<hl>
FYI, the oggs play fine using mplayer, firefox audio works fine, audacious audio works fine, etc. Just event sounds don't work
hook has quit [(Ping timeout: 240 seconds)]
hook has joined #nixos
mortum5 has quit [(Ping timeout: 255 seconds)]
<neonfuz>
tilpner: I already tried that
<tilpner>
Just to make sure, the "Notification Sounds" entry in the system settings is neither muted nor at a low value?
<neonfuz>
btw is it okay to define boot.kernelModules in both configuration.nix and hardware-configuration.nix?
<tilpner>
neonfuz - Should do nothing
<neonfuz>
and should I take the warning in the hardware config seriously? Will something ever overwrite that file?
<hl>
tilpner (IRC): Where can I find that?
<tilpner>
neonfuz - But lsmod | grep rt2800 doesn't list anything?
<neonfuz>
it does
<neonfuz>
it's loaded
<tilpner>
hl - No idea, some volume settings
<tilpner>
neonfuz - And the number
<neonfuz>
but only because I did modprobe rt2800usb
<tilpner>
Oh, that's what you mean by doesn't work
<neonfuz>
yeah
<neonfuz>
ifconfig shows no extra interfaces, dmesg just says new device found with no driver loading
<tilpner>
So your question "what packages do firmare come in?" was completely irrelavant because you didn't say it already loads fine :/
<neonfuz>
what loads fine?
<tilpner>
Well, you said rt2800usb loads fine
<neonfuz>
the module does, I have no idea if the fw is
<neonfuz>
does the module loading imply the firmware loading?
<tilpner>
I would hope so, if it doesn't log any errors
<neonfuz>
alright
<tilpner>
(But I don't know)
<tilpner>
Do you pass -a to ifconfig?
<hl>
The 'multimedia' settings page seems broken. If I run systemsettings5 in a terminal, I get these lines amongst others:
<neonfuz>
I asked what pkg fw came in because debian instructs to install firmware-ralink
<neonfuz>
I forgot I have all firmware on
<hl>
org.kde.kcoreaddons: Error loading plugin "kcm_pulseaudio" "The shared library was not found."
<hl>
WARNING: bool Phonon::FactoryPrivate::createBackend() phonon backend plugin could not be loaded
<hl>
void BackendSelection::load() "/run/current-system/sw/lib/qt-5.9/plugins/phonon4qt5_backend" does not exist (and the same line for a number of other search paths)
<neonfuz>
tilpner: no, I'll try that
<neonfuz>
yeah nothing still
<tilpner>
neonfuz - Wireless always worked out-of-the-box for me, so I don't really know what I'm doing, but try https://askubuntu.com/a/61547
<tilpner>
(Just the different detection methods)
MP2E has quit [(Quit: leaving)]
<neonfuz>
brb rebooting
<neonfuz>
still nothing
<tilpner>
neonfuz - Don't modprobe and check dmesg
schoppenhauer has quit [(Read error: Connection timed out)]
<tilpner>
(Later modprobe to check dmesg again)
ris has quit [(Ping timeout: 255 seconds)]
<tilpner>
... wow
<tilpner>
neonfuz - Did you try networking.enableRalinkFirmware ?
<adelbertc>
Ran a `nix-build` (from a shell script) on macOS which gave me a `result` symlink as usual, but when I try to `rm result` I get a Permission denied - is this a bug or am I missing something ?
MP2E has joined #nixos
<adelbertc>
nvm i know what happened
<gchristensen>
what happened?
iqubic has joined #nixos
hook has quit [(Ping timeout: 240 seconds)]
<neonfuz>
I think the problem is udev not figuring out the right driver
<neonfuz>
or the driver not recognizing the card?
<neonfuz>
but my specific device id is listed on the debian wiki as supported by rt2800usb
Supersonic112 has quit [(Disconnected by services)]
Supersonic112_ has joined #nixos
Supersonic112_ is now known as Supersonic112
eacameron has joined #nixos
LnL has quit [(Quit: exit 1)]
sigmundv has quit [(Ping timeout: 258 seconds)]
mudri has quit [(Ping timeout: 260 seconds)]
LnL has joined #nixos
eacameron has quit [(Ping timeout: 248 seconds)]
erictapen has quit [(Ping timeout: 260 seconds)]
cpennington has joined #nixos
lambdamu_ has joined #nixos
lambdamu has quit [(Ping timeout: 255 seconds)]
hellrazo1 has joined #nixos
mbrgm has quit [(Ping timeout: 240 seconds)]
mbrgm has joined #nixos
hellrazor has quit [(Ping timeout: 240 seconds)]
eacameron has joined #nixos
jsoo has joined #nixos
jsoo has quit [(Client Quit)]
eacameron has quit [(Ping timeout: 255 seconds)]
vengefulpickle has joined #nixos
cpennington has quit [(Read error: Connection reset by peer)]
oida has quit [(Ping timeout: 248 seconds)]
oida has joined #nixos
<adelbertc>
gchristensen: my `deploy` script was just `cp`-ing files out of `result` so i assume they jsut retained their permissions in the Nix store
<adelbertc>
which is root
<adelbertc>
my `deploy` script is what was calling `nix-build`, i incorrectly assumed it was nix's fault :-)
<adelbertc>
separate question, is anyone familiar with `haskellPackages.developPackage` ? I'm currently trying to use it with `haskellPackages.developPackage { root = "./." }` but am getting "*** found zero or more than one cabal file ([]). Exiting." even though I can confirm I have exactly 1 cabal file in the current directory
<adelbertc>
shlevy ^ ?
<adelbertc>
oh wait i think i know why. it uses the basename of the folder to look for the cabal file which in this case is... `.` i guess
<adelbertc>
so i have to do "../<my-proj-name>" which is... annoying
atorres1985 has quit [(Quit: Leaving)]
<copumpkin>
if I'm in a double single quoted string and I want to type the literal '${ into it
<copumpkin>
how do I do that?
Wizek_ has quit [(Ping timeout: 248 seconds)]
<copumpkin>
normally to escape ${, I need ''${
<copumpkin>
but with the single quote ahead of it, I end up with '''${
<copumpkin>
which no longer gets tokenizsed properly as an escaped ${
<copumpkin>
I suppose I can create an intermediate value
<copumpkin>
feels awkward though
<copumpkin>
or another layer of ${}
<copumpkin>
I guess I'll go with '${"${"}' for now
<fearlessKim[m]>
networkmanager_l2tp sets in its makefile "dbusservicedir = $(sysconfdir)/dbus-1/system.d" which ends up being /etc/dbus-1/systemd . On build this generates "cannot create directory '/etc/dbus-1/system.d': Permission denied", where should this folder typically go ?
<fearlessKim[m]>
iqubic yes. doesn't it ?
<iqubic>
Haven't tried
csingley has quit [(Ping timeout: 246 seconds)]
<iqubic>
Yes, that seems to have worked.
<iqubic>
Except for this error: The option value `environment.variables.KEYTIMEOUT' in `/etc/nixos/configuration.nix' is not a string or list of strings.
<iqubic>
Seems simple enough to fix.
<iqubic>
but will it work I use string with the char 1?
pxc has quit [(Ping timeout: 255 seconds)]
pxc has joined #nixos
iqubic has quit [(Ping timeout: 260 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peterhoeg pushed 1 new commit to master: https://git.io/vdguy
<NixOS_GitHub>
nixpkgs/master 4817454 Peter Hoeg: wireguard: wg-quick systemd unit was referencing /usr/bin
NixOS_GitHub has left #nixos []
iqubic has joined #nixos
<iqubic>
Is it possible to chroot into a nixos install?
<iqubic>
I managed to screw up my nixos system somehow. Not even too sure what I did.
<isHavvy>
Yes. But I don't remember how.
<iqubic>
Basically the machine boots to systemd and I can select which os I want. Once I boot into nixos, I get the big wall of text scroll by. Then when my display manager starts, all I get is a black screen.
<isHavvy>
iqubic: Do you have previous boots you can go back to?
<iqubic>
No. I just collected garbage to get rid of the prior boots.
<iqubic>
Which I guess was a silly thing to do.
<iqubic>
"Nix-collect-garbage -d" followed by "nixos-rebuild boot" leaves me with a single non working nixos generation
<iqubic>
What can I do to fix this?
<iqubic>
I have the nixos installation media, so I can boot into that and try to fix this if possible.
sary has quit [(Ping timeout: 248 seconds)]
Nobabs27 has joined #nixos
<iqubic>
I really don't want to have to reinstall nixos
<iqubic>
That's the config I was most recently trying to get working.
<iqubic>
Anyone know how I can fix these issues?
sary has joined #nixos
pxc has joined #nixos
<fearlessKim[m]>
iqubic could you log into an alternate tty ?
<iqubic>
I didn't try that.
<iqubic>
Let me try that now.
<iqubic>
Be back soon.
<iqubic>
probably should fire up IRC on a different machine come to think of it.
<iqubic>
As opposed to on the live system
iqubic_ has joined #nixos
<iqubic_>
switching to a different tty and back to my display manager tty fixed the issues.
<iqubic_>
I wonder why that worked
<iqubic_>
Also, freenode still thinks that my name is in use
iqubic has quit [(Ping timeout: 260 seconds)]
moet has joined #nixos
<moet>
so the powertop changes that i was working on the other day.. i'm not sure that they work
<moet>
i've applied them to my own system using a packageOverride of postPatch on powertop
<moet>
and powertop now creates the appriate (empty) file in the correct location.. but it never writes anything to it
<moet>
so.. /shrug thanks for everyone's help yesterday!
<moet>
(i've tried running `sudo powertop --calibrate` several times)
iqubic_ is now known as iqubic
Nobabs27 has quit [(Quit: Leaving)]
rydnr has joined #nixos
<rydnr>
Hi all!
hamishmack has quit [(Quit: hamishmack)]
<rydnr>
I'm trying to install extraterm (https://github.com/sedwards2009/extraterm). I have no experience writing Nix packages. I've cloned the repo, run node2nix, created a new 'extraterm' folder into my local nixpkgs repo (under tools/misc), copied there all *.nix files generated by node2nix, and tried to install it with nixos-rebuild switch -I nixpkgs=[path-to-my-nixpkgs]
ebzzry has joined #nixos
hamishmack has joined #nixos
ebzzry has quit [(Ping timeout: 240 seconds)]
<rydnr>
It failed with an error message that wasn't particularly meaningful to me: error: "The option value `environment.systemPackages.[definition 3-entry 11]' in `/etc/nixos/packages/misc.nix' is not a package.". The 11th entry is precisely "extraterm".
ebzzry has joined #nixos
<rydnr>
So I tried to run nix-build directly on the extraterm repo, after running node2nix, and it failed because it couldn't find "fontconfig.h". I installed fontconfig globally with nixos-rebuild, but it apparently made no difference.
<fearlessKim[m]>
rydnr: node2nix might generate valid files for a nix-shell but not for integration in the repo. Could you make extra term work in a nix-shell ? I don't know node2nix
endformationage has quit [(Quit: WeeChat 1.9)]
<moet>
rydnr: the missing fontconfig stuff needs to be available in the build environment.. i'm not sure how to do that, but you probably need to modify the nix expression to indicate fontconfig is a dep
MP2E has quit [(Quit: leaving)]
<rydnr>
Thank you both. I'll read more about nix-shell, since I've never really used so far. And the fontconfig as dep makes a lot of sense.
<moet>
i mean i guess powertop adjusts other things too, so i'll leave it alone
moet has quit [(Quit: reboot with pstate)]
kwork has quit [(Ping timeout: 260 seconds)]
Wizek has quit [(Quit: Connection closed for inactivity)]
<phI||Ip>
Do services run under a different user in 17.09? Have a service that seems to no longer function - curious if its permission
moet has joined #nixos
<moet>
yey. ok. figured out why i couldn't just set the powersave governor
<moet>
i'd spelled the nixos option wrong
<moet>
fix'd
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vbgl pushed 2 new commits to master: https://git.io/vdg2B
<NixOS_GitHub>
nixpkgs/master 2870979 Vincent Laporte: ocamlPackages.ounit: fix for OCaml 4.06
<NixOS_GitHub>
nixpkgs/master 7013164 Vincent Laporte: ocamlPackages.oasis: remove unneeded dependencies
NixOS_GitHub has left #nixos []
<mudphone>
What is the `{ pkgs, stdenv, fetchurl,... }:` declaration at the top of a nix module?
<mudphone>
What is it called?
<MichaelRaskin>
A module is a function, and this is the function argument list.
<srhb>
mudphone: Those are the arguments.
<mudphone>
oh, interesting... I'm trying to understand what the dependencies of the erlangR20 package are, so I can install it via ASDF-vm, rather than via the nix package. Is that lunacy?
<mudphone>
(asdf-vm is a version manager, which works with Erlang)
slack1256 has quit [(Remote host closed the connection)]
<srhb>
In general you get most out of Nix if you go all in, and will probably have to struggle a bit compared to on other distros if you forgo the Nix way.
MercurialAlchemi has joined #nixos
<srhb>
That said, it's usually doable.
<mudphone>
that sounds like good advice...
<mudphone>
what is the nix way to handle multiple versions of a package?
<srhb>
Just having multiple version :-)
<srhb>
For instance, if you're developing, you'll usually bring up a development shell that brings all of your dependencies "into scope" within that shell.
<srhb>
Like so, nix-shell -p package -p package2 -p package3
<srhb>
(Usually you define that in a shell.nix file)
<srhb>
When you do like this, it's no problem simply bringing another version into scope
<srhb>
... -p package2_v3
<mudphone>
that sounds cool, where can I learn more about `nix-shell`? I've been reading the manual, but haven't seen much about it
<srhb>
Most of it is probably buried under language specific sections in the nixpkgs manual, unfortunately.
<srhb>
But you can glean a lot from that.
<mudphone>
Ah, okay, thanks. I've been reading the nixos manual.
<srhb>
mudphone: Ah, yeah, you want the nixpkgs one for this
<srhb>
Try section 9.1.6.2
<srhb>
In fact, read the entire Erlang section I guess :-)
<tilpner>
Allows for things like (rustChannelOf { channel = "nightly"; date = "2017-09-04"; }).rust
<Judson>
Cool...
reinzelmann has joined #nixos
<Judson>
Okay, but if I want to *distribute* something in Rust, we're still a few versions behind stable in nixpkgs
<tilpner>
Judson - buildRustPackage doesn't work with newer cargo, but rustc needs cargo to build
<Judson>
So there's an ongoing problem with a nix builder?
<tilpner>
Judson - There are efforts to replace/update buildRustPackage, not sure what the current status is
<tilpner>
Not sure how you mean that question
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
Mutter has joined #nixos
<Judson>
It looks like there's a PR with (what sounds like) a big improvement over the old buildRustPackage. No idea what the status is there.
<Judson>
Although it sounds like the existing PR breaks all the existing rust packages... They need to be touched to give them new hashes.
Mutter has quit [(Remote host closed the connection)]
<mudphone>
gleber_: are "autoreconfHook makeWrapper" packages that I need to lookup somewhere else? or functiona?
<mudphone>
*functions?
thc202 has joined #nixos
<gleber_>
mudphone: these are packages but represented as functions
<gleber_>
mudphone: You can usually look their definitions up in all-packages.nix
<gleber_>
makeWrapper is something 100% nix-specific and not interesting for your use case
<gleber_>
mudphone: autoreconfHook is a nix-specific wrapper around autoreconf
<mudphone>
ah, I see, it calls `autoreconf.sh` to install... etc.
<mudphone>
so, it looks like this is how I find out how the packages are built... but, if I want to manage multiple versions of Erlang for development purposes, I should use nix-shell.
proteusguy has joined #nixos
Ivanych has joined #nixos
<gleber_>
mudphone: that's correct
acarrico has quit [(Remote host closed the connection)]
<gleber_>
mudphone: `nix-shell --pure` establishes an environment where only explicitly specified packages are available
<otwieracz>
I've got a software which is not nix-based and requires libssl.so. Is there anything I can do?
<otwieracz>
(nixos)
<mudphone>
thanks, I'll take a closer look at nix-shell
<gleber_>
mudphone: given that all packages in nix are always self-contained and there are no "global" directories like `/usr/lib`, there's no problem in having multiple Erlangs defined in `nixpkgs`.
<gleber_>
mudphone: enjoy! Nix is one of the most elegant solution to a standard software problem out there :)
<mudphone>
I see, so if I run multiple shells with different erlang versions, they can exist side by side
<gleber_>
mudphone: correct
<mudphone>
and I'm assuming they don't have to be downloaded every time I boot a shell up?
<gleber_>
mudphone: power of `asdf-vm` is a strict subset of power of `nix`
<mudphone>
erlang takes a while to download and build
<mudphone>
ah, cool that's a clear way to put it
ertes has quit [(Ping timeout: 248 seconds)]
periklis has joined #nixos
<otwieracz>
Actually its `sbcl` requiring some additional libraries. How can I expose those to `sbcl`?
<gleber_>
mudphone: the packages built by nix-shell will stay `/nix/store` until someone runs nix GC
<MichaelRaskin>
you can set LD_LIBRARY_PATH, or you could add stuff to CFFI search list
<adisbladis>
mudphone: Things will be cached in the nix store
<M1k3y>
hello. Can I write a nix expression that gets validated when using nixos-rebuild, but is ignored when running nixos-install from the live system?
<MichaelRaskin>
otwieracz: Or you can build lispPackages."cl+ssl" and use its launcher called cl+ssl-lisp-launcher.sh
<mudphone>
gleber_: ah, okay, thanks... looks like `--add-root` is the thing to search for
pxc has quit [(Ping timeout: 240 seconds)]
<gleber_>
mudphone: yes, it will add a derivation into gcroots, which will prevent it from being collected
b has quit [(Quit: Lost terminal)]
<otwieracz>
MichaelRaskin: Seems like '+' is invalid in systemPackages…
<MichaelRaskin>
You need to quote attribute name with ""
<otwieracz>
Ah ok.
<MichaelRaskin>
Maybe + should be added to name-escaping list, though.
<MichaelRaskin>
(most of the packages are autogenerated from QuickLisp)
<MichaelRaskin>
But that's not a suggestion for you to implement, and this won't happen today.
MichaelRaskin has quit [(Quit: MichaelRaskin)]
<mudphone>
so this `nix-shell` stuff is really cool. I had not heard of it until today. I came for the reproducible OS builds.
mortum5 has joined #nixos
pxc has joined #nixos
kkini has quit [(Remote host closed the connection)]
kkini has joined #nixos
<gleber_>
mudphone: hehe, all of nix ecosystem is built on top of very elegant and powerful principles. A drawback is that to get all of them, you need to buy in into all of it, which is a non-trivial upfront cost. IMO it pays off pretty quickly afterwards
schoppenhauer has quit [(Read error: Connection timed out)]
jb55 has quit [(Ping timeout: 255 seconds)]
ertes-w has joined #nixos
<sphalerite>
Is building macOS apps (application bundles) which will run independently of nix feasible with nix?
schoppenhauer has joined #nixos
kwork has joined #nixos
iqubic has quit [(Ping timeout: 260 seconds)]
jensens has joined #nixos
kwork has quit [(Changing host)]
kwork has joined #nixos
Itkovian has joined #nixos
pxc has quit [(Ping timeout: 264 seconds)]
<jluttine>
what iso image writers there are available (for kde) in nixos? unetbootin requires root access..
<mudphone>
gleber_: I'm surprised people don't talk more about nix-shell.
peacememories has joined #nixos
<musicmatze[m]>
They all require root, that's because how Linux works, not because how the image writer works
peacememories has quit [(Client Quit)]
bin7me has joined #nixos
zzamboni has joined #nixos
phreedom has joined #nixos
<sphalerite>
jluttine: unetbootin isn't a good idea anyway nowadays
<sphalerite>
jluttine: you shouldn't need root for them if you give yourself write access to the device node though
<jluttine>
i just want to write an iso image to usb...
<sphalerite>
For USB sticks, dd is usually the wya to go
<sphalerite>
in that case, find the device node for the USB stick (lsblk can help with that)
<sphalerite>
then setfacl -m u:yourusername:rw /dev/sdz (replace z as appropriate)
peacememories has joined #nixos
<sphalerite>
^ as root
<sphalerite>
then you can `dd if=image.iso of=/dev/sdz bs=4M oflag=direct status=progress` to write the image to the USB
<sphalerite>
just be very careful so oyu don't overwite your system!
bkchr has joined #nixos
<otwieracz>
Does nixos install header files somewhere?
<otwieracz>
Like, when curl is installed, am I able to find curl.h somewhere?
<clever>
nixos doesnt install header files
<clever>
you must use nix-shell to get an environment that can use headers
<otwieracz>
I am building FHSUserEnv.
<clever>
compiling isnt meant to run under the FHSUserEnv
<otwieracz>
The problems is that I need environment where I am able to *build* things.
<otwieracz>
For my work.
<clever>
otwieracz: "nix-shell -p curl stdenv", this will give a shell that can build things
<otwieracz>
But the problem is that I need „regular” /
schoppenhauer has quit [(Ping timeout: 240 seconds)]
<otwieracz>
So it's impossible?
<clever>
the stdenv in nixos isnt really made to work under an FHS env, you will be fighting the normal code all the way
<clever>
it will be far simpler to work with if you do it the right way
schoppenhauer has joined #nixos
<otwieracz>
OK, but in nix-shell -p where is, for example, libcurl?
katyucha has joined #nixos
<otwieracz>
Probably it has to use pkgconfig?
<otwieracz>
I mean, the issue is that one tool I use (and I am unable to fix) does not look at environment variables when searching for shared libs.
<otwieracz>
It expects it in few predefined places.
<otwieracz>
I am unable to „nixify” it, I have to use it as is.
<otwieracz>
But I am afraid that's just impossible in Nix.
proteusguy has quit [(Remote host closed the connection)]
<clever>
otwieracz: nix will set $NIX_CFLAGS_COMPILE to include the right -I flags
<clever>
otwieracz: and the gcc in nix will obey that search path
<otwieracz>
Yes, but that's not gcc.
<clever>
so when you #include <curl.h>, it just works
<otwieracz>
That's Lisp's Common Foreign Function Interface.
<clever>
ah
<clever>
does that accept the -I flag?
<otwieracz>
The problem is that it's not calling GCC (in this specific moment).
<otwieracz>
But just loads shared library to create lisp bindings.
lsix has joined #nixos
jb55 has joined #nixos
<otwieracz>
That's why it needs library to be under /lib/lib{ssl,curl}.so
<clever>
you can set LD_LIBRARY_PATH to change the search path
<trikl[m]>
I'm trying to use bundix 2.3.1 to create a gemset, but extconf breaks when trying to install nokogiri 1.8.1, it can't find the zlib headers required to build libxml2
<hyper_webirc>
so, tried to not use zfs from unstable.... after copying back my home folder (100gb), there were like 200 corrupted files according to scrub
<hyper_webirc>
zfs is not usable for me - back to ext4
<gchristensen>
trikl[m]: okay lol
<gchristensen>
hyper_webirc: zfs stable lead to 200 corrupted files out of 800g?
zzamboni has quit [(Quit: Leaving.)]
<hyper_webirc>
no, 100gb home folder... rsyncing from back to zfs --> over 200 corrupted files
zzamboni has joined #nixos
<gchristensen>
oh sorry 200 out of 100g were broken by zfs stable?
<hyper_webirc>
rsync -avp /tmp/backup/home /mnt/ --> over 200 corrupted files according to zpool scrub
<gchristensen>
dude, your computer has something wrong with it
<hyper_webirc>
I checked one file (email)... in the backup it had like 50k lines but only 5k lines on the zfs
<hyper_webirc>
gchristensen: and what has my computer to do with it? on ext4 I never had such issues
<hyper_webirc>
badblocks - even with -w option - couldn't find any problem with the ssd
<gchristensen>
you'd never know, zfs is unique in that it can tell you where problems are
darlan has joined #nixos
<gchristensen>
have you done a deep memtest?
peacememories has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<LnL>
are you sure the disk is ok, it's not because ext doesn't complain that there's no problem
<hyper_webirc>
gchristensen: the files are fine on the backup
<hyper_webirc>
LnL: badblocks -w
<gchristensen>
have you run memtest?
<hyper_webirc>
no
<gchristensen>
zfs uses memory much more extensively than ext4, this couldbe related
<NixOS_GitHub>
[nixpkgs] maximedenes opened pull request #30248: pr template: make values of platform field less ambiguous (master...PR_template_nit) https://git.io/vdgxL
NixOS_GitHub has left #nixos []
nslqqq has quit [(Ping timeout: 248 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 2 new commits to master: https://git.io/vdgxc
<NixOS_GitHub>
nixpkgs/master 9ca1b56 Maxime Dénès: pr template: make values of platform field less ambiguous
<NixOS_GitHub>
[nixpkgs] shlevy pushed 1 new commit to master: https://git.io/vd2JR
<NixOS_GitHub>
nixpkgs/master f6858e5 Shea Levy: Reserve uid/gids for kanboard
NixOS_GitHub has left #nixos []
hyper_webirc has quit [(Ping timeout: 260 seconds)]
freusque has quit [(Quit: WeeChat 1.7.1)]
freusque has joined #nixos
proteusguy has joined #nixos
dywedir has joined #nixos
Itkovian has quit [(Remote host closed the connection)]
mgttlinger has joined #nixos
Itkovian has joined #nixos
silver_hook has quit [(Quit: Konversation terminated!)]
silver_hook has joined #nixos
silver_hook has quit [(Changing host)]
silver_hook has joined #nixos
pie_ has quit [(Remote host closed the connection)]
pie_ has joined #nixos
<flyx>
I have a custom mkDerivation that creates a lot of executables in $out/bin. is it possible to tell nix to just make a few of them available on the path when installing this?
<the-kenny>
flyx: I don't think so. The usual approach is moving them to $out/share/<package>/bin/... and symlink them to $out/bin/.
<symphorien>
maybe it is easier to do another derivation which contains symlinks to the executables you want
<avn>
(although I would like to see generalized solution for it myself)
<avn>
I use that scheme for readelf/strings from binutils, w/o pulling all binutils to system path
reinzelmann has quit [(Quit: Leaving)]
jtojnar has quit [(Read error: Connection reset by peer)]
jtojnar has joined #nixos
asdasodfi has joined #nixos
peacememories has joined #nixos
peacememories has quit [(Client Quit)]
fendor has joined #nixos
pie_ has quit [(Ping timeout: 240 seconds)]
jtojnar has quit [(Remote host closed the connection)]
zzamboni has quit [(Quit: Leaving.)]
Tucky has joined #nixos
pie_ has joined #nixos
anderslundstedt has quit [(Quit: leaving)]
jtojnar has joined #nixos
civodul has quit [(Quit: ERC (IRC client for Emacs 25.3.1))]
zzamboni has joined #nixos
iyzsong has joined #nixos
ThatDocsLady has joined #nixos
zzamboni has quit [(Quit: Leaving.)]
ertes-w has quit [(Quit: Bye!)]
keta_suki has quit [(Remote host closed the connection)]
keta_suki has joined #nixos
<silver_hook>
Is the option programs.mosh.enable there as the Mosh server or the Mosh client?
<hyper_ch>
adisbladis: input/output error
rmrfroot has quit [(Remote host closed the connection)]
lord| has quit [(Quit: WeeChat 1.9.1)]
ThatDocsLady has quit [(Remote host closed the connection)]
ThatDocsLady has joined #nixos
mortum5 has joined #nixos
superherointj has joined #nixos
dywedir has quit [(Remote host closed the connection)]
<srhb>
Errr. How are you supposed to use use the riot web client?
<srhb>
Opening the index.html file directory just serves a bunch of "missing translation" errors
seppellll has joined #nixos
b has joined #nixos
zzamboni has joined #nixos
<srhb>
Oh, looks like that's just chromium. Weird.
bkchr has quit [(Ping timeout: 240 seconds)]
bkchr has joined #nixos
michiel_l has quit [(Ping timeout: 255 seconds)]
michiel_l has joined #nixos
taktoa has joined #nixos
<gchristensen>
srhb: can you run: ls -l $(nix-instantiate --find-file nixpkgs) and share the output with me?
iyzsong has quit [(Ping timeout: 255 seconds)]
<nh2>
how man I supposed to use `sudo` from a nixos service? /run/wrappers/bin/sudo hardcoded?
vandenoever has joined #nixos
superherointj has quit [(Remote host closed the connection)]
<tilpner>
${pkgs.sudo}/bin/sudo, but I don't know why
<tilpner>
All the services use it, and I agree that it should require suid
iyzsong has joined #nixos
<tilpner>
Or maybe all those services start as root, so nobody noticed they're wrong
<sphalerite>
silver_hook: mosh has a server? I thought it just went through ssh
<etu>
sphalerite: it auths over ssh and starts mosh-server and then connects over udp to that server
<Yaniel>
sphalerite: it logs in through ssh and starts its own server then
<Yaniel>
And prints the port and presumably a secret to ssh stdout
<sphalerite>
right. So it doesn't have a long-lived server process as such
bkchr has quit [(Ping timeout: 240 seconds)]
<etu>
nah
<Yaniel>
Yeah, only as long as the session
<etu>
It goes away when you do a proper disconnect
<sphalerite>
right, and all the module does is add mosh to systemPackages and opens the firewall ports
<tilpner>
nh2 - gitlab uses PermissionsStartOnly and preStart with pkgs.sudo, and that works fine. preStart is executed as root here, and the rest of ExecStart will be done as User
anderslundstedt has joined #nixos
fendor has quit [(Ping timeout: 240 seconds)]
b has quit [(Remote host closed the connection)]
tmaekawa has quit [(Ping timeout: 260 seconds)]
b has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] domenkozar pushed 1 new commit to master: https://git.io/vd2OU
<NixOS_GitHub>
nixpkgs/master a1dfb19 Domen Kožar: haskellPackages.double-conversion: remove upstreamed patch...
<hodapp>
does anyone use Citrix Receiver here? I am using 'wfica' on some launch.ica, and it pops up with a box to say it's connecting, then just sits there at the commandline with nothing on-screen and no messages aside from some GTK warnings.
fendor has joined #nixos
<hodapp>
which makes it a royal pain in the ass to search for anything, because I have to find a way to articulate no-error-at-all-but-nothing-working to a search engine
<NixOS_GitHub>
nix/master fda7b95 Eelco Dolstra: Fix a hang in ThreadPool...
NixOS_GitHub has left #nixos []
<hodapp>
interesting...
<tilpner>
schoppenhauer - That's not very specific. Any errors?
<ertes-w>
if i import the <nixos/modules/profiles/all-hardware.nix> module *and* define boot.initrd.availableKernelModules, will the two get merged, or will my own definition override the one from all-hardware?
<tilpner>
ertes-w - Should get merged
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 1 new commit to master: https://git.io/vd2cg
<schoppenhauer>
(I should take the time to fully understand nix. but it's as always … busy with other things and half-knowledge is enough for most things…)
<catern>
gchristensen: though this reminds me - you're a committer for the website, right? would you accept a pull request for a "Friends of Nix" page ala Rust as we were discussing in this channel a couple weeks ago?
<tilpner>
schoppenhauer - That's the bootstrap ghc required to build the new ghc. Try passing ghc802
cmacrae has joined #nixos
<tilpner>
schoppenhauer - You can find this information in pkgs/top-level/haskell-packages.nix, around +83
<gchristensen>
catern: I'm not able to commit, but I'd thumb it :)
<catern>
gchristensen: hmm, who can commit it?
<gchristensen>
catern: eelco and domen
<cmacrae>
Hey LnL o/ I've come across a bit of a weird issue with nix-darwin. It seems in the latest version, the resulting `set-environment` file for zsh isn't executable, so when `/etc/zshrc` sources it, we get a `permission denied`
<cmacrae>
Happy to show you my config, and the output in the shell for the error
<gchristensen>
can a few people run `ls -l $(nix-instantiate --find-file nixpkgs)` and pastebin the results?
<LnL>
oh! you shouldn't need to add all that stuff again
<cmacrae>
What stuff? :) The PATH stuff? 'Cause I'm appending some stuff in there (Go binaries & Hashicorp utils)
<LnL>
cmacrae: check /etc/zshrc I'm pretty sure you'll see the path/environment stuff twice
<cmacrae>
Hmm, looks like it's just once. However: it does look like this is my fault anyway. I could just add `source` in front of `${config.system.build.setEnvironment}` and it should work :p
<schoppenhauer>
tilpner: where does the "packages" namespace in that file come from?
<cmacrae>
I say "my fault", but this did in fact work before. However, easily fixable by me
<schoppenhauer>
tilpner: in nixos, theres only one ghc package. there is no ghc802.
<copumpkin>
huh
<LnL>
cmacrae: or right, the environment stuff is in /etc/zshenv you can just remove it
<tilpner>
schoppenhauer - Can you even clone it...? :/
<tilpner>
schoppenhauer - tweag did some weird things with their remotes
<cmacrae>
LnL: No luck, I'm afraid
<LnL>
is it running?
<tilpner>
schoppenhauer - I was trying with tx0.co/yj0a.nix , but I can't even git clone it manually
<schoppenhauer>
tilpner: I have no idea, but nix-instantiate doesn't come that far
<tilpner>
(And I will not do their git config --global stuff)
<schoppenhauer>
at least I had no error message when I cloned it yesterday
<tilpner>
But this could be a challenge to just clone
<cmacrae>
LnL: It is indeed
fendor has quit [(Ping timeout: 240 seconds)]
jtojnar has quit [(Remote host closed the connection)]
jtojnar has joined #nixos
<schoppenhauer>
hm.
<LnL>
cmacrae: not sure then, no problems here
tmaekawa has joined #nixos
<cmacrae>
LnL: Weird! I'll poke about. Thanks for your help :)
<tilpner>
schoppenhauer - Ping me if you manage to clone it with Nix
<LnL>
cmacrae: sorry for breaking your config :( changed some types to match the nixos options
<schoppenhauer>
tilpner: ok, but currently, I have no idea how to get nix to even trying to clone it …
<tilpner>
schoppenhauer - Did you not see my paste...?
fendor has joined #nixos
tmaekawa has quit [(Client Quit)]
<schoppenhauer>
tilpner: ok sorry, I should change my irc client -.-
<tilpner>
No, weechat is fine, you just need to configure it properly
<cmacrae>
LnL: Eh, it's fine! Wouldn't have lovely declarative config of macOS without this awesome project - so I forgive ya ;)
<tilpner>
(Although highlighting should work by default)
<tilpner>
(Unless you're on some relay client...)
gandreani has joined #nixos
zzamboni has quit [(Quit: Leaving.)]
zzamboni has joined #nixos
<schoppenhauer>
tilpner: I use glowing bear.
jtojnar has quit [(Remote host closed the connection)]
<schoppenhauer>
tilpner: and I need to change some css data for highlighted messages
<tilpner>
Hmm, never tried it, but heard lots of good about it
jtojnar has joined #nixos
eacameron has joined #nixos
csingley has quit [(Ping timeout: 240 seconds)]
<cmacrae>
LnL: Weird... I've just mimicked what the launchd job for khd is doing in a shell (set the PATH, ran the same command) and it works perfectly fine...
<cmacrae>
So, it seems only when it's loaded as a launchd job that it isn't behaving
jtojnar has quit [(Remote host closed the connection)]
Itkovian has joined #nixos
<silver_hook>
So, what is the programs.mosh.enable option for then?
jtojnar has joined #nixos
whald has joined #nixos
zzamboni has quit [(Quit: Leaving.)]
zzamboni has joined #nixos
jtojnar has quit [(Remote host closed the connection)]
jtojnar has joined #nixos
pie_ has quit [(Ping timeout: 240 seconds)]
erictapen has quit [(Remote host closed the connection)]
reinzelmann has quit [(Quit: Leaving)]
erictapen has joined #nixos
zzamboni has quit [(Quit: Leaving.)]
zzamboni has joined #nixos
zzamboni has quit [(Client Quit)]
zzamboni has joined #nixos
zzamboni has quit [(Client Quit)]
zzamboni has joined #nixos
mgttlinger has quit [(Quit: Connection closed for inactivity)]
<viric>
does anybody know of anything regarding qt5 and nix-shell, and a directory named "__nix_qt5__"?
<cmacrae>
Also LnL: I really like the changelog :) It's nice to have
charlycoste has quit [(Ping timeout: 255 seconds)]
<nh2>
tilpner: hmm, when I use "${pkgs.sudo}/bin/sudo ..." it complains that that sudo binary is not suid
acarrico has joined #nixos
<tilpner>
nh2 - Are you root at that moment?
freusque has quit [(Quit: WeeChat 1.7.1)]
<tilpner>
viric - Only that it's very annoying and a reason I don't use kde applications :/
<nh2>
tilpner: no, I'm the `consul` user when I call sudo
<tilpner>
nh2 - Did you read what I said about gitlab and preStart?
<tilpner>
Rather, do you have to be consul for that?
<viric>
tilpner: what is that thing?
<nh2>
tilpner: yes, using PermissionsStartOnly= you can make the preStart be executed by root, however what I need to run is not in a systemd unit at all, it's in a consul "check" (a script it runs regularly)
<nh2>
tilpner: correspondingly there's a sudoers entry that permits consul to run a specific command as root, but it seems that when I'm not actually root I can't use the sudo from the nix store directly
ixxie has joined #nixos
<ij>
I get an error on nixos-install. How do I get a stack trace or the file/line number?
<clever>
nh2: only /run/wrappers/bin/sudo is setuid root
michiel_l has quit [(Ping timeout: 255 seconds)]
<srhb>
nh2: The real sudo has a security wrapper
<nh2>
the location I posted works but I'm not sure if I should hardcode that or if there's a ${...} expression I should use instead to make its location more robust
<tilpner>
viric - A qtbase development hook creates it as a temp directory
<clever>
nh2: nix doesnt allow any setuid files within /nix/store/, so nixos has to create wrappers at bootup, that add the setuid in
<viric>
tilpner: aah
<tilpner>
ij - "--show-trace"
Sonarpulse has joined #nixos
<clever>
viric: and every time you leave nix-shell, you have to remove it to let nix-shell work again
<nh2>
clever: is /run/wrappers/bin/sudo a stable location I can rely on to contain sudo for the foreseeable future? Or is there some expression that would return its path?
<ij>
tilpner, accepted the flag, but didn't print a trace :(
<viric>
clever: right
<tilpner>
nh2 - config.security.wrapperDir IIRC
<viric>
clever: I guess I only experienced the annoyance; I have to find out the benefit, now
<clever>
nh2: it has changed at least once since i began using nixos
michiel_l has joined #nixos
<tilpner>
ij - What's the error?
<ij>
tilpner, It tries to put a symlink of /nix/store/…-nixos-system-… into /boot, which is vfat.
<clever>
ij: can you pastebin the full error, and several lines around it
<clever>
makefu: yeah, things have improved with linode, but the old guide is still handy to forcibly install nixos-install on any distro
<clever>
makefu: i have played with "nixos-rebuild build-vm" on gentoo before
<pop>
So my current setup is I have a file in /etc/nixos/configuration.nix which installs packages system-wide. Is there any way to only install packages for a specific user (and so only install packages in /home/username/*)?
<pop>
*running nixos.
<makefu>
clever: maybe we should add the info to the new wiki then (instead of linking to archive.org)
<clever>
makefu: that would help a lot
<tilpner>
pop - nix-env installed packages are user-local by default
<pop>
tilpner: Can I use a config file like configuration.nix for nix-env?
<tilpner>
Sort of, it's more of a workaround
<pop>
I've only seen examples of `nix-env -i <package name>`
<gchristensen>
there seems to be a strange symlink in the channel of nixpkgs to `.`, so I can do: /nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs/nixpkgs/nixpkgs/nixpkgs/nixpkgs forever. I've seen this now on two boxes. any ideas what's up?
<gchristensen>
is there a way to copy the channel version in to a drv? I can only seem to manage to copy the `.` symlink and not the actual channel data
<clever>
gchristensen: <nixpkgs> returns the nixpkgs directory of the nixos channel
<NixOS_GitHub>
[nixpkgs] peti pushed 4 new commits to master: https://git.io/vd225
<NixOS_GitHub>
nixpkgs/master a02e53a Peter Simons: hackage-packages.nix: automatic Haskell package set update...
<NixOS_GitHub>
nixpkgs/master d0e8806 Peter Simons: haskell-tinc: mark build as broken...
<NixOS_GitHub>
nixpkgs/master d98909e Peter Simons: hackage2nix: disable broken builds on Hydra
NixOS_GitHub has left #nixos []
<clever>
gchristensen: this will fetch a specific version of nixpkgs (specified in a json file), and use that for the eval, and also embed that nixpkgs into the build and use it at runtime
<clever>
yikes, the nixops AMI has a configuration.nix file!
<gchristensen>
yes :)
<clever>
$ echo $NIX_PATH
<clever>
nixpkgs=/run/current-system/nixpkgs
<clever>
gchristensen: ah, but nixos-config is entirely mia in the search path
<clever>
it wont find it
<gchristensen>
and so does any other nixos system you've setup and then took over with nixops
<gchristensen>
ah right, only because you're specifying the nixPath there
<gchristensen>
if you don't set the nixPath it'll have one and it'll be set to have a nixos-config
<clever>
yeah
jtojnar has quit [(Quit: jtojnar)]
<gchristensen>
clever: the "actual result" is a thing I actually did last week :)
<tilpner>
That's cheating! Using internal options like that... :c
<gchristensen>
tilpner: what internal options?
<tilpner>
extraSystemBuildsCmds
<tilpner>
*extraSystemBuilderCmds
<clever>
system.extraSystemBuilderCmds = ''
<clever>
ln -sv ${./.} $out/nixcfg
<clever>
tilpner: this puts a snapshot of all of my nixos config into every build
<gchristensen>
<3
<gchristensen>
goldmine right there
<boxofrox>
what does ${./.} represent/evaluate to?
<clever>
boxofrox: it will import the directory containing this nix file, into /nix/store/ (in my case, a snapshot of /etc/nixos/nixcfg) and then expand to that storepath
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 1 new commit to master: https://git.io/vd2aM
<NixOS_GitHub>
nixpkgs/master 2c21f3d Danylo Hlynskyi: readme.md: reincarnate wiki (#30243)...
NixOS_GitHub has left #nixos []
<tilpner>
Huh, so systemBuilders runs before all activationScripts. I'm stealing that
<clever>
extraSystemBuilderCmds runs at nix-build time, when creating the nixos build
pxc has quit [(Ping timeout: 264 seconds)]
<tilpner>
Just before activation, in a nixos-rebuild switch
<ij>
Did anyone see my whining about the docker-container profile in the mailing list?
<clever>
tilpner: and unlike activation, it wont run again on boot
* tilpner
puts "Look through all of clevers config" on todo
<gchristensen>
ij: calling it whining is not productive, please be considerate of the people in the community, even if they're unhappy or having a bad time
<ij>
I'm talking about myself.
<gchristensen>
oh _MY_ whining that is a slightly different story :P
<clever>
gchristensen: have you heard my idea about adding my kexec trick to nixops?
<gchristensen>
I have :)
<gchristensen>
I got half way to implementing it before Packet.net added proper NixOS support
<clever>
heh
<clever>
gchristensen: i can also see it being handy for non-standard partitioning/filesystems on existing platforms, like aws
<clever>
gchristensen: for example, zfs root
<tilpner>
gchristensen, clever - While I have you two here, I was wondering where I need to bring up a change proposol, and if this has already been raised. The problem is that the argument lists of many packages include configuration options that are not build inputs. callPackage will auto-fill them, even though the package author didn't intend that. This leads to nixpkgs dying with quite unhelpful error messages when
<tilpner>
packages with the name of these config options are added/overlayed to nixpkgs. The examples I encountered are "unicode" and "vanilla"
<gchristensen>
definitely
<gchristensen>
tilpner: yikes
<tilpner>
I would first try a two-stage loading of packages as a convention, first providing build inputs with callPackage, then providing config options with defaults from all-packages
schoppenhauer has quit [(Read error: Connection timed out)]
<tilpner>
But this touches so many packages, and could only be implemented gradually as a "do this during package updates" convention. Or the responsible team could decide it's not worth the effort...
<tilpner>
Overriding would need a little thought for that two-stage thing
<zimbatm>
it might also have quite a bit of impact on performance
<zimbatm>
nix is not that fast yet :)
<tilpner>
Alternatively we could just gradually rename these config options. "unicode" -> "supportUnicode" (vanilla -> ?)
<tilpner>
(That's not as clean, but wouldn't incur additional overhead
<tilpner>
)
pxc has joined #nixos
erictapen has quit [(Ping timeout: 255 seconds)]
schoppenhauer has joined #nixos
seppellll has quit [(Quit: Leaving)]
pxc has quit [(Ping timeout: 240 seconds)]
csingley has joined #nixos
fragamus has joined #nixos
jb55 has joined #nixos
pie_ has joined #nixos
erasmas has joined #nixos
<cbarrett>
trying to install nixos (under virtualbox) and it fails while “setting up /etc…”
<cbarrett>
spews a whole bunch of stuff from /etc/resolvconf.conf about various commands not being found
<cbarrett>
hopopt, icmp, igmp, gpg, ipv4, st, tcp, cbt, egp, there’s pages and pages of this
digitus has joined #nixos
<tilpner>
cbarrett - Installing from a NixOS ISO? Please paste the used commands and errors (preferably to gist.github.com)
<cbarrett>
ya
<tilpner>
(And probably your /etc/nixos/configuration.nix)
Neo-- has joined #nixos
<cbarrett>
i’ve just been following the instructions in the manual
<cbarrett>
so nothing special there i think. no way to copy and paste configuration.nix from vbox unfortunately
<sphalerite>
Is there a place that describes the historical context behind all modules in nixpkgs/nixos being evaluated when building a system config and usually being gated on an enable option, rather than having the user enable them by adding them to the imports, making evaluation a lot speedier?
<sphalerite>
Well, not necessarily just historical context but actual reasons for it
<sphalerite>
seems to me like it makes system evaluation slower while also making it harder to disable some default behaviours
<cbarrett>
srhb: it fails mounting my /vbox system and then goes into emergency mode and I can’t type at the prompt at all
<srhb>
cbarrett: I was wondering about the device, but since I don't know much about vbox I didn't say anything. Should have probably spoken up.
<srhb>
cbarrett: Might want to just disable it in the installer.
<cbarrett>
k
<srhb>
You can always add it from a functioning system
<cbarrett>
makes sense
<tilpner>
cbarrett - Can you use the grub line editor?
<tilpner>
(Or does that not show in VB?)
<srhb>
(Having booted once you get one system you can always get back to)
<cbarrett>
yeah grub shows up but i’ll just boot back into the CD
<srhb>
Well, barring Bad Things. O:-)
<cbarrett>
actually I think i fixed it, I didn’t have vbox set up right to mount that folder
<srhb>
Oh ok :)
<cbarrett>
hurrah I can run uname -a
<cbarrett>
I can ping google
<cbarrett>
jawesome
<cbarrett>
thx tilpner & srhb
b has quit [(Remote host closed the connection)]
<cbarrett>
I’ll look into that ticket in the next few days, hopefully tonight
<srhb>
Woo :)
ertes has joined #nixos
<srhb>
I think I might take a stab too, seeing as I've never managed a NixOS contribution yet.
<tilpner>
(my router died, or something like that)
<srhb>
Looks like find * from ${pkgs.tzdata}/share/zoneinfo/right might provide all the valid options
<srhb>
So I guess I just need to build an enum from that.
b has joined #nixos
<cbarrett>
srhb: sweet!
endformationage has joined #nixos
<cbarrett>
srhb: happy to pair with you on it as well, you should go for it!
<srhb>
Of course, this is one of those things that will break in really hairy and obscure ways if I get it wrong. :-P
<srhb>
Cool, thanks cbarrett :)
<cbarrett>
i definitely think that it should just fail if the tz doesnt exist. something like cat /share/timezones | grep “${sys.timeZone}”
<cbarrett>
or whatever
<srhb>
Yeah.
<tilpner>
Maybe you can somehow use "${systemd}/bin/timedatectl list-timezones" instead
<cbarrett>
yeah totally. was just pseudo code
<srhb>
No, that's insufficient
Judson1 has joined #nixos
Judson1 is now known as Guest24636
<srhb>
It's lacking things like GMT-0, Universal, Zulu, ...
<Isorkin>
How to change services.fail2ban.jails defaul ignoreip. Need add custom ip ?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] globin pushed 1 new commit to master: https://git.io/vd2Ha
<NixOS_GitHub>
nixpkgs/master 4bba665 Robin Gloster: liboauth: fix for new nss header layout
NixOS_GitHub has left #nixos []
leat has joined #nixos
<tilpner>
What happens when setting services.fail2ban.jails.DEFAULT?
phreedom has quit [(Ping timeout: 255 seconds)]
<tilpner>
Isorkin - ?
<Isorkin>
tilpner: generated jail.conf - [DEFAULT] ignoreip = 192.168.2.0/24 ignoreip = 127.0.0.1/8 - the last value is used
sary has joined #nixos
<tilpner>
Seems like there is no mechanism for that then. Maybe manually copy and import an altered version of security/fail2ban.nix
<infinisil>
or just write a systemd.service directly
<makefu>
^ this is what i would recommend
<csingley>
Yes that sounds wise.
<tilpner>
MichaelRaskin - Did you mean imports, or is require a thing?
<csingley>
However, the custom service requires custom pkgs outside the tree, which I'd like to also handle in configuration.nix
<csingley>
How to handle that reference?
cmacrae has joined #nixos
<MichaelRaskin>
require attribute is a thing.
<srhb>
Bugger, the is, like, a perfect use of lib.pathExists, but it;s impure :-P
endformationage has quit [(Quit: WeeChat 1.9)]
<srhb>
type = types.nullOr (types.addCheck types.str (tz: lib.pathExists "${pkgs.tzdata}/share/zoneinfo/right/${tz}")
<srhb>
Q_Q
endformationage has joined #nixos
cmacrae has quit [(Ping timeout: 240 seconds)]
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
griff_ has joined #nixos
griff_ has quit [(Client Quit)]
sigmundv has joined #nixos
mudri has quit [(Ping timeout: 240 seconds)]
<hyper_ch>
I suspect my zfs issues are due to ram... will be interesting to see what memtest will say
adisbladis has quit [(Ping timeout: 240 seconds)]
pie_ has quit [(Remote host closed the connection)]
pie_ has joined #nixos
<gchristensen>
<3 sounds good, hyper_ch
jb55 has joined #nixos
sigmundv has quit [(Ping timeout: 240 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] acowley opened pull request #30256: easyloggingpp: init at 9.95.0 (master...easyloggingpp-init) https://git.io/vd2b8
NixOS_GitHub has left #nixos []
<srhb>
Hmm.
[0x4A6F] has joined #nixos
<srhb>
Are there any examples of nixos services conceptually doing something like checking that an option matches a filename in some (other) store path?
<srhb>
I mean, as long as it's a dependency, is that really impure?
<srhb>
I feel like it's morally pure. :-P
fragamus has joined #nixos
griff_ has joined #nixos
<MichaelRaskin>
It's more like incompatible with dry run
<gchristensen>
those aren't store paths, but do go on :)
<MichaelRaskin>
5D
<pxc>
you're right. What should we call the base path for the store?
<pxc>
the Nix prefix
<pxc>
?
* gchristensen
checks the glossary
<MichaelRaskin>
Well, NixOS definitely assumes that Nix prefix is /nix/
<pxc>
yes, and that is fine
<pxc>
I would like to make it possible for the Fish package (whether this should be enabled by default can be decided later) to automatically set up the Nix environment for users, as early in init as possible
<gchristensen>
I think the nix build just calls it the prefix
<pxc>
in NixOS we do this by having it hook in to a file that NixOS puts in a certain place on the filesystem
<pxc>
for non-NixOS, we could just as easily have it source /nix/var/nix/profiles/default/etc/profile.d/nix.sh (through fenv) or nix-daemon.sh
<pxc>
so that whenever you install Fish _through Nix_, no configuration is required to make it pick up your NIX_* environment variables
MP2E has joined #nixos
<pxc>
this could be done easily, but Nix can be built to use different prefixes
<pxc>
NixOS doesn't seem to do it, but is there any way to look up the Nix prefix inside a package build?
<pxc>
that way I can add this in a generic way
ShalokShalom has joined #nixos
<pxc>
so that you don't have to run NixOS or set up nix-darwin or nix-home or some kind of module system to get your shell to initialize properly
<LnL>
pxc: $NIX_STORE should be set inside the build environment
<pxc>
LnL: so I can get the first component of that and use it?
<MichaelRaskin>
Well, there can be more than one component
[0x4A6F] has quit [(Ping timeout: 240 seconds)]
[0x4A6F]1 is now known as [0x4A6F]
betaboon has joined #nixos
fendor has joined #nixos
ixxie has joined #nixos
<pxc>
MichaelRaskin: what do you mean?
<pxc>
all I need to know is <where>/var/nix/profiles lives
<MichaelRaskin>
<where> can be /home/user/nix
<betaboon>
hello #nixos. I have a question about the autoUpgrade module: i would have expected it to set nixpkgs in nixpath, but it doesnt. is that somewhat counterintuitive or do i have a weird way of thinking?
taktoa has quit [(Remote host closed the connection)]
fendor has quit [(Ping timeout: 258 seconds)]
<pxc>
MichaelRaskin: I guess I'll just grab everything prior to the `/store` component. Should I care if what I want to do would break if someone's NIX_STORE is set to /nix/store/nix/store/nix/store/hahaha/nix/store?
<pxc>
but that's a good point, thanks
<pxc>
what I just said is dumb but I figured out how to actually do what I want, sorry for thinking out loud here. I'll try to tone it down lol
<MichaelRaskin>
Why not $NIX_STORE/.. ?
<pxc>
lol that'll do
metaphysician has quit [(Ping timeout: 240 seconds)]
metaphysician has joined #nixos
* ij
found the derivation nixos-system-…-… derivation that he was asking about earlier
mithsv has quit [(Read error: Connection reset by peer)]
mithsv has joined #nixos
<gleber_>
Have anyone ever seen both / and /nix/store be mounted with the same device? And yet these two directories have different content... O_O
<pxc>
yeah, I forget why it looks like that
<srhb>
Probably because /nix/store wants to be ro
<srhb>
Despite being on the same partition
<gleber_>
oh, so this is normal in NixOS?..
<pxc>
yes
<srhb>
Sounds normal to me.
<gleber_>
I almost got a heart attack :D
<srhb>
try to avoid that! :)
<LnL>
lol
<pxc>
for no real reason, I love the strangeness of NixOS sometimes
<pxc>
even the strangeness of seeing those crazy store paths for the first time, or discovering there's hardly a /usr/bin
fragamus has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<pxc>
there's a feeling of adventure in it :-D
griff_ has quit [(Quit: griff_)]
<srhb>
I think libc missing was mine.
dywedir has joined #nixos
zarel has joined #nixos
zzamboni has quit [(Ping timeout: 240 seconds)]
erictapen has quit [(Ping timeout: 258 seconds)]
fendor has joined #nixos
mudri has quit [(Ping timeout: 260 seconds)]
Arcaelyx has quit [(Ping timeout: 240 seconds)]
<cement>
or having a friend suggest nixos even though I hadn't run a POSIX *nix for long enough to know why nixos was better
<srhb>
That sounds depressingly underwhelming. Should not be attempted. :-P
<srhb>
On the other hand, switching to Ubuntu afterwards will be really... !!FUN!!
erictapen has joined #nixos
hyper_webirc has joined #nixos
<hyper_webirc>
so, running memtest86+ now... but it seems so slow :)
<gchristensen>
hyper_webirc: it is slow, it is a very extensive test :)
fendor has quit [(Ping timeout: 248 seconds)]
Arcaelyx has joined #nixos
<hyper_webirc>
will be interesting to see the results
griff_ has joined #nixos
<hyper_webirc>
how many passes should I do?
zzamboni has joined #nixos
Unode has quit [(Read error: Connection reset by peer)]
Unode_ has joined #nixos
<wilornel>
Hey #nixos! I need ot install eclipse platform 4.7. https://nixos.org/nixos/packages.html shows that ` eclipses.eclipse-platform` is the attribute I should use. However, when I do that, it installs eclipse-platform-4.6
Unode_ is now known as Unode
<hyper_webirc>
ok, memtest has become unresponsive...
<wilornel>
what was the command I had used last time to see the mapping between package name and attribute?
<hyper_webirc>
doing now failsafe mode
civodul has joined #nixos
<srhb>
wilornel: nix-env -qaP ?
<hyper_webirc>
memtest becoming unresponsive seems bad :(
<wilornel>
alright, and I did `nix-env -u home-manager` and then I tried `home-manager switch` but that did not update eclipse. Seems like home-manager is getting the old nixpkgs
<srhb>
wilornel: nix-channel --list ?
<srhb>
I'm assuming home-manager just imports nixpkgs, so if you have multiple channels I'm not sure exactly how it chooses.
<srhb>
In case that's too confusing, perhaps having just one channel will be easier :)
<Alling>
srhb: It seems a bit ... experimental. No way to do it in configuration.nix? I understand that private keys should be added manually, but it would be nice to have the rest of .ssh/config in my system configuration.
<Alling>
srhb: And I guess I want everything in ~/.ssh, not /etc/ssh
jacob__ has quit [(Quit: Lost terminal)]
<srhb>
Alling: Not really, no.
<srhb>
Alling: From the NixOS perspective, yo home yo troubles.
<srhb>
(At least for now)
<Alling>
srhb: Oh. Not even possible to have NixOS create a file with some specified content from configuration.nix?
<srhb>
Alling: Absolutely, writeFile will do that for you.
<srhb>
Alling: It will write it to the store of course, world readable, and you'll have to link it on your own.
<srhb>
Alling: There might even be helpers that can approximate the ssh config format for you, converting attribute sets.
<Alling>
srhb: Oh. Possible to make NixOS run some specified bash script on each build, doing the linking? :P
<srhb>
Alling: You could prooobably set up a systemd oneshot thing doing that.
<srhb>
I'm not sure if there's an easier way.
<srhb>
Alling: (By the way, you're not reimplementing home-manager) :-)
<srhb>
now*
<Alling>
I'm not too familiar with this tbh. Trying to configure my server for automated backups to an SSHFS and absolutely love NixOS' declarativeness.
<srhb>
Yeah :)
<Alling>
Been looking for a way to run a script on each build. system.activationScripts seemed relevant, but I was not allowed to use it since it already existed. :s
<srhb>
There's a way around that, I'm sure.
<srhb>
But it probably involved something scary like mkForce or whatnot. I don't know I'm afraid :)
[0x4A6F]1 has joined #nixos
[0x4A6F] has quit [(Ping timeout: 248 seconds)]
[0x4A6F]1 is now known as [0x4A6F]
cmacrae has quit [(Ping timeout: 264 seconds)]
<pxc>
x
Ivanych has quit [(Ping timeout: 260 seconds)]
<sphalerite>
Alling: you can just add to system.activationScripts. However, it's probably not what you want since they get called fairly early on at boot time, and as such may break your system
<Alling>
srhb: OK, hm ... is it possible to declare directories that should exist in Nix? Like /mnt/backup
<sphalerite>
Alling: typically you'll want a one-shot systemd service instead. Something like systemd.services.ensure-mnt-backup = {script = "mkdir -p /mnt/backup"; serviceConfig.Type = "oneshot";};
<Alling>
sphalerite: Ooh!
<sphalerite>
also, using mkForce for activationScripts is definitely a bad idea and very very likely to break your boot!
<srhb>
sphalerite: Thanks for clarifying :P
<sphalerite>
there are some cases where activationScripts are the right thing, but they're usually not
ebzzry has joined #nixos
<Alling>
sphalerite: Not gonna use activationScripts then, thanks!
<Alling>
Will a oneshot be executed every boot?
<sphalerite>
yes
<sphalerite>
well, if it's wanted by the target that you're booting to
<sphalerite>
so actually I think you need to add wantedBy = ["multi-user.target"]; to the definition
<Alling>
sphalerite: Hm, is it enough to put that nix expression in my config?
<sphalerite>
yes, should be
<sphalerite>
it'll also get run when you switch configurations from A to B if A has a different version of the unit than B
<Alling>
Sounds reasonable.
<sphalerite>
There's a bunch of options. See man systemd.{unit,service,exec} for the detials of what does what and https://nixos.org/nixos/options.html#systemd.services for nixos-specifics
<sphalerite>
you can also e.g. set a unit to not get restarted when switching configurations. This is used for the display manager to prevent user sessions from being killed on a configuration switch
taktoa has joined #nixos
<Alling>
sphalerite: It should be OK to just have a stupid "mkdir", right? Would anything bad happen if /mnt/backup is being used as an SSHFS mountpoint and I do mkdir /mnt/backup?
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<sphalerite>
if you use mkdir without -p it will fail because the directory already exists. -p gets it to ignore if it already exists
<sphalerite>
in addition to having it create any necessary parent directories
<sphalerite>
e.g. if /mnt doesn't exist it'll create that too
<sphalerite>
see man mkdir :)
<Alling>
Ye, did. :)
<sphalerite>
but yeah it shouldn't have any strange interaction with mount
<Alling>
Auto parent dirs is desired, but of course I don't want my backup to be overwritten or anything should it happen to be mounted there.
<Alling>
Seems that it doesn't overwrite anything so should be fine.
[0x4A6F]1 has joined #nixos
Sonarpulse has quit [(Remote host closed the connection)]
<kuznero>
Hi All! Should I defined `nix.useSandbox = true;` in my configuration.nix file or I can do it somehow for a specific `nix-env -f . -i ...`?
<sphalerite>
kuznero: iirc it doesn't work specified on the command line
<sphalerite>
So yes, setting the option makes sense
<kuznero>
sphalerite: so, I should just modify my configuration.nix file and rebuild and after that `nix-env -i` will be sandboxed?
oida has quit [(Ping timeout: 255 seconds)]
<sphalerite>
Providing you rebuild and activate the new config, yes
<sphalerite>
well, more specifically any builds caused by nix-env -i will be sandboxed
<kuznero>
ok, thanks! I am trying to repro problem that shows up on travis ci for coreclr now... hope it will help
<wilornel>
how to get nix-env to show the location of a package?
<sphalerite>
Yeah, you'd almost always want the sandbox on really. I don't know why it's still not default, since the performance of the relevant kernel features has increased significantly
m0rphism has joined #nixos
<sphalerite>
wilornel: What do you mean? The location of the expression, the derivation, or the output path?
<wilornel>
the output path
zarel has quit [(Quit: Leaving)]
<kuznero>
sphalerite: thanks for the tip - will turn it on by default for me then.
simukis has quit [(Ping timeout: 258 seconds)]
oida has joined #nixos
<sphalerite>
wilornel: nix-build '<nixpkgs>' -A tmux --no-out-link
<sphalerite>
if you don't specify --no-out-link it will create a result symlink in the cwd and add it as a gc root, which is useful if you need to keep the path around
<sphalerite>
replace tmux with whatever package you're looking for of course
civodul has quit [(Remote host closed the connection)]
civodul has joined #nixos
<wilornel>
thanks sphalerite
ixxie has quit [(Quit: Lost terminal)]
<wilornel>
lol, how to deal with scripts doing #!/bin/bash ?
<kuznero>
patchShebang
griff_ has quit [(Quit: griff_)]
<wilornel>
hmmm but that would be in the expressions/derivations, right?
JosW has quit [(Quit: Konversation terminated!)]
<wilornel>
alright, I fixed it manually for that one script
<kuznero>
wilornel: yep, that should be part of configure step or preBuild where you call to `patchShebang .` or on a specific script. How to call it as an expression with nix-env or nix-shell - have no idea. sphalerite can you help here?
[0x4A6F] has quit [(Ping timeout: 258 seconds)]
<sphalerite>
nix-shell -p stdenv
<sphalerite>
Then you can use patchShebangs interactively or just add --run 'patchShebangs <file>'
<sphalerite>
However, this won't add GC roots and the resulting script will break eventually
civodul has quit [(Quit: ERC (IRC client for Emacs 25.3.1))]
<sphalerite>
As that version of bash gets garbage collected
<sphalerite>
So using a portable shebang like #!/usr/bin/env bash or packaging your script is better
pxc has quit [(Ping timeout: 255 seconds)]
pxc has joined #nixos
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
Alling has quit [(Quit: Page closed)]
pxc has quit [(Ping timeout: 248 seconds)]
<kuznero>
sphalerite: is there any way when working on my package I can `nix-shell -f . -p coreclr`? It complains on `-f .`...
<dash>
kuznero: what do you want it to do?
<kuznero>
dash: I need to troubleshoot package logic...
<kuznero>
call build phase steps and see how it behaves
<dash>
ok?
gandreani has quit [(Ping timeout: 258 seconds)]
<dash>
so what do you expect '-p coreclr' to do there?
<sphalerite>
Just nix-shell -A coreclr should do it
<sphalerite>
If coreclr is the package you're trying to debug
<kuznero>
sphalerite: yes, I am still getting confused with -i vs -A...
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] globin pushed 1 new commit to master: https://git.io/vdant
<kuznero>
sphalerite: but I need coreclr package definition from current git branch I am working on...
<kuznero>
error: unknown flag ‘-f’
<dash>
you pretty much always want -A
<dash>
-p is for when you aren't working on a nix expr
<sphalerite>
I use -p plenty
<kuznero>
dash: thanks, I am trying `nix-shell -f . -A coreclr` to point it to my new definition, but nix-shell has no clue on what -f is
<sphalerite>
Yeah -f is only for nix-env
<dash>
ah right, because -f is a param for nix-env/nix-build
<sphalerite>
Most of the other tools just accept the path to the file as a parameter
<sphalerite>
You don't need -f with nix-shell
<sphalerite>
It'll try ./shell.nix then ./default.nix
<kuznero>
so, should it be possible then to say `nix-shell -A pkgs/development/compilers/coreclr/default.nix`?
<dash>
no, -A takes an attribute name
<sphalerite>
Just nix-shell -A coreclr
<sphalerite>
If you're in nixpkgs
<kuznero>
so, nix-shell works from current folder?
<sphalerite>
If you're not in nixpkgs but want to get the build env for coreclr from it, nix-shell ~/nixpkgs -A coreclr
<sphalerite>
It takes a path as a positional parameter
<sphalerite>
If none is given, it will try shell.nix then default.nix
joepie91 has quit [(Read error: Connection reset by peer)]
joepie91 has joined #nixos
joepie91 has quit [(Changing host)]
joepie91 has joined #nixos
<kuznero>
ok, thanks
joepie91 has quit [(Read error: Connection reset by peer)]
joepie91 has joined #nixos
joepie91 has quit [(Changing host)]
joepie91 has joined #nixos
<kuznero>
that was first step :) then while I am in nix-shell -A coreclr I need to invoke package phases as defined in the derivation for my coreclr version. Is it possible?
<kuznero>
Ah, please diregard...
<kuznero>
Interestingly `$(uname -r)` gives `unknown` when running nix-env -i... Is this by design?
anderslundstedt has quit [(Remote host closed the connection)]
jb55 has quit [(Ping timeout: 260 seconds)]
<kuznero>
sorry, please disregard...
<wilornel>
Maybe #nixos can help me with this! I declared pkgs.eclipses.eclipse-platform to install eclipse-platform-4.7. It seems like that install it without PlatformUI
<wilornel>
maybe I should just install pkgs.applications.editors.eclipse?
Aisenp has joined #nixos
hiratara has quit [(Ping timeout: 264 seconds)]
<Aisenp>
Hello just a question, anybody know how to disable slim in configuration file ?
<NixOS_GitHub>
[nixpkgs] Ekleog opened pull request #30261: fcron module: fix use with hardlink-optimized store (master...fcron-hardlink) https://git.io/vdaCW
NixOS_GitHub has left #nixos []
ris has joined #nixos
<Aisenp>
services.xserver.displayManager.slim.enable = false; d'ont work "The option `services.xserver.displayManager.job.execCmd` is used but not defined"
<Eisfreak7>
Hi. I just started using autorandr, and I discovered that it continuously writes "/bin/sh: xrandr: command not found" to the journal
<Eisfreak7>
I guess the problem is, that the python file just executes "xrandr" but it needs an absolute path, right?
<Eisfreak7>
Can someone guide me through fixing this?
<gchristensen>
Aisenp: you need to enable some display manager
<Aisenp>
so how i can do this ?
<Aisenp>
can i do it with a displayManager = "none" ??
<ToxicFrog>
Eisfreak7: if xrandr is in /run/current-system/sw/bin, telling it to run `/usr/bin/env xrandr` instead might be sufficient
<gchristensen>
you don't want one at all?
<Aisenp>
yes
<gchristensen>
do you want X to start?
<ToxicFrog>
If it's not and you don't want it to be, you'll probably need some kind of build step for autorandr that splices in the absolute path
<Aisenp>
it's just want to startx for spectrwm
<Eisfreak7>
ToxicFrog: How would I go about fixing the nix expression to patch the path?
<wilornel>
How can I install eclipse-platform with a ui?
<rycee>
wilornel: eclipse-platform includes the UI.
<kuznero>
What should I do if package install script wants to inspect /etc/os-release file on the system?
<ToxicFrog>
You probably want a call to substituteInPlace that replaces "xrandr" with "${xorg.xrandr}/bin/xrandr", except you'll want to be a bit more circumspect than that so you don't e.g. replace it in comments and help text and stuff as well
<wilornel>
rycee: doing "find /nix/store/prcp7k3r9ws97llr0srqp86f4kca056z-eclipse-platform-4.7/ | grep org | grep platform | grep -E "\bui\b"", I can't find platformui. Something must be not installing it
<ToxicFrog>
Eisfreak7: that said, you could probably just add `xorg.xrandr` to config.systemPackages and it would start working
<wilornel>
I also ran `home-manager switch` from a SSH session into the VM, not from a Display of the VM
<gchristensen>
Aisenp: what if you left slim enabled, but set services.xserver.displayManager.slim.autoLogin.user ?
<Aisenp>
yes it could be a solution but it's take more ram than disabling slim ?
cement has quit [(Ping timeout: 246 seconds)]
fendor has quit [(Ping timeout: 240 seconds)]
<gchristensen>
do you have very tiny amounts of ram?
<Aisenp>
nop it's just for the fun ^^
<gchristensen>
ah
<gchristensen>
I don't know if you can easily turn off the display manager part :/
<Eisfreak7>
ToxicFrog: Okay I'll check for other usages of substiuteInPlace, thanks for the pointer
<rycee>
wilornel: Not sure how to get it going under eclim but I've been using eclipse-platform quite extensively and it does show an UI :-)
<Eisfreak7>
ToxicFrog: Is there documentation for stuff like that?
<Eisfreak7>
ToxicFrog: I do have it in my systemPackages
<Aisenp>
no problem, i'll take this solution for the moment and if found another way i'll notice you
<ToxicFrog>
Eisfreak7: yeah, in the gigantic Nix manual
<ToxicFrog>
Or maybe the nixpkgs manual
<MichaelRaskin>
Doesn't SLiM _exit_ after starting autologin anyway?
<gchristensen>
iirc yes, MichaelRaskin
<ToxicFrog>
Or the NixOS manual, but probably nixpkgs
<wilornel>
rycee: yes, I can also launch eclipse with a ui
<MichaelRaskin>
Then RAM usage of slim is irrelevant anyway.
<ToxicFrog>
Eisfreak7: to be honest, my approach would just to be to add xrandr to SystemPackages and call it a day
sigmundv has quit [(Ping timeout: 255 seconds)]
ebzzry has quit [(Ping timeout: 240 seconds)]
justbeingglad has joined #nixos
justbeingglad has left #nixos []
<Eisfreak7>
ToxicFrog: It is systempackages. Also, autorandr is kinda working (I do have an issue but I don't know if it is related). But fixing it seems cleaner
* ToxicFrog
nods
<wilornel>
rycee: I'm asking #eclipse if there's a flag I need to pass to the installer to tell it I have a UI and for it to install platformui
<rycee>
wilornel: There is a PlatformUI in /nix/store/prcp7k3r9ws97llr0srqp86f4kca056z-eclipse-platform-4.7/eclipse/plugins/org.eclipse.ui.workbench_3.110.0.v20170612-1255.jar
<rycee>
wilornel: Maybe that's what it is looking for?
<wilornel>
how did you list its contents?
<rycee>
`unzip -l`
<rycee>
For example. `unzip -l /nix/store/prcp7k3r9ws97llr0srqp86f4kca056z-eclipse-platform-4.7/eclipse/plugins/org.eclipse.ui.workbench_3.110.0.v20170612-1255.jar`
<wilornel>
yeah, I found it in there (with `jar tf file`)
<kuznero>
So, /etc/os-release is not accessible in sandboxed builds. Should I try generate it on the fly while building?
<kuznero>
What is the best practice?
<kuznero>
There is something in `nixos/modules/misc/version.nix` about generating /etc/os-release. But I am not sure how is it possible to use it
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 2 new commits to master: https://git.io/vdalk
<NixOS_GitHub>
nixpkgs/master 1afd97a Léo Gaspard: fcron module: fix use with hardlink-optimized store
<Aisenp>
<gchristensen> I have found a trick to bypass the DisplayManager you just need to add displayManager.job.execCmd = ''''; after displayManager.slim.enable = false; and it is okay !!
<kuznero>
Mic92: So, it will be: 1) I generate temp file with the right content, 2) export NIX_REDIRECTS=/etc/os-release=/path/to/temp/etc/os-release, right?
<Eisfreak7>
Nevermind, I figured it out (pass the line services.autorandr.enable to the config when importing my nixpkgs)
jtojnar has joined #nixos
<Mic92>
kuznero: yep
<Mic92>
Eisfreak7: yeah, should just work. I use it daily
<Mic92>
it hooks into udev, so no service is running actually
dfranke has quit [(Ping timeout: 264 seconds)]
Aisenp has quit [(Quit: Leaving)]
mortum5 has quit [(Ping timeout: 255 seconds)]
<Eisfreak7>
Mic92: Yeah it works for me, it just doesn't execute the postswitch script and complains about not finding xrandr in the journal
<kuznero>
Mic92: do you always need to LD_PRELOAD when using NIX_REDIRECTS?
<Mic92>
kuznero: yes
<Mic92>
Eisfreak7: also postswitch works for me, I do a dbus-send to my windowmanager there
<Eisfreak7>
Mic92: Thats weired. It works if I manually trigger another profile, but it doesn't if I just plug something in (even though it recognises the new profile)
<Mic92>
kuznero: yes
dywedir has quit [(Ping timeout: 255 seconds)]
<Mic92>
Eisfreak7: how did you save the current profile with the other stuff not plugged in?
hiratara has joined #nixos
<Mic92>
s/how//
<Eisfreak7>
Mic92: Yes
<Eisfreak7>
Mic92: I have two profiles, "mobile" and "tv". My postswitch script just consists of "echo $(date) /tmp/log"
<Mic92>
Eisfreak7: is it executable?
<Mic92>
what is the shebang?
oida has quit [(Ping timeout: 240 seconds)]
<Eisfreak7>
Yes, #!/bin/sh
<kuznero>
new wiki looks pretty cool
gandreani has joined #nixos
<Mic92>
I put it just in ~/.config/autorandr/postswitch
oida has joined #nixos
<Eisfreak7>
Mic92: Yeah me too. As I said, if I do `autorandr -f -l tv` it is executed
csingley has quit [(Ping timeout: 240 seconds)]
<kuznero>
Is it ok to reference pkgs from stdenv.mkDerivation?
gandreani has quit [(Ping timeout: 248 seconds)]
<Mic92>
Eisfreak7: I have to look, if I forgot to backport something to 17.09 regarding autorandr
fragamus has joined #nixos
sigmundv has joined #nixos
<Mic92>
kuznero: what are you trying to do?
avn has quit [(Ping timeout: 240 seconds)]
vandenoever has joined #nixos
vandenoever has quit [(Changing host)]
vandenoever has joined #nixos
<makefu>
is there a way to find out which of the parts of my configuration or packages requires it to build the ghc?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 1 new commit to master: https://git.io/vdaBJ
<NixOS_GitHub>
nixpkgs/master 3541b9a Joerg Thalheim: autorandr: put xrandr into PATH
NixOS_GitHub has left #nixos []
<Mic92>
Eisfreak7: ^
serhart has quit [(Remote host closed the connection)]
serhart has joined #nixos
serhart has quit [(Remote host closed the connection)]
<Mic92>
makefu: nixUnstable has why-depends
<kuznero>
Mic92: I am trying to reuse nixos/modules/misc/version.nix
<Mic92>
kuznero: for what?
<kuznero>
To generate content for /etc/os-release and the use NIX_REDIRECTS before calling to ./build.sh for coreclr
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 1 new commit to release-17.09: https://git.io/vdaBn
<NixOS_GitHub>
nixpkgs/release-17.09 9a70d3c Joerg Thalheim: autorandr: put xrandr into PATH...
NixOS_GitHub has left #nixos []
oida has quit [(Ping timeout: 240 seconds)]
gandreani has joined #nixos
<Mic92>
kuznero: in nixpkgs we cannot import nixos
<Eisfreak7>
Mic92: Ah, thats what I was currently trying to solve with substituteInPlace. Is a wrapper generally considered "nicer"? Shouldn't all the information be available in the package itself?
<Eisfreak7>
Mic92: Its certainly more robust though
Oida has joined #nixos
nix-gsc-io`bot has quit [(Client Quit)]
<kuznero>
Mic92: then it is code duplication, which might be ok for this purpose I guess.
mc1 has joined #nixos
<Mic92>
kuznero: yes, should be. what is the file for anyway?
<Mic92>
*used for
<kuznero>
Mic92: /etc/os-release is used to decide the branch logic in the build script.
avn has joined #nixos
<Mic92>
kuznero: but I suppose nixos is not detected there anyway
<kuznero>
It is detected as Linux $(uname -r) - this is enough. Then it requires some more details from /etc/os-release.
<Mic92>
kuznero: just put something there, which makes it build
<Mic92>
did it not build before btw?
<kuznero>
It did, but produced some warning. And Travis CI fails.
<kuznero>
Travis might fail due to different reason, so I am trying to make it more predictable so to say
<Mic92>
what does travis fails with?
<Mic92>
I stopped carrying about travis at some point
<kuznero>
trying to invoke wget and fails to download something. btw, it does not do it when I run locally with sandboxing
<Mic92>
oh, did maybe I build coreclr on the machine, before I had sandbox builds enabled
<Mic92>
I take a look on a different one
gandreani_OG has joined #nixos
<obadz>
there's a whole range of commits on the master branch after 17.03 forked that I cannot run VM-based tests on. Really annoying as that means I can't git bisect a bug :-(
<obadz>
starting VDE switch for network 1
<obadz>
vde_switch: Could not set the VDE ctl directory '/tmp/nix-build-vm-test-run-ecryptfs.drv-0/vde1.ctl' permissions: Operation not permitted
<Eisfreak7>
Mic92: Do you know how I can convince nix to use the new version of autorandr now? I have a local clone of nixpkgs imported, but when I activate the service it just takes the one from the stable channel.
<Eisfreak7>
gandreani: Yes but I want the newest commit for autorandr.
<Mic92>
kuznero: mhm, it starts to build on my machine without using wget
<Mic92>
Eisfreak7: git cherry-pick
<kuznero>
Mic92: exactly same on my machine
<Eisfreak7>
Mic92: I think packageOverrides comes closest to what I want, thanks. But is there now way to just set the `services.autorandr.enable` attribute on my imported version of nixpkgs?
<Eisfreak7>
Mic92: I currently have a line like this in my config: `let repo = import <unstable> { config = { allowUnfree = true; useSandbox = true;};}; in`
<obadz>
Eisfreak7: cherry-pick might be more appropriate imho as it will share dependencies with your main system
<obadz>
Eisfreak7: but packageOverrides might work too
<kuznero>
Mic92: but if it is "ok" to have Travis CI fail while it is possible to build in sandbox locally, then it is probably ok. WDYT?
<Mic92>
kuznero: yes
<Mic92>
kuznero: if it builds for me, I usually don't care about travis
<kuznero>
Mic92: good, then I will start looking into corefx
<gandreani>
Eisfreak7: I got another one besides git cherry-pick, git checkout master -- nixpkgs/nixos/modules/services/misc/autorandr.nix
<obadz>
gandreani: cherry-pick is better is it will capture the full, presumable coherent commit
<obadz>
in case other files were touched
Khetzal has joined #nixos
<gandreani>
Hmm that's true
<Eisfreak7>
Mic92: Thanks, and how can I reference the `services.autorandr.enable` option on that local checkout then (without changing my whole system to the local checkout)?
<Mic92>
Eisfreak7: either you use packagesOverrides or an overlay
mudri has joined #nixos
<obadz>
Eisfreak7: why not just locally checkout the exact commit you've installed locally? then nothing will get rebuilt or redownloaded other than what you're cherry-picking
<Eisfreak7>
Mic92: obadz: Because I would prefer staying on the channel (and having auto-update work). Intuitively I would just do `repo.services.autorandr.enable = true`, but that doesn't work. What am I misunderstanding?
<Mic92>
Eisfreak7: I gave you the solution, I would say
<obadz>
injecting a module via overrides takes some fiddling
<Mic92>
overlays are respected by nixos
<Eisfreak7>
Mic92: Yes, you gave me a solution, thakns for that. I'm just trying to further my understanding to why my intuitive solution doesn't work.
mc1 has quit [(Ping timeout: 255 seconds)]
<Eisfreak7>
I thought `services.autorandr` just desugars to `pkgs.autorandr`, doesn't it?
kuznero has quit [(Remote host closed the connection)]
<Mic92>
Eisfreak7: what have you tried?
joepie91 has quit [(Excess Flood)]
<Mic92>
repo = import <unstable> will not affect pkgs
<Mic92>
and use overrideAttrs instead of overrideDerivation
roconnor has quit [(Ping timeout: 255 seconds)]
digitus has quit [(Quit: digitus)]
roconnor_ has joined #nixos
roconnor has joined #nixos
goibhniu has quit [(Ping timeout: 255 seconds)]
schoppenhauer has quit [(Ping timeout: 240 seconds)]
roconnor_ has quit [(Ping timeout: 255 seconds)]
Neo-- has joined #nixos
iqubic has joined #nixos
vandenoever has quit [(Read error: Connection reset by peer)]
schoppenhauer has joined #nixos
michiel_l has joined #nixos
lsyoyom has joined #nixos
<Eisfreak7>
Mic92: Which attrs are I supposed to override?
<Ralith>
Is there a way to tell llvmPackages.stdenv to use the libstdc++ from gcc7 without rebuilding llvm+clang?
<Ralith>
that is, I want it in the stdenv, I don't care what it was built with itself
markus1189 has joined #nixos
<Ralith>
i.e. I want a stdenv that has clang 4 and libstdc++ from gcc 7
markus1199 has quit [(Ping timeout: 255 seconds)]
<Mic92>
Eisfreak7: of autorandr
<Mic92>
Ralith: libcxxStdenv.mkDerivation maybe?
<Mic92>
so libcxxStdenv instead of stdenv
<Ralith>
Mic92: that gives me libc++, not libstdc++.
<Ralith>
I want libstdc++.
Neo-- has quit [(Remote host closed the connection)]
<Mic92>
Ralith: does clang in the normal stdenv not use libstdc++ anymore?
<Ralith>
clang in the normal stdenv uses libstdc++ from the normal stdenv, which is not the one I want.
<Ralith>
I want the libstdc++ from gcc 7.
anderslundstedt has joined #nixos
<Mic92>
Ralith: I think this was encoded in the compiler wrapper
<Ralith>
yes; I'm asking if there's a way to override that without overriding the libstdc++ the compiler was built with (and thereby having to rebuild the compiler)
betaboon has quit [(Ping timeout: 260 seconds)]
<Mic92>
Ralith: you can probably do cc-wrapper.override ({stdenv= ...;})
<Mic92>
Ralith: grep for ccWrapperFun in top-level/all-packages.nix