<elvishjerricco>
Is it possible to get the container IP as an argument to the `configuration.nix` of a nixos-container (imperative)? I'm trying to make a container that starts a tor client service, but it needs the container IP in `socksListenAddress`.
<elvishjerricco>
Obviously I could copy paste it in, but then the container would not be portable. i.e. I couldn't give that nix file to a friend and be sure he'll get a working container.
<tilpner>
Maybe containers.<name>.extraVeths.<name>.localAddress can help?
endertux has quit [(Remote host closed the connection)]
bennofs has quit [(Ping timeout: 258 seconds)]
<tilpner>
Oh wait, there's actually containers.<name>.hostAddress and containers.<name>.localAddress
endertux has joined #nixos
mkoenig_ has joined #nixos
mkoenig has quit [(Ping timeout: 246 seconds)]
sigmundv_ has joined #nixos
FreakingOut1987 has quit [(Remote host closed the connection)]
<elvishjerricco>
tilpner: Well those are used in a system's `configuration.nix` to configure declarative containers. I don't believe you can use those in an imperative container's `configuration.nix` to find the container's IP.
bennofs has joined #nixos
<tilpner>
Oh, how are you creating containers completely imperatively?
cpennington has quit [(Remote host closed the connection)]
Piece_Maker has joined #nixos
tmaekawa has quit [(Quit: tmaekawa)]
Acou_Bass has quit [(Ping timeout: 248 seconds)]
Piece_Maker is now known as Acou_Bass
<sirkha>
if nixos-rebuild dry-build --show-trace shows just "building the system configuration" and nothing after that, does that mean there are no changes?
<srhb>
ben: Sorry, I don't think I know what you mean. Yes, nix-env -u will upgrade packages from all channels (provided the channels have been upgraded)
<srhb>
oh yay channel update
<srhb>
wait, wrong channel...
<ben>
srhb: I was curious if channels might be more magical than I assumed but they're not, it's all good
Mateon1 has quit [(Ping timeout: 255 seconds)]
Mateon3 is now known as Mateon1
<srhb>
:)
<ben>
oh anyway i think i'm missing a bunch of gstreamer plugins, do we only package a subset or something?
<srhb>
No idea. Which ones are you missing?
<ben>
rtmp, the rtmpsink element in particular
<ben>
oh i guess we might not package librtmp
orivej has joined #nixos
<srhb>
Sorry, I don't know anything about gstreamer. :)
ertes has quit [(Read error: Connection reset by peer)]
<hyper_ch>
but I assume you wanted to say 512gb files?
<clever>
hyper_ch: first, run "truncate image-1.img -s 512m"
<hyper_ch>
but that gives you like a gazillion files on a 8tb drve
hamishmack has joined #nixos
<hyper_ch>
clever: also, I saw that people use just whole block devices like /dev/sdx and not partitions.... is that recommended?
<sphalerite>
Yes
<clever>
zfs will just automatically create gpt partitions, i believe
<hyper_ch>
:)
kungfukenny has joined #nixos
Wizek has joined #nixos
nathan has quit [(Ping timeout: 240 seconds)]
mrkgnao has quit [(Ping timeout: 240 seconds)]
<sphalerite>
colabeer: I think xautolock might be what you want
<sphalerite>
Wait no, it doesn't seem to have suspend actions
<hyper_ch>
when can we snapshot brains... that would make nixos-life so much simpler ;)
callmecabman has joined #nixos
<sphalerite>
lluchs: oh hi, didn't notice you'd joined here :D
<sphalerite>
eqyiel: did you find out more about the nix-repl issue you were having?
detran` has joined #nixos
<sphalerite>
eqyiel: I think it's just that nix-repl's a = b bindings aren't recursive, so if you have a previously defined a in the expression b it'll use that rather than the a you're defining
<sphalerite>
so I think the answer is "use more let"
<sphalerite>
Also I'm impressed that you managed to type all that stuff at the interactive prompt x) I would have just stuck it in a file and done :l relation.nix or something
detran has quit [(Ping timeout: 248 seconds)]
<chris|>
does anyone have a working override derivation for rust 1.20 to share?
hamishmack has quit [(Quit: hamishmack)]
<MichaelRaskin>
hyper_ch: until top managers make it mandatory for middle managers to snapshot their mind before talking to tech people/project managers and to restore after sending a meeting report — to prevent embedding of dangerous ideas like actually budgeting for a backup.
<hyper_ch>
:)
<hyper_ch>
well, it would be could if I could just snapshot clever's brain for all his zfs knowledge ;)
<jonge>
hi there. i am a newbie in nixos. however i would like to explore what kind of configuration parameters every package i want has. is there any nicely browsable documentation generated from the nix files or is the only way finding and reading the nix files of every package?
<Leo`>
I understood “yourThing” as an existing service.
<Leo`>
Ok.
<srhb>
jonge: Those are for nixos options
<srhb>
Leo`: Oh, sorry!
<srhb>
Leo`: That will autogenerate a yourThing service with the options you specify :)
<jonge>
srhb: what is the difference? are there no package-specific things but only general system things or what does it mean?
<srhb>
Leo`: You can see examples of doing this with most database packages in NixOS, they will set up a oneshot service to do initial configuration.
<Leo`>
That should do the trick.
<Leo`>
Let me try.
<srhb>
jonge: NixOS modules are special cases. They use various nixpkgs to create an OS. But a package is nixpkgs doesn't necessarily correspond to a NixOS module (far from it)
<srhb>
jonge: For instance, weechat is not a NixOS service, but an irc client. It has various build options, and your best bet to browse those is looking at the nixpkgs sources for the weechat derivation.
<srhb>
** But a package *IN* nixpkgs [...] *
<srhb>
Typing is hard.
leat has joined #nixos
<sirkha>
trying to get an openvpn client going, turned off autoStart, but there is no service listed in systemctl -a
<jonge>
srhb: hm i see. i am a bit lost in understanding where to look for what. but then i will look up those flag/option descriptions myself in the folders. thx
<srhb>
jonge: Perhaps if you have an example, it will clarify things.
<srhb>
Basically, the options to configure your NixOS are on that page.
<srhb>
Various package build options are not.
<srhb>
When will haskellPackages advance to GHC 8.2 ?
<jonge>
srhb yeah well my use case is: system install. i sit here with a booted live cd and try to figure out how to configure all the user, ssh, desktop etc. stuff. i have seen that there is valuable documentation in all the packages, but then i wondered if there is a central page that collects all these options and their descriptions in one browsable HTML page or so.
katyucha has joined #nixos
<srhb>
jonge: Yes, that's generally NixOS options as you found :)
<jonge>
srhb: by "nixos options" you mean /etc/nixos/configuration.nix?
<srhb>
jonge: No, I mean the page you linked to.
<srhb>
jonge: Or man configuration.nix -- or the options addendum in the NixOS manual
<sphalerite>
Any zfs wizards know of a way to see what is in a snapshot that isn't in the filesystem it came from anymore, i.e. the files constituting the "used" figure for it?
<hyper_ch>
sphalerite: clever is asleep and infinisil isn't here either
<sphalerite>
I know, but there are other zfs wizards :)
<hyper_ch>
and you're the 3rd one in my round of zfs-gurus in #nixos
<sphalerite>
and -d will cause nix-collect-garbage to do two things — delete old generations for any profiles it has write access to, and then gc as normal
<Leo`>
I'm not short on disk space, but I'd really like to get rid of those plain-text passwords hrmf.
hamishmack has quit [(Quit: hamishmack)]
<sphalerite>
hyper_ch: i3 and dwm are mutually exclusive
<sphalerite>
MichaelRaskin: haha not quite, but…
<hyper_ch>
sphalerite: they are? ok..... how to get i3 running?
<sphalerite>
hyper_ch: you can do it per-user by installing i3 into your profile one way or another and then write an xsessionrc that runs that instead of whatever desktop environment you currently run
<hyper_ch>
Leo`: why not remove all old generations?
phg has joined #nixos
<hyper_ch>
sphalerite: and how to run it globally?
<sphalerite>
Leo`: nix-store -q --roots
hamishmack has joined #nixos
<sphalerite>
or you can set services.xserver.windowManager.i3.enable to make it available as an option for all users, to select at login time
<Leo`>
hyper_ch: Well I'm just afraid to break something.
<sphalerite>
Leo`: also I'd recommend just leaving it there and changing the password :p
<sphalerite>
Leo`: seems sensible to me to consider any password that's been exposed in an environment that you don't want it to be exposed in as broken and needing to be changed
<Leo`>
sphalerite: The server is brand new, no one but me and another user who I trust accessed it yet.
<sphalerite>
But you're clearly not comfortable leaving it there.
<Leo`>
Yes because I will have other users eventually. :p
<sphalerite>
So changing it would probably be the best solution :)
<Leo`>
It's more like a future-proofing thing.
nh2 has joined #nixos
phreedom has joined #nixos
<hyper_ch>
Leo`: are you booted into the last generation?
<hyper_ch>
if so, does it work alright? if so, then there's no need to keep old generations
<Leo`>
I don't think so. But changes since the last boot are minimal.
<Leo`>
But yeah I should maybe try rebooting first...
<hyper_ch>
do that
<hyper_ch>
if it reboots, and services are up fine etc...
<hyper_ch>
then nuke the old generations and get rid of the password inside the config
Wizek has quit [(Quit: Connection closed for inactivity)]
<Leo`>
Sounds goot.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to master: https://git.io/vdZbx
<NixOS_GitHub>
nixpkgs/master a58120c Vladimír Čunát: ibus: hack-fix build after glib update...
NixOS_GitHub has left #nixos []
<Leo`>
good*
<Leo`>
Brb then.
Leo` has quit [(Quit: WeeChat 1.7.1)]
<hyper_ch>
sphalerite: care to share your i3 config?
<sphalerite>
If I could configure KDE using nix like I do with i3, and get all my shortcuts set up the way I like them (e.g. super+3 to get to the workspace that I have my browser on, and start the browser if it's not already runnign) I could just as well use KDE
thc202 has joined #nixos
<bennofs>
can kde do all the tiling that i3 can?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 1 new commit to release-17.09: https://git.io/vdZNk
<NixOS_GitHub>
nixpkgs/release-17.09 5ac8389 volth: nerdfonts: derivations is not fixed output...
NixOS_GitHub has left #nixos []
<sphalerite>
but I'm not sure if the former part is possible, and the latter part is probably possible but would involve a bunch of fiddling with settings in GUIs and be hard to reproduce, based on what Iknow about kde
<hyper_ch>
:)
<hyper_ch>
each one his own DE/WM
<sphalerite>
bennofs: I doubt it, but I don't really use that much of i3's tiling. The furthest I usually go is a vertical split between two windows, sometimes a vertical split where the right or left-hand side is a horizontal split between two windows
<sphalerite>
I think all that can be done pretty easily with kwin too
<bennofs>
i recently started using scratch windows
<bennofs>
for example, putting an editor window into the scratchpad allows you to quickly show/hide it with some key combo, no matter where you are
<eqyiel[m]>
sphalerite: thanks for clearing that up (nix-repl thing)!
<sphalerite>
np :)
mudri has quit [(Ping timeout: 240 seconds)]
mortum5 has joined #nixos
takle has quit [(Remote host closed the connection)]
<sphalerite>
hyper_ch: also plasma takes long to start
<sphalerite>
takes like 4 seconds on my pretty powerful laptop
takle has joined #nixos
<hyper_ch>
long? what are 4 seconds during the course of 15-20bn years ;)
<hyper_ch>
(I know)
<sphalerite>
i3 is up imperceptibly quickly :)
<hyper_ch>
are you one of those rare people who actually poweroff the notebook?
<hyper_ch>
I noticed with 2x external screens attacked, plasma does take quite a bit longer than just with notebook display
<sphalerite>
no
<sphalerite>
but I do reboot occasionally, and I log out of the account that I use plasma on frequently
<sphalerite>
(I use i3 on my main account and plasma on my games/proprietary software account)
<hyper_ch>
logout? why?
<hyper_ch>
ah ok
<hyper_ch>
can't you use games and proprietary stuff on i3?
<sphalerite>
I can, but I don't want to let that stuff into my main user account htat has access to all my precious data
<sphalerite>
I have no particular reason to prefer plasma for it
takle has quit [(Ping timeout: 248 seconds)]
<sphalerite>
I just used it mainly because I wanted to try it
<hyper_ch>
:)
<sphalerite>
I do like how shiny it is :p
<hyper_ch>
everybody loves shiny things
mudri has joined #nixos
<sphalerite>
why does steam need to build mono from source?
iyzsong has joined #nixos
<hyper_ch>
why does using vbox extension package require vbox built from source :)
<hyper_ch>
it's just as things are
<srhb>
This was not always the case for Steam though
<hyper_ch>
no idea :)
<sphalerite>
hyper_ch: because the extension pack is proprietary software and legally can't be distributed something something
<sphalerite>
hyper_ch: mono is free software and should be fine
<hyper_ch>
sphalerite: still doesn't force to recompile it all
<hyper_ch>
on other systems you just download the ext package
<hyper_ch>
and open it with vobx
<hyper_ch>
no compiling required
<srhb>
It's not the same issue at all.
<sphalerite>
I don't know how it works
<hyper_ch>
I really need to gcj patch in unstable :(
Leo` has joined #nixos
<Leo`>
Re.
<Leo`>
Hmmm.
<Leo`>
I really should fix that firewall issue...
<hyper_ch>
what issue?
<hyper_ch>
you know you can put firewall rules into the configuration.nix
<Leo`>
So. When I booted up the server into the current configuration, everything worked fine except... The firewall would reject all incoming connections, even on opened ports.
<Leo`>
No idea why.
<Leo`>
So I live-booted a rescue system, chrooted in it, disabled the firewall, and rebooted and voilà.
<hyper_ch>
configuration.nix or I won't believe it
<Leo`>
And now if I enable the firewall again it works just fine.
<Leo`>
Except that every now and then it starts dropping incoming connections again until I just restart firewall.service.
<Leo`>
It's pretty weird...
<hyper_ch>
configuration.nix or I won't believe it
<gchristensen>
sphalerite: a blinking caps lock means your kernel panicked
mog has quit [(Quit: im quiting!)]
<aanderse>
17.09 final out? if so... anyone having issues with steam this morning?
<sphalerite>
gchristensen: yeah, the journal says there was a null pointer deref
<sphalerite>
gchristensen: latest kernel seems to work though
<gchristensen>
nice
<sphalerite>
aanderse: I was running steam just 5min ago just fine, on 17.09
<hyper_ch>
aanderse: it's not officially out
<aanderse>
hyper_ch: ok, thanks
<aanderse>
sphalerite: i'm running 17.09 and just got this fun: mkdir: cannot create directory ‘/nix/store/2ba7iv5swbnnciiyr4gaah555r86kyj9-steam-fhs/steamrt/pinned_libs_64’: Read-only file system
<Leo`>
Ok, so I think I found the source of the firewall problem (cc hyper_ch).
<Leo`>
I have a container which runs an OpenVPN server, and because of that it must have the NET_ADMIN capability.
<hyper_ch>
between keyboard and chair?
<Leo`>
So on boot, it sets its own firewall rules (which are set to the default values), but since it has NET_ADMIN, it replaces the rules for the host as well.
<hyper_ch>
:)
<Leo`>
The solution is giving it a privateNetwork = true, I guess.
<Leo`>
(Or, giving it the same firewall config, idk...)
<Leo`>
(The latter has the advantage of not having to write even more routing rules...)
<iMatejC>
hyper_ch: what if he is using a standup desk
<iMatejC>
?
<Leo`>
she*
<iMatejC>
sorry
<Leo`>
NP. :)
<hyper_ch>
iMatejC: pebkaw [wall]
<Leo`>
Pebkaf? (Floor.)
eacameron has quit [(Ping timeout: 255 seconds)]
<hyper_ch>
I doubt leo is outside, so there's a wall somewhere
<Leo`>
Actually I'm sitting on a chair.
<hyper_ch>
so pebkac :)
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
Itkovian has joined #nixos
S0rin has quit [(Ping timeout: 240 seconds)]
erictapen has quit [(Ping timeout: 246 seconds)]
Leo` has quit [(Quit: WeeChat 1.7.1)]
S0rin has joined #nixos
Leo` has joined #nixos
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<Leo`>
Now getting back to: deleting old generations.
katyucha has quit [(Ping timeout: 240 seconds)]
Itkovian has joined #nixos
<hyper_ch>
either delete them or not
<Leo`>
Currently deleting.
oida has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] makefu opened pull request #29985: fix build for proot on aarch64 (master...pkgs/proot/aarch64-fix) https://git.io/vdnfJ
NixOS_GitHub has left #nixos []
<gchristensen>
hyper_ch: please don't be a pain
Arcaelyx_ has joined #nixos
phreedom has quit [(Ping timeout: 248 seconds)]
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
Arcaelyx has quit [(Ping timeout: 240 seconds)]
oida_ has quit [(Ping timeout: 248 seconds)]
Itkovian has joined #nixos
Itkovian has quit [(Client Quit)]
Itkovian has joined #nixos
fritz09 has joined #nixos
<fritz09>
hi
Fare has joined #nixos
<fritz09>
I’ve booted with the latest graphical image, followed the instalation guide and want to install nixos but I get “copy-from-other-stores.pl: warning the group ‘nixbld’ specified in ‘build-users-group’ does not exists”
<fritz09>
do I need to add the group to configuartion.nix? or does the group is missing in the image?
JosW has quit [(Quit: Konversation terminated!)]
Yotam_ has joined #nixos
<makefu>
fritz09: sounds like a bug in the installer to me
<fritz09>
so I download the previous release and update after installation
<makefu>
yes this may work
<fritz09>
@makefu: is there a way to add the missing group in the running system?
<Yotam_>
Can I add options like tab width to my emacs.nix? if so is there a list somewhere with all of the options?
<Leo`>
fritz09: I had this issue when installing from a non-Nix live image. I just dynamically added a group nixbld and a user nixbld0 in this group, and as far as I can tell it worked properly.
<Leo`>
Added on the live image, I mean.*
<Leo`>
Not on the NixOS system being installed.
<Leo`>
So yeah, basically, from the live image, run addgroup and adduser.
<fritz09>
@Leo`: thanks I will try that
<Leo`>
It should work.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peti pushed 2 new commits to master: https://git.io/vdnJt
<NixOS_GitHub>
nixpkgs/master d0fbe1b Peter Simons: hackage-packages.nix: automatic Haskell package set update...
<NixOS_GitHub>
nixpkgs/master b92f590 Peter Simons: haskell-hspec-core: enable test suite again
NixOS_GitHub has left #nixos []
<shlevy>
fpletz: ping
<fritz09>
“useradd: cannot deterime your username”
<fritz09>
same with groupadd
<gchristensen>
what's up shlevy?
* gchristensen
is about to head out
<shlevy>
gchristensen: Was going to ask fpletz how he's using kanboard ;)
<Leo`>
Typically, in that case, the path to the standard includes and libs and all.
<Leo`>
paths*
<Leo`>
nix-env just creates symlinks AFAICT.
<orivej>
ee1943: unusable how?
<ee1943>
orivej: clang doesn't have the search paths it should, so it fails to find simple headers like iostream, stdio.h, etc
<shlevy>
Hmm how are you installing clang?
<ee1943>
Leo`: what's the problem with enabling those paths for the regular shell too?
ixxie has quit [(Ping timeout: 258 seconds)]
<Leo`>
ee1943: I don't know.
<shlevy>
Yeah, looks like 'nix-env -i clang' installs the *unwrapped* clang
<ee1943>
shlevy: I've tried basically all of the packages... clang, llvmPackages.libcxxClang etc
<shlevy>
You want nix-env -iA nixpkgs.clangUnwrapped
<ee1943>
just dropping it in my configuration.nix
<shlevy>
erm, never mind
<shlevy>
Ah, OK
<shlevy>
I wrote the wrong one in that command too :D
<shlevy>
bizarre
<ee1943>
:)
<shlevy>
Lemme test
<ee1943>
sure :)
<ee1943>
I've noticed a couple issues related to this in the issue tracker but they seem to be resolved. Not sure what's going on over here
<ee1943>
on the nixos-unstable channel FWIW
<shlevy>
Hmm it just worked for me
<shlevy>
did nix-build -A clang
<shlevy>
And then used the resulting clang to compile hello world
<shlevy>
what does: readlink -f $(which clang) give you?
<shlevy>
ee1943: ^
Arcaelyx_ has quit [(Read error: Connection reset by peer)]
Arcaelyx has joined #nixos
<globin>
shlevy: hmm we should probably highPrio the wrapped clang if this is the case
<shlevy>
globin: Yes, we should :)
<orivej>
ee1943: currently clang wrapper does not include all the necessary search paths. I believe this was done because clang needs different paths in different contexts (e.g. building with libstd++ vs libc++). However, standalone clang should be supported.
<orivej>
You can work around this by exporting NIX_x86_64_unknown_linux_gnu_CFLAGS_COMPILE="-I/run/current-system/sw/include -I/run/current-system/sw/include/c++/6.4.0 -I/run/current-system/sw/include/c++/6.4.0/x86_64-unknown-linux-gnu"
<ee1943>
using this I can check what it's search paths are: echo | clang++ -x c++ -v -E - 2>&1 | sed -n '/^#include </,/^End/s|^[^/]*\([^ ]*/include[^ ]*\).*$|-I\1|p'
<sphalerite>
Celebrated the usb-c hub working too soon...
<sphalerite>
Don't know what happened but anything IP-related started hanging
<shlevy>
globin: Honestly though I'd rather we jsut stopped pretending nix-env -i by-name is a sane way to use the tool :P
<sphalerite>
Including stuff like ip a :(
<dtzWill>
hmm, can confirm re:not working when installed vs nix-shell re:clang c++ headers
<sphalerite>
shlevy: ++
<ee1943>
orivej: I see, that's interesting
<ee1943>
shlevy: well the problem is that g++ shows this: -I/nix/store/g65l985579ri87i8m3906a2nmyrl24a2-gcc-6.4.0/lib/gcc/x86_64-unknown-linux-gnu/6.4.0/../../../../include/c++/6.4.0
<Leo`>
Just saw that. But it's not possible to search by option name, is it?
<infinisil>
i use man configuration.nix for that
<Leo`>
Oh.
<Leo`>
That'll do.
<infinisil>
Also you can do nix-instantiate --eval '<nixpkgs/nixos>' -A options.boot.loader.systemd-boot.enable.description
<infinisil>
or config.boot.loader.systemd-boot.enable to get its value
peacememories has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
peacememories has joined #nixos
peacememories has quit [(Client Quit)]
peacememories has joined #nixos
<ee1943>
orivej: I've made a comment on that issue detailing the problem
<ee1943>
honestly the repro is basically just a hello world problem, since the issue hinges on clang not being able to find the system headers at all. If it can find the headers, the problem goes away
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vcunat pushed 2 new commits to master: https://git.io/vdnkY
<NixOS_GitHub>
nixpkgs/master dd200f8 Jan Tojnar: librsvg: fix thumbnailer path...
<orivej>
ee1943: Thank you. An important question is how clang should find the headers. For example if we properly wrap clang, it will work, but not the tools that parse with libclang.
<ee1943>
orivej: how so? I figured that if clang can find all the headers, then all of the libclang clients could too since libclang is basically just a API for clang
<ee1943>
i apologize if I'm completely wrong. I only have a cursory understanding of how clang et al integrate with nix
grumble has quit [(Quit: me)]
erictapen has joined #nixos
erictapen has quit [(Remote host closed the connection)]
erictapen has joined #nixos
<orivej>
libclang does run clang, it's more clang "runs" libclang
<orivej>
oops, libclang DOESN't run clang
grumble has joined #nixos
<orivej>
so, our clang wrapper adds additional arguments for clang, clang parses them and invokes code from libclang. clang without libclang is basically just a command line parser. AFAIK the only generic way to add search paths to libclang is to recompile it.
<jtojnar>
how much rebuilds is a mass rebuild?
<LnL>
I think making libcxx first class (similar to libc) in the cc-wrapper it the correct way to solve this
<infinisil>
jtojnar: see how pretty much every package in nixpkgs uses mkDerivation? When stdenv changes (where mkDerivation comes from), such as the setup.sh script, that's a mass rebuild
darlan has joined #nixos
<sphalerite>
infinisil: but other stuff can be considered a mass rebuild too even if it's not stdenv
<sphalerite>
infinisil: like the python interpreter
<infinisil>
Yeah
<infinisil>
or the same with other language modules, such as haskell
mortum5 has quit [(Ping timeout: 248 seconds)]
<ee1943>
orivej: I see. That's a bummer.
darlan has quit [(Client Quit)]
fritz09 has quit [(Quit: fritz09)]
<jtojnar>
I was asking because of the change in librsvg, which is a dependency of wrapGAppsHook and will rebuild ~500 derivations. Apparently that is not considered a mass rebuild.
<orivej>
ee1943: yes, but we can (1) build multiple libclang's, each with enough search paths to compile c++, (2) patch libclang to respond to a certain environment variable, and wrap every program that links with libclang (clang, probably rtags) such that it initializes that variable with proper search paths
<infinisil>
Well, there is no strict definition of a 'mass rebuild', it's a spectrum, and this is somewhere inbetween
<sphalerite>
Yeah I'd consider it a mass rebuild too but ¯\_(ツ)_/¯
<ee1943>
orivej: suppose option 2 is selected. rtags and irony now work fine, but wouldn't compiling a project from the command line still fail?
<ee1943>
* regular command line, not nix-shell or anything
<orivej>
ee1943: clang will be wrapped in the same way
cfricke has joined #nixos
<orivej>
LnL: I agree that first class libcxx is good, but will account only for clang, not for libclang linked to other programs (include-what-you-use, rtags)
<ee1943>
orivej: I see. That seems like a good solution
JosW has joined #nixos
Itkovian has joined #nixos
<LnL>
fixing just libclang won't help for other nix dependencies, I think those projects need a wrapper similar to our cc-wrapper
<LnL>
dtzWill: ping
Fare has quit [(Ping timeout: 264 seconds)]
<orivej>
LnL: unless we imlement the whole cc-wrapper as a patch for libclang :)
peacememories has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
peacememories has joined #nixos
jensens has joined #nixos
peacememories has quit [(Client Quit)]
<mguex>
argh...the rust/cargo situation in nixpkgs is really annoying
moet has joined #nixos
<srhb>
mguex: What's wrong with it? I was about to dive in earlier, but another project stole my time.
<srhb>
(dive into Rust, not the Nix ecosystem around Rust)
<orivej>
LnL: This is half joking, because I don't think that other programs that link to libclang need the full functionality of cc-wrapper. I expect that projects that use cmake and pkg-config will be satisfied when libclang can compile standard c++ .
<LnL>
orivej: could be, I just know that rtags won't find dependencies in a nix-shell unless you generate a json file based on NIX_FLAGS_COMPILE
<mguex>
srhb: yes
koserge has quit [(Ping timeout: 258 seconds)]
deepfire has quit [(Ping timeout: 258 seconds)]
<mguex>
a lot of crates fail to build
<srhb>
Yeah, I was under the impression that most people use the Mozilla overlay.
<moet>
hi nixos, i run a fairly low-powered machine (~1ghz cpu, 1gb ram) .. is this below the requirements for nixos, or are there configuration changes i can make to keep linux snappy?
<srhb>
moet: It's mostly about what applications you run, not which distro you choose.
<srhb>
I've never tried with anything that slow, it might be a bit tough if you're intending to run a graphical environment on it
<moet>
srhb: i was thinking about defaults of the kernel vm system and whatnot, but .. i'm currently running firefox and irssi, lol
jensens has joined #nixos
<moet>
fwiw, firefox isn't running any javascript most of the time
hc_ has joined #nixos
<orivej>
LnL: right. at least those who can and do use clang outside nix-shell will be happy because there is no NIX_CFLAGS_COMPILE there. I would have been one of them when I worked on a C++ project and used include-what-you-use.
hc has quit [(Ping timeout: 258 seconds)]
moet has quit [(Quit: leaving)]
eacameron has joined #nixos
hc_ has quit [(Ping timeout: 258 seconds)]
hc has joined #nixos
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<NixOS_GitHub>
[nixpkgs] bachp opened pull request #29992: glusterfs service: fix issues with useRpcbind (master...glusterfs-rpcbind) https://git.io/vdn3W
NixOS_GitHub has left #nixos []
zraexy has quit [(Ping timeout: 258 seconds)]
jellowj has quit [(Ping timeout: 246 seconds)]
Fare has joined #nixos
erictapen has quit [(Ping timeout: 258 seconds)]
Yotam_ has joined #nixos
ixxie has quit [(Quit: Lost terminal)]
Yotam_ has quit [(Client Quit)]
zzamboni has joined #nixos
zzamboni1 has joined #nixos
zzamboni has quit [(Read error: Connection reset by peer)]
JohnMH has joined #nixos
<tilpner>
sphalerite - So booting with acpi_rev_override=5 doesn't require turning on the GPU before starting X?
zarel has quit [(Quit: Leaving)]
<sphalerite>
tilpner: yep
<hyper_ch>
sphalerite: my zfs expert.... :) encountered a problem :)
<tilpner>
sphalerite - Nice! I'll try that on the next reboot :)
<sphalerite>
seems to vary across hardware though so YMMV
<sphalerite>
hyper_ch: did you now
<hyper_ch>
yes :)
<sphalerite>
"Run nix-store --optimise in case you really need some a bit of space. It is a lengthy operation, and after multiple-outputs PR landing, it became less useful. " how does multiple-outputs affect --optimise?
<hyper_ch>
I was working with snapshots... so I did zfs send pool1/dataset@snapshot1 | zfs receive pool2/dataset@snapshot1 -- that worked fine. So I tried to send a second snapshot: zfs send pool1/dataset@snapshot2 | zfs receive pool2/dataset@snapshot2 -> but I got: cannot receive new filesystem stream: destination 'pool2/dataset' exists must specify -F to overwrite it..... so how can I send subsequent snapshots then?
sivteck has joined #nixos
sivteck has quit [(Client Quit)]
<LnL>
you can send the incremental changes from snapshot1 to snapshot2
<hyper_ch>
LnL: what's what I tried... but can I do that
<MichaelRaskin>
gchristensen: well, it only works in the form of saving-throw defense, stockpiling is complicated.
<solenodic>
hmm... what am I missing
<MichaelRaskin>
Well, you can always go the explicit way with git checkouts.
<solenodic>
okay so running `nix-channel --update` without "nixos-unstable" at the end
<solenodic>
downloaded a new file that wasn't downloaded before
<solenodic>
so I think it will do the trick
<MichaelRaskin>
Of course, git checkout takes a bit more space than a channel snapshot…
<solenodic>
yep
<solenodic>
thanks LnL
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] LnL7 pushed 1 new commit to release-17.09: https://git.io/vdnnt
<NixOS_GitHub>
nixpkgs/release-17.09 c3176ee Will Dietz: llvm: init 5.0...
NixOS_GitHub has left #nixos []
<gchristensen>
MichaelRaskin: I know :( limitations apply. the system I upgraded yesterday was interesting in that in order to handle a mass rebuild, I had to disable almost all the services, rebuild, GC, update the channel, rebuild, GC, enable the services, rebuild
<gchristensen>
replacing that system's 8GB / with a 500GB / only bought probably 50-75 sanity points
<NixOS_GitHub>
[nixpkgs] peti pushed 1 new commit to master: https://git.io/vdncV
<NixOS_GitHub>
nixpkgs/master 2784bd7 Peter Simons: haskell-json-autotype: jailbreak to fix build within LTS 9.x...
NixOS_GitHub has left #nixos []
katyucha has joined #nixos
sigmundv_ has quit [(Ping timeout: 258 seconds)]
rauno has joined #nixos
takle has quit [()]
sivteck has quit [(Quit: user missing.)]
takle has joined #nixos
silver_hook has quit [(Ping timeout: 258 seconds)]
drakonis has quit [(Quit: Leaving)]
radivarig has joined #nixos
jb55 has joined #nixos
Ivanych has joined #nixos
katyucha has quit [(Ping timeout: 246 seconds)]
kuznero has joined #nixos
<radivarig>
What's the best way to find out the package path of a binary to call from config? I can get echo path with "${pkgs.coreutils}/bin/echo" which I found in an example but how to find for any package? For example 'shutdown', whereis shutdown links to /nix/store/hash..-system-path/bin, but there are many other binaries as well, and no system-path package
<hyper_ch>
ls -al /run/current-system/sw/bin/echo
<hyper_ch>
if it's a binary
<LnL>
you can also use nix-index to what package contains the binary you're looking for
<radivarig>
awesome, thanks
oida_ has joined #nixos
oida has quit [(Ping timeout: 260 seconds)]
sigmundv_ has joined #nixos
JohnMH has left #nixos ["Bye!"]
nh2 has joined #nixos
kuznero has quit [(Remote host closed the connection)]
_rvl has joined #nixos
sigmundv_ has quit [(Ping timeout: 240 seconds)]
Fendor has joined #nixos
Guest32082 has quit [(Quit: Lost terminal)]
nh2 has quit [(Ping timeout: 258 seconds)]
<bachp>
Is there a way to increase the root disk of a vm that is created to run a nixos/tests test?
<LnL>
I seem to recall something about disk size, but it might have been for an extra disk
<disasm>
how can I have a nixos module write a config file to a state directory?
<disasm>
I don't want this file written to the store at all (ldap creds inside)
<bachp>
LnL: I could also be an extra disk
jsgrant has left #nixos []
<bachp>
It's just that new versions of minio only run if they have at least 1GiB of free disk in their data directory
<NixOS_GitHub>
[nixpkgs] mboes opened pull request #30001: git: Properly wrap Git commands that are implemented in Perl. (master...git-add-interactive) https://git.io/vdnRm
<tilpner>
iqubic - Why do you think it should run?
<tilpner>
Nobabs27[m] - Good guessing, mostly :/
<tilpner>
(Or locate/find, if it's already installed)
MP2E has joined #nixos
<iqubic>
tilpner: I think it should run because fetchFromGithub needs to run in order to give me an outPath.
<tilpner>
iqubic - But what makes you think fetchFromGitHub in any way even looks at your extraPostFetch?
<tilpner>
iqubic - AFAICT it doesn't, and is not supposed to
<tilpner>
Or does fetchzip do that?
garbas has joined #nixos
<iqubic>
tilpner: If it doesn't even look at the extraPostFetch, then why is it that I can it to a value?
<tilpner>
fetchFromGitHub accepts anything, try passing the argument fooBarQuux
<Mic92>
Are there nix unstable binaries somewhere? I would need aarch64
<iqubic>
So how can I get a certain snippet of shell code is run after fetchFromGithub is run?
<tilpner>
extraPostFetch could actually work, I didn't see passthruAttrs being passed to fetchzip
<iqubic>
Perhaps this is not working because I had fetched this repo once before without the extraFetchPost being present.
<tilpner>
Your extraPostFetch runs for me
<iqubic>
How can you tell?
<iqubic>
Also, had you fetched this repo before?
<tilpner>
I had not
<iqubic>
That might be my issue.
<iqubic>
Nix is refusing to fetch this repo a second time
<tilpner>
I replaced your script with a bogus "foobar" and waited for corresponding /nix/store/6zngq1rdh0ans9qyckqimqibgnlvlfrm-stdenv/setup: line 114: foobar: command not found
<iqubic>
Ah. I see.
<tilpner>
I'm a little confused by the '' included in "mkdir $out/''${repo}"
<iqubic>
I was getting errors about repo not defined until I did that.
yegortimoshenko has joined #nixos
<tilpner>
I know what's going on... remove the '' in your extraPostFetch and put a rec on your argument set
<tilpner>
"pkgs.fetchFromGitHub rec { owner ..."
<iqubic>
I have done that.
<tilpner>
That should make repo available as a variable in extraPostFetch
<iqubic>
Wait, I put rec in the wrong place
jensens has quit [(Ping timeout: 240 seconds)]
<Nobabs27[m]>
yes! finally got it to work!
<tilpner>
Right now you're escaping the ${}, so you're putting a literal ${repo} in your script
<iqubic>
Right, and now stuff is working a bit better.
<iqubic>
However, I'm still not sure that my extraPostFetch was run.
<iqubic>
Where is that supposed to move the repo to?
<tilpner>
Try changing it to foobar :P
<iqubic>
I will.
yegortimoshenko has left #nixos ["ERC (IRC client for Emacs 27.0.50)"]
* tilpner
doesn't get what your extraPostFetch is supposed to achieve
<iqubic>
However, the extraPostFetch thing is not being run at all.
fusion809 has joined #nixos
<fusion809>
Hi folks how does one install a whole DE after the system is already installed? Like using nix-env? I have installed NixOS already and I'd like to install KDE Plasma without going back and re-installing NIxOS all over.
<fusion809>
Is it a matter of running nix-install after editing configuration.nix to install KDE?
<tilpner>
fusion809 - You change your configuration.nix, then call "nixos-rebuild switch" as root
<fusion809>
Ah, I was close, thanks.
bkchr has joined #nixos
<fusion809>
Is GNOME available for NixOS? Or is it KDE Plasma or not DE at all?
vandenoever has quit [(Ping timeout: 260 seconds)]
<iqubic>
No. All I did was change the extraPostFetch to "foobar" and then ran "nixos-rebuild test"
<iqubic>
I got no errors at all.
<tilpner>
iqubic - tx0.co/UCP7
<iqubic>
So it put the fetched repo into $(pwd)?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] WilliButz opened pull request #30006: nasm: run checkPhase (master...nasm-run-tests) https://git.io/vdngn
NixOS_GitHub has left #nixos []
<tilpner>
No
<tilpner>
That's just to show that your thing actually works like I think you want it to
fusion809 has quit [(Quit: Konversation terminated!)]
jbaum98 has joined #nixos
<jbaum98>
i've installed nixpkgs with a multi-user setup on darwin with the new install script and building anything causes my system to freeze sometimes. I can't get it to happen consistently enough to give a minimal example. is this a known issue?
<iqubic>
tilpner: Yeah, it does work, when it can be run.