgs] @matthewbauer pushed to staging-next « mesa: make sure $drivers output gets created »: https://git.io/fh3cc
<mdash>
if it's a nix expr bundled in a git repo obviously it's intended to build whatever version it's accessed at
<jackdk_>
mdash: yep, that's right, so then what's a sensible version number to put in the derivation?
<{^_^}>
[nixpkgs] @samueldr merged pull request #53435 → nixos/version: add LOGO to /etc/os-release → https://git.io/fh3cU
<jackdk_>
I suppose I could just go back to mudcore-git-${lib.substring 0 7 (lib.commitIdFromGitRepo ./.git)}, which would mean that building no longer needs to pull in git and all of its deps just to calculate version number for a dev build
<jackdk_>
and the overlay can pass in a specific revision and date string
<mdash>
you don't need version numbers for a dev build though, that's the thing
<simpson>
> lib.commitIdFromGitRepo
<{^_^}>
<LAMBDA>
<simpson>
Wow. TIL.
<mdash>
that's really only important for things like nixpkgs that need to track specific versions of code
<jackdk_>
what happens if you call mkDerivation with version = null?
<mdash>
mkDerivation doesn't look at version.
<samueldr>
AFAIK only nix-env actively cares about the version, and even then, it uses the one in the name attribute
<jackdk_>
huh. TIL.
<mdash>
jackdk_: all that matters is the inputs
<wedens>
when I create a module, should I add it to `imports` list of all of my other modules to access config of the module?
<jackdk_>
thanks all for your help.
Rusty1 has quit [Quit: Konversation terminated!]
<{^_^}>
[nixpkgs] @samueldr opened pull request #53436 → nix-universal-prefetch: init at 0.1.0 → https://git.io/fh3cP
prooftechnique has quit [Quit: Page closed]
<mabel>
cool! it took more fighting, but I was finally successful at stepping through a python program with neovim
<mabel>
the pypi ended up being broken :\ so I found someone's github with 3 stars where they had repackaged the code as wheels, and that worked fine
<simpson>
Nice.
pie_ has quit [Remote host closed the connection]
pie_ has joined #nixos
<ldlework>
so
<ldlework>
when i run discord now i just get a message saying
<{^_^}>
[nixpkgs] @FRidh opened pull request #53440 → stdenv: make checkInputs native → https://git.io/fh34z
<eeva>
Friend of mine is starting to learn nix and regularly sends me beginner questions. Everytime, I try and check if these are answered in the manual. What if so stuff is missing, where can I report it? Moreover, do we have a beginners 101 book/manual about nix the language somewhere?
<srhb>
eeva: The nixpkgs issue tracker on github is a great place to file issues about missing docs
<eeva>
Ok
<srhb>
eeva: There's no centralized book. There's the nix wiki, the nix pills, a number of blog post and the manuals, but that's about it.
orivej has quit [Ping timeout: 258 seconds]
the-kenny has joined #nixos
jackdk_ has quit [Remote host closed the connection]
init_6 has joined #nixos
thc202 has joined #nixos
alex`` has joined #nixos
dermetfan has joined #nixos
jmgai has joined #nixos
jtojnar has joined #nixos
<{^_^}>
[nixpkgs] @FRidh pushed 6 commits to python-unstable: https://git.io/fh3Br
badmutex[m] has quit [Remote host closed the connection]
pinage404[m] has quit [Remote host closed the connection]
icetan has quit [Remote host closed the connection]
schmittlauch[m] has quit [Remote host closed the connection]
exarkun1 has quit [Read error: Connection reset by peer]
dtz has quit [Read error: Connection reset by peer]
thefloweringash has quit [Remote host closed the connection]
philipp[m] has quit [Remote host closed the connection]
t[m] has quit [Read error: Connection reset by peer]
Ralith has quit [Read error: Connection reset by peer]
bpye has quit [Remote host closed the connection]
bkchr has quit [Remote host closed the connection]
codyopel has quit [Read error: Connection reset by peer]
WilliamHamilton[ has quit [Write error: Connection reset by peer]
Seb[m] has quit [Remote host closed the connection]
PlasmaStrike[m] has quit [Remote host closed the connection]
clacke[m] has quit [Remote host closed the connection]
timokau[m] has quit [Remote host closed the connection]
florianjacob has quit [Read error: Connection reset by peer]
timclassic has quit [Remote host closed the connection]
bennofs[m] has quit [Read error: Connection reset by peer]
siraben has quit [Read error: Connection reset by peer]
mica[m] has quit [Remote host closed the connection]
infinisil[m] has quit [Remote host closed the connection]
bake[m] has quit [Remote host closed the connection]
moredread[m] has quit [Remote host closed the connection]
Guest17278 has quit [Remote host closed the connection]
Ericson2314 has quit [Remote host closed the connection]
sevcsik[m] has quit [Read error: Connection reset by peer]
tokudan[m] has quit [Read error: Connection reset by peer]
chreekat[m] has quit [Read error: Connection reset by peer]
yangm97 has quit [Remote host closed the connection]
hjulle has quit [Remote host closed the connection]
stammon has quit [Remote host closed the connection]
ejpcmac has quit [Remote host closed the connection]
roberth has quit [Remote host closed the connection]
sphalerit has quit [Read error: Connection reset by peer]
attente[m] has quit [Remote host closed the connection]
vaibhavsagar has quit [Remote host closed the connection]
leons has quit [Remote host closed the connection]
MarkOtaris has quit [Remote host closed the connection]
ptotter[m] has quit [Read error: Connection reset by peer]
bachp has quit [Write error: Connection reset by peer]
nh2[m] has quit [Write error: Connection reset by peer]
rnhmjoj has quit [Remote host closed the connection]
Dandellion[m] has quit [Remote host closed the connection]
olejorgenb[m] has quit [Remote host closed the connection]
Guest41637 has quit [Write error: Connection reset by peer]
Vskilet has quit [Remote host closed the connection]
nyanloutre[m] has quit [Write error: Connection reset by peer]
BinkyTheClown has quit [Remote host closed the connection]
craige[m] has quit [Remote host closed the connection]
kaychaks[m] has quit [Write error: Connection reset by peer]
rycee has quit [Write error: Connection reset by peer]
fgaz has quit [Write error: Connection reset by peer]
simbergm has quit [Remote host closed the connection]
slabity has quit [Read error: Connection reset by peer]
cornu has quit [Remote host closed the connection]
edrex has quit [Read error: Connection reset by peer]
zurdo has quit [Read error: Connection reset by peer]
irdr has quit [Max SendQ exceeded]
irdr has joined #nixos
<pie_>
nix-env ignores NIX_PATH or something, so how can I get it to install something like if i was passing -I
erictapen has joined #nixos
florianjacob has joined #nixos
jmgai has quit [Ping timeout: 256 seconds]
orivej has joined #nixos
<{^_^}>
[nixpkgs] @orivej-nixos pushed commit from @orivej to master « clang-tools: override llvm version in all-packages »: https://git.io/fh3Rn
<edef>
pie_: -f
cfricke has joined #nixos
<edef>
pie_: like, i've given up on the whole ~/.nix-defexpr thing and generally just do nix-env -if '<nixpkgs>' -A foo
dermetfan has quit [Ping timeout: 246 seconds]
<pie_>
oh. I havent really figured out -f yet x) i almost never use nix-env either >_>
<edef>
so hardware.opengl.enable plus the 32-bit one
<edef>
what do glxinfo / glxgears / etc say/do
<edef>
(just trying to rule out baseline gfx issues, hoping it isn't down to the unstable/stable mix)
<pie_>
seems to be fine OpenGL core profile version string: 4.5 (Core Profile) Mesa 18.1.7
<edef>
i personally run nixos-unstable on my laptop (and i suspect most nixos devs do), so i'm not super experienced with the failure modes that might arise
<{^_^}>
[nixpkgs] @FRidh pushed 5 commits to python-unstable: https://git.io/fh3uD
<exarkun1>
Somehow making vala use graphviz-nox means I have to build libX, cairo, pango, libGL, mesa, and twenty other packages. joy.
fendor has joined #nixos
realrokka has joined #nixos
<{^_^}>
[nixpkgs] @timokau opened pull request #53449 → patchutils: various fixes → https://git.io/fh3ub
<MichaelRaskin>
I guess it comes with too much GTK integration support
<realrokka>
How does one run a 32-Bit Binary on NixOS? I tried PatchELF with no look. I would like to play this crackme http://pwnable.kr/bin/bof , I throws "File not found Error", because it cant find the 32bit interpreter or so
<exarkun1>
seems to be something like that, yea :/
<manveru>
patchelf can't really convert a 32bit binary to 64bit
mmlb has joined #nixos
eadwu has joined #nixos
phaebz has joined #nixos
ma9e has joined #nixos
kisonecat has joined #nixos
phaebz has quit [Ping timeout: 250 seconds]
<manveru>
seems like purescript package is broken :(
silver has joined #nixos
<manveru>
isn't it built by hydra?
<wedens>
how can I set up openvpn with split dns? (local subnet to local dns server, everything else to vpn-provided dns server) currently I use NM for the network, but it's changeable if needed. I had this set up on Arch using some netctl+openpnv+dnsmasq hacks.
<frigate_freedom>
Hello guys! I'm trying to create package for ksh. It builds with meson. https://pastebin.com/JiJWksfC <- here is my default.nix. When I'm installing package I get error:
<frigate_freedom>
src/lib/libast/meson.build:20:0: ERROR: File /build/source/build/src/lib/libast/comp/conftab.c does not exist.
<frigate_freedom>
So how can I deal with this problem? I'm noob in Nix, git and programming :D
<manveru>
so i'd probably add gcc or clang to nativeBuildInputs
orivej has quit [Ping timeout: 250 seconds]
orivej has joined #nixos
<ambro718>
Suppose I have a python script (or another interpreted language) that uses a nix-shell shebang. That script itself may need to run programs from nix packages. Clearly it can do that by invoking "nix-shell" on its own. However, is there a method that allows using the same Nix instance for performance reasons, to avoid re-evaluating lots of nixpkgs?
<ambro718>
for example on my computer it takes half a second for nix-shell to start a python.
<ambro718>
so if the python script itself calls nix-shell, I am already losing 1 second for nix-shell to do its thing
<frigate_freedom>
manveru: added gcc and binutils, but error stil the same
<simpson>
ambro718: There's not a slick way, no. The nix-shell shebang is at a local maximum of slickness.
_lawlesseel has joined #nixos
<frigate_freedom>
as I can see from 'nix-env -f . -iA ksh' output -- gcc works ok and problem is in missing file conftab
<gchristensen>
iirc someone was thinking about a nix evaluation daemon
<etu>
gchristensen: Opinions on my nix-mode PR? :)
<{^_^}>
[nixpkgs] @gnidorah opened pull request #53451 → bitwig-studio2: support pulseaudio → https://git.io/fh3ow
<manveru>
<MichaelRaskin>
nix-evaluation daemon sounds like something that should use unix domain socket and only accept requests from the same user
b has joined #nixos
ma9e has quit [Quit: Lost terminal]
<wedens>
etu: I've checked out your branch and indentaton behaves weirdly with electric-pairs. IIRC it worked fine in your previous PR branch, but was broken in master
markuskowa has joined #nixos
boothead has joined #nixos
<gchristensen>
etu: it looks fine to me, but I dunno what I'm doing :)!
<gchristensen>
etu: just doing what I can by posting weird indentation behavior reproductions
<boothead>
Anyone know the fix for "error: serialised integer 8679964202353849210 is too large for type 'j'"? I found this but it's not quite the same error message https://github.com/NixOS/nixpkgs/issues/37287. clever do you have the command you ran to determine what the integer mean?
<{^_^}>
#37287 (by msteen, 41 weeks ago, open): SSH substituter with nologin as shell should not result in serialized integer error
<gchristensen>
boothead: nix runs `ssh user@remote nix-store --serve --write` -- if you do that, does it work?
<wedens>
etu: looks like something is wrong with empty lists and sets: {\n} or [\n]
<gchristensen>
wedens: please leave feedback like that on the PR -- it is hard to keep track of it all otherwise
<wedens>
ok
<boothead>
gchristensen: nope :-(
<boothead>
I get the same error
<etu>
wedens: yeah, post it there and I'll look into it
<gchristensen>
boothead: you get the j error?
<etu>
wedens: we need more corner cases
<FRidh>
ambro718: within a package we never use nix-shell, so it's unlikely you have one nix-shell script calling anoterh
<boothead>
I tried it from the target machine and got this: error (ignored): bad archive: input doesn't look like a Nix archive error: cannot add path '/nix/store/0yp9ry6r8mrjwc8v3xr178y47nasdl0b-etc-hosts' because it lacks a valid signature
<boothead>
gchristensen: yes, it looks like exactly the same error.
freeman42x[nix] has quit [Quit: Leaving]
<boothead>
One thing that might be of note is that when I ssh into the target machine I get a prompt: Ignore insecure directories and continue [y] or abort compinit [n]
<boothead>
However I can still ssh and run a command like ls no problem
<etu>
wedens: I wounder if that worked before the breakage... I'll test later when I get home.
<ambro718>
FRidh: True, but I'm doing something a bit unusual. It's a script for a microcontroller SW project to program the device, and it would be nice if based on what type of device is used (several completely different families supported), it should use the right tool to do that (e.g. avrdude, bossac, teensy-loader-cli, openocd).
<wedens>
etu: I believe it did
<boothead>
gchristensen: before I get the error: cannot add path '/nix/store/0yp9ry6r8mrjwc8v3xr178y47nasdl0b-etc-hosts' because it lacks a valid signature it does seem to be copying some stuff over.
<ambro718>
FRidh: it would be a waste if when you run the script all of those tools are installed, when you only need one.
<etu>
wedens: Great, have to do more digging then :) thanks for the testcases
<boothead>
e.g. copying path '/nix/store/0yp9ry6r8mrjwc8v3xr178y47nasdl0b-etc-hosts' from 'ssh://ben@office'... so it looks like it starts doing things then fails because of the signature... Is there some set up I'm missing maybe?
<FRidh>
ambro718: but those tools can be proper packages
<FRidh>
so no nix-shell inside those
<ambro718>
FRidh: they are, but my python script that is a generic programmer decides which tool to use
<ambro718>
my script is like, ./flash.py --device <device type> --file <file to flash>
<ambro718>
based on <device type> the script calls the right tool
dermetfan has joined #nixos
<{^_^}>
[nixpkgs] @FRidh pushed 2 commits to python-unstable: https://git.io/fh36e
kim0 has joined #nixos
civodul has joined #nixos
dermetfan has quit [Ping timeout: 252 seconds]
frigate_freedom has left #nixos [#nixos]
Makaveli7 has quit [Quit: Leaving]
jtojnar has quit [Read error: Connection reset by peer]
kp__ has joined #nixos
phaebz has joined #nixos
endformationage has joined #nixos
phaebz has quit [Ping timeout: 272 seconds]
kp__ has quit [Quit: WeeChat 2.3]
<ivegotasthma>
hello, I'm getting the wrong system font on firefox. I don't think anything changed between boots, any idea how to fix this? https://i.imgur.com/LmJdSms.png
kp__ has joined #nixos
<ivegotasthma>
I installed nixos yesterday, so it's all new to me and I'm not sure how to navigate to debug the issue
<wedens>
is there nix function to list directory content and filter by extension?
<realrokka>
manveru: thx, but I should have asked more precise: How to run/debug a 32Bit in GDB? I guess I'll just stick with an Kali VM for this sort of things :/
lostman has quit [Quit: Connection closed for inactivity]
<{^_^}>
[nixpkgs] @FRidh pushed 4 commits to python-unstable: https://git.io/fh3Px
<ivegotasthma>
I'm compiling several python 3.6 wheels in a virtualenv. I seem to be missing the zlib headers, but I've installed nixos.zlib.dev. Any idea why the virtualenv wheels cannot find the headers?
<FRidh>
because with nix they're in non-standard paths, and if you're not in a built environment, then likely your env vars aren't setup either
Mateon1 has quit [Ping timeout: 250 seconds]
Mateon1 has joined #nixos
<ivegotasthma>
FRidh: what do I have to read to understand how to fix it
<silver_hook>
Any idea how to enable this on NixOS (to be persistent)?
<{^_^}>
[nixpkgs] @FRidh pushed 5 commits to python-unstable: https://git.io/fh3Mk
<{^_^}>
[nixpkgs] @FRidh merged pull request #53453 → wine: add perl to buildInputs to fix winemaker → https://git.io/fh3X5
<{^_^}>
[nixpkgs] @FRidh pushed commit from @tadeokondrak to master « wine: add perl to buildInputs to fix winemaker »: https://git.io/fh3Mm
kisonecat has joined #nixos
<ivegotasthma>
FRidh: I got this nix build env, but it still doesn't find the zlib headers, any idea what could be wrong? I copied this almost 1 to 1 from the manual and added my own deps. https://dpaste.de/yPhr/raw
shibboleth has quit [Remote host closed the connection]
shibboleth has joined #nixos
tzemanovic has quit [Remote host closed the connection]
tzemanovic has joined #nixos
tzemanovic has quit [Remote host closed the connection]
tzemanovic has joined #nixos
tzemanovic has quit [Remote host closed the connection]
tzemanovic has joined #nixos
<sg2002>
Hello. I've been trying to update my old 32 bit laptop to a newer nixos for quite a while, but finally I've found the actual bug on the tracker - https://github.com/NixOS/nixpkgs/issues/36947 . But apparently there is not activity on it for quite a while. Is anyone aware when we can expect it to get fixed? There's some talk there on updating the bootstrap tools, but was this done?
<{^_^}>
#36947 (by dtzWill, 42 weeks ago, open): libgcc_s is from bootstrap tools??
tzemanovic has quit [Remote host closed the connection]
tzemanovic has joined #nixos
tzemanovic has quit [Remote host closed the connection]
tzemanovic has joined #nixos
tzemanovic has quit [Remote host closed the connection]
Dedalo has joined #nixos
<nodyn>
Hi everyone, I'm looking for a bit of help/advice. I'm trying to setup a LDAP server and I was hoping to have Nix handle the generation of slapd.d, but slapadd checks whether or not database directories exist (which Nix won't allow, as the database is outside the Nix store so that I can write to it), is there a way to fake it (maybe buildFHSUserEnv?) or is that a lost cause?
phaebz has joined #nixos
cyounkins has joined #nixos
Makaveli7 has joined #nixos
griff_ has quit [Quit: griff_]
cyounkins has quit [Remote host closed the connection]
<Orbstheorem>
Hello, I'm importing nixos-unstable from my configuration.nix to install some bleeding-edge packages (such as mopidy or youtube-dl), in the case of mopidy, since I have to run on root to execute switch-to-configuration it refuses to evaluate because pyspotify (dep of my mopidy closure) is unfree, even if I have set `{ nixpkgs.config.allowUnfree = true; }` in my `configuration.nix` How can I pass an override
<Orbstheorem>
when I `import <nixos-unstable> {}`?
<Orbstheorem>
I worked-around by having `{ nixpkgs.config.allowUnfree = true; }` in my `/root/.config/nixpkgs/config.nix`, but ugly...
<rycee>
shazow: Feel free to try Home Manager out. But I should note that it currently doesn't have a convenient way to uninstall. I am planning to add an uninstall command in the near future.
agander has joined #nixos
<Church->
Oh hey rycee
<ottidmes>
Orbstheorem: I use: unstable = import /wheel/fork/nixpkgs-unstable { config = { allowUnfree = true; }; overlays = []; }; so I explicitly pass it a config allowing unfree packages
phaebz has joined #nixos
<Orbstheorem>
ottidmes: thanks!, So, which file gets imported when you import a channel? I though it was nixpkgs/nixos/default.nix
<Orbstheorem>
But I see no config attribute as input
<Orbstheorem>
Oh, I guess it's `pkgs/top-level/default.nix`
mabel has joined #nixos
phaebz has quit [Ping timeout: 252 seconds]
<shazow>
rycee: noted, thank you!
niksnut has joined #nixos
<ottidmes>
Orbstheorem: I am just using normal Nix to import my unstable checkout, so it would just import /wheel/fork/nixpkgs-unstable/default.nix, so default.nix in the root of nixpkgs, which if you check its contents will just import pkgs/top-level/impure.nix, which in turn calls pkgs/top-level/default.nix after filling in the missing arguments based on external information (environment variables and the existance of
<ottidmes>
certain folders/files on disk) hence it is called impure.nix
eadwu has quit [Ping timeout: 252 seconds]
sg2002 has quit [Quit: ERC (IRC client for Emacs 26.1)]
<{^_^}>
nix-darwin#117 (by kyren, 1 week ago, closed): Ability to disable specific checks in modules/system/checks.nix
<yl[m]>
this did not work for me. ` system.extraSystemBuilderCmds = ''ln -sv ${pinnedNixpkgs} $out/nixpkgs'';` I also tried `system.activationScripts.pinnixpkgs = '' ... '';` with no success
<kyren>
yl[m]: it's a trick to make it so that doing an upgrade doesn't change NIX_PATH so you don't have a weird situation where you have to log out and log back in, another entry in my configuration places nixpkgs there, let me show you
<yl[m]>
it did not work but I started to suspect something might be wrong in my nixos-rebuild wrapper since none of them seems to have worked. I need to dig in deeper, thx kyren!!
aswanson has joined #nixos
<kyren>
no problem, do you want help with your nixos-rebuild wrapper?
<kyren>
I can give you an example since I've done this all pretty recently
palo has quit [Quit: WeeChat 2.2]
<yl[m]>
kyren: I found the problem. I did -I nixos_config instead of nixos-config.
palo has joined #nixos
<yl[m]>
kyren: sure why not, it always helps to see someone else's work!
<kyren>
oh, nice
<kyren>
it's not exactly complex
<yl[m]>
do you have yours opensource?
<kyren>
er, sorry I meant that to mean, my example is not very complex so it's hardly a great example
<aswanson>
anyone have a good understanding of multiple monitors within nixos? I've been futzing with `services.xserver.xrandrheads` to try and correct my monitor arrangement but they are still out of order
<kyren>
I keep secrets in my system configuration so no, it's on my own private gitlab, but I can show you the build script really fast
<yl[m]>
kyren: let me show you mine, maybe I can use a few comments on my PR :)
<{^_^}>
kalbasit/shabka#81 (by kalbasit, 3 days ago, open): Pin the nixpkgs version used to build shabka
<dmj`>
is there a optionalList = x: arg: if x then arg else []
<dmj`>
in the standard library
<ottidmes>
dmj`: lib.optional
<dmj`>
ottidmes: I don't think that's the same
<ottidmes>
dmj`: lib.optionals then
<ottidmes>
> optionals false [ 1 ]
<{^_^}>
[ ]
<ottidmes>
> optionals true [ 1 ]
<{^_^}>
[ 1 ]
agander_ has joined #nixos
[Leary] has joined #nixos
agander has quit [Ping timeout: 240 seconds]
agander__ has joined #nixos
<kyren>
yl[m]: well, ti's so simple that it's not worth showing, it's just a scirpt that sets NIXPKGS NIX_DARWIN and HOME_MANAGER env variables to https urls like: https://github.com/LnL7/nix-darwin/archive/master.tar.gz, then pass those into darwin-rebuild / nixos-rebuild
Lears has quit [Read error: No route to host]
<yl[m]>
kyren: I have another PR working for nix-darwin but I got stuck last time with the checks which BTW thank you for getting that one out of the way!
rfold has joined #nixos
<dmj`>
ottidmes: thanks
<dmj`>
that worked
agander_ has quit [Ping timeout: 244 seconds]
<kyren>
yl[m]: one thing that I'd like to do in the near future is use nix-prefetch-git or something similar to update small files which only contain git urls and versions and their sha, I forget how I planned it out but if there's not a tool I want to write one, where it just outputs a simple nix expression that is the latest git revision at a url, so you don't have to point to master all the time and can "revert" by using version control, almost like channels
<kyren>
but they're in a file in VC
<kyren>
yl[m]: oh, and you're welcome for the bug report :)
<kyren>
oh, that's it exactly, yeah it's nix-prefetch-git with the json option, right
<kyren>
obviously you already know that then! I hadn't done that yet but I was planning on it, but in between then and now I forgot that nix-prefetch-git has json output and that's obviously the best way to do it
<kyren>
that's super useful, thanks, I might use this right now in fact
<kyren>
I like this technique *so much*, and I really don't like global channels, part of me wishes that all the nix tooling was built to use this style as the default
griff_ has joined #nixos
<kyren>
well, not that it's super *super* hard to do yourself, but it took me a while to discover that you could do it and how it all worked
kisonecat has joined #nixos
julm has quit [Ping timeout: 245 seconds]
<yl[m]>
I don't either but I was pinning nixpkgs in NIX_PATH to the source and it bothered me a lot when logging out/in. but when I saw your issue on nix-darwin it was like an aha moment so I am switching back out of channels again
<Myrl-saki>
elvishjerricco: If the backup is still running, but the last machine is not part of the backup, it will mark it as incomplete instead of "running"
<Myrl-saki>
The easiest fix would probably be to just check if `backup_status` is running before setting it to "incomplete".
<Myrl-saki>
Alternatively, change line 411 to "incomplete".
<MichaelRaskin>
gchristensen: to be honest, for $HOME I am still on the mere DVCS side, unlike shlevy. It is /etc/ where I have a store path
<gchristensen>
MichaelRaskin: replacing $HOME seems a bit risky, I think it is likely a lot of FDs will be pointing to it?
<gchristensen>
updating /etc seems much more reasonable
<ldlework>
shealevy's blog is cool but would benefit HUGELY from centering the content with a reasonable max-width
<ldlework>
i meant to email them before but they probably dgaf :)
<shlevy>
ldlework: :)
<ldlework>
oh hai :P
<shlevy>
I need to put my blog generation code in a repo somewhere
<ldlework>
reading super long lines makes my eyes-cross :D
<shlevy>
If it's an easy fix I'd happily do it... Otherwise I'd happily take a patch if it doesn't get too far away from my one little corner of semantic web :P
<MichaelRaskin>
ivan: it is work to benchmark this fragmentation effect even on a single system with a single load profile, generic guesses are very likely to be off-mark
<shlevy>
(I know literally no one has user stylesheets, but theys hould damnit)
griff_ has joined #nixos
<MichaelRaskin>
gchristensen: I expect FDs to referto symlink targets anyway
<gchristensen>
ah hm yeah
<MichaelRaskin>
And GC respects open FDs
<MichaelRaskin>
ldlework: consider reader mode in your browser?
<gchristensen>
I put my /home on a zfs dataset so I could move it around more easily and play with this.
<gchristensen>
ivan: probably depends on your use case
<steveeJ>
has anyone encountered fetchGit failing with "fatal: not a tree object"?
<elvishjerricco>
steveeJ: Usually means the rev isn't on the default branch IIRC
<elvishjerricco>
You can use `ref` to point to a branch the rev can be found on I think
<shlevy>
gchristensen: I'd expect fds to behave just fine
<shlevy>
Not sure what the risk you're worried about
<shlevy>
gchristensen: Speaking of though the weird provides/requires thing I've got going there is part of a "capability"-like model for system config... Which I eventually hope to mirror in with a capability-based runtime. All programs have their own view of the world, ideally through capabilities directly but realistically most programs will need a shim namespace to fake it :D
<Myrl-saki>
I just realized that all of etc stuff don't directly point to /nix/store, is this new?
<elvishjerricco>
Confirms it defaults to HEAD for ref. rev defaults to tip of ref. rev is found under ref
<Myrl-saki>
Instead, it goes /etc/name -> /etc/static/name
<Myrl-saki>
And /etc/static points to whatever the /nix/store/*etc.
<elvishjerricco>
Myrl-saki: I believe that's so it can accurately determine which etc files to delete (since people can make their own symlinks to the store in etc)
<shlevy>
Myrl-saki: That's for more-atomicish updates
<shlevy>
Since you have a mix of mutable and immutable there
lopsided98 has joined #nixos
<shlevy>
Only file adds/deletes are non-atomic in this model, file content changes all happen atomically
<shlevy>
(this isn't new)
<gchristensen>
is it not actually atomic? I think it is
<shlevy>
gchristensen: It's not atomic if you add a new file
<shlevy>
Or remove one
<gchristensen>
ah
<gchristensen>
but an a->b swap is
<MichaelRaskin>
It is non-atomic, and slow
<shlevy>
gchristensen: Don't know what you mean by a -> b
<MichaelRaskin>
But yes, swapping the /etc/static target is asingle operation
<gchristensen>
v1 to v2
<MichaelRaskin>
And it swaps all the targets at once
<shlevy>
gchristensen: Of a single file? Yeah that's all atomic
countingsort has joined #nixos
<shlevy>
Another reason to prefer an opt-in mutability model ;)
<gchristensen>
You don't need to sell me
<shlevy>
Though I still need to move my /etc from tmpfs to actual store dir... That'll take more mucking in NixOS
<MichaelRaskin>
If only passwd was more willing to negotiate
<Myrl-saki>
Oh, it's even lines 82-84 on my checkout of nixpkgs.
<MichaelRaskin>
I am a bit annoyed I cannot experiment with unreadable store
<Myrl-saki>
# For every file in the etc tree, create a corresponding symlink in
<MichaelRaskin>
(without patching Nix)
<Myrl-saki>
# /etc to /etc/static. The indirection through /etc/static is to make
<Myrl-saki>
# switching to a new configuration somewhat more atomic.
_lawlesseel has quit [Remote host closed the connection]
ddellacosta has quit [Remote host closed the connection]
_lawlesseel has joined #nixos
Acou_Bass has quit [Ping timeout: 272 seconds]
gagbo has quit [Quit: I'm out !]
Acou_Bass has joined #nixos
<babic>
I'm not sure of the best way to package something. I want to add a program that depends on a library that depends on one of pyqt4/pyqt5/pyside.
ck3d has quit [Quit: WeeChat 2.2]
<babic>
The library isn't in nixpkgs, so I have to add it and I'm not sure how to do it. If I make it available for other programs how to handle the pyside/pyqt5/pyqt4 dependency?
<babic>
I currently have it just depending on pyqt5 (which is what my program wants) but that doesn't seem like a good thing to commit
<babic>
It doesn't build without one of those, so I can't just leave it to downstream
<simpson>
I'm not sure how the Python tree maintainers feel, but I'd think that that would be something that people could pick freely somehow. Annoyingly, the only current switch is `python27Packages` vs `pypyPackages` etc. which is not very multi-dimensional.
<simpson>
It would seem progressive/helpful to at least have a configurable optional parameter `pyqt` which defaults to pyqt5, or something like that.
kisonecat has quit [Ping timeout: 258 seconds]
<babic>
Yeah, that sounds as improvement
<babic>
I think it's good enough for creating a pull request and see if the maintainers have any suggested improvements
<simpson>
Yes.
<pbb>
Hi, how does nixos handle kernel modules..: If I don't reboot after a kernel upgrade, will the modules for the booted version still be available?