<mmlb[m]>
Hey all, I'm having issues with my openvpn service config all of a sudden. I'm running nixos-unstable and I tried updating yesterday and all of a sudden my `up = "...";` script is causing an error with the following error message: "cannot coerce a list to a string, at /nix/var/nix/profiles/per-user/root/channels/nixos/nixos/modules/services/networking/openvpn.nix:16:18" and if I try changing it to `up = [ "..." ];` the error
<mmlb[m]>
changes to: The option value `services.openvpn.servers.admin.up' in `/etc/nixos/openvpn.nix' is not of type `strings concatenated with "\n"'
<{^_^}>
#97438 (by pbogdan, 1 day ago, merged): nixos/openvpn: path now requires conversion to a string
<mmlb[m]>
infinisil: welp :/ thanks. My github-search-fu totally failed me
<{^_^}>
[nixpkgs] @Infinisil pushed commit from @pbogdan to release-20.09 « nixos/openvpn: path now requires conversion to a string »: https://git.io/JUCfA
<juliusdeane>
how would I go about installing MATLAB on nixos? it's proprietary and doesn't seem to be in nixpkgs.
<juliusdeane>
haha I tried the default linux installer already
<juliusdeane>
it's definitely clueless about how things work on nixos lol
alchemist___ has joined #nixos
mariatsji has quit [Ping timeout: 244 seconds]
alchemist___ has quit [Remote host closed the connection]
orivej has quit [Ping timeout: 258 seconds]
<CyberManifest>
juliusdeane: can you do an install on nixos as you would similar to arch? I don't use NixOS so I'm not familiar ... but if you can then this may be helpful: https://wiki.archlinux.org/index.php/MATLAB
sigmundv has quit [Read error: Connection reset by peer]
sigmundv has joined #nixos
stiell has quit [Ping timeout: 258 seconds]
alchemist___ has joined #nixos
Blackraider has left #nixos ["WeeChat 2.7.1"]
<alchemist___>
I just received a wifi usb adapter and I really don't know how to make it work on nixos
<juliusdeane>
CyberManifest: not quite. Unlike arch, the paths to everything aren't what a normal linux program would expect, which is part of the reason why we need special nix configurations
stiell has joined #nixos
<juliusdeane>
alchemist___: what type is it?
<alchemist___>
juliusdeane: It's a TP-link and the model is TL-WN821N
<CyberManifest>
juliusdeane: can things not be repackaged in nix fashion ?
<Jake[m]>
Is there a way to disable lvm2 on NixOS?
<CyberManifest>
right, so NixOS is FHS compliant or FHS-like, similar to how Linux is POSIX compliant where BusyBox is POSIX-like compliant
<CyberManifest>
?
<samueldr>
not at all
<CyberManifest>
Not at all FHS compliant ?
<samueldr>
you'll have to read more about what POSIX defines (multiple things) to see where it fits, and what busybox and coreutils are
<CyberManifest>
so it's not really GNU/Linux ?
<samueldr>
you're getting concepts awfully mixed here
<CyberManifest>
I know what BusyBox, and coreutils are
<samueldr>
NixOS does not respect the FHS, by design
Blackraider has joined #nixos
Blackraider has quit [Remote host closed the connection]
<CyberManifest>
so it isn't fairly GNU/Linux certified
<samueldr>
I'm not sure you do considering you're getting the FHS mixed up with Busybox, and there is no reason to
<infinisil>
Jake[m]: It's "enabled" by default?
jumper149 has quit [Quit: WeeChat 2.9]
<infinisil>
What does that mean?
<CyberManifest>
samueldr: I do analogy comparisons to comprehend other new things
<simpson>
The FSF doesn't certify many distros; https://en.wikipedia.org/wiki/GNewSense is the only one, I think. We meet RMS's definition of GNU/Linux, *when* a GNU userland is used. So it's the user's choice as to whether they'll have GNU/Linux or a custom userland.
<samueldr>
GNU/Linux certified is not really a thing
<Jake[m]>
infinisil: It's enabled in the sense that `lvm2-activation.service` is a dependency of `systemd-udev-settle.service`, and it gives me a message when I boot `lvm2-activation-generator: lvmconfig failed`
<Jake[m]>
I'm tryna make my computer start up fast
<CyberManifest>
only that userland fails with proprietary mess where other user lands succeed
<CyberManifest>
adoption of standards makes sense
<CyberManifest>
pave your own path and you pay the price
<infinisil>
Jake[m]: `disabledModules = [ "tasks/lvm.nix" ];` seems to do it
<kandinski>
I don't know what I did and how I did it, but it seems I've managed to break nix: https://termbin.com/a5vq
<Jake[m]>
infinisil: you freaking rock
<infinisil>
:)
<infinisil>
Jake[m]: Btw I recommend looking at `systemd-analyze plot > plot.svg; firefox plot.svg` to see what's slowing boot down
alchemist___ has quit [Remote host closed the connection]
<Jake[m]>
Yeah that's how I grew suspicious of udev settling
<Jake[m]>
We'll see how it shakes out after I rebuild and reboot.
<kandinski>
Advice appreciated.
<infinisil>
,fancyUninstall kandinski
<{^_^}>
kandinski: Fancy way to uninstall packages, needs fzf installed: nix-env -q | fzf | xargs -I{} nix-env -e {}
<{^_^}>
infinisil: Did you mean fancy-uninstall?
<infinisil>
Maybe try that
<kandinski>
infinisil and {^_^}, thanks!
cript0nauta has joined #nixos
zeta_0 has joined #nixos
alchemist___ has joined #nixos
Darkmatter66 has joined #nixos
ddellacosta has quit [Quit: WeeChat 2.8]
maier has joined #nixos
alchemist___ has quit [Remote host closed the connection]
CyberManifest has quit [Quit: Leaving...]
ddellacosta has joined #nixos
<{^_^}>
[nixpkgs] @worldofpeace merged pull request #97608 → lxpanel: fix build by switching to gdk-pixbuf-xlib → https://git.io/JUcjC
alchemist___ has joined #nixos
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/JUCTz
maier has quit [Ping timeout: 246 seconds]
<alchemist___>
juliusdeane: so I installed the driver as mentionned in the issue you linked. I see my current wifi adapter (the one my laptop has) but the usb one is not listed as a network interface
<alchemist___>
So i used to see it but now the output of lsub is weird: https://ibb.co/NyCQvNd
h0m1 has joined #nixos
<alchemist___>
I added `boot.extraModulePackages = with config.boot.kernelPackages; [ rtl8192eu ];` to my config since i did lsusb for the first time... Idk if it messed some stuff up
Darkmatter66 has quit [Quit: ZNC 1.7.5 - https://znc.in]
<Rusty1>
2357:0107 is the adapter
<Rusty1>
does lsmod show the driver ?
<alchemist___>
well it's weird, when i plug and unplug the adapter, the list lsusb outputs doesn't change...
<Rusty1>
see if dmesg says anything useful
CyberManifest has joined #nixos
jtmcx has joined #nixos
alchemist___ has quit [Remote host closed the connection]
alchemist___ has joined #nixos
juliusdeane has quit [Remote host closed the connection]
<Rusty1>
try nixpkgs.linuxPackages.rtl8192eu
<alchemist___>
Rusty1: so doing `lsmod | grep rtl8192eu` doesn't show anything
<alchemist___>
Rusty1: do you mean that I install it as a package and not as an extraModulePackage?
zeta_0 has joined #nixos
<zeta_0>
is it possible to get a file back the way it was on a previous generation ?
<Rusty1>
try that instead i suppose alchemist___
jonringer has quit [Ping timeout: 256 seconds]
<zeta_0>
i forgot to make a copy of my xmonad.hs config, so i'm trying to get my old config file back, any tips ?
<alchemist___>
Rusty1: ok i'll try that! Thanks for bearing with me! I'm not used to hardware problems haha
alchemist___ has quit [Remote host closed the connection]
<lordcirth>
Steam can't launch Dead Cells - it complains about a ton of missing libraries. Anyone run into this? https://nixos.wiki/wiki/Steam doesn't say much
o1lo01ol1o has quit [Ping timeout: 260 seconds]
orivej has joined #nixos
<xe4>
I am trying to follow along with https://www.tweag.io/blog/2020-05-25-flakes/ and when I add experimental-features = nix-command flakes to ~/.config/nix/nix.conf I get warnings when I use nix, "warning: unknown setting 'experimental-features'"
<xe4>
also "nix shell" doesn't work for me, but "nix-shell" does
<lordcirth>
xe4, you need nix unstable
<lordcirth>
xe4, on nixos, configuration.nix: nix.package = pkgs.nixUnstable;
<moet>
occasionally my system will refuse to sleep. instead, if i close the lid, it cycles between sleep and wake states indefinitely.. how can i find out what is waking up the computer?
<lordcirth>
moet, wake events should be logged in dmesg
proofofkeags has quit [Ping timeout: 260 seconds]
<moet>
lordcirth: ah.. ok, i was looking in journalctl and not finding things.. i'll check dmesg
<moet>
is there a standard string to look for?
<lordcirth>
I don't remember
<lordcirth>
journalctl should include dmesg output, though
<moet>
hmm..
Supersonic112 has joined #nixos
<{^_^}>
[nixpkgs] @endgame opened pull request #97627 → aws-sam-cli: fix version bounds → https://git.io/JUCmF
pjt_tmp has quit [Ping timeout: 256 seconds]
pjt_tmp has joined #nixos
<pjt_tmp>
moet: hold on, I've woke my device recently. I think I can find the string
Supersonic has quit [Ping timeout: 240 seconds]
Supersonic112 is now known as Supersonic
<pjt_tmp>
moet: look for something like "ACPI: Waking up from system sleep state 64"
<pjt_tmp>
also did I miss anything? I disconnected for a minute
<moet>
pjt_tmp: that's super helpful.. i'll look around those ACPI wake messages for the cause.
<moet>
pjt_tmp: you didn't miss anything
<lordcirth>
Trying to run Dead Cells; ldd shows that it requires libhl.so, and there is libhl.so in the same directory. I tried patchelf --replace-needed, with both the relative and absolute path. Anything else I can try?
<lordcirth>
steam-run gives a more helpful error but doesn't work
<lordcirth>
Hmm, LD_LIBRARY_PATH=$LD_LIBRARY_PATH:. steam-run ./deadcells starts but fails due to lack of steam connection
<lordcirth>
Setting steam_appid.txt seems to work!
<CMCDragonkai>
Previously I had connected both monitors to the first GPu
<CMCDragonkai>
Multimonitor setup worked
<{^_^}>
[nixpkgs] @Mic92 pushed to release-20.09 « nixos/installer: enable sshd by default »: https://git.io/JUC3e
<CMCDragonkai>
But now I connected the second monitor to the second GPU
mariatsji has joined #nixos
<CMCDragonkai>
The second monitor doesn't automatically update. And says there's no signal.
<CMCDragonkai>
The `xrandr` command doesn't show that there's a second monitor. But `nvidia-settings` shows that there is a second monitor but it's disabled.
<CMCDragonkai>
What's the right way to enable the second monitor and make them share the same X screen, rather than creating a whole new X screen for the second monitor on NixOS?
cript0nauta has joined #nixos
<jasom>
How can I install a 64-bit version of wine?
<CMCDragonkai>
Because xrandr doesn't even show the second monitor. I'm thinking the NVIDIA 2nd GPU is not even enabling it.
<lordcirth>
jasom, do you need 64bit wine, or just a 64bit WINEPREFIX? Have you tried creating a prefix with WINEARCH=win64 ?
<jasom>
lordcirth: I need a 64 bit WINEPREFIX, when I try WINEARCH=win64 it tells me it's a 32 bit prefix, even with a fresh directory
mariatsji has quit [Ping timeout: 272 seconds]
<jasom>
lordcirth: e.g. this results in claiming a 32-bit installation: rm -rf ~/newwineprefix mkdir ~/newwineprefix; WINEARCH=win64 WINEPREFIX=$HOME/newwineprefix wine foo.exe
<jasom>
I assume that's because wine is a 32-bit executable, but I may be wrong
<jasom>
lordcirth: and if I take a pre-existing win64 prefix I get: "wine: '/home/aidenn/.wine' is a 64-bit installation, it cannot be used with a 32-bit wineserver."
<Ke>
Mic92: sometimes I wonder, if packages must be in the same set as all the auxiliary functions
<Mic92>
Ke: probably not.
<Mic92>
But I have not the time right now to do this refactoring
<Ke>
right now obviously any changes would break many people's configs obviously
<Ke>
that too
maier has quit [Ping timeout: 258 seconds]
<{^_^}>
[nixpkgs] @treed opened pull request #97630 → rpm: add openmp on darwin → https://git.io/JUCs6
kreyren_ has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @vbgl pushed to release-20.09 « coqPackages.paramcoq: init at 1.1.2 for Coq 8.12 »: https://git.io/JUCsS
kreyren_ has joined #nixos
<notgne2>
is there some way of using `nix repl` without any of the logging or the header? `nix-instantiate` doesn't seem to do building. I'm just trying to get the result of a Nix expression as JSON (it is not a derivation, but does contain derivations)
<notgne2>
is the best way just to wrap the output in a dummy derivation and use `nix-build`?
<notgne2>
I guess `--quiet | tail -n '+3' | head -n '-1' | sed 's/\x1b\[[0-9;]*m//g' | jq -r` is also technically an option
<veleiro>
are filesystem reads blocked during nixos-rebuild or how does that work?
<veleiro>
I never imagined such a thing would affect infrastructure but i may have forgotten
<Ke>
veleiro: the old stuff never goes away, only some symlinks change
<Ke>
but yes, I sometimes wonder, how magically it works even then
<samueldr>
notgne2: the issue is that you're using it assuming a usage that's not prescribed :)
<samueldr>
notgne2: so it might work _for now_ :D
<samueldr>
veleiro/Ke, the magic all happens with switch-to-configuration, at the very end of nixos-rebuild if needed, and if requested
<veleiro>
so the difference between build/switch/boot/test etc?
<veleiro>
when you run a program, its of the generation that is currently activated
sangoma has joined #nixos
<veleiro>
that's not hard to understand for me
<samueldr>
build only builds, no switching; switch builds, switches, and adds to boot menu; boot builds, and adds to boot menu; test builds and switches
<Ke>
samueldr: but someone could have written software that constantly rereads it's configs and gets a missmatch between it'
<Ke>
its config files that have different generations
<veleiro>
but im not sure what blocking filesystem reads would provide for an
<veleiro>
immutable system
<notgne2>
well, services are managed by NixOS too so that can often be prevented
<samueldr>
the /run/current-system symlink changes in one fell swoop, but yeah, services can be messy
<notgne2>
and I think wherever possible, configurations are given via CLI arguments pointing to a Nix store rather than using `/etc`
<Ke>
but if your service uses config directly from the /nix/store and not via symlinks you'll always be fine
<samueldr>
yeah
<samueldr>
it's a case-by-case problem
<veleiro>
hmm i havent had any problems with this
<notgne2>
I wonder if maybe some weird chroot magic can be done to make all applications point to the Nix store for configuration even if they don't like it
sangoma has quit [Read error: Connection reset by peer]
<notgne2>
you could of course also patch the software but that's a lot more difficult
<{^_^}>
[nixpkgs] @marsam opened pull request #97633 → opencv: enable videoio on darwin → https://git.io/JUCGR
<Ke>
veleiro: the update of the symlinks is still quite a speedy operation, so you'll be unlikely to hit the intermediate state anyway
<veleiro>
so services are restarted, but when could they be messy?
<samueldr>
it definitely sounds like a case of YAGNI, and when you'll need it, then you implement in a way that is better suited to the service
<veleiro>
when they dont restart immediately on rebuild?
<veleiro>
i sorta see that
otulp has joined #nixos
<Ke>
I think there is some variable that describes whther things are restarted on a switch
<Ke>
things like d-bus in general can not be updated without reboot
<veleiro>
this conversation stemmed from mutable vs immutable systems
<veleiro>
and i have until now assumed nixos to be immutable
<Ke>
it is not impossible for someone to program a nasty program that would just not play nice with nixos service management, I believe
<veleiro>
like a container is the only immutable system? that's boring
<samueldr>
store paths are, but the system is implemented on top of mutating as close as possible to the root of the FS as convenient
<samueldr>
and well, state is not immutable :)
<Ke>
veleiro: well in any case I have rootfs on tmpfs and only a few non-volatile paths for myself
<Ke>
there is runtime state, but it's not persistent
<Ke>
and then there is immutable state and some state that use case needs, maybe eg. home is on persistent mutable mount
<Ke>
not all my hosts even need a persistent home
<veleiro>
i dont undestand enough about the state part, do you mean user/program files?
<veleiro>
i think so
<veleiro>
so when an app has new state after switching, but the old state is what is modifying
<veleiro>
what happens?
<veleiro>
user files (as ive seen) are still present in /var (postgres or whatever)
<veleiro>
is that what the system.stateVersion is for?
<Ke>
well basically when you install packages they just end up as static data in /nix/store, when you enable them as service some things get symlinked or otherwise referred to from the root
<Ke>
references can be things like wrapper scripts or just string references in configs
<Ke>
or the symlinks
<Ke>
I guess studying your rootfs might give you a best idea
<Ke>
and indeed some software wants to have state on disk, normally nixos does not manage that too much, I believe
<veleiro>
is this and "unsolved issue" or something that is accepted as a fault?
<Ke>
I think that's maybe a bit philosophical question
<Ke>
stateVersion manages some incompatibilities and maybe migrations, I believe, never had to use it myself
<Ke>
sometimes I had related issues, but I mostly just fixed it by deleting old state, as the old state was not of value
<Ke>
like we had nscd caches that were incompatible or some symlinks that were created by old version of systemd
<Ke>
in those cases having tmpfs rootfs helps a lot
<Ke>
because you can ignore old state that you do not explicitly want to preserve
waleee-cl has quit [Quit: Connection closed for inactivity]
<veleiro>
exactly
saschagrunert has joined #nixos
<veleiro>
so systemd programs would KILL if they were restarted
<veleiro>
that's something ive seen and understand
<dminuoso>
veleiro: Think of it as a feature that we dont just restart your program.
<colemickens>
well the units are stopped, swapped, and then the units are restarted with their updated store paths to new binaries.
<dminuoso>
It's because we don't have to. It's no trouble having two or more simulatenous programs.
<colemickens>
but like, firefox isn't restarted.
<colemickens>
And not all services auto-restart, but most do if you change their config.
eliumir has joined #nixos
<veleiro>
dminuoso: thats a very good point
<veleiro>
there could be 1000s systemd units of the same program
<dminuoso>
Well in case of systemd units its different
<dminuoso>
Because they are *named*
sangoma has joined #nixos
<dminuoso>
And they usually occupy shared resources
<dminuoso>
That is, we cant sensibly distinguish them with hashes like we can in the nix store
<dminuoso>
That is, we expect `systemctl status postgresql` to make sense anywhere, and it uses the shared resource of the listening port.
<dminuoso>
Perhaps there was ways to make this work with linux namespaces, this might be interesting to explore. But at the end, this is the design choice made by nixos.
<dminuoso>
(Or actually, not namespaces, but rather by attaching these services to a bridge, and then having some declarative way of re-exposing services)
<dminuoso>
Similar to how docker networking works
<veleiro>
it sounds to me that its more the fault of the apps than the system itself
<dminuoso>
I wouldn't call it a "fault".
<dminuoso>
But yeah.
mariatsji has joined #nixos
<dminuoso>
Most systemd units tend to run things that must be reachable by some means. To discover them, you'd have to uniquely find them on a port, or maybe via some configuration under /etc
<dminuoso>
(Or maybe a unix domain socket that is always located under a conventional path inside /run)
alp_ has quit [Ping timeout: 272 seconds]
<dminuoso>
So there can only be one such service running at any time.
<veleiro>
hmm interesting, thanks for the explanation
<veleiro>
i'm very convinced by nix about how to approach things
<dminuoso>
:)
<veleiro>
i was just thinking about the super attachement to systemd though
<veleiro>
obviously there's guix
<dminuoso>
The main difficulty in explaining nix/nixos to people is similar to how explaining pure, declarative and lazy functional programming languages is difficult to traditional imperative programmers.
<veleiro>
if i try to explain haskell to others, its a lost cause
<dminuoso>
Well, Haskell is not necessarily hard to explain, but it takes experience to teach well.
<veleiro>
but lazy evaluation makes perfect sense when evaluating the state of an OS
<dminuoso>
I found teaching Haskell to my coworkers works well enough.
<dminuoso>
Some aren't comfortable enough to write algorithms in a declarative fashion yet, but they can certainly read them at least.
<dminuoso>
It depends on who you are talking to, and why they are listening to you.
<dminuoso>
If it's just to convince "random java fanatic" that Haskell is great, that's a conversation I'd skip.
<veleiro>
well you can create haskell's type of pure functionality in many languages right?
<veleiro>
so its just a guide
<dminuoso>
Hardly
<veleiro>
but it makes more sense for the state of an OS
<dminuoso>
I mean, the main thing that I find majorly annoying and frustrating in nix, is the lack of a type system.
<dminuoso>
In a pure functional lazy setting, having an expressive and rigid type system is really important to me.
<veleiro>
can you give an example of the lack of a type system?
<dminuoso>
nix simply has none.
<dminuoso>
Not sure what example you want
<veleiro>
so nix has no types? maybe ive never noticed
<sphalerite>
nix has two type systems I can think of.
<dminuoso>
You haven't noticed you aren't able to write `let x = 1 : Int; int ...` ?
<dminuoso>
err
<dminuoso>
Drop the `t` at the end of int there. Bad muscle memory. :p
<sphalerite>
1) the language built-in one, which doesn't have any rules for declaring the type of something and can't be checked statically
<sphalerite>
2) the module system.
<veleiro>
nope, but i see what you mean
<sphalerite>
(the nixos module system)
<sphalerite>
Of course, whether they're good type systems is another question :D
<cole-h>
The module system is genius, tbh. Who came up with it, again?
<sphalerite>
nbp
<cole-h>
nbp++
<{^_^}>
nbp's karma got increased to 0b11
<dminuoso>
sphalerite: Well, a type system with just a single uni type is arguably not very useful. :)
<cole-h>
It's not as good as actual types a la Rust and Haskell, but it's an OK substitution, IMO
<sphalerite>
dminuoso: I'd argue it isn't that.
<dminuoso>
cole-h: Sure, the nixos module type system is better than nothing. :)
aborsu has joined #nixos
<dminuoso>
But at the end it's just runtime tagging. The second you start doing higher order shenanigans, it becomes less than helpful.
<sphalerite>
dminuoso: it does define which operations can be performed on which types of values. The trouble is that it can't be checked statically.
<dminuoso>
But anyway, didn't want to start a rant here, it's a well known fact. And I know there's work being done on adding a proper type system to nix.
<sphalerite>
The manual describes a bunch of types informally in a haskell-like way :D
<veleiro>
dminuoso: have not heard the rant. i want to know everything fyi
<dminuoso>
sphalerite: When I say a type system, then I mean it in the traditional type theoretic sense. This dynamic runtime tagging and checking is *not* a type system.
<sphalerite>
not sure that type system stuff is still happening.
<dminuoso>
In that sense anyhow.
<cole-h>
Yeah, I don't think it's still happening in the sense that man hours are being put into it...
<dminuoso>
And I dont want to argue about "what is a type system", just accept my definition for the purpose of this discussion, that should end it quickly. :
<sphalerite>
https://github.com/regnat/ptyx is this what you mean by "work being done on adding a proper type system to nix."?
<dminuoso>
That seems to be one of the attempts.
<sphalerite>
because my impression is that that was a bachelor's thesis project and basically stopped at the bachelor's thesis
<sphalerite>
are you aware of any more recent/active attempts?
<dminuoso>
I think I've seen another, but I dont recall the details
<colemickens>
bqv: there are links to a guide/tutorail at the bottom. I'm going to just build a VM with their Nix and see how it goes. I don't really quite understand IPLD or Nix internals well enough yet to grok everything. :S
<veleiro>
so what do you mean by module system? a type of derivation?
<dminuoso>
veleiro: nixos modules are to nixos, as nix files are to nix expressions, roughly.
<dminuoso>
Every time you use some options like `services.foo = 1;`, those options and their behaviors are defined in a nixos module
<bqv>
colemickens: my bad, apparrently my web reader chopped that off
<veleiro>
thanks for the explanation
sputny has joined #nixos
<bqv>
colemickens: my issue may have been trying to build using the flake.nix, rather than release.nix
pjt_tmp has quit [Ping timeout: 260 seconds]
lotharn has quit [Ping timeout: 256 seconds]
domogled has quit [Quit: domogled]
lotharn has joined #nixos
mallox has joined #nixos
<bqv>
This is exciting, though. I'm tempted to use my main store for it
<bqv>
Nix is reproducible, after all
<bqv>
I have a spare machine for it but it's slightly incapacitated
cybrian has quit [Ping timeout: 256 seconds]
o1lo01ol1o has joined #nixos
<bqv>
colemickens: actually, using this might be how that machine broke, if it one day accidentally worked and did corrupt the store :D
<{^_^}>
[nixpkgs] @vcunat pushed to master « knot-dns: libbpf is Linux-only »: https://git.io/JUCcw
<cptrbn>
Is it possible to build an ARM7 image from macOS (with nix-build), or do I need to fire up a VM? Trying to make a custom headless image for a RPi2
<karantan>
I have several .nix configuration files which I include in configuration.nix. I one file (extra-configuration.nix) I define custom php package and I add it to the `environment.systemPackages`. I also set this custom php package ( named "phpPackage") to be used for phpfpm ("services.phpfpm.phpPackage = phpPackage;").
<karantan>
Then in the second file where I have WordPress, Apache and phpfpm settings I configure phpfpm pools. But the problem is that it's not using my custom php package. Instead it's using default php 7.3.
<graf_blutwurst>
Hi everyone, did anyone elses xfce just kill itself? I can't even open up some setting managers anymore and the font is throughly screwed up
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
<graf_blutwurst>
oh `fc-match` doesn't find anything anymore. uff how does that work in nixos?
<Mic92>
It does for me
o1lo01ol1o has quit [Remote host closed the connection]
<Mic92>
I have a cache in ~/.nix-profile/lib/fontconfig/cache
<Mic92>
I think it is maintained by home-manager
thc202 has joined #nixos
<graf_blutwurst>
hmm interesting. I'll check home-managers default font config. maybe i killed it on accident because i have no font configs in home-manager
<{^_^}>
[nixos-org-configurations] @garbas pushed 2 commits to master: https://git.io/JUC0m
ptottermtemp is now known as ptotter[m]
<eyJhb>
I know this is not a NixOS question, but if I wanted to import home-manager as a path, wouldn't a cloned repo and doing `imports [ ./path/to/home-manager/default.nix ];` not work? I am getting `anonymous function at /state/home/projects/nixos/nixus-setup/channels/home-manager/default.nix:1:1 called with unexpected argument 'name', at
<hyper_ch2>
finally 2x 16TB drives have arrived :)
__monty__ has joined #nixos
o1lo01ol1o has joined #nixos
sigmundv has joined #nixos
sangoma has quit [Ping timeout: 258 seconds]
jb55 has quit [Remote host closed the connection]
jb55 has joined #nixos
<{^_^}>
[nixpkgs] @erikarvstedt opened pull request #97647 → maintainers/build: skip packages that don't evaluate → https://git.io/JUCEV
justanotheruser has quit [Ping timeout: 246 seconds]
Thra11 has joined #nixos
sangoma has joined #nixos
malook has quit [Quit: malook]
<ldlework>
Can anyone help me think about how to make a derivation that essentially pops out a script that wrap emacs such that it has a number of binaries on its path?
voidcontext has quit [Ping timeout: 258 seconds]
voidcontext has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<makefu>
ldlework: there is writeShellScript and you can create your path PATH=${pkgs.makeBinPath [ my-binaries ]}
<ldlework>
isn't there some kind of "makeBinPath" so I only have to add one path to the wrapped program's path?
cr4y1 has quit [Read error: Connection reset by peer]
<ldlework>
and some kind of makeWrapper helper
cr4y1 has joined #nixos
<ldlework>
can't quite figure it out from the docs
cr4y1 has quit [Read error: Connection reset by peer]
cr4y1 has joined #nixos
hyper_ch2 has quit [Ping timeout: 258 seconds]
hyper_ch2 has joined #nixos
czer00 has quit [Read error: Connection reset by peer]
sangoma has joined #nixos
kolaente_ has joined #nixos
<kolaente_>
Has anyone else here seen issues with nix-collect-garbage and btrfs lately? As in, every time I try to run it it crashes and my root partition gets mounted as read only
<kolaente_>
And I'm not really sure how to debug this
<Ke>
very first of all, check kernel log with dmesg command
<Ke>
I use btrfs and I can assure you, this is not normal
<kolaente_>
Ke: I get "unable to find ref byte nr 14772121600 parent 0 root 5 owner 18540358 offset 0" and a bunch of debug output
<Ke>
kolaente_: are you using quotas or some other dangerous substances
<kolaente_>
Not that I'm aware of
<Ke>
anyway that's almost 100% certainly not a nixos issue
<Ke>
you can run btrfs check --readonly from some recovery media, but it all depends on how much time you want to spend reporting this issue to btrfs devs
<Ke>
easiest solution to get on with your life is possibly just recover files and reinstall
<kolaente_>
Probably yeah
<kolaente_>
Good thing that's easy with nix
<kolaente_>
I also thought about a hardware issue but that seems unlikely since the ssd that's running my installation is only four months old
<Ke>
nowadays btrfs check --repair occasionally fixes corruptions, but you might just want to reinstall anyway
sangoma has quit [Read error: Connection reset by peer]
rprije has quit [Ping timeout: 265 seconds]
<Ke>
kind of, right now many people think that modern btrfs mostly corrupts due to faulty hw. but I have seen no quantitative proof of that
tmaekawa has quit [Quit: tmaekawa]
<Ke>
this is only, if you stay with the safe features, like not quota
tmaekawa has joined #nixos
kolaente_ has quit [Ping timeout: 244 seconds]
kolaente_ has joined #nixos
tmaekawa has quit [Client Quit]
kolaente_ has quit [Client Quit]
<ldlework>
How do you escape things like ${RED} in a nix string that is actually bash?
<lassulus>
''${RED}
acarrico has quit [Ping timeout: 256 seconds]
<lassulus>
if its inside a '' multiline string
<qyliss>
''${RED} in a '' string, or \${RED} in a " string
civodul has quit [Quit: rebooting Emacs...]
<ldlework>
thx
civodul has joined #nixos
sangoma has joined #nixos
<damjan>
what's the practise for nixpkgs pull requests, do I squash rebase improvements while it's reviewed, or is the pull request squashed at the end?
<ldlework>
What's the shortest one-liner for someone to try out a nix derivation stored on github?
<eyJhb>
vifon: Someone in here will be a lot more qualified to answer it :p
xd1le has quit [Remote host closed the connection]
<vifon>
So, hello everyone. I'd like to stay on the stable channel but include one package from unstable. I've added the unstable channel as nixos-unstable. I've prefixed my configuration.nix with "let unstablePkgs = <nixos-unstable>; in" (after the arguments of course) and then I included unstablePkgs.$PACKAGE in my environment.systemPackages.
xd1le has joined #nixos
dbmikus has quit [Ping timeout: 265 seconds]
<vifon>
This is the error:
<vifon>
error: value is a path while a set was expected, at /etc/nixos/configuration.nix:49:16
<eyJhb>
I have the {} suggestion in your let, which might make the difference
<siraben>
I get something like this: Unable to find libstd under RUST_SRC_PATH. N.B. RUST_SRC_PATH variable needs to point to the *src* directory inside a rust checkout e.g. "/home/foouser/src/rust/src". Current value ""/nix/store/1h2i69b7vljmgng9cxzrvkdj5ixb9hak-rust-src-1.48.0-nightly-2020-09-09-e2be5f568/libstd""
<siraben>
racer error^
simonpe^^ has joined #nixos
saschagrunert has quit [Remote host closed the connection]
<simonpe^^>
I have a derivation (which is impure) that i build with 'nix-build --option sandbox false', I need the buildPhase script to "see" a couple of environment variables from my shell environment to figure out what urls to use as cache. I don't want to use getEnv() because that would change the sha of the result
<simonpe^^>
so I guess the question is, how do I thread a couple of enviroment variables through to the mkDerivation build scripts?
<jaen>
Hi, should I expect the `amdgpu` work on `nixost-unstable`? When I do `vainfo` on 20.03 it works and reports a (somewhat short) list of supported formats. On unstable it fails with a binary symbol mismatch. Am I doing something wrong (basically just plugging in "amdgpu" into kernel modules as I'm not seeing any explicit libva package like for
<jaen>
intel) or is that broken curently?
dbmikus has joined #nixos
CyberManifest has quit [Quit: Leaving...]
superherointj has quit [Quit: Leaving]
alp_ has joined #nixos
dbmikus has quit [Ping timeout: 256 seconds]
dermetfan has quit [Ping timeout: 260 seconds]
<{^_^}>
[nixpkgs] @Ma27 opened pull request #97666 → nixos/nextcloud: fix `nginx`-config for Nextcloud 19 and older → https://git.io/JUCDL
<sterni>
simonpe^^: why not use them as inputs for your package (i. e. arguments) and getEnv?
<kahuna>
Hey all, Need some help to install nixos plz. Following instreuction, I must mount mount /dev/disk/by-label/nixos /mnt but i can do it : msg is mount: /mnt: /dev/sdb1 already mounted on /iso
<kahuna>
I use a version in USB
<simonpe^^>
sterni: because they impact the hash of the derivation, so if I change one of them I can't reuse my cache
eoli3n_ has joined #nixos
<{^_^}>
[nixpkgs] @gebner opened pull request #97668 → mlton: use $TMPDIR → https://git.io/JUCD5
knupfer has quit [Remote host closed the connection]
knupfer has joined #nixos
dbmikus has quit [Ping timeout: 256 seconds]
karantan has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Mic92>
kahuna: is /dev/disk/by-label/nixos supposed to be your system partition?
<Mic92>
It might be that the installer already uses the same label
<Mic92>
In that case you might call lsblk to inspect all your partitions and mounted via direct path instead of via label
kyren_ has joined #nixos
<{^_^}>
[nixpkgs] @ttuegel opened pull request #97669 → plasma-workspace: fix path to qdbus → https://git.io/JUCym
<kyren_>
hey all, suppose I wanted to use nixpkgs to cross compile from linux to mingwW64, where the program in question at the top-level is written in rust, what's the current likelihood of out of the box success with this?
<{^_^}>
[nixpkgs] @ttuegel opened pull request #97670 → kconfigwidgets: install plugin in .out → https://git.io/JUCyG
<kyren_>
I've tried several things and I can go over what I've tried, but most likely they're either wrong or I should just not expect it to work right now, so rather than asking a bunch of potentially dumb questions, just saying whether I should be able to figure it out myself might be enough
<Mic92>
kyren_: there is some mingw support in nixpkgs, but might not work out for rust yet. I think it was mostly introduced for some folks doing haskell
<Mic92>
I did cross-compile direnv for windows at some point
<{^_^}>
[hydra] @andir opened pull request #805 → LDAP support continued (with missing packages & tests) → https://git.io/JUCyX
<{^_^}>
[nixpkgs] @andir opened pull request #97673 → perlPackages: init Catalyst::Authentication::Store::LDAP at 1.016 (and dependencies) → https://git.io/JUCyM
<kyren_>
Mic92: hmm okay, I'm gonna keep trying to get a better sense of the state of things anyway, and maybe report back with better questions ready
<kyren_>
but thank you!
<Mic92>
kyren_: If you just want something working maybe rustup is an option
<bbigras>
kyren_: also maybe ask on the forum. I'm also interested in this.
<kyren_>
that's actually just what I want, I think that already works via nixpkgs-mozilla overlay as well, but I have some nixpkgs dependencies too like e.g. openssl, but I'm also a beginner at this so I'm trying to figure out if like, I need to start by setting crossSystem to mingwW64 or if that's just not going to work
<Mic92>
kyren_: if you just need a CC set to mingwW64, you could just try a nix-shell with stdenv from that
fendor_ has joined #nixos
fendor has quit [Read error: Connection reset by peer]
<Mic92>
kyren_: i.e.: pkgsCross.mingwW64.stdenv
<Mic92>
We should even have that in the binary cche
<Mic92>
*cache
<kyren_>
oh right I keep forgetting that things are so flexible like that
fendor_ has quit [Remote host closed the connection]
<sterni>
simonpe^^: you can just make your cache as a whole a fixed output derivation if doesn't change even if you change the nev vars
jonringer has joined #nixos
fendor_ has joined #nixos
ransom has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
sigmundv has quit [Read error: Connection reset by peer]
<kyren_>
okay, I think my problems are much more basic than this, I'm trying to use mkShell and list compiler tools in nativeBuildInputs rather than using stdenv.mkDerivation or possibly buildRustPackage, maybe that's my problem
sigmundv has joined #nixos
drewr has quit [Quit: ERC (IRC client for Emacs 27.1)]
<kyren_>
okay maybe I was wrong, the simplest possible crate doesn't build when setting crossSystem, so maybe the strategy of just building individual non-rust deps using pkgsCross is better, anyway I'll report back if I have any success, thanks!
<simonpe^^>
sterni: it's too big to do that, it's a yocto build and it overlaps quite a bit with nix
growpotkin has joined #nixos
<simonpe^^>
but its basically as reproducible so I can pretty much trust it do do the same thing given the same inputs, disregarding it currently runs inside of docker because it expects a peculiar environment which I couldn't get working insdie a buildFHSUserEnv
<simonpe^^>
so its (nix (docker (bitbake)))
sherub has joined #nixos
dermetfan has joined #nixos
omneh has joined #nixos
omneh has joined #nixos
omneh has quit [Changing host]
<{^_^}>
[nixpkgs] @ryneeverett opened pull request #97677 → nixos/docker: load more required kernel modules → https://git.io/JUC96
omnipotententity has quit [Ping timeout: 265 seconds]
<eyJhb>
Is it possible to do `users.users.root = ...` without ruining everything? Want to add a ssh key to root
<cole-h>
Yes
<{^_^}>
[nixpkgs] @Jiehong opened pull request #97681 → sublime-merge: use 2033 as dev and 2032 as stable versions → https://git.io/JUC7V
<eyJhb>
cole-h: Trying to figure out the best way to deploy using Nixus. And allowing for sudo without PW on my main user seems careless, adding a ssh key to root seems "not well". My only option seems to be adding a user only used for deploying?
<eyJhb>
only option left*
<eyJhb>
Not sure if you have a better approach cole-h
<eyJhb>
The search issue atm. with not displaying example correct
<eyJhb>
Didn't look at it
<eyJhb>
Was there anything else malformed cole-h ?
<jasom>
Can I have multiuser nixpkgs on linux without systemd?
<cole-h>
eyJhb: Not that I noticed from a quick glance
<bqv>
jasom: clever made "not-os" which uses runit
<eyJhb>
Great, lets see how much I can break!
<bqv>
But beyond that, no
<cole-h>
eyJhb: The great thing about NixOS is that you can rollback ;)
<bqv>
I am trying to build an s6-based nixos though…
<eyJhb>
True! Just hoping it won't disturb my tv show atm. cole-h :p
<jasom>
bqv: I guess I'll dig into what it is that nixpkpgs wants from systemd and see how much work it would be to reimplement as not-systemd
<bqv>
A lot.
<bqv>
(almost everything...)
<cole-h>
Yeah, NixOS is extremely reliant on systemd
<bqv>
I'm starting as trying to get a basic container working
<jasom>
not nixos, but nix on a hosted linux
<bqv>
That way its not insurmountable
<cole-h>
Oh
<bqv>
Oh
<eyJhb>
Oh
<cole-h>
That should be easier. You just need a daemon service, really. I don't think there's anything in the pkgs part of nixpkgs that relies on systemd?
<cole-h>
Aside from various packages, of course
<cole-h>
{lib,}udev, etc
<bqv>
Yeah it's just derivations, then
alexherbo2 has quit [Read error: Connection reset by peer]
alexherbo2 has joined #nixos
<jasom>
looks like 1 service and 1 socket is all it puts in systemd
knupfer has joined #nixos
knupfer has quit [Client Quit]
knupfer has joined #nixos
simonpe^^ has quit [Remote host closed the connection]
fendor has quit [Ping timeout: 260 seconds]
kahuna has joined #nixos
domogled has quit [Remote host closed the connection]
domogled has joined #nixos
superherointj has joined #nixos
hiro99 has quit [Quit: hiro99]
<{^_^}>
[nixpkgs] @kevincox pushed 0 commits to kevincox-chrony-state: https://git.io/JUC5x
<emily>
password in the plain on the root volume or use the system keychain, but it's not entirely clear to me how to put all the pieces together
<emily>
like, that Ruby snippet doesn't seem to set up a login hook or anything to mount the volume? does just injecting the password into the system keychain and putting it in fstab manually work, or is that part just left to the reader to figure out?
<typetetris>
Hmm I have a call to `nix-shell` in CI, where `NIX_PATH` isn't set to something useful. But the shell.nix should be self contained. But still, I have to user `-I nixpkgs=https://<some nixpkgs url>` because it insist on getting bash from the nixpkgs in the NIX_PATH, can I avoid that?
<energizer>
i am seeing only boxes instead of fonts and gtk apps all say "(hexchat:14393): Pango-CRITICAL **: 12:12:17.166: pango_font_get_hb_font: assertion 'PANGO_IS_FONT (font)' failed". how to debug?
<bqv>
Alt+: for eval
<bqv>
Or M-x ielm
<aidenholmes>
i'm using spacemacs, i need to restart it to change configuration
<bqv>
For interactive eval
Kloenk_ has quit [Quit: WeeChat 2.9]
<bqv>
Yeah, but you can run lisp at runtime
Kloenk_ has joined #nixos
<bqv>
Changing some configs temporarily, on the fly
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
knupfer1 is now known as knupfer
<{^_^}>
[nixpkgs] @vcunat pushed to master « Revert "Merge #96844: nixos/nfsd: run rpc-statd as a normal user" »: https://git.io/JUCAc
<{^_^}>
[nixpkgs] @flokli pushed commit from @NinjaTrappeur to release-20.09 « test-driver.py: fix VM state directory deletion »: https://git.io/JUCAC
<energizer>
What does this mean? strace -e trace=file hexchat -> openat(AT_FDCWD, "/nix/store/c4rdlbksy4y8i4y86bsrkd439lqfy1xh-gtk+-2.24.32/lib/libpangoft2-1.0.so.0", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
<sherub>
Hey folks, I am facing an issue with glX extensions on my newly setup NixOS. Anyone here who can help me understand what I am missing? Is this even the forum to ask questions like that?
aidenholmes1 has quit [Quit: aidenholmes1]
<bqv>
Someone clever will be along shortly to assist
<simpson>
I'm not at all clever, sadly. How have you diagnosed your problems so far, and do you know your GPU vendor and/or driver?
<simpson>
sherub: I'm not an nv expert, but why aren't you using nouveau?
<sherub>
I believe that it is a driver issue. When I don't use these drivers, alacritty does work. but then the whole display gets screwed (everything gets squished vertically to have)
<sherub>
simpson I don't know what nouveau is
<energizer>
i think it means it can't find this function pango_font_get_hb_font in the pango shared lib, yes?
<energizer>
but that is fixed, so maybe my problem is different
<simpson>
sherub: nouveau is the reverse-engineered open driver for nVidia cards. https://nouveau.freedesktop.org/wiki/FeatureMatrix/ suggests that your "old" NV50 chipset should work fine as long as you don't need TV out.
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/JUCxY
<sherub>
Docs mentioned that usually proprietary drivers give better rendering. And when I left it to default everything used half the screen with everything vertically squished.
<sherub>
I got to nvidia's site to see the appropriate drivers for the GPU I have and got it was 340 and then looked that in nixos search
<{^_^}>
[nixpkgs] @matthewbauer pushed commit from @arcnmx to release-20.09 « Fix arch eval error introduced in #61019 »: https://git.io/JUCxn
<{^_^}>
[nixpkgs] @SuperSandro2000 opened pull request #97700 → cod: init at unstable-2020-09-10 → https://git.io/JUCxC
<simpson>
"better 3D performance" comes at the cost of 2D performance (Web browser, video, PDF reader) and, as you're learning, they're very brittle and require the entire driver to be loaded or unloaded, all the way up the stack. In particular you *need* to reboot after switching to/from "nvidia".
lunik1 has quit [Ping timeout: 256 seconds]
<simpson>
The docs don't say how to explicitly choose nouveau. I'd guess writing "nouveau" would work, but also I think that it might be autodetectable if you simply remove the proprietary "nvidiaLegacy340".
zupo has joined #nixos
<sherub>
removing nvidiaLegacy340 makes it so that only half my display is used.
<bqv>
The middle half?
<simpson>
Oh! I see, I misread your earlier message. Okay, I guess you're stuck with the proprietary drivers. Let's figure out your GLX, then. Are you using `glxinfo` to check?
<sherub>
When I try to run glxgears it gives outError: couldn't get an RGB, Double-buffered visualWhen I try to run glxinfo it gives outname of display: :0Error: couldn't find RGB GLX visual or fbconfig
<bqv>
Imgur probably easiest, but ix.io etc support images
<bqv>
Either way, is it not just a resolution issue?
<simpson>
sherub: To answer some of your Reddit questions, GLX is GL (graphics language, like OpenGL) for X11. GLX vendors include nVidia and Xorg. Your GLX is loaded when Xorg starts, so pastebinning /var/log/X.0.log would let us double-check which drivers are being used.
<bqv>
Including the name of that screen, and it's available resolutions
<sherub>
output of xrandr is
<sherub>
Screen 0: minimum 8 x 8, current 1920 x 1080, maximum 8192 x 8192DVI-I-0 disconnected primary (normal left inverted right x axis y axis)VGA-0 disconnected (normal left inverted right x axis y axis)DVI-I-1 disconnected (normal left inverted right x axis y axis)HDMI-0 connected 1920x1080+0+0 (normal left inverted right x axis y axis) 600mm x 340mm
<sherub>
plugged in? monitor. then yes. I am currently using it
<bqv>
Oh
<sherub>
also simpson what am I looking for in X.0.log?
<bqv>
Ok, well, `xrandr --output HDMI-0 --mode 1600x900` should do it iirc
<simpson>
sherub: GLX, nVidia, errors. I can read it quickly if you pastebin it, but it's not rocket science. Be aware that a lot of warnings are completely normal.
karantan has joined #nixos
<energizer>
is there a cache that has built unfree nixpkgs? i dont want to build them myself
<bqv>
Beyond that, I'm sorry, I haven't a clue
<bqv>
energizer: ipfs is coming™
<energizer>
bqv: but who will build those packages?
<bqv>
You
<gchristensen>
who do you trust to build those packages?
<bqv>
Me
<gchristensen>
=)
<bqv>
Everybody!
<energizer>
yes, there are a lot of people i trust to build packages
ixxie has joined #nixos
<simpson>
bqv: Oh! I see what you saw now! Nice find.
<sherub>
yeah it is a 4k capable monitor my GPU can probably push more than FullHD, I plan to get to that issue too.
<simpson>
Heh, indeed, the log corroborates the idea that the TV's EDID has trashy modes. (EDID is how monitors tell the GPU which modes they support.)
LogicalOverflow has quit [Remote host closed the connection]
<simpson>
Thank you for saying that now. Let me relate an anecdote. To drive the typical TV at 4k, you'll need a *custom mode*. This isn't super-hard, you'll need to stare at the manpage for `cvt` for a while. But beware: You may find yourself unhappy with the framerate, because you might not be able to drive it at 60fps over your GPU and TV combination.
<leungbk>
What is the exact command used to obtain the hashes in the nixpkgs repo? I've tried some variants of "nix-hash --type sha256 --base32 ." on a git-checkout of debugpy at the version in nixpkgs (python-modules/debugpy/default.nix) but cannot get a matching hash.
<simpson>
I recall driving a TV at Sears with a laptop using a custom mode. There just wasn't enough clock for all those pixels; the best mode was 15fps.
<sherub>
This GPU is a temp solution. I was waiting for the next gen GPU launch from both nvidia and AMD and will switch in coming months
<sherub>
so i am okey with this issue for now.
<bqv>
simpson++ I've had to do that before, though last time I had that issue was on windows >_>
<{^_^}>
simpson's karma got increased to 30
<simpson>
If the TV's EDID is bad, then it's the TV's issue.
<sherub>
Also "noticed that your highest-DPI resolution isn't necessarily your screen-filling resolution." I did not really understand exactly what you meant. Can you help me understand better
<simpson>
(And, again, TV EDIDs are *commonly* bad. Maybe three out of five of the TVs I've owned, I've needed custom modes or some other tweak.)
<simpson>
DPI is a measure of how densely the pixels are packed on your screen. It's possible to have a high-DPI mode which doesn't use all of the (usually vertical) space.
<simpson>
I don't see any problem with your nVidia GLX. Indeed, your log has commemorated you running glxgears and glxinfo by repeatedly discovering that your TV doesn't support 3D glasses.
aidenholmes has quit [Ping timeout: 244 seconds]
<sherub>
When you say TV, do you mean any general display? or something specific?
<simpson>
sherub: Hold on. Did you reboot before running xrandr? I think that we may have missed a key step. nVidia's drivers are an all-or-nothing setup; you have to have the kernel drivers loaded *and* be using their GLX as well.
<simpson>
"TV" is my name for the market segment that this sort of monitor comes from. Your "LG Electronics LG HDR 4K". I don't know whether it was literally marketed as a television.
<sherub>
If I understand correctly you mean that I first comment out the nvidia line
<simpson>
...Hm, but the kernel drivers would be there, and GLX did load...
<sherub>
the run the xrandr to pin resolution at a value
<sherub>
right?
<simpson>
Yes.
<sherub>
and the commenting out nvidia will mean reboot there
<simpson>
Yeah. For most things, just nixos-rebuild will work. For nVidia, you need nixos-rebuild *and* reboot.
<sherub>
ok cool lemme give it a shot.
<simpson>
I think that's about all the help we can give for now. Hopefully you understand a bit more about the different drivers. Good luck.
zupo has joined #nixos
<sherub>
I am new to ircs too. How do I ++ you guys
Darkmatter66 has joined #nixos
<bqv>
Just say the name: sherub++
<{^_^}>
sherub's karma got increased to 1
knupfer has quit [Quit: knupfer]
knupfer1 has joined #nixos
<{^_^}>
[nixpkgs] @treed opened pull request #97704 → [20.09] rpm: add openmp when building with clang → https://git.io/JUCp5
knupfer1 has quit [Client Quit]
knupfer has joined #nixos
Darkmatter66_ has quit [Ping timeout: 256 seconds]
<sherub>
@bqv++ and simpson++ thanks so much folks.
<{^_^}>
simpson's karma got increased to 31.000000000000007
sherub has quit [Remote host closed the connection]
knupfer has quit [Client Quit]
knupfer has joined #nixos
<energizer>
the answer to my problem was `fc-cache -r`
Blackraider has joined #nixos
<energizer>
thanks jtojnar (found your comment on github)
<stteevveen>
Hi! I don't know if I do things wrongly, but when I use fetchgit, if I do ls -a in the unpackPhase, there is the .gitignore file listed, but not the .git file. Consequently I cannot use the git command in the unpackPhase so as to checkoutand set current the needed branch of a project. Is it normal that I should not see the .git file after a
<sherub>
Hi folks thanks for all the help earlier. The display properly works after running `xrandr --output HDMI-0 --mode 1600x900`. Now how do I turn this into a declarative nixos config? I looked through options and tried something like, xrandrHeads = [ { output = "HDMI-1"; primary = true; } ]; and resolutions = [{ x = 1600; y = 900; }]; but it
<sherub>
did not work
<{^_^}>
[nixpkgs] @Lassulus merged pull request #97452 → zandronum: use new hg url → https://git.io/JUnie
fendor has quit [Read error: Connection reset by peer]
<srid>
I'm trying to install NixOS on my X1 carbon, but it just is not recognizing the USB boot drive, which I create using the official image using `dd if=.. of=/dev/sda`. What could be wrong?
camsbury has quit [Remote host closed the connection]
<srid>
I see the drive in the Boot Menu, but selecting it goes back to the boot menu.
<{^_^}>
[nixpkgs] @Ekleog pushed commit from @saschagrunert to master « cargo-kcov: init at 0.5.2 (#97691) »: https://git.io/JUWeX
leungbk has joined #nixos
<energizer>
"The option home.activation defined in impermanence/home-manager.nix does not exist" what's this about?
kahuna has quit [Ping timeout: 264 seconds]
cript0nauta has quit [Remote host closed the connection]
cript0nauta has joined #nixos
<leungbk>
What is the exact command used to obtain the hashes in the nixpkgs repo? I've tried some variants of "nix-hash --type sha256 --base32 ." on a git-checkout of debugpy at the version in nixpkgs (python-modules/debugpy/default.nix) but cannot get a matching hash.
<{^_^}>
[nixpkgs] @maxxk opened pull request #97715 → lldb_5, lldb_6: add patch for >=python-3.7 → https://git.io/JUWeh
<edef>
leungbk: nix-prefetch-git for fetchgit
<edef>
leungbk: nix-prefetch-url for fetchurl
<edef>
leungbk: nix-prefetch-url --unpack for fetchzip
<{^_^}>
[nixpkgs] @jonringer pushed commit from @drewrisinger to master « python3Packages.parso: disable on python 2.7 »: https://git.io/JUWv4
spudly- has joined #nixos
spudly- has quit [Changing host]
spudly- has joined #nixos
<energizer>
i think it wanted me to move imports=[impermanence/home-manger.nix] into home-manager.users.energizer instead of being top-level in my config
<energizer>
once i do that, it complains attribute sanitizeDerivationName missing
Izorkin has joined #nixos
<energizer>
which was indeed added after 20.03
<cole-h>
Yep
<energizer>
that means i need to switch to unstable in order to use impermanence/home-manager?
sputny1 has joined #nixos
spudly has quit [Ping timeout: 260 seconds]
spudly- is now known as spudly
<seku>
is there some impermanence thingie in home manager now?
<cole-h>
No
sputny has quit [Ping timeout: 260 seconds]
sputny1 is now known as sputny
<energizer>
seku: impermanence has a hm thing in it