graphene has quit [Remote host closed the connection]
pie_ has joined #nixos
graphene has joined #nixos
worldofpeace has quit [Quit: worldofpeace]
<acowley>
If a build produces a binary that it then wants to run later during the build, and that binary needs a shared library also built during the build, how best to instrument things such that it can find the binary?
<acowley>
ugh, "... such that it can find the library?"
<acowley>
I'm updating Paraview, and it builds its own VTK, and at some point it wants to read some XML using a binary produced during the build, and that binary can't find a shared VTK library produced earlier in the build. There are many ways to skin this cat, but I'd prefer something genteel.
<acowley>
They have at attempt at resolving this in their cmake, so ideally I'll be able to tweak that
worldofpeace has joined #nixos
wpcarro has joined #nixos
blankhart has joined #nixos
mkoenig has quit [Ping timeout: 260 seconds]
tenten8401 has quit [Quit: Konversation terminated!]
<tenten8401>
does anyone else have CA certificate issues in the Wine Staging that's in nixos-unstable? I'm launching the Battle.net installer and it's failing to verify the digital signature, but it's only with the wineWow build
<tenten8401>
If I look up the Battle.net error code it's asking me to import a CA certificate
<tenten8401>
it verified correctly and let me past with the regular wineStaging
kiloreux has quit [Read error: Connection reset by peer]
<aszlig>
tenten8401, infinisil: IIRC the battle.net updater uses bittorrent for fetching the content, maybe there's a way to use a native bittorrent client instead?
Supersonic112 has joined #nixos
Supersonic has quit [Disconnected by services]
Supersonic112 is now known as Supersonic
igo95862 has joined #nixos
romildo has quit [Quit: Leaving]
udkyo has joined #nixos
Coyin has quit [Remote host closed the connection]
Coyin has joined #nixos
udkyo has quit [Ping timeout: 265 seconds]
Coyin has quit [Client Quit]
sam__ has quit [Ping timeout: 240 seconds]
wpcarro has joined #nixos
sigmundv has quit [Ping timeout: 264 seconds]
trevthedev has joined #nixos
trevthedev has quit [Changing host]
trevthedev has joined #nixos
wpcarro has quit [Ping timeout: 244 seconds]
silver has quit [Read error: Connection reset by peer]
trevthedev has quit [Quit: WeeChat 2.0]
derped has quit [Ping timeout: 244 seconds]
<nh2>
can somebody explain me what exactly `strictDeps` does?
aanderse-laptop has quit [Ping timeout: 276 seconds]
<tertle||eltret>
maybe il install nix on my other drive
acarrico has quit [Ping timeout: 240 seconds]
<tertle||eltret>
are yall gonna release an enterprise version like RHEL?
<buhman>
> implying RHEL is desirable
<{^_^}>
undefined variable 'implying' at (string):159:1
derped has joined #nixos
<buhman>
tertle||eltret: so, having obnoxious "LTS" policies and doing whatever unreasonable thing any customer with a sufficient amount of money asks for?
<tertle||eltret>
im not implying, im just curious about the future of it?
<tertle||eltret>
or who it is aimed at or if it 'just makes sense'
<tertle||eltret>
im setting up a linux box on my LAN so i can remote in from my main box
<tertle||eltret>
is there a specific program that you would use for this?
<nbathum>
I'm running it as my main desktop right now
cryptomonoid has joined #nixos
<nbathum>
there are definitely some folks using nixos at their work
<nbathum>
to varying degrees I'd speculate
vandenoever has quit [Ping timeout: 260 seconds]
<buhman>
tertle||eltret: I would hope that future is a fundamental paradigm shift, not morphing to the whims of corporate politics
<drakonis>
hm yes you're opinionated
<drakonis>
perhaps you should stop using linux
<drakonis>
or computers at all
<samueldr>
tertle||eltret: it would be interesting to see longer support for a given release
<buhman>
I agree, this is an excellent idea
<samueldr>
though right now the issue is with the amount of hands working
wpcarro has joined #nixos
<samueldr>
and if you're looking for certification for hardware/software, that's probably not something that's thought of...
<drakonis>
tertle||eltret, RHEL is pretty much kept afloat by the amount of hands working
<samueldr>
... though there may be some nix-using entities that could work on that
<drakonis>
which is a very high number
<buhman>
drakonis: isn't that effort better spent towards radical innovation, rather than legacy maintenance?
<drakonis>
they also do radical innovation
mizu_no_oto has joined #nixos
<Raybih>
Oneness Pentecostalism (also known as Apostolic or Jesus' Name Pentecostalism) is a movement within the Christian family of churches known as Pentecostalism.
<drakonis>
idgi
<samueldr>
tertle||eltret: just for the record, when you said "like RHEL" what did you have in mind?
<samueldr>
like longer support for stable branches?
<samueldr>
or stable ABIs for everything for an insane amount of time?
wpcarro has quit [Ping timeout: 256 seconds]
<tertle||eltret>
idk like maybe something like that. im thinking about getting a linux cert for my resume and im just curious about it all
<tertle||eltret>
i dont really know exactly what a RHEL sub entails tbh
<samueldr>
ah, don't think there are any cert that even come close to anything nix~y :/
<tertle||eltret>
i will probably install nix on my mechanical drive and have RHEL on the SSD
<tertle||eltret>
i also enjoyed funtoo
<tertle||eltret>
it made a bit more sense to me
<tertle||eltret>
im just stoked that I finally have a spare machine to do it with
<tenten8401>
aszlig: I haven't even gotten to that part yet
<tenten8401>
I'm just trying to install the launcher
<drakonis>
nix is too much of an departure
<drakonis>
still has limitations missing from regular distros
<tenten8401>
infinisil: If the Lutris PR ever started working we may see an easily installable overwatch
<tenten8401>
I was currently trying to set it up manually since Lutris kept throwing random errors
<tenten8401>
it'd be interesting to see a lutris rewrite with nix integration
Raybih has quit [K-Lined]
<tenten8401>
but yeah if anyone would like to test #42075 and help fix it (pretty sure it doesn't need much at this point), it'd be greatly appreciated
<tenten8401>
y'know, I wonder if the real problem is the lack of 64-bit wine available to Lutris
<tenten8401>
one sec
<drakonis>
that's a problem
<drakonis>
no 64 bit wine nor WoW wine
<tenten8401>
would also explain the "taskset/cpuset not available on your platform!"
<drakonis>
hour long wine build and it sometimes doesn't work properly
<tenten8401>
what's the different between the WoW wine and 64-bit?
<drakonis>
you can run 32 bit binaries
<tenten8401>
I installed + built the wow wine earlier but was having issues with certificate authorities
<drakonis>
WoW provides 32/64 instead of either 32 or 64
lassulus has quit [Ping timeout: 256 seconds]
lassulus_ is now known as lassulus
<iqubic>
Why is wine being a bit of a pain.
<iqubic>
??
<drakonis>
because stupid defaults
<tenten8401>
is there anything that needs to be added to build wineWow to get actual CA certificates imported by default? I'm like 90% sure that's the issue as it's what the Blizzard article was suggesting you do
<aszlig>
tenten8401: yeah, ideally instead of fetching random stuff from lutris.net that would need to be patchelf'ed (or buildFHSUserEnv'ed) there should just be a NAR cache on lutris.net
graphene has quit [Read error: Connection reset by peer]
baconicsynergy has joined #nixos
graphene has joined #nixos
hiroshi has joined #nixos
Sonarpulse has quit [Ping timeout: 240 seconds]
jperras has quit [Ping timeout: 264 seconds]
<{^_^}>
[nixpkgs] @orivej-nixos merged pull request #31551 → cudnn*_cudatoolkit*: use public NVIDIA mirror → https://git.io/vFww7
<{^_^}>
[nixpkgs] @orivej-nixos pushed commit from @lukeadams to master « cudnn*_cudatoolkit*: use public NVIDIA mirror (#31551) »: https://git.io/fNG3z
<{^_^}>
[nixpkgs] @matthewbauer opened pull request #43718 → make-tarball: also check for licenses & versions → https://git.io/fNG3o
<ryantm>
samueldr: Do you know who is up and can fix it? graham is probably asleep
<samueldr>
not sure anyone else has access to the infra yet, though if anyone does, LnL may be the next one up who can check?
iqubic has left #nixos ["ERC (IRC client for Emacs 25.3.1)"]
<samueldr>
just looked at the backlog on #nixos-borg and yeah, don't think anyone else has access
drakonis has quit [Remote host closed the connection]
wpcarro has joined #nixos
andrewrk has quit [Ping timeout: 240 seconds]
wpcarro has quit [Ping timeout: 244 seconds]
jackdk has quit [Ping timeout: 264 seconds]
andrewrk has joined #nixos
cryptomonoid has joined #nixos
sbdchd has quit [Remote host closed the connection]
endformationage has quit [Quit: WeeChat 1.9.1]
cryptomonoid has quit [Ping timeout: 244 seconds]
iqubic has joined #nixos
jperras has joined #nixos
<kalbasit>
I’m on an EFI device, with an ESP partition for boot and a btrfs for the FS. I want to install Arch on a separate subvolume. Can systemd-boot boot Arch? Do I have to switch to Grub with the os prober?
<iqubic>
No. You will be able to use systemd-boot for that.
<iqubic>
I'm not sure exactly how you set that up, but I know it can be done.
<kalbasit>
how to tell it where things are? I couldn’t find a guide for dual boot, besides a windows one
<kalbasit>
hmmmm, I’ll check arch wiki then, maybe some indication there
<iqubic>
kalbasit: You should ask you question on the arch linux IRC channel.
vmandela has joined #nixos
<iqubic>
It general has more people around so you'll get an answer faster than asking here.
<iqubic>
But yeah, do also consult the arch wiki.
<kalbasit>
will do, thx
jperras has quit [Ping timeout: 248 seconds]
vandenoever has joined #nixos
vandenoever has quit [Changing host]
vandenoever has joined #nixos
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
freeman42x]NixOS has quit [Ping timeout: 260 seconds]
<|Leary>
kalbasit: to manually register your Arch install you'd use efibootmgr, but the Arch install of systemd-boot may well hook that in for you.
fleaz has joined #nixos
toby1851 has joined #nixos
<kalbasit>
|Leary: I'm using LUKS as well and I prefer Nix opening the partitions, I'll probably have to do some R&D
<|Leary>
Oh, actually I think that's to register systemd-boot itself with the bios, rather than an OS with systemd-boot... forget that.
silver_hook has joined #nixos
silver_hook has quit [Changing host]
silver_hook has joined #nixos
tertle||eltret has quit [Quit: Connection closed for inactivity]
wpcarro has joined #nixos
<adisbladis[m]>
systemd-boot lists the directory with config and loads all of them
phreedom has joined #nixos
lambdamu has quit [Ping timeout: 268 seconds]
wpcarro has quit [Ping timeout: 244 seconds]
sjourdois has joined #nixos
nikki93 has joined #nixos
Forkk has joined #nixos
heatm1s3r has joined #nixos
bitonic has joined #nixos
mahalel_ has joined #nixos
ihar has joined #nixos
Cale has joined #nixos
jophish has joined #nixos
oharvey has joined #nixos
aanderse has joined #nixos
r0bby has joined #nixos
gleber_ has joined #nixos
newhoggy has joined #nixos
jasom has joined #nixos
kyren has joined #nixos
mbrgm has joined #nixos
andi- has joined #nixos
hhes has joined #nixos
{^_^} has joined #nixos
Orbstheorem has joined #nixos
angerman has joined #nixos
xplat|work__ has joined #nixos
Tritlo has joined #nixos
philipcristiano has joined #nixos
kitemikaze_ has joined #nixos
flokli has joined #nixos
georgew has joined #nixos
Smithx10 has joined #nixos
teozkr has joined #nixos
zielmicha_ has joined #nixos
sorear has joined #nixos
rodarmor has joined #nixos
johnhamelink has joined #nixos
Wizek has joined #nixos
benkolera has joined #nixos
Kim has joined #nixos
shabius has joined #nixos
PyroLagus has joined #nixos
iqubic` has left #nixos ["ERC (IRC client for Emacs 25.3.1)"]
Wizek has quit [Max SendQ exceeded]
toby1851 has quit [Ping timeout: 244 seconds]
iqubic has joined #nixos
spear2 has quit [Remote host closed the connection]
spear2 has joined #nixos
Wizek has joined #nixos
hamishmack has joined #nixos
<iqubic>
I figured out why I wanted busybox in my systemPackages. I thought that would be the way to get "killall"
freeman42x]NixOS has quit [Ping timeout: 240 seconds]
<clever>
,locate killall
<{^_^}>
Found in packages: psmisc, killall, busybox, netbeans, playonlinux, bash-completion
<clever>
iqubic: its also in the psmisc package
<Lisanna>
wow, what a useful command
<Lisanna>
what are all the other ofborg cmds?
<Lisanna>
or whatever that bot is called
<Lisanna>
,help
<{^_^}>
Use `,` to list all commands, `,foo = Foo!` to define foo as "Foo!", `,foo =` to undefine it, `,foo` to output "Foo!", `,foo somebody` to send "Foo!" to the nick somebody
<iqubic>
I know. I just went ahead and got psmisc installed. I ran into issues with busybox the other day.
<LnL>
samueldr: morning, yeah don't think I can do anything
<clever>
Lisanna: and ,locate just uses nix-index, which you can install locally
<Lisanna>
oh ok
<iqubic>
busybox is not meant to be installed on a system after it is installed.
<iqubic>
It screws with many standard commands.
<rizary>
so every time i do nix-build on my ubuntu on windows, i keep getting `killing process 21150: No such process` at the end of build. The build it successful (perhaps?) though. Is this intended behavior?
<emily>
iqubic: you might also want to use "pkill -x foo", which doesn't require an extra package and doesn't kill *every process* without prompting on many Solaris/Illumos machines :p
<iqubic>
emily: Sometimes I want to kill a certain program.
<emily>
(I almost installed killall recently, but decided to take the opportunity to retrain for whatever the portability/fewer dependencies is worth.)
<emily>
"pkill -x" is almost identical to killall(1)
bennofs has joined #nixos
derped has quit [Ping timeout: 256 seconds]
<iqubic>
So I can run "pkill -x vim" And vim and only vim will be killed?
<{^_^}>
[nixpkgs] @CMCDragonkai opened pull request #43720 → pythonPackages.imgaug: init at 0.2.6 → https://git.io/fNGnA
wchresta has joined #nixos
derped has joined #nixos
cryptomonoid has joined #nixos
Supersonic has quit [Ping timeout: 260 seconds]
tenten8401 has quit [Ping timeout: 268 seconds]
crmlt has joined #nixos
Supersonic has joined #nixos
graphene has quit [Remote host closed the connection]
graphene has joined #nixos
cryptomonoid has quit [Ping timeout: 240 seconds]
wchresta has quit [Ping timeout: 240 seconds]
johnw_ is now known as johnw
johnw has quit [Changing host]
johnw has joined #nixos
udkyo has quit [Quit: udkyo]
schjetne has joined #nixos
udkyo has joined #nixos
alter-schjetne has quit [Ping timeout: 248 seconds]
udkyo has quit [Client Quit]
bennofs has quit [Ping timeout: 265 seconds]
wpcarro has joined #nixos
ajs124 has joined #nixos
noam has joined #nixos
qwerty_asdf has quit [Quit: WeeChat 2.0]
crmlt has quit [Quit: Lost terminal]
crmlt has joined #nixos
wpcarro has quit [Ping timeout: 276 seconds]
cryptomonoid has joined #nixos
fractal has joined #nixos
reinzelmann has joined #nixos
jperras has joined #nixos
lo_mlatu has joined #nixos
<lo_mlatu>
question: How can I remove old grub boot entries?
bennofs has joined #nixos
<etu>
First you have to remove old generations, there are different ways of doing that. After that you have to do a nixos-rebuild boot/switch to rewrite the grub menu
<etu>
The nix utils for removing generations aren't "nixos aware" so to speak, so they don't clean up boot stuff
<lo_mlatu>
So there does not exist a nixos-* tool for doing this?
Tucky has joined #nixos
jperras has quit [Ping timeout: 276 seconds]
<clever>
lo_mlatu: just the normal nix-collect-garbage
<etu>
Not that I'm aware of, I usually remove generations that I don't want to keep for whataver reason, then I do a nix-collect-garbage to clean out the store and the next time a make a new generation with nixos-rebuild it removes my boot entries
<lo_mlatu>
clever: I mean the tool to remove old generations, is "sudo nix-env -p /nix/var/nix/profiles/system --delete-generations" a good way?
<etu>
lo_mlatu: That's what I do, not sure it's the "best" or "most correct" way though:)
<clever>
yeah, that will work fine
<lo_mlatu>
Got it, thank you etu and clever : )
<etu>
lo_mlatu: Just make sure to run nix-collect-garbage after you deleted the generations to free up space:)
bennofs has quit [Ping timeout: 256 seconds]
tzemanovic has quit []
lo_mlatu has quit [Ping timeout: 252 seconds]
<crmlt>
Hi, i forgot... how I'm supposed to reference the package binary within configuration.nix to get its full path... It was something like "<nixpkgs.chgrp>" I just can't remember...
<kalbasit>
clever: seems `config.networking` is also not available in nixpkgs?
<clever>
kalbasit: correct, that is only in nixos
<clever>
kalbasit: your configuration.nix has to conditionally change the overrides
<clever>
not the nixpkgs layer
<kalbasit>
clever: there's no way to communicate down to nixpkgs, or do you mean I will have to use an overlay in the configuration.nix for this `hadesFixup` I have?
<kalbasit>
s/Fixus/Fixes/
<clever>
kalbasit: you can make it an argument to the package, that is boolean, and then override that from nixos
<clever>
{ stdenv, doHadexThing ? false, ... }:
<clever>
and then in nixos, nixpkgs.overlays = [ (self: super: { something = super.something.override { doHadexThing = true; }; }) ];
<kalbasit>
oh, so I can just send down the hostname instead? `{stdenv, myHostname ? "", ... }` correct?
<clever>
you could also do that
<kalbasit>
awesome
<kalbasit>
clever: do you have an example of that? haven't seen any yet
<clever>
it would just be a variant of the example i just gave, with strings instead of bools
wpcarro has joined #nixos
<kalbasit>
clever: I'm sorry for the confusion, but I meant to ask about passing down any variable from NixOS to Nixpkgs
toby1851 has quit [Read error: Connection reset by peer]
<adisbladis[m]>
What's a good way to go from a .drv to it's output paths?
<{^_^}>
[nixpkgs] @johanot opened pull request #43725 → nixos/kubernetes: Added missing dependency, awk, to kube-addon-manager path → https://git.io/fNGgI
<adisbladis[m]>
The moment after I wrote that I realised `nix-store -q` is exactly what I want -.-
<ldesgoui[m]>
system-units is failing to build because systemd-238 is missing example/systemd/system/system.slice, any way to fix this?
<le_jonge>
does anyone have experience with "stackage2nix"? i tried to nix-build the wrapper there, but it compiles for hours and it looks like it actually compiles all the stackage packages instead of just building an index of them??
<ldesgoui[m]>
Fails using nixpkgs-unstable and nixos-unstable
* woffs
read "unusable" for unstable :-D
<ldesgoui[m]>
haha. might as well just back-port the package i need, don't feel like fighting all of this :)
<manveru>
le_jonge: yeah, it's really compiling all of hackage...
<manveru>
it's the reason i went with stack instead
<clever>
manveru: try importing that file in nix repl
<clever>
manveru: it may be an attrset, containing an attr for every single package
<clever>
and since you didnt give it an attr, it builds all
<manveru>
clever: you can tell le_jonge :)
<manveru>
i haven't done haskell in months
<le_jonge>
manveru: generally i am happy with `stack2nix`. but then i am unable to change the nix expressions in a way that i have `ghcid` in my nix-shell environment, because stack2nix does some magic that i don't understand. i already opened an issue on the stack2nix github repo. besides that i wanted to play with stackage2nix.
<clever>
le_jonge: add .env to the attribute and ghcid works fine
<le_jonge>
but if stackage2nix is such an effort compiling everything from lts... i'd rather stick to stack2nix
crmlt has quit [Read error: Connection reset by peer]
humanoyd has joined #nixos
<clever>
le_jonge: i use ghcid with stack2nix all the time
<le_jonge>
clever: oh THAT is nice to hear. i guess i am doing it completely wrong. i have a line like `if pkgs.lib.inNixShell then drv.env else drv` but it does not include ghcid
<le_jonge>
if i don't install ghcid system wide it won't work
<clever>
ghcid is installed system wide, too much of a bother to override it into literally 100's of different derivations
<le_jonge>
hm. my whole imagination of a nix project is "you just clone the repo and EVERYTHING YOU NEED is available in nix-shell"
<le_jonge>
otherwise i am again at "ok so for the following workflow you first need to install...."
<le_jonge>
and it's also not possible to say `nix-shell -p ghcid myproject.nix`, right?
<clever>
le_jonge: -p cant be mixed with nix files
<clever>
-p auto-generates a nix file and loads it
<le_jonge>
right.
toby1851 has joined #nixos
<le_jonge>
i just want something like `if pkgs.lib.inNixShell then ( do something that adds ghcid and maybe other developer tools to the environment).env else drv`
<clever>
le_jonge: i try to avoid pkgs.lib.inNixShell, it tends to just cause more confusion
trylist has joined #nixos
<le_jonge>
it would be pretty cool to be able to do that even with a --pure shell. so is your opinion about that "it's not the right way" or "it's too complicated"?
<clever>
for example, it makes it imposible to run `nix-build` inside a `nix-shell`
<clever>
it automatically switches to the wrong derivation
<le_jonge>
what is the usecase/reason for running nix-build in a nix-shell?
<sphalerite>
yeah it's nicer to just have a separate attr for the dev environment
<clever>
not wanting to close the nix-shell because re-opening causes a 2 hour recompile
<le_jonge>
hm
hsribei has quit [Quit: Connection closed for inactivity]
<le_jonge>
sphalerite: a separate attr/nixfile would be also fine for me. i am happy if i can do `nix-shell --pure --something --something` and have ghcid available although i don't have it installed systemwide
<clever>
le_jonge: you would need to create a new default.nix that overrides mkDerivation to add ghcid to every single package in your tree
<crmlt>
anyone here uses enlightenment DE?
<clever>
le_jonge: which also means rebuilding every single package
<clever>
le_jonge: line 47 loads the stack2nix output, 51 passes it a set of overrides, and and 99 overwrites mkDerivation
<clever>
le_jonge: and the new mkDerivation can append ghcid to the inputs
<le_jonge>
clever: that sounds like a lot of work to just get another package in the PATH. is that really the only way?
<clever>
le_jonge: you could maybe use mapAttrs in shell.nix, to mutate every single attribute, and do it in a shallow manner, so the deps arent modified
trylist has quit [Ping timeout: 276 seconds]
<clever>
le_jonge: this is why i just installed ghcid system wide
<clever>
le_jonge: ghc itself is not installed, so i dont need --pure very often
<le_jonge>
hm, i understand.
goibhniu has joined #nixos
crmlt has quit [Quit: Lost terminal]
silver has joined #nixos
knupfer has joined #nixos
<{^_^}>
[nixpkgs] @srhb opened pull request #43728 → Don't lose rbac submodule with no explicit options → https://git.io/fNGVQ
civodul has joined #nixos
Lisanna has quit [Remote host closed the connection]
anderslu1dstedt has quit [Quit: leaving]
wpcarro has joined #nixos
tobiasBora has quit [Ping timeout: 244 seconds]
<le_jonge>
clever: in which repo is the sourcecode of nix-shell (or is nix-shell just a wrapper script around something), if i want to learn from it?
anderslundstedt has joined #nixos
<clever>
le_jonge: nix-shell and nix-build are the exact same binary, both are in the nix repo
<le_jonge>
thx
<clever>
le_jonge: nix-build will check the value of argv[0] to decide what mode to run in
ckauhaus has quit [Remote host closed the connection]
<le_jonge>
oh nice this is even just a single file.
<srhb>
The only thing I've experienced recently is oddness with 64 bit steam games and the 32 bit libprimus, iirc.
<srhb>
Notably Rimworld.
<tilpner>
When I bought this laptop, nvidia and amd were said to be about equally bad. With all the mainline amd drivers now though, I wouldn't buy another nvidia laptop
slyfox has quit [Ping timeout: 256 seconds]
<tilpner>
(And I've only heard bad things about CUDA on NixOS, if that's something you intend to use)
<goibhniu>
any idea if external GPUs work with thunderbolt on linux reliably these days?
<srhb>
goibhniu: There's lots of youtube videos on the topic at least. I'd say it's an enthusiast thing, but it works.
slyfox has joined #nixos
<goibhniu>
that sounds promising
Ridout has joined #nixos
<itsok>
Hi folks, I'm having issues getting KDE Plasma to start when installed from the nix package manager. I'm not using NixOS (nix package manager on Arch instead). Trying to start Plasma with startx and the according .xinitrc, but always getting "start_kdeinit: No such file or directory". kdeinit5 is in my PATH though...
wpcarro has joined #nixos
<srhb>
itsok: But is start_kdeinit? Note that I've no idea whether that can reliably work outside of NixOS at all, but that appears to be the specific error.
<FRidh>
I don't think anyone ever tried using it outside of NixOS
<itsok>
I didn't even know start_kdeinit was a binary...
<itsok>
Which package holds it, does anyone know?
<joepie91>
,locate start_kdeinit
<{^_^}>
Found in packages: kdeFrameworks.kinit.bin
<joepie91>
itsok: ^
<srhb>
itsok: Looks like kdeFrameworks.kinit
<srhb>
oh, rught..
<cocreature>
I really need to figure out how to get nix-locate to work locally :)
<itsok>
kdeFrameworkds.kinit is installed
<{^_^}>
[nixpkgs] @srhb merged pull request #43728 → k8s: Don't lose rbac submodule with no explicit options → https://git.io/fNGVQ
<srhb>
cocreature: Is this legit in Haskell land sans Nix?
<srhb>
I can see that it could be...
<srhb>
Never seen it before though.
<cocreature>
srhb: I think since cabal switched to per-component builds it should be legal
<cocreature>
but I’m not entirely sure
<srhb>
Oh ugh, that's gonna be a while out to fix for us then...
<cocreature>
iirc there were some attempts to also support per-component builds in nix but I don’t know the current status of that
<srhb>
Yeah, I know. It's a good thing, but.. Probably not a tomorrow thing.
<cocreature>
yeah sadly :/
<cocreature>
as a quick fix, we could probably just slap a dontCheck on spatial-rotations?
alex`` has joined #nixos
<srhb>
Yep.
xy2_ has joined #nixos
freeman42x]NixOS has joined #nixos
<{^_^}>
[nixpkgs] @srhb opened pull request #43730 → haskellPackages.spatial-rotations: dontCheck for cyclic dependency → https://git.io/fNG1R
<cocreature>
srhb++
<{^_^}>
srhb's karma got increased to 10
<srhb>
Hm, does that even work...
<srhb>
srhb--
<srhb>
:P
<_dbe>
Whoa. A karma bot.
_dbe is now known as dbe
<gchristensen>
{^_^}++
<{^_^}>
{^_^}'s karma got increased to 135
<gchristensen>
:O
orivej has joined #nixos
<dbe>
{^_^}: ++++
<dbe>
{^_^}++++
<{^_^}>
{^_^}++'s karma got increased to 1
<dbe>
whoops.
tzemanovic has joined #nixos
<gchristensen>
haha
<cocreature>
srhb: I have no idea how it works but you definitely deserve more karma :)
<srhb>
haha, thanks.
<srhb>
It's just broken for other reasons too...
init_6 has joined #nixos
acarrico has joined #nixos
baconicsynergy_ has quit [Ping timeout: 248 seconds]
baconicsynergy has quit [Ping timeout: 240 seconds]
<johanot>
srhb: does karma have half-life I wonder? if so, can it ever be zero or negative? :P
<gchristensen>
it can't go down unless you try to game it
<srhb>
johanot: I believe a buddhist would tell you that karma does not increase or decrease but through your actions in all lives.
<srhb>
johanot: You share office with someone who I think is an authority on the matter, so don't take my word for it.
<srhb>
I think past-life-karma is still unimplemented in nixbot though...
<Taneb>
srhb: maybe I should make a PR
<johanot>
it already got way too complicated to implement using simple arithmetics.. karma-AI to the rescue!
<{^_^}>
[nixpkgs] @kwohlfahrt closed pull request #31832 → kerberos_server: allow choosing MIT or Heimdal → https://git.io/vFHdD
<{^_^}>
[nixpkgs] @kwohlfahrt reopened pull request #31832 → kerberos_server: allow choosing MIT or Heimdal → https://git.io/vFHdD
ericsagnes has joined #nixos
derped has joined #nixos
<gchristensen>
hey so master takes >8G to evaluate now on ofborg, causing all PRs to fail since those machines don't have so much RAM
<infinisil>
Haha wow
<gchristensen>
do we (1) buy bigger machines? (2) fix memory usage? somehow (3) throw away ofborg?
<gchristensen>
what do we do?
<gchristensen>
note this is a very time sensitive question, "fixing nix" isn't the answer
<dbe>
I've gotten the habit of adding a default.nix-file to my projects, then I can run nix-shell or nix build to build the project. Sometimes, I have multiple targets though. Does anyone have a neat way to solve that? Maybe by adding multiple derivations to default.nix? Or having multiple .nix-files? Or a more "central" package store like an overlay to patch into?
<infinisil>
Maybe use this lower memory usage branch of nix I've heard about
<srhb>
infinisil: Isn't that just for constant space fetches?
<infinisil>
Oh right..
<srhb>
Not sure how much if at all that affects ofborg...
<infinisil>
Yeah nvm
<srhb>
Cheap to try I guess :)
<gchristensen>
I've shut down the evaluators since it is uselessly marking PRs as broken until we have an idea
<FRidh>
Do we know what parts contribute significantly to memory usage and evaluation time? I know NixOS module system is costly when evaluating a NixOS system. How is it with just evaluating Nixpkgs?
Shell has joined #nixos
<sphalerite>
does "all of nixpkgs" for ofborg include haskellPackages?
<gchristensen>
the eval which is failing for me is the part which lists all the out paths for all the packages across 4 archs
<gchristensen>
yes
<sphalerite>
Because it might just be a matter of sheer numbers
<sphalerite>
I wouldn't be surprised. There is a lot there
<gchristensen>
well also there are bugs in the GC and our fixed-point means nothing can go out of scope
<srhb>
That's why I'm wondering if we could split it up
<srhb>
Like, everything sans haskellPackages, then just haskellPackages
ckauhaus has quit [Remote host closed the connection]
wpcarro has quit [Ping timeout: 244 seconds]
<{^_^}>
[nixpkgs] @worldofpeace opened pull request #43732 → elementary-cmake-modules: add version → https://git.io/fNGyQ
hyper_ch2 has quit [Quit: Page closed]
asymmetric has joined #nixos
<srhb>
rizary: Are you using justStaticExecutables for your binary (re. servant/docker/nix)
<rizary>
srhb: not yet.
<srhb>
rizary: Try that :)
ckauhaus has joined #nixos
<gchristensen>
it looks like setting the initial heap to 10g works ok
<asymmetric>
i have a hard time understanding the difference between buildInputs, propagatedBuildInputs and nativeBuildInputs. in the nixpkgs manual it says that buildInputs is for build-time deps, and propagatedBuildInputs for run-time. but then it also says "build time only dependencies are declared in `nativeBuildInputs`"
<rizary>
srhb: okaaaay :)
<gchristensen>
I don't like it though :/
<sphalerite>
gchristensen: my guess is it doubles the heap size whenever it runs out of space, and it goes from 4 -> 8 -> 16 -> 32 (and 32 is too much) while 10 -> 20 works and is enough
<sphalerite>
or something like that
<gchristensen>
ouch maybe
<sphalerite>
if so, definitely just a stopgap
<srhb>
asymmetric: The details are in section 3.3, and are understood by only a select few.
aarvar has quit [Ping timeout: 240 seconds]
Ridout has quit [Ping timeout: 260 seconds]
<asymmetric>
yeah, got that impression by reading that section ;)
<srhb>
asymmetric: Yeah, sorry. :
__Sander__ has joined #nixos
<cocreature>
asymmetric: I think build-time for nativeBuildInputs only applies to tools like preprocessors whereas buildInputs is for libraries that you link against
ThatDocsLady_ has joined #nixos
<sphalerite>
asymmetric: as far as I understand it, depsFooBar are the dependencies that *run* on Foo and *generate stuff* that runs on Bar
<Shell>
I think nativeBuildInputs are e.g. executables you run during build, buildInputs are libraries your thing will be using once built, and propagatedBuildInputs is for stuff that needs to be in the path of things that depend on your thing.
<srhb>
:-)
<cocreature>
we really need to find someone who actually knows how this works and get them to document it :)
iqubic` has left #nixos ["ERC (IRC client for Emacs 25.3.1)"]
<srhb>
I share sphalerites understanding, fwiw.
<srhb>
Which I think succinctly encapsulates it, if a bit mind-bendingly...
<mightybyte>
Is there a way to try to fetch something from a cache without actually building it?
<srhb>
mightybyte: --max-jobs 0
<sphalerite>
if we had proper multiarch support in tools (*cough cough gas*) it would all be a lot simpler, we'd only have to care about what the tools run on
ThatDocsLady has quit [Ping timeout: 276 seconds]
<srhb>
mightybyte: Hm, unless you meant "only copy, don't realise"
<srhb>
nix copy will do the same thing, too..
nuncanada has joined #nixos
<mightybyte>
srhb: Oh, maybe that's what I mean
<mightybyte>
I just want to verify that a cache is working
<asymmetric>
Shell: thanks, that makes snese
<srhb>
mightybyte: I'm not sure such a thing actually exists outside of just curling. *thinks*
<mightybyte>
curling is a bit lower level than I want
<srhb>
mightybyte: Would copying to a different store work?
<mightybyte>
I want to verify that my nix.conf is correct
<asymmetric>
Shell, cocreature buildInputs is not just for libraries though, it's also for test binaries
<srhb>
mightybyte: If you don't care about a specific path, nix ping-store might do the trick?
<mightybyte>
What I'm really trying to do is set up an ssh cache.
<sphalerite>
mightybyte: try ping-store with a nonexistent one?
<FRidh>
asymmetric: propagatedBuildInputs is abused by Python packages for declaring runtime dependencies. In general, what it does is add the items defined there to the buildInputs of whatever package adds this package to *buildInputs.
<sphalerite>
ssh caches baaaasically don't work currently
<sphalerite>
AFAIK
<sphalerite>
they should in principle but don't in practice
<srhb>
sphalerite: What's broken about them? I believe I'm using one. It's brittle but works...
<asymmetric>
FRidh: that definition seems to have little to do with runtime dependencies. which consuses me even more :) so what's the non-abusive way to specify a runtime dep?
<sphalerite>
srhb: half the time they give you "Broken pipe"
<sphalerite>
actually more than half probably
<mightybyte>
sphalerite: Ok, so the non-existent one does fail. But the reason I'm doing this at all is because I built something that should have been cached, but it didn't seem to be.
<asymmetric>
FRidh: ok i see. and why do interpreted langs need this workaround?
<sphalerite>
angerman: you could use symlinkJoin or buildEnv
<sphalerite>
symlinkJoin is a bit simpler and probably sufficient for what you need
dramforever has joined #nixos
<mightybyte>
srhb: Hmmm, I'm having trouble with that.
<srhb>
mightybyte: Ok, I'm making all sorts of assumptions about the "shape" of your derivation :-)
<srhb>
mightybyte: Like that it takes `pkgs ? ...`
<infinisil>
,runtimeDeps
<{^_^}>
In order of preference: Patch source OR ((if it uses PATH -> wrap with new $PATH) AND (if it uses dlopen, (patchelf --set-rpath in postFixup OR wrap with new LD_LIBRARY_PATH)))
<infinisil>
This is actually for runtime deps, not the same as propagatedBuildInputs
<srhb>
mightybyte: Oh, you can just import that directly then
<srhb>
mightybyte: '(import ./.default.nix)'
<srhb>
Er...
<srhb>
./default.nix
<angerman>
sphalerite: let's see :-)
<FRidh>
asymmetric: how would you expect interpreted languages to find their dependencies? Consider Python, how should one package find another? Remember, we're not using FHS .
<mightybyte>
srhb: error: cannot import '/nix/store/dm1si4asap2hgi55vlqb36dixkwfy807-source', since path '/nix/store/6f032wpiqczqsnvinr0knxzc0a2yp3vi-source.drv' is not valid, at /Users/mightybyte/work/pact/default.nix:12:4
<srhb>
hmm, ifd shenanigans...
<asymmetric>
FRidh: the difference being that programs written in non-interpreted languages have their dependencies statically linked?
<srhb>
mightybyte: I don't know how to solve that except for instantiating the path first.
<srhb>
Which, with IFD's, can be pretty huge in itself.
<le_jonge>
clever: when i use `ghcid --command="runghc Setup.hs project-test" --test "Main.main"` then this correctly reloads and reruns tests as soon as i edit test files. i understand that if i want to have tests and the library watched, i would need to provide the repl command multiple targets. but it says that it does not support it. any advise?
ckauhaus has quit [Remote host closed the connection]
<mightybyte>
srhb: What's IFD?
<FRidh>
asymmetric: actually, there are dynamically linked but using absolute paths
<srhb>
mightybyte: It's when you're generating a derivation from a derivation itself.
<srhb>
mightybyte: import from derivation
<mightybyte>
Ahh
toby1851 has quit [Ping timeout: 245 seconds]
<cocreature>
le_jonge: you need to invoke ghci directly in that case. Cabal (used via Setup.hs) doesn’t support opening repls for multiple components
<mightybyte>
sphalerite: Your comment was to remove the whole ~/.cache/nix directory?
<le_jonge>
cocreature: so you mean `ghcid --command="ghci" --test "Main.main"` ?
<le_jonge>
sorry i ask stupid questions but i am a bit lost how this even works
<dramforever>
cocreature: It doesn't?
<cocreature>
dramforever: afaik no, do you have other information? I know that new-repl definitely doesn’t and afaik that limitation comes from the underlying repl command in Cabal
<dramforever>
uhh
<dramforever>
My 'other information' would be that 'stack repl' *does* support multiple targets
toby1851 has joined #nixos
<cocreature>
le_jonge: I generally add a .ghci file that sets up ghc options (your are probably going to need at least something like ":set -isrc") and has a ":l test/Main.hs" or something like that at the end
<srhb>
mightybyte: or just the binary-cache-* within. Don't think it matters much.
<cocreature>
in general, loading multiple components isn’t possible since they could require conflicting options. however, in practise stack’s approach of just merging flags works really well and you can replicate it using a .ghci file
<srhb>
sphalerite: It was --all on a very, very large store
<mightybyte>
I'm personally fine with --all
<mightybyte>
But it never seemed to work
<sphalerite>
oh
<sphalerite>
huh?
jperras has joined #nixos
<d1rewolf_>
guys, why does nixos-env -qa | grep network | grep manager show "network-manager", but I have to use networkmanager (without dash) in environment.systemPackages?
<dash>
because systemPackages wants an attribute name
<gchristensen>
d1rewolf_: use -qaP instead
<dash>
rather than whatever is in the "name" field in the derivation
<gchristensen>
copumpkin / contrapumpkin any chance you're around at all today?
<d1rewolf_>
gchristensen, -P is the same as --preserve-installed?
<gchristensen>
no
<gchristensen>
well yes, but not in the -q command
<gchristensen>
-P means --attr-path
<le_jonge>
cocreature: great, this works for me with the following lines in .ghci: ":set -isrc -itest" and ":load test/Spec.hs", but this would mean that i can't run `ghci` without running the tests now, right?
<d1rewolf_>
gchristensen, I see. thanks
jperras has quit [Ping timeout: 264 seconds]
<le_jonge>
cocreature: my bad. it does _not_ run the tests. i now use `ghcid --command="ghci" --test="Main.main"` and this works.
hxrts_ has joined #nixos
<copumpkin>
gchristensen: hi sorry yes am just on a trip and on mobile about to commute to work. Will check in on desktop later today, sorry for being hard to get!
<le_jonge>
cocreature: this is great, i love it. thank you very much for your help!
<gchristensen>
copumpkin: it is ok :) thank you a lot. I want to ask questions about one of your favorite issues.
npmccallum has joined #nixos
<d1rewolf_>
is anyone aware of a rufi package for nix?
<d1rewolf_>
manveru, ah...rofi does exist for nix! for some reason my head this morning I had rufi in my head. was searching for the wrong thing Need...coffee....
<d1rewolf_>
manveru, which bar are you using out of curiosity?
jtojnar has quit [Read error: Connection reset by peer]
graphene has quit [Remote host closed the connection]
periklis` has joined #nixos
<d1rewolf_>
manveru, if I wanted to use your dotfiles, home-manager is the way to do so?
graphene has joined #nixos
<d1rewolf_>
manveru, could you share your configuration.nix?
spear2 has quit [Quit: Leaving]
<manveru>
i'm using it via home-manager directly
<d1rewolf_>
manveru, ok, I'll need to read up on that. thx
<manveru>
and no, don't have configuration.nix public yet... would need some serious cleaning :P
<d1rewolf_>
manveru, ;)
<mightybyte>
sphalerite: Oooh, this is looking promising...
<d1rewolf_>
no worries. Your dotfiles are very helpful
periklis has quit [Ping timeout: 244 seconds]
<asymmetric>
any idea why this expression fails with `/nix/store/qjbvbd8yk7ywwn28wqz41myj6vvz5rv0-rofi-unwrapped-1.5.1/include/rofi/mode.h:30:10: fatal error: cairo.h: No such file or directory`
bmikedesk has quit [Remote host closed the connection]
<sphalerite>
asymmetric: if you add NIX_DEBUG=1 you can get the full compiler command line and see if it's passing any -Is that should be pointing it to cairo
jperras has joined #nixos
Piraty has left #nixos ["--"]
bmikedesk has joined #nixos
cryptomonoid has quit [Ping timeout: 264 seconds]
humanoyd has quit [Quit: WeeChat 2.2]
<asymmetric>
sphalerite: as in `NIX_DEBUG=1 nix-build -A rofi-calc`?
cryptomonoid has joined #nixos
<sphalerite>
asymmetric: no, put NIX_DEBUG = 1; in the derivation
ckauhaus has quit []
wucke13 has joined #nixos
<wucke13>
What do you use to prettify/lint/autoformat your nix expressions?
graphene has quit [Remote host closed the connection]
wucke13 has quit [Client Quit]
wucke13 has joined #nixos
<sphalerite>
wucke13: I don't currently use it much but hnix can do it
graphene has joined #nixos
Guest80558 has quit [Ping timeout: 256 seconds]
<wucke13>
Is there something already available in the nixpkgs which dies autoformat etc. ?
<d1rewolf_>
manveru, ;)
<d1rewolf_>
woops. sorry, delayed
<d1rewolf_>
nix-env -qaP gnome-settings-daemon returns "nixos.gnome3.gnome-settings-daemon gnome-settings-daemon-3.26.2", but nix-env -i gnome3.gnome-settings-daemon matches no derivations. what's the appropriate "name" for this to place both nix-env -i <here> and in systemPackages?
<d1rewolf_>
sphalerite, what about for placing it in configuration.nix?
<sphalerite>
d1rewolf_: systemPackages = with pkgs; [ gnome3.gnome-settings-daemon ]
<mightybyte>
sphalerite: Looks like that did it, thanks!
<mightybyte>
Also thanks srhb
<d1rewolf_>
sphalerite, ok, thank you. so most of the time, tools expect the attribute name
<srhb>
mightybyte: :)
<sphalerite>
d1rewolf_: yep
<sphalerite>
mightybyte: \o/
<sphalerite>
d1rewolf_: nix-env is the only thing that searches by name and it's a bit of a terrible behaviour which AFAIK only exists for backwards compatibility nowadays
<d1rewolf_>
alright, off to go do my second machine install in the last 12 hours. So far it's been an amazing experience. I'm so....so....happy ;) nixos is awesome
<adamt>
Hiya. Did somebody implement lib.strings.toInt in a less slow way? (4ms pr call, probably due to using builtins.fromJSON, kinda adds up)
wucke13 has quit [Quit: WeeChat 2.0]
<adamt>
(I'm actually calculating whether an IP address belongs to a CIDR, so if somebody implemented that instead it would be mighty nice.)
Guest80558 has joined #nixos
graphene has quit [Remote host closed the connection]
<ryantm>
adamt: try using builtins.match
graphene has joined #nixos
jtojnar has joined #nixos
<adamt>
ryantm: How? Can it transform the matches to a different type?
<adamt>
(i already use it to split the CIDR into strings)
<ryantm>
adamt: I haven't thought about it too carefully, but I think it ought to be possible to make a regular expression that matches strings of IP adddresses in a CIDR.
qwerty_asdf has joined #nixos
<adamt>
Sure, depending on the regex implementation anything goes. :P
init_6 has quit [Ping timeout: 260 seconds]
hxrts_ has quit [Ping timeout: 276 seconds]
srdqty has quit [Ping timeout: 260 seconds]
cacatoes has quit [Remote host closed the connection]
<nbathum>
hrmm. I built GHC 8.6.1 using alpmestan's ghc.nix... silly question, how do I install this?
srdqty has joined #nixos
<nbathum>
im in a nix-shell, but calling installPhase from my user can't create the nix store directory
Shell has left #nixos ["Be back later..."]
<nbathum>
can I carry my env over to sudo so I can put this into the store?
<srk>
is it possible to disable sandbox for nixops? or just for specific build/derivation?
<infinisil>
LnL: Probably gonna make this work via ,find instead
<bkchr[m]>
Anyone know when logind setups all required permissions?
<LnL>
infinisil: yeah that's probably better
srdqty has joined #nixos
<sphalerite>
bkchr[m]: when a user logs in on a console
magnetop` has quit [Ping timeout: 276 seconds]
alex`` has quit [Quit: WeeChat 2.1]
<bkchr[m]>
sphalerite: I'm trying to get plasma wayland working. I can start the session from my X11 session, but if I use it directly with SDDm, it does not work. I assume that maybe some permissions are not set correctly
<sphalerite>
bkchr[m]: hm interesting. I'm guessing logging in on a text console and starting it from there also works?
andi- has joined #nixos
cryptomonoid has quit [Ping timeout: 260 seconds]
<bkchr[m]>
sphalerite: no
<sphalerite>
huh. Sway worked that way last time I tried it
alex`` has joined #nixos
camsbury has joined #nixos
<sphalerite>
Do the logs suggest it's a permissions issue, or why do you think that's what it is?
dbmikus has joined #nixos
mounty has quit [Ping timeout: 268 seconds]
<worldofpeace>
bkchr[m]: ^
mounty has joined #nixos
<nbathum>
infinisil: thanks. apparently the file has 'nobuild', I think it is more meant for ghc development. I might go check in their channel
<das_j>
Hey, just a short question: How do I override a unit type? Just setting serviceConfig.Type results in the variable being redefined, but using mkForce results in "cannot coerce a set to a string"
<sphalerite>
das_j: hm. I'm not 100% sure but I think this might be because serviceConfig isn't a submodule, and you'd need to use mkForce on serviceConfig
<sphalerite>
which would obviously be quite a pain if there's other stuff predefined in it :/
<das_j>
sphalerite: Would this disable automatic submodule merging?
<infinisil>
das_j: I thought this should work, can you gist the nix file?
winem_ has quit [Remote host closed the connection]
wpcarro has quit [Remote host closed the connection]
<gchristensen>
"Why the heck do I still have /nix/store/hash-something ???" nix-store --gc --print-roots | awk -F '-> ' '{print $2}' | xargs -n1 -I{} nix why-depends {} /nix/store/hash-mypath
* infinisil
checks the type of this service module
lambdamu has joined #nixos
<infinisil>
das_j: Oh, yeah sphalerite is correct, it's not a submodule, and that's why mkForce doesn't work
<infinisil>
It's more like a `serviceConfig = mkForce { Type = "forking"; }` because of that
<infinisil>
Not sure how to make it work..
stepcut has quit [Remote host closed the connection]
<bkchr[m]>
sphalerite: The logs do not indicate something.. I looked into strace
<das_j>
well, I'll just copy the other few dict keys
<infinisil>
LnL: Well I said I wouldn't touch the bot for a bit yesterday, but I quickly changed the <path> syntax to ,find (as already suggested by gchristensen recently :))
cpc26 has joined #nixos
<infinisil>
,find
<{^_^}>
Use ,find to get a GitHub link to the master version of files in nixpkgs, e.g. ",find hello/default.nix" or ",find all-packages.nix"
<gchristensen>
nice help!
<srk>
infinisil: was there a PR with your znc changes?
<infinisil>
My PR only changes what's done when the config file is regenerated each time
FRidh has quit [Quit: Konversation terminated!]
<infinisil>
Nice, me and ldlework too I think are using this PR already successfully
<infinisil>
:)
xy2_ has joined #nixos
<das_j>
Hm, still doesn't work with the forking type. Does anyone know why systemd would ignore ExecStop? It just kills the processes
<infinisil>
das_j: "After the commands configured in this option are run, it is implied that the service is stopped, and any processes remaining for it are terminated according to the KillMode= setting."
<infinisil>
So maybe the command doesn't properly terminate it?
<infinisil>
This is from man systemd.service btw
<das_j>
infinisil: If I execute the command manually, it seems to work
<das_j>
That's why I'm asking
<infinisil>
Nothing in the logs?
asymmetric has quit [Ping timeout: 240 seconds]
<adamt>
das_j: Make the comand output something and it should end up in the journal. Then you can at least verify that it ran.
joshie has joined #nixos
<acowley>
What's the current best practice for dealing with QT_PLUGIN_PATH? i.e. an application works in KDE when installed with nix-env, but if run in nix-shell -p it can't find the xcb plugin.
<acowley>
I know this has been in flux for the past year
cpc26 has quit []
<LnL>
infinisil: ah, must be what I was thinking of
<das_j>
admin[m]4: If I type in an invalid command, systemd complains about it so I assume it works
<infinisil>
das_j: You have access to the main pid via $MAINPID
<infinisil>
Maybe check if that one exited in the stop script after executing the stop command
<das_j>
infinisil: I don't need the main PID. ExecStart starts tmux with the daemon in it and the stop command just types "exit" into that tmux session
<infinisil>
das_j: Hmm, maybe try a 1 second sleep or so, dunno :P
<das_j>
Oh I think I got it
<infinisil>
Maybe it's an async operation somewhere and needs a bit of time
<das_j>
Yes, tmux send-keys types the keys and then ExecStop is over and systemd kills the remaining processes
hxrts_ has joined #nixos
hxrts_ has quit [Remote host closed the connection]
cryptomonoid has joined #nixos
Mateon3 has joined #nixos
Mateon1 has quit [Ping timeout: 248 seconds]
Mateon3 is now known as Mateon1
<infinisil>
das_j: And??
<das_j>
Still testing
<das_j>
But `${getBin pkgs.coreutils}/bin/tail --pid=$MAINPID -f /dev/null` is what I'm currently using to wait
<das_j>
Yep, it works
johanot has quit [Quit: Leaving.]
<infinisil>
das_j: Ah, so it was asynchronous so some extend
<infinisil>
das_j: That is a fancy way to wait for a pid to die, didn't know of that :o
<das_j>
Just found it on SO and I love it
<das_j>
btw, what is getBin for? Why not just use pkgs.coreutils?
Tucky has quit [Quit: WeeChat 2.2]
<sphalerite>
for multiple-output packages
<das_j>
oh, makes sense
<infinisil>
,locate joker
<sphalerite>
coreutils.bin will only work on a multiple-output package with the bin output, whereas getBin checks for the output's presence and just returns the derivation as is if it doesn't exist
<{^_^}>
Found in packages: joker.bin, profanity
<infinisil>
joker has its binary in the bin output as an example
wpcarro has joined #nixos
<das_j>
That's nice, I have never worked with multiple output derivations before
d1rewolf_ has joined #nixos
<d1rewolf_>
Hi guys. Graphical iso written to USB with dd
<d1rewolf_>
On Dell 5520 laptop, immediately boots into Dell support assistant and then powers
<d1rewolf_>
Off. Any ideas?
<d1rewolf_>
No errors from dd. Filesystem looks right
<sphalerite>
d1rewolf_: try checking the firmware settings for boot options
<d1rewolf_>
In Dell bios?
<sphalerite>
secure boot might be preventing it, or fast boot or something like that
<sphalerite>
yes
<d1rewolf_>
Ok, thx will look after this meeting
baconicsynergy has joined #nixos
<d1rewolf_>
Appreciate it. Is there a way to validate a USB drive after dd'ing?
baconicsynergy_ has joined #nixos
cryptomonoid has quit [Ping timeout: 260 seconds]
<d1rewolf_>
To ensure it's burnt properly
<d1rewolf_>
?
wpcarro has quit [Ping timeout: 245 seconds]
<d1rewolf_>
Er... written properly?
<ToxicFrog>
d1rewolf_: dd it back into an image file, compare it to the original image you dd'd to it
rui has joined #nixos
<sphalerite>
you could do something like head -c $(stat -c %s image.iso) /dev/sdX | sha256sum
<sphalerite>
and compare it to the image's sha256sum
<d1rewolf_>
Toxic frog: that'd dd the whole drive (64gb)
<ToxicFrog>
d1rewolf_: I'm assuming proper use of the bs= and count= options
graphene has quit [Remote host closed the connection]
Swant is now known as birthdayhat
graphene has joined #nixos
<srk>
rui: is pulseaudio running? if so it probably takes sound card for itself
<rui>
srk: It probably is. Let me see what happens if I stop it
<srk>
you can do 'pulseaudio -k'
<srk>
press 'Start' in qjackctl
<srk>
need to be quick as pulse restarts automatically
<sphalerite>
to avoid that you can put `autospawn = no` in ~/.config/pulse/client.conf
<srk>
good to know
<rui>
srk: no, it still doesn't work
d1rewolf_ has quit [Ping timeout: 252 seconds]
<srk>
(I'm not using any scripts in qjackctl, pulse automatically connects to jack after I do this)
<rui>
and I know I was fast enough because rerunning `pulseaudio -k` tells me there's no such daemon
<srk>
you can run `pavucontrol` as well to see the state of pulseaudio
<rui>
srk: Yes, when I have pavucontrol open, pulseaudio restarts automatically
<srk>
it does anyway unless you set what sphalerite mentioned
asymmetric has joined #nixos
<rui>
if it does, it's way slower
__Sander__ has quit [Quit: Konversation terminated!]
<srk>
anyway you want 'Jack sink' listed in outputs not the soundcard directly (after jack starts..)
<rui>
srk: maybe I screwed up some configuration. On that link with the instructions I followed, it's supposed to leave stuff set up so that Jack and Pulseaudio can live happily together.
<rui>
I should put it all back to default somehow
<wpcarro>
Has anyone published any books on Nix that might be worth reading?
nD5Xjz has joined #nixos
<srk>
rui: right, bit more complex setup but it shouldn't matter - it just tells pulseaudio to unhook the soundcard and hook to jack instead afaik - but you can do that manually at first to see if jack even works
<srk>
not even, just loads jack modules
<srk>
I remeber something like that.. not sure how to prevent pulseaudio to take soundcard for itself automatically
<sphalerite>
wpcarro: Eelco Dolstra wrote a very nice PhD thesis on it ;)
<sphalerite>
wpcarro: other than that I'm not aware of any long-form material on nix
<wpcarro>
sphalerite: good call... I'll check that out
<sphalerite>
but the thesis is definitely worth reading
<hodapp>
uhm... so I built Boost without setting enablePython = true, and I'm now seeing that by default it's false, yet somehow I did a build that required Boost with Python bindings...
<hodapp>
I am very confused.
<hodapp>
in a "how did this build ever work?" sort of way
<joepie91>
ahh, "it works and I have no idea why", the absolute worst of possible software states :)
<hodapp>
when Boost is involved usually the most I can offer is "I beat on it for days until the template errors stopped"
<sphalerite>
Is it at all possible to have swap on an nbd device without deadlock/thrashing making the system unusable?
<hodapp>
nbd?
<sphalerite>
(and with the swap actually in use)
<sphalerite>
hodapp: network block device
<hodapp>
hmm, well, I've had swap on really slow SD cards, had the system in use, and had it not deadlocked/thrashing
<hodapp>
so a faster network block device should be okay
<sphalerite>
well the issue is that nbd involves a program in userspace
<srk>
sphalerite: think I've had these as swapfiles on nbd dev
<srk>
can't remember
<sphalerite>
that program being swapped out is undesirable for obvious reasons, and I don't know how to avoid that happening
toby1851 has quit [Ping timeout: 256 seconds]
<hodapp>
oh, I thought it was referring to something kernel-level, didn't realize
<joepie91>
I feel like latency is going to be a problem there
worldofpeace has joined #nixos
worldofpeace has quit [Client Quit]
<srk>
nbd is kernel level, with few userspace utils
<hodapp>
I thought swap was usually dealt with in large enough blocks that latency usually wasn't the problem
worldofpeace has joined #nixos
<sphalerit>
Actually I'm XYing here aren't I? The reason I want to do this is that ghc won't build in 2GiB of RAM (on a scaleway C1 ARM server)
<sphalerit>
And scaleway servers' storage is only accessed via the network
<hodapp>
just try it and see, perhaps
<srk>
create a swapfile then, shouldn't be any different to running nbd
<sphalerite>
srk: yes that's what I was going to do, shouldn't make any difference from just using the nbd directly AFAIU
<sphalerite>
also, fun RAM-eating tool: `sort /dev/zero`
<sphalerite>
anyway, what I've found tends to happen (even with no swap) is that eating up the RAM on these machines makes them completely unresponsive
<infinisil>
Oh no, this freezed my pc! Damn you
periklis` has quit [Ping timeout: 244 seconds]
<sphalerite>
no ping, no SSH, nothing
<sphalerite>
waiting a couple days also doesn't seem to help
selaux has quit [Quit: Leaving]
<infinisil>
I guess this is because of the zfs swap
<samueldr>
infinisil: I was scared to let it fill all the ram, but wow, was it fast
<sphalerite>
you have zfs swap?
<sphalerite>
samueldr: :D
<samueldr>
in what, under 5 seconds, it filled 47GB
<samueldr>
(I don't know if the OOM would have hit other things before)
<sphalerite>
yes, /dev/zero is quick to read from :D
<infinisil>
sphalerite: yeah, i should turn it off because of the freezes, but if i turn it off i often can't rebuild certain things :(
<sphalerite>
infinisil: how about swap that's not on zfs?
qwerty_asdf has quit [Quit: WeeChat 2.0]
sigmundv has quit [Ping timeout: 276 seconds]
<infinisil>
Would be quite annoying to repartition
<infinisil>
Although, that should work
<samueldr>
more annoying than freezy builds?
<infinisil>
Heh good point
<sphalerite>
or you could use a USB stick for swap :p
wpcarro has quit [Remote host closed the connection]
<infinisil>
Actually my sdd is only used as a zfs cache device, so i don't even have to move any data
jmeredith has quit [Quit: Connection closed for inactivity]
<sphalerite>
oh yeah, you might also want to look into zram to squeeze more out of the RAM you have
<patrl>
since it belongs in nodePackages, I ran generate.sh to rebuild the nix expressions using node2nix, but this of course results in a massive merge conflict
<patrl>
otherwise it's a trivial PR
<patrl>
@tilpner ah ok, good to know
habnabit is now known as _habnabit
igo95862 has joined #nixos
cryptomonoid has quit [Ping timeout: 240 seconds]
puck is now known as pucc
goibhniu has quit [Ping timeout: 264 seconds]
hhes has quit [Ping timeout: 276 seconds]
ixxie has joined #nixos
philippD has joined #nixos
ntqz has joined #nixos
ntqz has quit [Read error: Connection reset by peer]
joshie has quit [Remote host closed the connection]
joshie has joined #nixos
wpcarro has joined #nixos
patrl has quit [Quit: WeeChat 2.0]
<philippD>
Is there a way to tell nix to ignore version restrictions in Cabal files for Haskell packages?
ThatDocsLady_ has quit [Ping timeout: 268 seconds]
<pstn>
Can somebody explain to me why this configuration leads the resulting wg unit to fail when it tries to add the routes?
wpcarro has quit [Ping timeout: 245 seconds]
<pstn>
Failing ip command: `ip route replace 10.23.42.1/24 dev wireguard-xndr table main`
milanos has left #nixos [#nixos]
reinzelmann has quit [Quit: Leaving]
<srk>
try putting sleep 1 just before it
<srk>
something similar happened to me few weeks ago iirc
patrl has joined #nixos
TFTIO has joined #nixos
<andi->
pstn: the networking code is async. It does't guarantee that teh device exists when your call returns. It might still not be created at that point. The sleep sounds like a practiacal solution.. It is ugly thought.
<srk>
I've recently added bunch of patches to rtnetlink-hs (haskell netlink implementation) to be able to monitor devices/addresses/routes https://github.com/sorki/network-status/
<srk>
it basically reconstructs the kernel state based on netlink messages
<srk>
I might add a way to change stuff as well
graphene has quit [Remote host closed the connection]
<srk>
this could address a lot of shortcomings of ip when matured
graphene has joined #nixos
mounty has quit [Read error: Connection reset by peer]
mounty has joined #nixos
milanos has joined #nixos
<milanos>
Hey, I need some help with sudo permissions
asymmetric has quit [Remote host closed the connection]
<milanos>
I want to do the equivelent of creating a new file in suoders.d called 99_alcove and add give it sudo permission. I also want to add this Defaults!/path !requiretty to it
<milanos>
joepie91: Ok, I see how to give it sudo permission. How would I execure the Defaults! /path !repuiretty part?
rihards has joined #nixos
<joepie91>
milanos: not sure about that part
<milanos>
joepie91: Well thanks for the help :)
drakonis has quit [Remote host closed the connection]
<joepie91>
milanos: it's worth googling for something like "sudo.extraRules defaults" or so, btw :) lots of people publish their system configs
<milanos>
joepie91: Can't find it. Do you think its possible to manually create a sudoers.d and just exec it as normal
cryptomonoid has joined #nixos
joko has quit [Ping timeout: 240 seconds]
orbisvicis has joined #nixos
joko has joined #nixos
pie_ has quit [Ping timeout: 244 seconds]
<milanos>
joepie91: Incase you are curious, its under security.sudo.extraConfig
<milanos>
joepie91: Thanks for pointing it out
cryptomonoid has quit [Ping timeout: 245 seconds]
wpcarro has joined #nixos
orbisvicis has left #nixos [#nixos]
<mightybyte>
Anyone know what causes "nixos-rebuild switch" to fail with the following error?
<mightybyte>
error: opening file '[': No such file or directory
<milanos>
mightybyte: did you load in any other files
<srk>
milanos: you can also look up this stuff with `man configuration.nix`, sometimes the fastest way if you're not sure about some stuff is to look at the module implementation in nixpkgs (git grep ftw)
<reygoch>
Hi, I'm having trouble seting up croatian keyboard layout in i18n module. Using "loadkeys croat" does the trick, but if I put "hr" or "croat" in i18n.consoleKeyMap than the build fails. "localectl list-keymaps" reports "Failed to read list of keymaps: No such file or directory".
<srk>
d1rewolf_: do you have boot.loader.grub.efiSupport enabled?
<d1rewolf_>
pastebin appears blocked for me at the moment
<reygoch>
Anyone else had such problems?
<d1rewolf_>
srk: I do not
<d1rewolf_>
srk: I'll try that
<srk>
d1rewolf_: gpt most probably requires efi
<d1rewolf_>
srk: no, same error message
<d1rewolf_>
but followed by
<d1rewolf_>
grub-2.02/sbin/grub-install: error: embedding is not possible, but this is required for RAID and LVM install.
<srk>
ooh
<d1rewolf_>
I have lvm on top of luks
<d1rewolf_>
which worked ok on my other laptop
<d1rewolf_>
not sure why this one is giving me heartache
<milanos>
d1rewolf_: Do you have boot.loader.efi.canTouch.EfiVariables enabled?
<samueldr>
wow, I wish grub's messages were more verbose :(
<srk>
++
<samueldr>
looking at the source, there are three conditions for that error
<d1rewolf_>
milanos: I don't. i'll try it
<samueldr>
if at least it was one error message per condition, the exact cause would be easier to know
<d1rewolf_>
milanos: error: The option `boot.loader.efi.canTouch' defined in `/mnt/etc/nixos/hardware-configuration.nix' does not exist.
<{^_^}>
[nixpkgs] @timokau opened pull request #43755 → sage: use proper overlays → https://git.io/fNZgh
<pstn>
Okay so I found out what's wrong with my config now. The route creation fails when the interface IP is in the subnet that is allowed to connect.
<acowley>
I was looking at the current hydra evaluation error (https://hydra.nixos.org/jobset/nixos/trunk-combined#tabs-errors). It mentions an error when wanting to expand the heap by 10GB, and so I'm wondering, is RAM usage good given how big nixpkgs is, or is this something that could do with attention (beyond installing more RAM)?
<acowley>
Every time I've looked at it, the nix C++ code has struck me as pretty clean, so it's not the worst thing to dig into.
<manveru>
i wonder how hard it'd be to implement nix in other languages :)
<manveru>
i mean, just running derivations seems straightforward
toby1851 has quit [Ping timeout: 256 seconds]
xy2_ has quit [Ping timeout: 256 seconds]
<johnw>
manveru: like, say, Haskell maybe?
<manveru>
not sure that's a good idea...
cryptomonoid has joined #nixos
<sphalerite>
acowley: there was some talk about that earlier
<acowley>
johnw: Have you looked at all at how hnix's memory use compares?
<johnw>
not head to head, but it's not far off from what I've seen
bmikedesk has quit [Remote host closed the connection]
<acowley>
That PR with an 18% savings seems pretty great. I guess it's clear, though, that some amount of segmentation will be required. Something like: chunks of the nixpkgs graph are serialized and swapped out of RAM until needed.
<{^_^}>
[nixpkgs] @xeji pushed commit from @volth to master « xfce4-13.xfce4-screenshooter: 1.9.1 -> 1.9.2 (#43625) »: https://git.io/fNZVS
lonokhov has joined #nixos
udkyo has joined #nixos
d1rewolf_ has joined #nixos
<d1rewolf_>
guys, I can't get past this error: /nix/store/hl2ljrggl5whilzlp39mgd4w1kpjp178-grub-2.02/sbin/grub-install: error: embedding is not possible, but this is required for RAID and LVM install.
<{^_^}>
[nixpkgs] @xeji pushed commit from @oxij to master « libcardiacarrest: 11.0-6 -> 12.1-6 (#43697) »: https://git.io/fNZVF
stepcut has quit [Remote host closed the connection]
<d1rewolf_>
guys, I'm having problem with grub-install during the nixos-install process (even with EFI parition) and need to see exactly what the grub-install command is. how would I go about doing it?
npmccallum has quit [Quit: npmccallum]
hakujin has joined #nixos
alex`` has quit [Read error: Connection reset by peer]
<d1rewolf_>
and sphalerite, that's after creating the UEFI partition :(
pie_ has joined #nixos
<hakujin>
hi, I'm getting a strange error when evaluating a nixos module with a few options that are `type = lib.types.bool`: `The option value $OPTION in $FILE is not of type `null or string or path or package'
<samueldr>
hakujin: have you figured out which option it could be?
<infinisil>
hakujin: Well what's your nixos module? The error is certainly there
<hakujin>
samueldr: infinisil: let me sanitize and gist
<d1rewolf_>
samueldr: warning: Attempting to install GRUB to a disk with multiple partition labels. This is not supported yet.
igo95862 has joined #nixos
<d1rewolf_>
error: embedding is not possible, but this is required for cross-disk install. /nix/store/plnanfcgak2dcsrq8k7ci18am9xkpdxy-install-grub.pl: installation of GRUB on /dev/sda failed
doyougnu has quit [Quit: WeeChat 2.0]
<sphalerite>
d1rewolf_: huh. it should only have the GPT partitiob label
<d1rewolf_>
sphalerite: I'm talking in #grub and they said the only way to isolate the problem is to get the actual grub-install command line :-/
<samueldr>
d1rewolf_: just to be sure, your machine is EFI, right?
<samueldr>
(sorry if you were asked and answered before)
<d1rewolf_>
I would look at /nix/store/plnanfcgak2dcsrq8k7ci18am9xkpdxy-install-grub.pl, but it's gone by the time the error happens
<d1rewolf_>
samueldr: yes. It's a dell 5520 laptop
<d1rewolf_>
with efi
<d1rewolf_>
nixos-install --show-trace gives me nothing
stepcut has joined #nixos
wchresta has joined #nixos
<sphalerite>
d1rewolf_: it shouldn't really be trying to install on the device directly since this is EFI
<samueldr>
good, since it's grub, boot.loader.grub.device is set to "nodev" right d1rewolf_?
<infinisil>
You blended out the most important part.. ($OPTION)
<d1rewolf_>
samueldr: it wasn't set to nodev. stand by
<samueldr>
hakujin, infinisil, looks like types.bool can't be set in environment's attrset
<infinisil>
hakujin: Oh, but I see the problem
<hakujin>
infinisil: The option value `systemd.services.service-name.environment.REQUIRE_SSL' in `/Users/hakujin/src/nix/module.nix' is not of type `null or string or path or package'.
<d1rewolf_>
samueldr: bingo! ;-) lingering config item from another machine
<infinisil>
samueldr: Indeed, and I was the one who made that change :)
<hakujin>
ah!
<d1rewolf_>
fingers crossed...here goes a reboot
<samueldr>
infinisil: was about to link the commit
<samueldr>
hakujin: you probably need to *somehow* convert the bool into a string
<hakujin>
so I need to use `toString` on the bool, yeah?
<hakujin>
gotcha
<hakujin>
makes sense. thanks!
<infinisil>
> toString true
<{^_^}>
"1"
<hakujin>
oh, hmm. lol
<infinisil>
hakujin: You sure you want that?
<samueldr>
I say somehow, because it depends how your thin expects the value to be
<infinisil>
> lib.boolToStr true
<{^_^}>
attribute 'boolToStr' missing, at (string):160:1
<infinisil>
> lib.boolToString true
<{^_^}>
"true"
<infinisil>
hakujin: ^^
<hakujin>
aha
<hakujin>
infinisil: much appreciated
<infinisil>
:)
<infinisil>
I guess previously it used builtins.toJSON
<infinisil>
> builtins.toJSON true
<{^_^}>
"true"
<infinisil>
Which would've worked, but I think it's better like this
<samueldr>
toJSON is such a dirty trick :)
<samueldr>
(though it worked)
<infinisil>
Relevant pr is #38356, which still would need a fix for the other occurence..
<{^_^}>
[nixpkgs] @xeji pushed commit from @r-ryantm to master « podofo: 0.9.5 -> 0.9.6 (#43744) »: https://git.io/fNZMd
rihards has quit [Quit: rihards]
stepcut has joined #nixos
hakujin has joined #nixos
crmlt has joined #nixos
crmlt has quit [Remote host closed the connection]
<nh2>
dtz: I finally figured out one of the remaining big problems with static-haskell-nix. The problem is where you pass `--ghc-option=-optl=-static`. It works for many use cases but not for some (e.g. stack or intray-web-server). This is because you intend to pass that flag only to the final gcc/ld invocation that links the exe.
<nh2>
But `--ghc-option=-optl=-static` will tell GHC to use it for *any* linker invocation, even those for intermediate files like when TemplateHaskell is used. The solution is to pass that flag only in `ghc-options` of the `executable` section in the cabal file instead of as a general GHC option.
<nh2>
To do this conveniently from nix, we would need a cabal configure-time flag that does this, e.g. named `--enable-static-executables`.
pie_ has quit [Ping timeout: 256 seconds]
rtjure has joined #nixos
cryptomonoid has joined #nixos
stepcut has quit [Remote host closed the connection]
<Guest80558>
hello! I'm facing this error: `Nix daemon out of memory`. Does anybody know how find out which file is causing it? I read here (https://github.com/NixOS/nix/issues/418) to `strace on nix-instantiate --dry-run ...` but I don't know exactly what to do there
<{^_^}>
nix#418 (by domenkozar, closed): Nix runs out of memory
<samueldr>
Guest80558: doing what exactly does it run out of memory?
igo95862 has quit [Remote host closed the connection]
<infinisil>
Is there a practical difference between openjdk or oraclejdk?
simukis has quit [Ping timeout: 268 seconds]
<maurer>
Yes
<maurer>
well, their JREs at least
<maurer>
I don't know about the compilers they use
igo95862 has joined #nixos
<infinisil>
How are their jre's different?
<maurer>
There are at least some commercial java things (mostly config tools) I have that only run correctly on oraclejdk, though I don't know the why of it
<samueldr>
isn't that an historical note or is there still a difference?
<samueldr>
I seem to recall that oracle's is now openjdk, but with support?
<maurer>
There was a difference in the JREs as of a month ago
<samueldr>
good to know
<samueldr>
(I wasn't contradicting)
<maurer>
I don't know the details of why, only that I have software that NPEs on one but not the other
Guest80558 has joined #nixos
<maurer>
"Apart from the base code, Oracle JDK includes, Oracle’s implementation of Java Plugin and Java WebStart. Also includes third party closed source and open source components like graphics rasterizer and Rhino respectively." I'll bet this has something to do with it
<adelbertc>
so im hitting this issue because our K8s cluster sets a security policy to run containers under non-root which makes sense and afaik is common
<adelbertc>
but if i run containers as non-root of course i cannot access any files provided by Nix because tehy are chown'd by 0:0 per that line (afaict)
<{^_^}>
[nixpkgs] @dotlambda opened pull request #43770 → bittornado: init at unstable-2018-02-09 → https://git.io/fNZQd
mizu_no_oto has quit [Quit: Computer has gone to sleep.]
<sphalerite>
that makes sense, so yeah maybe in your case you'd want to change it as you said
<sphalerite>
I'm not sure if that applies in the general case
<sphalerite>
might be worth opening an issue about it
wpcarro has joined #nixos
<infinisil>
OHh..
<infinisil>
The launcher downloads its own libraries into ~/.minecraft ..
<infinisil>
Is that bad?
<sphalerite>
I think so
<infinisil>
What could happen?
<sphalerite>
does it do that even if they're available?
<infinisil>
I'll check that
stepcut has joined #nixos
<sphalerite>
I mean, it feels bad. It's probably perfectly normal and does work by the looks of it.
<infinisil>
I doubt it's gonna work when I use that
pie__ has quit [Remote host closed the connection]
<sphalerite>
oh or libpressureaudio
pie__ has joined #nixos
<infinisil>
Gonna try that one too then
stepcut has quit [Remote host closed the connection]
Ridout has joined #nixos
<infinisil>
Yeah, pulseaudio is needed
wchresta has quit [Ping timeout: 255 seconds]
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<sphalerite>
oh. "libpulse implementation that unconditionally fails to connect to the PulseAudio daemon and does nothing else"
<sphalerite>
(cardiacarrest0
<infinisil>
Yeah, makes sense to not work
shachaf_ has joined #nixos
<adelbertc>
aha i found out a possible reason - i was using it with name binding like `args@{ uid ? 1000, ... }` and then referencing `args` but learned that default arguments aren't included if theyr'e not explicitly passed in
Guanin has joined #nixos
shachaf has quit [Disconnected by services]
<adelbertc>
so `args` didnt include `uid` which still made it hit the upstream default of `0`
leo60228 has quit [Ping timeout: 252 seconds]
<sphalerite>
oh yeah that's an annoying one
shachaf_ is now known as shachaf
ajs124 has left #nixos ["Stream closed by us: Timed out waiting for stream resumption (connection-timeout)"]
Guest80558 has quit [Ping timeout: 244 seconds]
<sphalerite>
is there an alternative to nix-copy-closure for copying .drvs around yet?
<sphalerite>
oh I could use nix-instantiate with --store I guess
<sphalerite>
which isn't quite like copying .drvs but close enough
tertle||eltret has joined #nixos
<wpcarro>
I'm considering converting my dotfiles into a nix expression, but I'm wondering if this is something that Nix is good for? My gut says yes, but before I undertake everything, I wanted to consult the experts first. Trying to support things like alias and function defns, zsh settings, brew installs, etc
ajs124 has joined #nixos
<sphalerite>
wpcarro: brew installs probably not, since package management is something nix does rather well itself
<wpcarro>
sphalerite: right of course. I guess I should've clarified. Things that I currently am installing with brew I'd like to move to Nix
<wpcarro>
But I was wondering about aliases and function definitions...
<sphalerite>
wpcarro: as for shell config (aliases, functions, zsh settings), that's something you *can* build with nix but which may not make much sense to.
<sphalerite>
there is home-manager though which puts stuff together with nix and symlinks it all in as well
<sphalerite>
do we have a factoid for that?
<sphalerite>
,home-manager
<{^_^}>
home-manager allows you to manage your per-user configuration similarly to NixOS' configuration.nix: https://nixos.wiki/wiki/Home_Manager
<sphalerite>
we do!
<wpcarro>
interesting...
<wpcarro>
I've seen people Dockerize their configuration, so I was reasoning by analogy that perhaps Nix could be used to accomplish the same thing.
<{^_^}>
[nixpkgs] @xeji pushed commit from @volth to master « xfce4-13.parole: 0.9.2 -> 1.0.1 (#43765) »: https://git.io/fNZFl
<adelbertc>
i have a Nix overlay that makes it easy for me to install a bunch of packages in one fgo, like `nix-env -iA nixpkgs.personal.haskellTools` and then i check that into a git repo
<adelbertc>
same repo with the corresponding dotfiles
<adelbertc>
that eventually get symlink'd (with `stow`)
<adelbertc>
still manual but works well enough
<wpcarro>
adelbertc: haven't heard of stow... will look into
<wpcarro>
manveru: thanks for sharing
<wpcarro>
sphalerite: I will happily use any existing Nix expression that will work for my current make and model of my MBP... I just don't trust myself to write one without breaking my setup
<adelbertc>
its basically symlinking that accounts for corner cases like if folders already exists or if they dont exist, etc
sam__ has joined #nixos
sam__ has quit [Read error: Connection reset by peer]
<sphalerite>
wpcarro: there aren't usually many model-specific quirks that need to be taken into account in the config. I can recommend trying, you could always just install to an external USB HDD to try it out if you don't want to touch the OSX installation yet :)
<sphalerite>
(or a USB flash drive)
<wpcarro>
sphalerite: You're inspiring me...
<wpcarro>
sphalerite: this is dangerous rhetoric! Do you have an existing Nix expression anywhere that I can fork?
<sphalerite>
or try it in a VM to see how wonderful it is without having to worry about hardware-specifics yet
<sphalerite>
nixos-generate-config, which is part of the standard install process, genereates a config for you with lots of comments and suggestions in it
<sphalerite>
I'd recommend starting from that
fresheyeball has joined #nixos
<fresheyeball>
Is there a way I can set an env var in callCabal2nix?
lonokhov has quit [Ping timeout: 245 seconds]
dbmikus has quit [Quit: WeeChat 2.2]
{`-`} has joined #nixos
<adelbertc>
fresheyeball: in general if you just add an attribute it will be available in bash
<fresheyeball>
adelbertc: what do you mean by that?
<fresheyeball>
callCabal2nix ./. {}; <- how do I add an attribute?
ambro718 has quit [Quit: Konversation terminated!]
<kalbasit>
is there a Hydra as a service out there that I can use to test out my dotfiles (include machine builds)?
<adelbertc>
gah, so i also have an entrypoint script for the Docker image that calls programs by just interpolating the Nix attribute, e.g. `${tini}/bin/tini -s -- "$@"` but are not part of the `contents` explicitly so those are still owned by `root` in the image
<adelbertc>
which doesn't work if you run the container as non-root
<adelbertc>
D:
<gchristensen>
:(
wpcarro has quit [Remote host closed the connection]
<adelbertc>
hello gchristensen :)
<gchristensen>
hello adelbertc, long time no chat
wpcarro has joined #nixos
<LnL>
you can run the daemon inside a container :)
<sphalerite>
kalbasit: not that I know of. Would probably be overkill anyway though, is there any reason you can't just build it on your own machine?
<adelbertc>
oof
<adelbertc>
i see two ways forward.. one is to just carefully craft everything so i call commands directly and then make sure the appropriate packages are in `contents`
<adelbertc>
another is to try to `chmod` the whole `/nix` in `runAsRoot`.. i dont even know if that works actually
<adelbertc>
since depending on when `runAsRoot` is run it may not see all the things dumped in there
<kalbasit>
sphalerite: no reason, just lazy :p I'll have to do that once my laptop can take over my desktop's daily tasks then my desktop will become my server :)
<LnL>
adelbertc: you'll want to do that in the first layer since you might get 2 large layers otherwise
sam__ has quit [Ping timeout: 255 seconds]
<adelbertc>
"that" being the latter option?
<LnL>
adelbertc: and doing stuff afterwards with docker build might get tricky (if that's something you need to do)
<adelbertc>
id like it to be as Docker friendly as possible
<adelbertc>
i guess that points to the first option...
<LnL>
yeah
<adelbertc>
its effectively how Dockerfiles work anyways.. not that thats a good reason
<adelbertc>
😭
<LnL>
btw, I think I added an option to do chown like stuff without runAsRoot
tzemanovic has joined #nixos
<LnL>
since it's just the owner of files in the layer tarball
<adelbertc>
oh?
<adelbertc>
is that the `uid` and `gid` params?
<worldofpeace_>
kalbasit: it seems that sometimes the installer tests fail for a variety of reasons probably, I think it should pass next time though
<adelbertc>
or are you talking about something else
<LnL>
adelbertc: yeah those
<kalbasit>
worldofpeace_: I've noticed yea. I think it's flaky tests. Maybe consider running failed tests more than once? Bazel does that to identify flaky tests
<adelbertc>
yeah so im using those, i think that just `chowns` the top-level stuff, but not stuff inside `/nix/store`
<worldofpeace_>
kalbasit: I think it's just I/O :)
<LnL>
it doesn't chown anything, that's why you don't need qemu
<adelbertc>
"top-level" meaning the thing `dockerTools` does when it "flattens" all the Nix contents so `/bin`, `/lib` etc are mounted on `/`
<adelbertc>
right
<kalbasit>
worldofpeace_: oh I see, still it would help keep it moving forward :)
pie__ has quit [Remote host closed the connection]
<adelbertc>
it does in my case because i want to be able to call an executable in `/nix/store`
doyougnu has joined #nixos
<adelbertc>
or otherwise reference it (e.g. `PYTHONPATH` including packages in the Nix store)
<adelbertc>
while not running as root (Kubernetes security policy running non-root containers)
<adelbertc>
the `PYTHONPATH` use case kind of kills my first idea to carefully craft the Nix contents
<worldofpeace_>
kalbasit: It would help if the trunk-combined tests got a bigger share too (cuz QA) but I think that's up to some other people here maybe
hamishmack has joined #nixos
<adelbertc>
i can try to recursively `chmod` with `runAsRoot` then..
<LnL>
adelbertc: I remember now, I added that for purity before buildImage would leak the build user ids into the layers
<adelbertc>
ahh
<LnL>
alltho a side effect was supposed to be that you could configure the owner of all paths
<elvishjerricco>
samueldr: I think it might only occur when I log out, but not every time.
<samueldr>
hmmm, I never log out
<elvishjerricco>
Ah, well there goes that theory probably :P
<samueldr>
though, *something* could have happened that acts like when you logout
<samueldr>
dunno, a rebuild, then some service restarts, then bluetooth drops, then I try reconnecting and "???"
cryptomonoid has quit [Ping timeout: 240 seconds]
tenten8401 has joined #nixos
<elvishjerricco>
When I log out, something on my system disconnects my headset. Usually, I'll hear the disconnection tone on the headset. When I log out with them connected and don't hear that tone, that's when I stop being able to connect (after logging back in of course)
<{^_^}>
haskell/cabal#5446 (by nh2, open): Add `--enable-executable-static` flag for fully static executables
<elvishjerricco>
samueldr: Maybe something from my session is hanging somehow and causing the connection to never actually disconnect?
<samueldr>
I need to remember to login as root on a VT to check on logout if it's reproducible
<samueldr>
(e.g. if I don't hear a disconnect)
<tenten8401>
Does anyone here develop in Qt5 with NixOS? I'm currently trying to get started with a Rust Qt5 binding, but after installing qt5.full and extra-cmake-modules, I'm getting an issue that says "The imported target "Qt5::Gui" references the file /nix/store/whatever/libqicns.so, but this file does not exist"
<tenten8401>
and it definently doesn't exist
<samueldr>
tenten8401: what's the store path with hash?
<samueldr>
sorry if I got your hopes up, no idea, there are other image format plugins in qtbase, and otherwise that lib is available in `qt5.qtimageformats`
<tenten8401>
I'm just running the hello world example
wpcarro has quit [Remote host closed the connection]
doyougnu has quit [Ping timeout: 240 seconds]
<tenten8401>
should I try a different set of bindings?
stepcut has quit [Remote host closed the connection]
shlevy has joined #nixos
<adelbertc>
urgh changing dockerTools rsync command to use`${toString uid}:${toString gid}` doesn't seem to affect the Nix store
<adelbertc>
even added the `og` param
ericsagnes has quit [Ping timeout: 256 seconds]
<gchristensen>
anyone around who has contributed a NixOps module?
wpcarro has quit [Remote host closed the connection]
tenten8401 has quit [Quit: Konversation terminated!]
init_6 has quit []
hakujin has quit [Ping timeout: 245 seconds]
<adelbertc>
how does `dockerTools` "flatten" the contents when copying from the Nix store? like if I add `foo` to `contents and `foo` has `$out/a/1.txt` and `$out/b/2.txt` then `a/1.txt` and `b/2.txt` are also mounted at `/` in the image