<iddt>
fresheyeball: not sure, as i'm not sure how you're attempting to build this project. i think a screenshare chat would help me narrow it down, tho.
<judson>
At this point I feel like I've done all the tricks I can find online to keep my nix-shells from getting garbage collected, and it's getting frustrating to have to reinstall my build deps all the time.
<ekleog>
so on the --print-roots line, there's -> /nix/store/...-... ; can you nix-store -qR [that path] and check rspec is in it?
<judson>
To clarify: rspec is not itself in the gc-roots, but /home/judson/dev/shex-map/.nix-gc/shell.dep-6 -> /nix/store/nabdxm7k0qc27z74qnd2yg7cjsadflyf-wrapped-ruby-shex-map is and /home/judson/dev/shex-map/.nix-gc/shell.drv -> /nix/store/ji6g76i51qhdi60b33g88aw2xd848ip3-shex-map-interactive-environment.drv which should walk to rspec
goibhniu has quit [Ping timeout: 250 seconds]
<judson>
/nix/store/xl8k3fjbz5fy99ns0r7av12d1d8m379j-rspec-core-3.8.0.gem.drv is there.
<ekleog>
so either something is seriously broken, or your rspec is properly pinned until you move the file created by nix-instantiate
<ekleog>
(I don't know about your second attempt at pinning
<ekleog>
)
<judson>
The --print-roots stuff seems useful...
<ekleog>
yup, there's --print-live and --print-dead that are useful too :)
<judson>
Okay, that nix-store --add-root $(nix-store -q --references) thing seems worthless. The nix-instantiate roots one thing, and it's references pull in all those (seems obvious) but then I've got difference store paths for the others.
<judson>
Okay, actually, something is wrong.
<judson>
ekleog, it seems like the nix-instantiate from the wiki article builds one version of this shell.nix, and then nix-shell itself builds another.
Rusty1 has joined #nixos
<judson>
Bah. I can't be sure. It's a derivation/realization thing.
freedan42x has quit [Quit: Leaving]
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fhzfe
<ekleog>
judson: isn't it just a shell.nix vs. default.nix issue?
<mdash>
alright realizing I don't know as much about nix evaluation as i thought
<mdash>
i have been handling my deployments by putting password hashes, etc., in /etc/nixos/secrets.nix
<mdash>
which I then list as an import in my configuration.nix
<mdash>
nextcloud admin password is configured as 'services.nextcloud.config.adminpass',i put 'services.nextcloud.config.adminpass = "foo";' in secrets.nix and it works great
<mdash>
I now want to move that into a container
<mdash>
First thing I tried was 'containers = { nextcloud = containerCfg: { services.nextcloud = { ...; adminpass = config.services.nextcloud.config.adminpass; }; }; }'
<mdash>
the error I get is: 'The option `containers.nextcloud.services.nextcloud.adminpass' defined in `<unknown-file>' does not exist.'
<mdash>
is this a plausible thing to try to do? i thought that lazy eval would make this sort of thing work out ok
<nh2>
clever: the stack trace that suggests that the library does a mistake, I suspect that they actually changed something so that because in their own example `uri=srt://:8888/` the host string is empty, they somehow pass NULL
<nh2>
instead of the empty string
<clever>
nh2: having debug symbols would let you confirm the args to strchr
<nh2>
clever: where would I enable debugging to get full symbols in `libgstsrt.so` instead of libc? Because then I could look straight at the arguments it passes to the libc call
<clever>
nh2: whatever derivation is providing it
<clever>
,locate libgstsrt.so
<nh2>
(I mean I know I have to change my newly added `srt/default.nix` derivation, but can never remember what to pass there so symbols stay in and -O0 is passed properly)
<{^_^}>
Couldn't find in any packages
<clever>
nh2: ah, thats what enableDebugging does
ng0 has joined #nixos
<clever>
check the example above
<clever>
it automatically applies the right override flags for you
<nh2>
clever: so `dontStrip = true; enableDebugging = true;`?
justanotheruser has quit [Ping timeout: 250 seconds]
<nh2>
I wonder why it downloads `libuv` when I pass those
Rusty1 has quit [Ping timeout: 245 seconds]
<clever>
nh2: you may have GC'd libuv previously
Supersonic has quit [Disconnected by services]
Supersonic112 has joined #nixos
Supersonic112 is now known as Supersonic
<fresheyeball>
clever: learned something
<fresheyeball>
it builds fine on the CI server, if inside nix-shell
<fresheyeball>
but nix-build still fails
justanotheruser has joined #nixos
<clever>
fresheyeball: which CI system?
xkapastel has quit [Quit: Connection closed for inactivity]
<{^_^}>
#41340 (by nh2, 33 weeks ago, open): GCC fails because argument list created by nix is too long
bgamari has quit [Remote host closed the connection]
<timclassic>
I'm using nixops to deploy from x86_64 to aarch64 (and I have an aarch64 build host configured), and this generally works. However, when I deploy vault this way, the x86_64 binary gets installed instead of the aarch64 variant. Where should I look to debug this?
justan0theruser has joined #nixos
<timclassic>
If I install vault directly on the aarch64 system via nix-env, the correct arch gets installed.
<clever>
nh2: if you restart pulseaudio, then chrome looses the ability to capture any audio, until its also restarted
teto2 is now known as teto
d4g_ has quit [Ping timeout: 245 seconds]
justanotheruser has quit [Ping timeout: 245 seconds]
<clever>
nh2: i believe you want to set the nix var mesonBuildType
<clever>
nh2: running meson --buildtype=help in a nix-shell should give options
<timclassic>
clever: Ah, it worked! (I used `super.system` since I'm in an overlay) I should go put that in all of my imports of pkgs, then...
<timclassic>
clever: Thanks so much!
lassulus_ has joined #nixos
lassulus has quit [Ping timeout: 240 seconds]
lassulus_ is now known as lassulus
<fresheyeball>
clever: we still partying over here
<fresheyeball>
just fyu
<fresheyeball>
fyi
<clever>
fresheyeball: already in another voice chat
sigmundv_ has quit [Ping timeout: 244 seconds]
<fresheyeball>
haha
<nh2>
clever: what's very odd is that fresheyeball's setup uses the nix-dameon, yet the strace on nix-build apparently shows the E2BIG in question, which is unexpected (the strace shouldn't see what the nix-daemon does)
<muvlon>
is there a nixos party going on? :D
<clever>
fresheyeball: one min...
jasongrossman has quit [Read error: Connection reset by peer]
<clever>
lokado: nixos-install will ask for a root pw when the install is finished
<clever>
lokado: and it should allow a login
<clever>
fresheyeball: the only places nix can throw bad_alloc is in the evaluator, but nix-daemon never evals nix
dvim has joined #nixos
<clever>
so the bad_alloc must be coming from elsewhere
doyougnu has joined #nixos
<fresheyeball>
clever: I had a big log dump in that dir that was killing the deamon
<fresheyeball>
my debugging killed the debugging
<fresheyeball>
now I am dumping the strace
<clever>
oh
<clever>
the logs in the current dir, made ./. explode
<clever>
send the logs elsewhere
<fresheyeball>
yeah
pie___ has joined #nixos
pie__ has quit [Ping timeout: 240 seconds]
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Guest99275 has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @lopsided98 opened pull request #54402 → beohm-gc: fix cross build of dependent packages → https://git.io/fhzTx
countingsort has quit [Ping timeout: 240 seconds]
MP2E has quit [Remote host closed the connection]
slack1256 has joined #nixos
iddt has quit [Quit: bye]
jasongrossman has joined #nixos
<iqubic>
I'm looking for a screenshot utility that is lightweight, lets me select a portion of the screen to snapshot, and works with a tiling window manager.
<{^_^}>
[nixpkgs] @CMCDragonkai opened pull request #54403 → pythonPackages.grandalf: init at 0.6 → https://git.io/fhzkc
<simpson>
iqubic: I use `scrot`. It has various options for those things. You'll want to read the man page and determine how it can work for you.
<iqubic>
simpson: Will it let me select just a portion of a window?
<iqubic>
Like not the entire window, but just a select portion of it.
<iqubic>
Like, let's say I just want to screenshot this IRC message, and not anything else on my screen at this moment.
<{^_^}>
[nixpkgs] @worldofpeace pushed 37 commits to master: https://git.io/fhzLF
smolboye has joined #nixos
<teto>
,locate libpulse
<{^_^}>
Found in packages: qmmp, libao, apulse, libpulseaudio, libsForQt5.vlc, pulseaudioFull, pulseaudioLight, libcardiacarrest, libpressureaudio, libpulseaudio.dev, pulseaudioFull.dev, pulseaudioLight.dev, libcardiacarrest.dev, xfce.xfce4-pulseaudio-plugin, xfce4-13.xfce4-pulseaudio-plugin
gagbo has joined #nixos
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
dvim has joined #nixos
<aleph->
Apologies what's the command I want to call to build a local python package? nix build -E '(import <nixpkgs> {}).python.BuildPythonPackage ./default.nix'?
<clever>
aleph-: does it give aaaaaaaany error when you run that?
<aleph->
It does, I just can't remember the exact string I'm thinking of.
<aleph->
Didn't see it in the manual.
<clever>
aleph-: what error does it give?
<aleph->
Attribute BuildPythonPackage not found in (string):1:1
<aleph->
I'm thinking of the wrong package I know it.
<clever>
aleph-: poking around with tab completion in `nix repl '<nixpkgs>'` found pythonPackages.buildPythonPackage
<aleph->
Or I was close.
<aleph->
Let me go write that down finally since my history keeps crapping out.
Mateon3 has joined #nixos
dvim has quit [Ping timeout: 245 seconds]
<aleph->
Thanks clever
Mateon1 has quit [Ping timeout: 240 seconds]
Mateon3 is now known as Mateon1
drager has joined #nixos
jackdk has quit [Ping timeout: 244 seconds]
alp has left #nixos ["Leaving"]
alp has joined #nixos
vk3wtf has quit [Ping timeout: 268 seconds]
gagbo has quit [Ping timeout: 272 seconds]
<MichaelRaskin>
aleph-: for Pascal, sometimes just telling fpc to build it turns out to be enough…
smolboye has quit [Ping timeout: 245 seconds]
<aleph->
Good to know.
<aleph->
Just went with a sorta clone in Python
slack1256 has quit [Remote host closed the connection]
<iqubic>
So I use ZFS. Is it possible to make run a screen lock program under this condition.
<srhb>
iqubic: I don't think ZFS has much to do whether or not that's possible, so yes.
<iqubic>
I'm asking because I know that you can't hibernate when using ZFS.
<srhb>
Pretty sure you can do that too.
<iqubic>
And I wonder if there is a difference between screen lock and hibernating.
<srhb>
Or did they break it again...
<iqubic>
IDK.
<srhb>
I mean, at a bare minimum, a screen lock just makes it impossible to input anything or see what's on the machine. That's definitely possible regardless of zfs or not.
<srhb>
Suspend to RAM also works fine with/without zfs.
<iqubic>
So what screen lock/screen saver do you recommend?
<srhb>
iqubic: Literally more options related to physlock in NixOS than to slock.
<iqubic>
xss-lock is also a thing.
<neonfuz>
dtz: Hey, are you the maintainer of fractal?
<neonfuz>
I'm not getting any icons when I run it, a bunch of X icons everywhere
Ariakenom has joined #nixos
<neonfuz>
maybe I colud have asked generally first before pinging, sry lol
<etu>
iqubic: i3lock is quite common as well
<iqubic>
doesn't that only work if I am running i3?
<neonfuz>
I doubt it?
<neonfuz>
yeah I think it's just referencing to being in the style of i3
<neonfuz>
minimal, etc
<etu>
iqubic: nah, it works in other environments as well
lokado has joined #nixos
<lokado>
hi，how do you get the sha256 required for the fetchpypi function?
<iqubic>
,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.
<{^_^}>
[nixpkgs] @peti pushed 14 commits to haskell-updates: https://git.io/fhzmn
<lokado>
Thank you iqubic
Tucky has joined #nixos
<teto>
now able to generate nix derviations for both lua src.rock and rockspec. So we should be able to import all of luarocks.org
<{^_^}>
[nixpkgs] @worldofpeace opened pull request #54410 → [18.09] mate updates and downgrades → https://git.io/fhzOP
dermetfan has joined #nixos
v{}id has joined #nixos
Phillemann has joined #nixos
<jasongrossman>
teto: Awesome!
<Phillemann>
Where are topics regarding NixOS usually discussed? I feel IRC might be bad for my question, since the relevant people might just not be online at the moment.
<jasongrossman>
,ask
<{^_^}>
Just ask your question. It's the best way to know if anybody can help.
v{}id has quit [Client Quit]
<jasongrossman>
There's also reddit, and I think discord. But here is usually best. Just ask multiple times if necessary.
<Phillemann>
Okay, so I'm setting up a shell.nix for react native development. I need the android emulator, so I've enabled it. My shell, however, doesn't have "emulator" in $PATH.
<Phillemann>
I'm not sure if this is intentional.
<jasongrossman>
Phillemann: Now if I were you I'd copy that text somewhere, and paste it back in here at a different time of day - any different time of day, at random. We have experts in various time zones. (Not me, though.)
aleph- has quit [Ping timeout: 250 seconds]
<Phillemann>
jasongrossman: Okay. Seems really inefficient, but I'll do that.
<jasongrossman>
,ask = Just ask your question. It's the best way to know if anybody can help. Or look at other ways of getting in touch with the Nix community here: https://nixos.wiki/wiki/Get_In_Touch
<{^_^}>
ask redefined, was defined as Just ask your question. It's the best way to know if anybody can help.
inquisitiv3 has quit [Ping timeout: 268 seconds]
dvim has joined #nixos
hyper_ch2 has joined #nixos
hvemvet has joined #nixos
<teto>
is there some snippet somewhere that can generate the "fetchurl/fromGithub"
<teto>
from an url
<hyper_ch2>
?
civodul has joined #nixos
<teto>
let me rephrase, rockspec have their own format to describe where to fetch the sourcecode. I am trying to translate this into nixpkgs code. For instance if the rockspec file contains "url = "git://github.com/antirez/lua-cmsgpack.git"," I want to generate "fetchFromGithub { .... }"
<teto>
like nix-prefetch --generate-nix-code <URL> would be nice
<wedens>
teto: can you convert this format to json?
<immae>
Hey there! I’m writing some new packages, and I’m wondering how I can ensure purity when building them. For instance I have `bundle` or `composer` commands to build the things, and they "may" do some internet calls if some things are missing, but I’d like to forbid them from doing it (getting an error instead of fetching the packet when they try to do)
<symphorien>
if you enable the sandbox (default in recent nix versions) then internet access will be forbidden during build
__monty__ has joined #nixos
<symphorien>
you can see if the sandbox is enable in `nix show-config`
<immae>
Thanks symphorien
<{^_^}>
[nixpkgs] @jtojnar opened pull request #54415 → nixos/gnome3: split into common, flashback and 3 → https://git.io/fhzEZ
<immae>
It’s not, how can I enable it? (I’m running nixops from a non-nixos system)
<symphorien>
is it a multi-user of single-user system ?
<immae>
the local system is a multi-user one
<symphorien>
then add sandbox=true to /etc/nix/nix.conf
ThatDocsLady has joined #nixos
<immae>
Simple :D
<immae>
Thanks :)
<immae>
(I suppose it will trigger a rebuild of everything?)
<symphorien>
no it should not.
<immae>
ok
<immae>
So it’s only next time I build them that I’ll get the errors
<symphorien>
yes
<immae>
Cool
<immae>
Thanks a lot
<symphorien>
to force a rebuild use nix-build --check
<immae>
Incrementally is fine for now, but thanks for the command
<{^_^}>
[nixpkgs] @clefru opened pull request #54417 → electrum: 3.2.4 -> 3.3.2 plus new dependencies. → https://git.io/fhzuR
<ivegotasthma>
infinisil: thanks
simukis has quit [Client Quit]
simukis has joined #nixos
<infinisil>
ivegotasthma: Just reviewing package update PR's by testing them would help a great amount. Code reviews too if you're comfortable with them
<ivegotasthma>
infinisil: where should I look for the code reviews? So far I've only dabbled with packaging.
simukis has quit [Client Quit]
simukis has joined #nixos
<ivegotasthma>
I've been reading the threads from the issues / rfcs you sent, scaling issues are interesting
simukis has quit [Client Quit]
simukis has joined #nixos
<ivegotasthma>
testing out PRs on github is awkward
fusion809 has joined #nixos
ottidmes has joined #nixos
Phillemann has left #nixos ["WeeChat 2.3"]
<bennofs[m]>
Do we have any examples of a module for a python wsgi application in nixpkgs?
<__monty__>
ivegotasthma: Why is it awkward? With that last one ^ for example, you can get that by running `git fetch upstream pull/54417/head:local-branchname` then you checkout local-branchname and you can test it.
simukis has quit [Client Quit]
simukis has joined #nixos
<__monty__>
ivegotasthma: For code review click the `Files Change #N` link right under the PR's title. Then you can comment on specific lines or on the entire PR.
<__monty__>
ivegotasthma: With the little plus that appears next to the line numbers when you hover or the `Review changes` button at the top respectively.
<ivegotasthma>
__monty__: I get the review system, it's simple
<ivegotasthma>
__monty__: testing the changes is where the forked project PR workflow seems cumbersome
<ivegotasthma>
__monty__: thanks for the tip
simukis has quit [Client Quit]
fendor has joined #nixos
sanscoeur has quit [Remote host closed the connection]
<duairc>
What's the best way to globally add $HOME/.local/bin to the PATH for all users in NixOS?
simukis has joined #nixos
dermetfan has quit [Quit: WeeChat 2.3]
<qyliss^work>
duairc: environment.shellInit ?
xkapastel has quit [Quit: Connection closed for inactivity]
<laas>
I've noticed something odd, packages seem to build quite slowly from source, if I add an override to them
<laas>
e.g. I've added a very simple patch to kakoune, and it takes around 30 seconds to build the package
<laas>
Is this expected? My CPU load is very low the entire time
ilya-fedin has quit [Quit: Connection closed for inactivity]
eadwu has joined #nixos
<bennofs[m]>
why is the gitweb module defined in the same directory as nginx? (and it is a sub-service of nginx as well, it's config is services.nginx.gitweb.*)
<bennofs[m]>
Do we want to do this for services that integrate with nginx?
<gchristensen>
probably not
fendor has quit [Ping timeout: 246 seconds]
hvemvet has quit [Remote host closed the connection]
fendor has joined #nixos
<{^_^}>
[nixops] @domenkozar pushed commit from @roberth to master « fix another search path impurity »: https://git.io/fhzoi
<{^_^}>
[nixpkgs] @zimbatm pushed commit from @marsam to release-18.09 « aws-google-auth: init at 0.0.29 »: https://git.io/fhzoM
phreedom has quit [Remote host closed the connection]
phreedom has joined #nixos
fendor has quit [Ping timeout: 244 seconds]
<bennofs[m]>
also is there a tool I am missing to auto-generate simple python package definitions? (for inclusion in nixpkgs)
<srk>
pypi2nix?
<bennofs[m]>
which one? the one I remember generates a single big python-packages.nix that isn't really usable to for including in nixpkgs
<srk>
indeed
fendor has joined #nixos
fendor has quit [Ping timeout: 245 seconds]
<sphalerite>
laas: you can also pass --option cores 8 to nixos-rebuild :)
elibrokeit has quit [Quit: A random quit message]
<{^_^}>
[nixpkgs] @domenkozar pushed to master « stack2nix: distribute again »: https://git.io/fhzKj
<sb0>
okay. any information on how to do that? I went with the default settings, and hydra-init and hydra-create-user seemed to write something to some sqlite database
<sb0>
but then the web interface rejects the password I provided to hydra-create-user
<sb0>
also, for some reason, services.hydra.enable = true still installed postgresql
<gchristensen>
I think most things assume you will be using postgres
<gchristensen>
I think the sqlite support is mostly used in the test tooling, so maybe explore that
<sb0>
meh i'll just install postgres then
<lewo>
sb0: I'm pretty sure you need a postgresql DB
<hodapp>
only reason I ever messed with postgres was to deal with some GIS stuff that *had* to use PostGIS to do some basic file-to-file conversions >_>
<ivan>
# --help returns non-0 exit code even when working
<ivan>
signal-cli --version
snajpa has quit [Ping timeout: 268 seconds]
srk has quit [Ping timeout: 244 seconds]
aither has quit [Ping timeout: 246 seconds]
srk has joined #nixos
snajpa has joined #nixos
sanscoeur has joined #nixos
nomeata has joined #nixos
<nomeata>
I am trying to use nix-shell and pkgs.haskellPackages.shellFor to get an environment where I can build multiple packages using cabal new-build that we usually build using plain nix (and cabal2nix). Now I run into problems like this:
<nomeata>
<command line>: can't load .so/.DLL for: libgmpxx.so (libgmpxx.so: cannot open shared object file: No such file or directory)
<nomeata>
How do I find out what’s wrong?
aither has joined #nixos
dermetfan has quit [Quit: WeeChat 2.3]
<linkrage>
do we have "podman" package for NixOS? nix-env -qaP podman gets killed by OOM Killer I think (gotta upgrade)
<__monty__>
,locate podman
<{^_^}>
Couldn't find in any packages
<__monty__>
,locate pod2man
<{^_^}>
Found in packages: perl, perl526, perldevel, perl526Packages.podlators, perl528Packages.podlators, perldevelPackages.podlators
<__monty__>
There ya go, linkrage. I'm memory-constrained too so I usually join #bottest and ask {^_^} : )
<linkrage>
lewo, on NixOS in order to run containers with podman (which uses runc) without the need for docker daemon... do I have to install cri-o ? (I assume since we have runc package that's all I need -> podman+runc installed?)
ng0 has quit [Ping timeout: 256 seconds]
ng0 has joined #nixos
ng0 has quit [Client Quit]
<lewo>
linkrage: sorry, I don't know (i didn't play with podman yet)
<linkrage>
lewo, I see. There is small chance I'd have the time to test all of this on NixOS soon and if that's the case I'd be happy to report results/issues etc.
<lewo>
linkrage: yeah, it would be really nice:)
<linkrage>
lewo, If I have a chance to start on this I'll let you know on start as well :)
justan0theruser has quit [Ping timeout: 268 seconds]
fuzzy-id has quit [Ping timeout: 252 seconds]
<lewo>
linkrage: thx
dermetfan has joined #nixos
<nomeata>
elvishjerricco: when running `cabal new-build`. I can work around it by overriding the shellHook of the derivation produced by shellFor to set export LD_LIBRARY_PATH="${pkgs.gmp}/lib:$LD_LIBRARY_PATH". But this might be a local problem here, so maybe I shouldn’t escalate to the outside yet
<{^_^}>
[nixpkgs] @ivan opened pull request #54426 → signal-cli: init at 0.6.2 → https://git.io/fhzdv
orivej has quit [Ping timeout: 246 seconds]
<laas>
why is `nix-shell -p pkgs.some-package` different from `nix-shell -E 'let pkgs = import <nixpkgs> {}; in pkgs.some-package`?
<laas>
the first one doesn't get the compile-time dependencies
<gchristensen>
that gets you the build env of that package
<gchristensen>
which is the same as nix-shell -A some-package
<laas>
ah
<{^_^}>
[nixpkgs] @clefru closed pull request #54417 → electrum: 3.2.4 -> 3.3.2 plus new dependencies. → https://git.io/fhzuR
<lewo>
sb0: I don't think so. Maybe you can use the runCommand plugin to do that.
<Denommus>
I'm getting "error: while setting up the build environment: getting attributes of path '': No such file or directory" when I try to install any package
<Denommus>
I'm using nix over Arch Linux
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
<{^_^}>
[nixpkgs] @grahamc opened pull request #54431 → b612: init at 2019-01-21 → https://git.io/fhzx8
<_d0t>
hi. I wanna build some code from its source directory with nix-build. I get "variable $src or $srcs should point to the source" when I run nix-build. How do I fix this?
<_d0t>
Assuming I have dropped a default.nix in the source code dir.
<{^_^}>
[docker] @peti pushed to master « Update to nix version 2.2.1. »: https://git.io/fhzj9
<cransom>
you need a `src = ./.;` in your mkDerivation
<ottidmes>
_d0t: that is due to the unpackPhase expecting a src or srcs being defined by default, you can work around it in several ways, but the best way would depend on what you are packaging
Denommus has quit [Remote host closed the connection]
<_d0t>
Oh... silly me
<_d0t>
That was obvious
<_d0t>
Thank you cransom!
<_d0t>
ottidmes: the src hack worked just fine. All I want is to build the code in place.
<{^_^}>
[nixpkgs] @leenaars opened pull request #54433 → b612-font: init at 1.003 → https://git.io/fhzjj
<laas>
is there a way to force nix-build to build from source?
<gchristensen>
--check
<ottidmes>
_d0t: I assumed you did not have a src for your package, since setting src = something; seemed obivous to me :P if you ever have such a package, you can use e.g. buildCommand rather than using the normal phases
<laas>
thanks!
<_d0t>
another question. How do I call or pass flags to autoreconf?
<laas>
how do I make fetchFromGithub use the master branch?
<laas>
I can just change the revision, but I'd have to change the sha256 hash value every time the branch updates
<laas>
or is there a way to use a local path for a package?
<clever>
laas: src = ./.; or src = /home/clever/apps/foo;
<_d0t>
laas: funny. It's what I have just asked :)
hexo_ is now known as hovniatko
<__monty__>
Hmm, trying to test a python program that needs pygobject, on a mac. nix-shell -p 'python37.withPack...ps.pygobject3 ])' gnome3.gtk doesn't seem sufficient. Still get: `ValueError: Namespace Gtk not available`. Anyone can tell me what I'm doing wrong.
erictapen has quit [Ping timeout: 268 seconds]
<ar1a>
why in home-manager do we have to do programs.firefox.enable = true; instead of just adding it to home.packages?
<simpson>
ar1a: That said, I'm going to read your mind and guess that you want to know about `programs.firefox.package`, which will let you customize which Firefox package is used.
Ariakenom has joined #nixos
<laas>
where can one see the documentation for stdenv?
<laas>
I can see that there are documentation comments in the source, but I can't actually find a way to either generate them or any place to view pregenerated documentation
<ivan>
how can I make the font substitutions in my ~/.config/fontconfig/fonts.conf apply to everything except PDFs in Chromium
nschoe has joined #nixos
<tazjin>
simpson: we use the lib comments to generate docs with the tool that laas linked, the pull-request for it was merged during 35C3 and I think they're not in the public manual yet
<WilliamHamilton[>
clever: if I write `nix-shell -p "with import <nixpkgs> {}; pkgs.haskellPackages.ghcWithPackages (p: [(pkgs.haskellPackages.callPackage ~/local-package {}))) ])" "--run" "ghci"` then in the ghci session the package is present but hidden, and I have to type `:set -package local-package` to expose it. Why does this happen, and how could I make it exposed by default?
<laas>
tazjin: is there any online version of the docs?
<laas>
or is there a way to easily build them myself?
<ar1a>
how can I get command-not-found working? (i'm on nixpkgs-unstable) -- when i run `command-not-found sox` it spits out "cannot open database `/nix/var/nix/profiles/per-user/root/channels/nixos/programs.sqlite' at /run/current-system/sw/bin/command-not-found line 13."
<symphorien>
wucke13: is this driver different from epkowa ?
<ixxie>
tilpner: thanks that did the trick... silly me I had the fonts.fonts attribute in the same module so I should have known better
<symphorien>
(Epkowa is already in nixpkgs)
<tilpner>
wucke13: Fails with sed: -e expression #1, char 26: unknown option to `s' here, are you sure your repo is updated?
mmlb has quit [Ping timeout: 246 seconds]
<tilpner>
ar1a: Did you set programs.command-not-found.enable = true;?
<tilpner>
(Though it should be true by default)
linkrage has quit [Quit: Leaving]
<wucke13>
tilpner: sorry. Forgot about that attemp. Hang on a minute, I'll correct it.
<tilpner>
ar1a: Oh, are you using proper channels from nixos.org/channels, or from github?
<ar1a>
from nixos.org/channels
<tilpner>
Huh
<symphorien>
ar1a: nix-channel --update
<ar1a>
tried that
<ar1a>
ah, i have nix-index installed
<ar1a>
that might? be overwriting it
<symphorien>
With sudo as well ?
<ar1a>
oh, no its not
dermetfan has quit [Ping timeout: 245 seconds]
<tilpner>
And did you check if that database exists?
<ar1a>
`where command_not_found_handler` shows me the one from /etc/zshrc
<ar1a>
/nix/var/nix/profiles/per-user/root/channels/nixos/programs.sqlite: cannot open `/nix/var/nix/profiles/per-user/root/channels/nixos/programs.sqlite' (No such file or directory)
<ar1a>
and yes with sudo
<symphorien>
Strange
<wucke13>
symphorien: I'm just about to find out
kandinsk1 is now known as kandinski
<ar1a>
is it generated for nixPKGS-unstable? or only nixos-*
<ar1a>
symphorien:
<symphorien>
I don't know
<tilpner>
You can check by unpacking your channel
<tilpner>
Or... using the sqlite CLI to check if it's actually a valid database
Laalf has joined #nixos
<wucke13>
symphorien: with ` hardware.sane.extraBackends = with pkgs; [epkowa];` and `hardware.sane.netConf = "<printers ip in local subnet>";` no connection can be established.
<symphorien>
I have never tried epkowa with network scnaners
ng0 has joined #nixos
<symphorien>
The point is: epkowa includes a sotware calles iscan or image scan from epson
justanotheruser has joined #nixos
<symphorien>
software called*
<wucke13>
symphorien: Yeah. Epson links to the imagescan project if one searches for linux drivers on their page. Maybe the one is part of the other, though imagescan both offers a sane driver and some sort of GUI AFAIK
<symphorien>
Well maybe you don't need to completely duplicate the effort at least.
<lush>
hey everybody. Just started to update my nixos and for some reason it tries to compile libre office from source.. I let it try to do that for some hours now, but I'm kind of tired of waiting. Aren't there any precompiled binaries in the current cache?
<__monty__>
lush: The build'll probably fail.
<lush>
I don't need LO rn. Should I just delete it from my package list for a few days and try again later then?
<__monty__>
lush: Probably, unless you want to hunt for a channel revision that has a cached libreoffice.
<lush>
alright, thanks :) I don't need LO for the moment, I'll just reinstall it later
<lush>
worldofpeace: thx :)
ambro718 has quit [Quit: Konversation terminated!]
lush has quit [Quit: Page closed]
Denommus has joined #nixos
kklawikowski has joined #nixos
<kklawikowski>
Morning guys.
<Denommus>
I screwed up
<Denommus>
I erased my ~/.nixops
<kklawikowski>
Wassup
<Denommus>
can I again configure my existing instances with my files somehow?
<kklawikowski>
man, I screwed up even more
<kklawikowski>
just installed Windows 10 Enterprise Bloatware Free Edition
<kklawikowski>
that I like it
<kklawikowski>
and that's a sad thing
<kklawikowski>
such a snappy OS
<praetorg>
gross
<gchristensen>
kklawikowski: glad to hear you're happy with it
<kklawikowski>
just kidding guys, using this just for Adobe Creative Cloud stuff, just made new partition specially for nixOS <3
<praetorg>
i kid i have it on a partition to update my firmware every now and then
<betaboon>
Denommus: did that happen long ago? you might try to restore those deleted files with extundelete. but you have to umount the device. this if it is your root device you would have to boot from a usb-flash-drive
superherointj has quit [Ping timeout: 250 seconds]
<betaboon>
Denommus: assuming your on linux and on ext anyhow
<Denommus>
betaboon: it was this afternoon, but I already downloaded a bunch of stuff after that
<Denommus>
I bet it's already lost
<praetorg>
it might not be
<Denommus>
perhaps I could snapshot the disks, erase the instances, and create them again?
<ToxicFrog>
Does nixOS have any support for configuring Docker containers in the configuration? I rummaged around in options.html and couldn't find anything, but figured it was worth asking
<ToxicFrog>
(alternately, does anyone have a better way of running pihole on Nix, the installer is kind of a shitshow)
drakonis_ has joined #nixos
<betaboon>
Denommus: well yeah you could that. you can specify the snapshot id of which a ebs-volume is created. havent used that yet tho
<kklawikowski>
just asking, still no support for Adobe Creative Cloud on wine or something like this?
<betaboon>
Denommus: for the future: trash-cli
<ivan>
does nix hit the substituters in order, or in parallel?
<betaboon>
Denommus: or even better: put the statefiles into git and use git-crypt to encrypt them automatically (but encrypt them!)
acarrico has joined #nixos
Drakonis__ has quit [Ping timeout: 252 seconds]
bencahill has joined #nixos
<bencahill>
Hi! I'm having problems with a static IP I set not always being there. I think it is because sometimes the interface is eth0 and sometimes it is eth1 (config is set for eth0). I found the option networking.usePredictableInterfaceNames that looks like it would fix this, but it defaults to true and I have not changed it in my configuration. Any ideas what might be going on?
<betaboon>
Denommus: how many instances are we talking about ?
<Denommus>
betaboon: 2
<ivan>
bencahill: check journalctl -b for early boot messages on how network interfaces are being renamed by systemd. probably something really odd is going on with your hardware.
<betaboon>
well then thats not too much work
<betaboon>
but yeah, loosing state file is a buzzkill
ThatDocsLady has quit [Ping timeout: 246 seconds]
<Denommus>
betaboon: does nixops export work as a backup?
<kklawikowski>
praetorg thanks man
<ivan>
bencahill: "eth0" and "eth1" don't look like predictable interface names though
<kklawikowski>
did I forgot to mention that on Bloatware free edition the browser is missing?
<bencahill>
ivan: Thx. I see this: "kernel: ipw2200 0000:03:03.0 wlp3s3: renamed from eth0"
<betaboon>
Denommus: not that i know of. only those sqlite wal and shm files are there. but they are in the same location so most likely lost as well
<betaboon>
Denommus: and i dont know if you could restore from those
<bencahill>
ivan: That explains why it is sometimes eth1; wireless gets named eth0 and then renamed. The question is why doesn't eth0 get renamed? I don't see anything else indicative at a quick glance.
<ivan>
that would go in services.udev.extraRules = ''
reinhardt has quit [Quit: Leaving]
<ivan>
I don't know why something isn't getting renamed but systemd has options to increase verbosity if you want to go down that rabbithole
<bencahill>
ivan: I figured that would be my next step. I may just go the easy messy route and assign the IP to both eth0 and eth1 in configuration.nix so it will work either way.
<bencahill>
ivan: Oh I was searching for eth0...d'oh. Here are the lines about eth1:
<ivan>
some of these older drivers may be very poorly tested now
superherointj has joined #nixos
<ivan>
14 year old hardware may also be flaking
<ivan>
you can also compare two boot logs from a desired rename and undesired rename
<ivan>
journalctl -b -2 for the previous boot and so on
<bencahill>
Haha yep these are old machines. They work well for our purpose though. That is an awesome idea thx.
Denommus has joined #nixos
<bencahill>
I'm guessing it may just be a race issue and the network eth never gets renamed (since when it works it's just eth0)...just depends which the kernel recognizes first.
<ivan>
if it's an ordering issue maybe you can fix it by putting a kernel module into boot.initrd.availableKernelModules or maybe not
<ivan>
are you configuring eth0 and eth1 even though they have renamed names? is systemd letting you do that?
<bencahill>
ivan: Was configuring eth0. It just throws some errors when eth0 is not available (because it's eth1 this time). So I think if I configure both it will just throw an error for one each time.
<{^_^}>
[nixpkgs] @vcunat opened pull request #54438 → openssl_1_1: use the same default CA path as 1.0.* → https://git.io/fhgOy
<ivan>
bencahill: if systemd is giving you stable names, use those instead
<ivan>
e.g. wlp3s3 if that's the right one
<bencahill>
ivan: I think that's the wireless card. I need ethernet which is sometimes eth0 and sometimes eth1
<bencahill>
This is a laptop
<ivan>
:(
<ivan>
try the udev mac thing I guess
<bencahill>
I think I'm going to be lazy and do the eth0/eth1 thing because this is for a handful (6 or so) of machines that I want to be able to quickly rebuild with the fewest steps
<bencahill>
ivan: Thank you for your help. I guess I should have been able to figure those things out but it was nice to talk it through with someone. :-) I haven't used nixos for years and am just getting back into it for this project, and I am loving it, just as I did years ago. It really feels so "clean" after using Ubuntu etc.
<praetorg>
its great to have system config centralized and deterministic
<bencahill>
So nice. For this project I threw it in our version control which will be so nice for the future. And it's self-documenting which is huge too.
superherointj has quit [Quit: Leaving]
<laas>
Does anyone here use kakoune?
doyougnu has joined #nixos
<laas>
Syntax-highlighting doesn't seem to work for me for some reason
<laas>
even when I manually set it
<betaboon>
Denommus: is there any data on those nodes that you need to save ?
<Denommus>
betaboon: yes
<betaboon>
Denommus: take a snapshot of the esbs and recreate from those. or recreate instances from those snapshots, and get the data. the data is not list ;)
<Denommus>
there should be a way to associate an existing deployment with a new nixops setup
<betaboon>
Denommus: the problem is the stored privatekeys that are required to get into the existing ec2-instances
<betaboon>
Denommus: when those are lost there is nothing you can do
jomik has joined #nixos
jtojnar has quit [Quit: jtojnar]
<jomik>
Do we set `hardware.cpu.intel.updateMicrocode` in configuration.nix? I figured that it would fit better in hardware-configuration.nix. :o
<betaboon>
Denommus: i guess you have to do manual cleanup to do in the aws-console. like the subnets etc. I'm not quite sure if you can leave all the created resources in there and try to redeploy. i would realy suggest manually deleting all the stuff that is left there as to avoid conflicts
<Denommus>
I'll start to create btrfs snapshots for my work machine as well, to be safe
<Denommus>
anyway, gotta go home. Thanks for the attention, betaboon
<betaboon>
Denommus: sure. you're welcome
Denommus has quit [Read error: Connection reset by peer]
<ivan>
mdash: I don't like the formatting of hardware-configuration anyway
jomik has joined #nixos
<jomik>
Uh, I am trying to change my shell from fish to zsh. I set it to zsh in configuration.nix, but that does not seem to update my user..? Probably because I don't have immutable users?
<jomik>
How do I set it then?
<gchristensen>
did you log out / log in?
<ivan>
chsh?
<jomik>
I rebooted
<__monty__>
jomik: You're not gonna miss fish's history completion?
<gchristensen>
how did you set it?
<jomik>
__monty__: Probably - But there is similar in zsh.
<jomik>
I think the searchable auto completion may be missed more :p
<jomik>
gchristensen: I set it with `users.users.jomik.shell = pkgs.zsh;`
<jomik>
And I have `programs.zsh.enable = true;` set as well.
<__monty__>
zsh just can't do it as well as fish does though.
<jomik>
__monty__: Maybe not. But I got annoyed with having to explicitly tell programs to use bash instead of fish. Because they did not anticipate the user having a non-posix compliant shell.
<praetorg>
jomik did you reconfigure
<jomik>
praetorg: Elaborate
<praetorg>
jomik nixos-rebuild switch?
<jomik>
Of course :)
<__monty__>
jomik: That's why I `exec fish` in .zshrc : >
<jomik>
__monty__: cheaty :P
<praetorg>
and you shell is still not zsh?
<jomik>
Why have zsh then?
<__monty__>
jomik: Just so I don't run into programs that expect but don't find a posix shell.
<__monty__>
Oh, or why zsh rather than bash? It was the only shell on my arch.
<jomik>
Yeah, that :P I see.
ddellacosta has quit [Remote host closed the connection]
<jomik>
Seems a bit fishy to do that though.
<jomik>
I am going to give zsh a try.
<__monty__>
Why? It's best of both worlds. I never run into the problem you described and I get to use fish with its great defaults.
<praetorg>
jomik were you using fish 3.0?
<jomik>
Nah, 2.7.1 apparently.
<praetorg>
fish 3.0 is more posix compliant
<praetorg>
it's in unstable
<jomik>
Oh, they deprecated ^ in favor of 2> :O
<jomik>
And && ||
<jomik>
Ah, neato
<jomik>
Still doesn't solve my current issue though!
<jomik>
Why this does my shell not update when I change the shell in users.users? Is it because it is not immutable?
<jomik>
And what should I chsh to in that case?
<praetorg>
whats the contents of /etc/shells?
<jomik>
fish, bash and sh
<praetorg>
with zsh enabled as you say in you config it should be there
<jomik>
Honestly, fish should NOT be there, as I don't have fish enabled..
<jomik>
wait..
<jomik>
ah, nvm, `/nix/var/nix/profiles/system` is linked to my latest generation :D
jomik has quit [Quit: WeeChat 2.2]
danbst has joined #nixos
ixxie has quit [Remote host closed the connection]
bencahill has quit []
<jasongrossman>
I also sometimes have problems with programs expecting me to have a POSIX shell set as default. The most recent example is that if I use exwm and Alacritty (and fish) then my CPU usage goes up to 100%.
dermetfan has quit [Ping timeout: 246 seconds]
<__monty__>
o.O
dermetfan has joined #nixos
<praetorg>
scripts written without shebangs are a curse
<jasongrossman>
Oh wait. I just rememebered that it's worse than that. I couldn't get exwm to work at all when fish was my default shell. (And the probem above happens whether fish is default or not.)
<jasongrossman>
praetorg: Yes.
civodul has joined #nixos
jomik has joined #nixos
<jomik>
So! Apparently deleting my old generations helped.. Weird~
stranger___ has quit [Quit: Connection closed for inactivity]
<praetorg>
jasongrossman alacritty has scripts with explicit calls to /bin/bash. it's supposed to be fish compatible
<gchristensen>
seems much safer to leave the user's shell to bash/zsh, and have whatever starts your terminal run fish instead
<jasongrossman>
praetorg: I guess it's buggy then. Easy to work around though.
<jasongrossman>
gchristensen: Yes, but LESS ELEGANT!
<jasongrossman>
gchristensen: I agree that it's the best solution though.
<__monty__>
`exec fish` 1 - everyone else 0 : >
Maxdamantus has joined #nixos
<adisbladis>
Side note: Using fish as your login shell breaks emacs TRAMP (if you have fish on the remote)
<gchristensen>
especially for "could be useful" packages ... imo, nayway
<ottidmes>
I have been able to fix them easily so far, but this error is pretty scary "error: cannot coerce a function to a string, at undefined position", if I were to get and it was not caused by some recent changes I made, I would have to check all my code, yikes (--show-trace is enabled)
<danbst>
gchristensen: :thumbup:
perique has joined #nixos
knupfer has joined #nixos
Laalf has joined #nixos
ddellacosta has joined #nixos
o1lo01ol1o has joined #nixos
<{^_^}>
[nixpkgs] @matthew-piziak opened pull request #54443 → add textlint rules and plugins → https://git.io/fhgWL
<adisbladis>
ar1a: Hmm, how is that even possible? I guess you could hard-code /home/ar1a/.nix-profile/bin/emacs or even "${config.home-manager.users.ar1a.programs.emacs.package}/bin/emacs"
<adisbladis>
ar1a: Depending on how you use home-manager that is
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
kvda has joined #nixos
kvda has quit [Client Quit]
ddellaco_ has joined #nixos
simukis has quit [Quit: simukis]
jackdk has joined #nixos
ddellacosta has quit [Ping timeout: 250 seconds]
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
<ar1a>
that eventually got me there, /var/lib/unbound/unbound.conf
<ar1a>
thanks danbst
jasongrossman has quit [Remote host closed the connection]
gagbo has quit [Ping timeout: 268 seconds]
dvim has quit [Quit: WeeChat 2.3]
jasongrossman has joined #nixos
<jasongrossman>
symphorien: Do you have to do anything to get NixOS to pick up a ~/.Xcompose file question mark (I can't type a question mark at the moment because I just set that key to be my compose key)
<ar1a>
lol
<ar1a>
does ?? work
<jasongrossman>
ar1a: Not at the moment. No problem to sort that out later.
<jasongrossman>
ar1a: Once I've got custom compose sequences worked out.
<jasongrossman>
This is funny: a search for "nixos compose" gets lots of irrelevant hits, which I should have expected, since Nix is a functional language.
danbst has quit [Quit: Page closed]
jasongrossman has quit [Remote host closed the connection]
<ottidmes>
jasongrossman: just in case it might be relevant: services.xserver.xkbOptions = mkDefault "compose:menu"; is what I use for my compose key
jasongrossman has joined #nixos
iqubic has joined #nixos
<jasongrossman>
ottidmes: Yes, thanks. My problem is not setting the compose key but setting custom compose sequences.
carlosdagos has joined #nixos
melchips[m] has joined #nixos
eadwu has quit [Ping timeout: 250 seconds]
kvda has quit [Read error: Connection reset by peer]
doyougnu has quit [Ping timeout: 245 seconds]
rcshm has joined #nixos
<iqubic>
I have that issue too
<rcshm>
hi, i am trying to update the software by rerunning nix-env -f '<nixpkgs>' -iA devEnv. it was working before i updated nixos to 18.9. now it errors with nix-repl has been removed because it's not maintained anymore,
<cransom>
18:48:03 <iqubic> I have that issue too
<ar1a>
iqubic: i was searching around for nixos stuff and found logs from 2017 with you in there with my exact issue
<cransom>
oops, pardon
<ar1a>
do you have every issue?
<rcshm>
how do i get around this? thanks for any help.
<ar1a>
:P
<cransom>
i have a random clicking and pasting issue
<cransom>
rchsm: remove nix-repl from your default/shell.nix?
<colemickens>
ottidmes: cool, thank you
<rcshm>
infinisil, so instead of nix-env -f '<nixpkgs>' -iA devEnv what should i do?
<infinisil>
rcshm: As cransom said
<rcshm>
let me check cransom. but i don't recall any shell.nix.
<ottidmes>
colemickens: nix eval --raw nixpkgs.hello works too
<infinisil>
rcshm: You need to remove nix-repl from wherever devEnv is defined
<infinisil>
rcshm: This might be in ~/.config/nixpkgs/config.nix or ~/.config/nixpkgs/overlays/*
<cransom>
rcshm: you have at a least a default.nix. you don't need a nix-repl anymore. the system is telling you it doesn't exist anymore, you need to remove it from whatever is asking for it.
<infinisil>
Or somewhere completely different depending on how your setup is :)
<infinisil>
cransom: They don't need a default.nix really
ng0 has quit [Ping timeout: 256 seconds]
<rcshm>
i have confix.nix
o1lo01ol1o has quit [Remote host closed the connection]
<rcshm>
oh
<rcshm>
let me check
<infinisil>
rcshm: Yeah so devEnv might be defined there, you can remove nix-repl from it
o1lo01ol1o has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<rcshm>
got it infinisil.
<cransom>
yeah, thats true. i forget about creating envs in config.nix rather than a shell.nix setup
<iqubic>
What is the difference between shell.nix and config.nix?
<colemickens>
ottidmes: I see, thanks for that the nix eval and the link to the prefetcher work. I assume the focus is for use in automation? good to keep an eye on either way....
ng0 has joined #nixos
<rcshm>
that was it. thanks cransom, infinisil!
earldouglas has joined #nixos
<ottidmes>
colemickens: yes, it is meant for automation and just convenience when defining your own packages, do you have an example of a package that uses cargoSha256 that would be a good test case if it works with my tool?
<colemickens>
I have automated the update of packages there, other than for the Rust packages that need the cargoSha256.
<colemickens>
However, I've seen someone else in the Discourse building an auto-update mechanism that I'd rather collaborate with than keeping piling on a bash update.sh script.
<earldouglas>
Is there a way to install a package from nixpkgs that isn't listed in all-packages.nix? nix-build from that dir doesn't like the expression.
<colemickens>
I have (plans for) a few overlays repos and it would be nice to have a good layout/update script for suhc repos.