<worldofpeace>
ekleog: feel free to merge and backport this PR. It looks perfect to me, and I even had it on my todo list to fix.
maddo has quit [Quit: See ya]
<evils>
oh hi worldofpeace, got a question, kicad's failing on hydra because it exceeds its output limit, but it's not broken, will this get marked as broken at the release of 20.03?
gustavderdrache has quit [Quit: Leaving.]
jb55 has quit [Ping timeout: 240 seconds]
ddellacosta has joined #nixos
<worldofpeace>
evils: that means someone would have to build it from source no matter what, so it is at least broken on the builder, so yes it will be marked as broken. This issue is probably typical of the gcc upgrade I believe. It would be a good idea to open an issue for it.
<evils>
worldofpeace: you mean the switch to gcc 9 set a lower output limit?
vykook has quit [Ping timeout: 260 seconds]
<{^_^}>
[nixpkgs] @adisbladis pushed to master « emacs: Add adisbladis as maintainer »: https://git.io/JviwO
<evils>
worldofpeace: the root issue here is that a sub-component (the 3d packages) should not be built by hydra, but are (that's about 6gb), even though that's almost just a download operation
blaggacao has quit [Ping timeout: 258 seconds]
* evils
opens an issue
letty has left #nixos [#nixos]
bvdw has quit [Read error: Connection reset by peer]
bvdw has joined #nixos
vykook has joined #nixos
<lordcirth_>
What is the clean way to specify in NixOS / NixOps that a directory should exist? In this case /srv/tftp
<lordcirth_>
Do I make my own derivation that contains the contents of it, then install it like a package?
<adisbladis>
lordcirth_: That's not something a package creates, that's typically something dealt with by systemd.
h0m1 has quit [Quit: WeeChat 2.7.1]
h0m1 has joined #nixos
mjrosenb has joined #nixos
claudiii has quit [Quit: Connection closed for inactivity]
vykook has quit [Ping timeout: 258 seconds]
<atemu12[m]>
Yeah, the simplest way would be to create a systemd service that runs a script on startup which makes sure the dir exists
justanotheruser has quit [Ping timeout: 240 seconds]
nckx has joined #nixos
vykook has joined #nixos
<flokli>
lordcirth_: use systemd.tmpfiles to describe the folder that should exist, as well as permission etc.
<flokli>
nixpkgs has plenty of examples where we do this in other places :-)
<{^_^}>
[nixpkgs] @danderson opened pull request #82623 → nixos/dysnomia: explain what dysnomia is. → https://git.io/Jviwg
m0rphism has quit [Ping timeout: 240 seconds]
<worldofpeace>
evils: Thanks, hopefully someone can help you get that fixed.
<evils>
looking into doing the writeup, apparently it fails on nixos-unstable too, and has hydraPlatforms = [], not as meta.hydraPlatforms, is this something that changed at some point?
vykook has quit [Ping timeout: 268 seconds]
<adisbladis>
evanjs: Hah! I have no idea what fixed it, but I had to restart my desktop session anyway and now setting brightness works
bhipple has joined #nixos
`slikts has quit [Quit: No Ping reply in 180 seconds.]
chloekek has quit [Ping timeout: 255 seconds]
vykook has joined #nixos
thc202 has quit [Ping timeout: 256 seconds]
`slikts has joined #nixos
`slikts has joined #nixos
`slikts has quit [Changing host]
nuncanada2 has quit [Ping timeout: 268 seconds]
<{^_^}>
[nixpkgs] @jtojnar pushed to gnome-3.36 « squash! gjs: install test libs to installedTests »: https://git.io/JviwH
<iceypoi>
question... is it cheating to install LFS and then slap nix on top of it?
<iceypoi>
I.e. can you still honestly say "I'm running LFS"
<iceypoi>
:D
<danderson>
What stuff means is up to you. If LFS + Nix is what makes you happy, go for it :)
<tomberek>
iceypoi: not sure what you mean by cheating... hrm, i'd say yes. Your kernel is from scratch.
<danderson>
I'm curious why you wouldn't use nixos though. In my mind it's basically LFS where someone's done the manual work for you
<danderson>
... but I guess that's the same argument as every distro :)
<MichaelRaskin>
Actually, in a single run of A/B LFS manual work is also done for you
<MichaelRaskin>
It's upgrades that get you
<iceypoi>
I'm just somewhat tempted to do the LFS thing once in my life
<adisbladis>
Go for it :>
<Yaniel>
you can also build your own kernels in nixos so...
<danderson>
I'm always tempted to try LFS, until I read the instructions again
<iceypoi>
just 0 interest in chasing down dependency hell
<danderson>
and then I go "ugh so much work" and go back to something else :)
<adisbladis>
You'll learn a lot anyway
<MichaelRaskin>
A/B LFS does that for you!
<MichaelRaskin>
(automated + beyond)
<danderson>
oh, ALFS works now? It's been a long time since I looked at the ecosystem
<MichaelRaskin>
I am not sure if it works _still_
<MichaelRaskin>
Back before I started using Nix, it did
rogue_koder_ has quit [Ping timeout: 265 seconds]
<drakonis>
adisbladis: that is a beautiful image.
<iceypoi>
realistically, even gentoo is more or less LFS + package manager :P
<adisbladis>
drakonis: I take every chance I get to use it :>
<drakonis>
except gentoo automates most of the work!
<iceypoi>
honestly, I find nixos is an amazing trade-off between safety, speed and flexibility
<iceypoi>
so that's what I run on my machines quite happily
vykook has quit [Ping timeout: 255 seconds]
<drakonis>
its quite good and there's always room to improve
rogue_koder_ has joined #nixos
<{^_^}>
[nixpkgs] @jtojnar pushed to gnome-3.36 « squash! gjs: install test libs to installedTests »: https://git.io/JviwA
<iceypoi>
sure, there certainly is
<iceypoi>
sometimes the read-only nix store bites me
<iceypoi>
sometimes unusual paths, it's not perfect
<iceypoi>
but it's pretty solid
<{^_^}>
[nixpkgs] @matthiasbeyer opened pull request #82624 → mixx: Build with libshout 2.4.1 → https://git.io/Jviwx
<danderson>
hmm, fonts.enableFontDir is false by default. But the option description makes it sound desirable for X systems... Anyone know why I would not want it?
<iceypoi>
and I can actually realistically try and configure stuff like drivers
<jboyens>
Hello again. I'm attempting to get a systemd user service installed, but it's not being picked up by systemd. I see it in the nix store. I'm simply using systemd.user.services.<servicename>. The service in the store looks fine.
ajs124 has quit [Quit: killed]
Scriptkiddi has quit [Quit: killed]
das_j has quit [Quit: killed]
mrpi has quit [Quit: killed]
<jboyens>
I've also compared it with other systemd user units that are in Nixpkgs and it seems perfectly normal.
mrpi has joined #nixos
ajs124 has joined #nixos
Scriptkiddi has joined #nixos
das_j has joined #nixos
<tomberek>
cole-h: turns out that the rust build expects clang which brings in errno.h the right way and gcc doesn't. Adding clang to buildInputs makes the buildPhase succeed. Now just need to make the install phase work ....
vykook has joined #nixos
<jboyens>
The service even appears in the /etc/systemd/user directory for user units!
rsoeldner has quit [Remote host closed the connection]
rsoeldner has joined #nixos
<qyliss>
adisbladis cole-h MichaelRaskin: what we talked about at CCCamp was actually not virtio-fs, but then virtio-fs came along and seems to do (almost?) everything we wanted.
<MichaelRaskin>
Ah
<cole-h>
^ colemickens actually, I think :P
<tnks>
If I have a Nix expression that evaluations to a medium-sized attrset of derivations. What's the fastest way to get a list of which attrs are cached in substituters, without building anything that's not cached?
vykook has quit [Ping timeout: 240 seconds]
<Ashy>
i'm trying to follow these steps to get nvidia prime working: https://nixos.wiki/wiki/Nvidia#Optimus but i get an error: error: The option `hardware.nvidia.optimus_prime.offload' defined in `/etc/nixos/configuration.nix' does not exist.
<Ashy>
any ideas?
<Ashy>
i'm on channel 19.09 and also have the unstable channel for one or two packages
vykook has joined #nixos
<jboyens>
Ashy: it changed recently, I think it's just `prime`
<Ashy>
tomberek: hmm, i've managed to nixos-rebuild and reboot but nvidia-settings still fails with: ERROR: Unable to load info from any available system
bhipple has quit [Remote host closed the connection]
mexisme has joined #nixos
abathur has quit [Ping timeout: 258 seconds]
mexisme_ has quit [Ping timeout: 256 seconds]
ddellacosta has quit [Ping timeout: 265 seconds]
vykook has quit [Ping timeout: 255 seconds]
orivej has quit [Ping timeout: 255 seconds]
<tomberek>
Ashy: nvidia-smi?
Shouou has quit [Ping timeout: 272 seconds]
vishnusuresh has quit [Ping timeout: 256 seconds]
blaggacao has quit [Read error: Connection reset by peer]
<T_S_>
What's up with R packages in nixpkgs? The Bioconductor packages are from 3.8 while conda pulls packages from 3.10. About a year behind. Is it simply a matter of help needed or are there more technical constraints. Who best to discuss this with?
<simpson>
Probably nobody cares enough to do the work required to update the versions listed.
<simpson>
That's maybe too generic of an answer, not specific to the R subsystem.
<T_S_>
thanks @simpson. looks like there are some tools needed to update development/r-modules/bioc-packages.nix. Can't just update the bioconductor version at the top of the file.
<bqv[m]>
Found a fibbonacci bug in my github client though…
<bqv[m]>
Every time I "retry" the failed reopen, it repeats all previous retries too
vykook has joined #nixos
<{^_^}>
[nix] @Ericson2314 opened pull request #3418 → Generalize `isFixedOutput` in preparation for CA drvs → https://git.io/Jvi6i
sudoforge has joined #nixos
<{^_^}>
[nixpkgs] @delroth opened pull request #82636 → stubby: enable Type=notify for better startup sequencing → https://git.io/Jvi6P
dansho has joined #nixos
<lovesegfault>
I don't remember the last time I was this pissed off with software
<lovesegfault>
I give up trying to get a reasonable gdb setup with nix
jmeredith has quit [Quit: Connection closed for inactivity]
<lovesegfault>
this is a pretty crap experience
vykook has quit [Ping timeout: 255 seconds]
iyzsong has quit [Ping timeout: 272 seconds]
palo1 has joined #nixos
<simpson>
lovesegfault: Agreed. The C debugging experience, in general, is disappointing; having to dig around to find the right debugging info is yet one more irritation.
vykook has joined #nixos
<lovesegfault>
Yeah, this is Rust code but it links to C
<lovesegfault>
the problem is the source gets trashed after the pkg is built
<lovesegfault>
so ofc the debug info left behind is garbage
<lovesegfault>
and points to nonexistent files
palo has quit [Ping timeout: 258 seconds]
palo1 is now known as palo
sudoforge has quit [Ping timeout: 256 seconds]
vykook has quit [Ping timeout: 240 seconds]
vykook has joined #nixos
iyzsong has joined #nixos
Heirlung has quit [Read error: Connection reset by peer]
Heirlung has joined #nixos
noudle has joined #nixos
vykook has quit [Ping timeout: 258 seconds]
bvdw has quit [Read error: Connection reset by peer]
bvdw has joined #nixos
vykook has joined #nixos
jakobrs has joined #nixos
<jakobrs>
If I try to use btrfs on my root filesystem with nixos, am I going to run into problems?
ottidmes has joined #nixos
<colemickens>
probably not any more than if you use btrfs in any other scenario ;)
<delroth>
yes, but more because of btrfs than nixos
<colemickens>
note, I make no actual claims about btrfs's stability in this year of our lord.
<jakobrs>
Does the kernel used on nixos compile with the right flags, or do I need to compile the kernel myself?
<delroth>
it should be enabled as a module by default
<colemickens>
You shouldn't need to do anything besides use it as a filesystem in your config.
<kalbasit>
Nix modules: is there a way to set all options named 'enable' to true in config?
<jakobrs>
Alright, I'll try
vykook has quit [Ping timeout: 260 seconds]
xelxebar_ has quit [Remote host closed the connection]
xelxebar has joined #nixos
<jakobrs>
Mounting a btrfs file system worked
vykook has joined #nixos
<jakobrs>
Next step: entire root partition
<infinisil>
kalbasit: You should be able to look at the `options` argument and recurse into that to discover all options
<kalbasit>
infinisil: do you have an example by any chance?
<infinisil>
Nope
<infinisil>
Um
<infinisil>
mapAttrsRecursive will be perfect for this
<infinisil>
Oh yeah, but I guess the second argument should be more somethnig like `if the last element is "enable" then set that to true, otherwise don't do anything
<iceypoi>
in the end, I put my `wantedBy` to the wrong place
<iceypoi>
it was inside `serviceConfig` and since this is not strongly typed, it just put it into the file instead of moaning. Once I put it one level above it works
<lassulus>
maybe we should check if everything in serviceConfig is capitalized? afaik all option the unit file are
<iceypoi>
and/or check that any of the well-known ones like `wants` or `wantedBy` are NOT present.
<iceypoi>
because this is virtually guaranteed an error
<lassulus>
well we can give it no default value, then people always have to give it one
<iceypoi>
i mean inside `serviceConfig`, check that there is no `wants` or `wantedBy`.
<iceypoi>
you probably want a default value, since e.g. you usually will not fill-in both wants and wantedBy for example
<iceypoi>
point is serviceConfig is a dynamic set, as in it takes any properties.
<iceypoi>
whereas the parent it typed.
vykook has joined #nixos
reactormonk has joined #nixos
szicari_ has quit [Ping timeout: 256 seconds]
vykook has quit [Ping timeout: 255 seconds]
doyougnu has quit [Ping timeout: 256 seconds]
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has quit [Remote host closed the connection]
v88m has quit [Ping timeout: 256 seconds]
vykook has quit [Ping timeout: 268 seconds]
fendor has quit [Read error: Connection reset by peer]
<jluttine>
how can i find what in my configuration.nix causes some package to be installed? i haven't installed uritools explicitly but i'm getting: uritools-3.0.0 not supported for interpreter python2.7
<jluttine>
so i'd like to figure out what i should disable in my configuration..
vykook has joined #nixos
<adisbladis>
jluttine: You can do `nix why-depends` on the system closure
o1lo01ol1o has joined #nixos
<adisbladis>
But only after it's built.
<jluttine>
but i cannot build it because of that error..
<jluttine>
maybe i can try it on my current system
<adisbladis>
Try that first
<jluttine>
how can i use the system closure as an argument..
<symphorien>
/run/current-system
<adisbladis>
jluttine: The only thing depending on uritools in nixpkgs is mopidy
<jluttine>
adisbladis: how did you figure that out?
<adisbladis>
Just greped in nixpkgs
<jluttine>
:)
<adisbladis>
Hm... Mopidy packages seems like they need some love
<adisbladis>
They are not consistent in their python version
Shouou has joined #nixos
vykook has quit [Ping timeout: 260 seconds]
<adisbladis>
I don't know if they have to be though.. How do they talk to each other?
<adisbladis>
Down the rabbit hole we go :>
o1lo01ol1o has quit [Remote host closed the connection]
<adisbladis>
jluttine: Could you help me test a PR if I come up with a potential solution?
linarcx has quit [Ping timeout: 255 seconds]
<jluttine>
adisbladis: sure! i'm not familiar with mopidy/mpd as such though, i've just used it on my home server to play music :)
vykook has joined #nixos
o1lo01ol1o has joined #nixos
<adisbladis>
I really wish our modules also had maintainers...
chagra_ has quit [Ping timeout: 260 seconds]
knupfer has quit [Remote host closed the connection]
<adisbladis>
jluttine: the mopidy-local-* extensions fail to build. They are deprecated anyway so we should just remove them.
<{^_^}>
[nixpkgs] @magnetophon opened pull request #82652 → faustCompressors: temp fix for build → https://git.io/JviS2
<{^_^}>
[nixpkgs] @magnetophon closed pull request #81301 → faust2jaqt: make it work again → https://git.io/Jvzjb
<eyJhb>
adisbladis: Maybe yeah, but can you check the output files from the npm packages? E.g. see if the src that vimplugins point to is correct?
<eyJhb>
Also.. Did you change coc-go?
<{^_^}>
[nixpkgs] @B4dM4n opened pull request #82654 → wireshark: use relative cmake LIBDIR → https://git.io/JviSa
vykook has quit [Ping timeout: 265 seconds]
blaggacao has quit [Ping timeout: 258 seconds]
Rusty1 has joined #nixos
blaggacao has joined #nixos
bvdw has quit [Read error: Connection reset by peer]
blaggacao has quit [Read error: Connection reset by peer]
vykook has joined #nixos
<eyJhb>
Hate when master won't build
bvdw has joined #nixos
blaggacao has joined #nixos
ottidmes has quit [Ping timeout: 255 seconds]
<{^_^}>
[nixpkgs] @matthiasbeyer opened pull request #82655 → xmage: init at 1.4.42V6 → https://git.io/JviSy
jumper149 has joined #nixos
vykook has quit [Ping timeout: 256 seconds]
<jumper149>
Hi, I'm trying to install Nixos aarch64 on my RPi 3 and ran into weird errors. First of all nixos-rebuild started to compile everything on it's own instead of using binary caches (do I need to set the binary caches somewhere?) and then I ran into some audit errors, which caused nixos-rebuild to get stuck.
<FRidh>
Does nixpkgs offer any package for starting a container as a user? I want to avoid the docker daemon if possible. The use case is to run nix without interfering with the current system.
<FRidh>
nixos-container is only for root I think
<manveru>
FRidh: there was nixos-shell i think
<FRidh>
manveru: interesting. It does seem to mount the users profile which is exactly what I not want :)
vykook has joined #nixos
<manveru>
well, you can also just use `(nixos { configuration = ... }).vm` and run that in qemu
<FRidh>
Right, forgot about that. I see now that nixos-shell offers a module, where you can choose not to include home
<adisbladis>
Without that the performance is abysmal
<tnks>
I just built something that I expected would be cached, and I don't want to lose the build (took a while). Is there a good command to test if path exists in a Cachix cache without risking what I've built locally?
ambroisie has quit [Remote host closed the connection]
<tilpner>
tnks: You can probably query the cache directly, with curl
matthiaskrgr has quit [Quit: Free ZNC ~ Powered by LunarBNC: https://LunarBNC.net - currently broken?]
<tilpner>
That will tell you what information cache.nixos.org has on the store path with the hash 6k9cfcrpmiljvhnmmwyw9wc0dxc8sxip
ambroisie has joined #nixos
<FRidh>
adisbladis: if I am correct you're suggesting creating a docker image and running that with podman?
<GiGa>
Afternoon, I'm having a forgetful moment
vykook has joined #nixos
<adisbladis>
FRidh: Yes
matthiaskrgr has joined #nixos
<GiGa>
Running "nix-build -A gramps" yields "error: getting status of '/home/giga/NixPkgs/nixpkgsApr2019/pkgs/default.nix': No such file or directory". My current directory is "/home/giga/NixPkgs/nixpkgsApr2019/pkgs/"
alexherbo24 has joined #nixos
<GiGa>
If I run from the gramps directory I get "error: cannot auto-call a function that has an argument without a default value ('stdenv')"
o1lo01ol1o has quit [Remote host closed the connection]
zebrag has joined #nixos
<T_S_>
I recently was asked to help out some biologists with their software stack. Managed to get a mixed nixpkgs/conda environment working. Then I had the urge to get rid of conda and build an environment that depended only on nixpkgs. Got it working, but noticed that packages hosted under Bioconductor were a year or more out of date. Looking for tips on how to attack that problem specifically and/or thoughts on the broad
<T_S_>
er question of supporting biology/medical workflows using nix/nixpkgs.
<T_S_>
ok I see how the script in pkgs/development/r-modules/generate-r-packages.R is supposed to work. probably generates a lot of broken packages that need individual attention...hence the delay.
<GiGa>
Was about to report the Duplicity build issue, then found out someone had already reported it :)
noudle has joined #nixos
knupfer has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<tnks>
man... I'd love a switch to prefix every line of a Nix build with the name of the package being built... it's so hard to read these logs to answer that one question. Does anyone have a good workaround? Or is that something worth contributing as a PR?
o1lo01ol1o has joined #nixos
<clever>
tnks: you either want `nix build` (instead of `nix-build`) or `nix-build -Q`
<{^_^}>
[nixpkgs] @svanderburg pushed to master « nodePackages: regenerate with node2nix 1.8.0 »: https://git.io/JviQn
<clever>
tnks: if you know which drv is failing, you can also just `nix-store -r /nix/store/foo.drv` to re-build it without any parallelism
reallymemorable has joined #nixos
<{^_^}>
[nixpkgs] @svanderburg pushed to master « base16-builder: regenerate with node2nix 1.8.0, add supplement.json with node-pre-gyp to fix build »: https://git.io/JviQc
<tnks>
clever: that's kind of like too far the other direction... because I want to see the progress too.
<clever>
tnks: `nix build` will show progress, but only the most recent line, from everything combined, with the name as you wanted
<tnks>
clever: I guess that's true... still, this UX seems lacking. Like... maybe I want the full log dumped to a file... maybe that would help.
<clever>
tnks: nix saves the full log to /nix/var/log/
<clever>
tnks: you can read it with `nix-store -l` or `nix log`
jakobrs has joined #nixos
<tnks>
clever: okay, I guess that helps.
<jakobrs>
Is there any planned date for 20.03, or is it just when it's ready?
<bhipple>
I'm currently using the 20.03 beta and have had no problems, so if you're feeling up for it you can upgrade now (using the same process you'd use when it's officially released)
<jluttine>
jakobrs: thanks. i'll try rebooting to the previous generation to see if that still works
jluttine has quit [Quit: WeeChat 2.7.1]
drakonis has joined #nixos
jluttine has joined #nixos
LysergicDreams has quit [Ping timeout: 260 seconds]
bukkitgerman86 has joined #nixos
<jluttine>
yep, touchpad works with the previous version. have to check this in more detail later. but if anyone else is having similar issues, i'd be interested to hear. if it's something in nixpkgs, i'd expect not to be the only one..
<{^_^}>
[nixpkgs] @bjornfor pushed commit from @B4dM4n to master « wireshark: use relative cmake LIBDIR »: https://git.io/Jvi7B
<jluttine>
found this in journalctl: input: Logitech USB Receiver Mouse as /devices/pci0000:00/0000:00:1a.1/usb3/3-1/3-1:1.1/0003:046D:C52B.0002/input/input1
<jluttine>
so looks like it had found the mouse
<jakobrs>
Does switching to the previous configuration without rebooting make the touchpad work?
cyraxjoe has quit [Ping timeout: 256 seconds]
MightyJoe has joined #nixos
<shapr>
I had a weird issue where both ubuntu and nixos said they couldn't load a chunk of my wifi firmware. That was true across cold boots, etc. I fixed it by removing the battery and power plug and frantically shaking my laptop.
Soo_Slow has joined #nixos
<{^_^}>
[nixpkgs] @toonn opened pull request #82664 → wire-desktop: linux 3.12.2916 -> 3.15.2922, mac 3.12.3490 -> 3.15.3621 → https://git.io/Jvi7r
turion has quit [Quit: Leaving.]
civodul has joined #nixos
slack1256 has joined #nixos
<jakobrs>
If switching to the previous configuration without rebooting does work, I'd recommend bisection
<jakobrs>
/nix/var/nix/profiles/system-gen-link/bin/switch-to-generation test, where gen is the generation number
<kenran>
Hi everyone, I'm trying to build a rust project (not nixified) on NixOS. The project wants to link against SDL2, and `ld` cannot find that. I added SDL2 and SDL_image to my `environment.systemPackages`. Is that the wrong place?
<LnL>
kenran: ^ yeah, systemPackages (intentionally) has no influence on builds like that
zupo has joined #nixos
<kenran>
LnL: cool, thanks! I thought it might be something like that but must have missed with my googling.
* cole-h
has no idea why trying to tab-complete `builtins.` segfaults in `nix repl '<nixpkgs>'`, but not in `nix repl`
<cole-h>
(in a VM)
<LnL>
kenran: simplest solution in this case would be running the build inside of nix-shell -p SDL2 SDL_image
<kenran>
LnL: yeah, I've done that now and wrote a simple shell.nix for it!
<LnL>
yeah, that's better if you don't want to figure this out again next time you want to build it :)
nextloop has joined #nixos
<kenran>
btw, is it "good practice" to strive and have my nix-shell environments be usable with --pure?
justanotheruser has quit [Ping timeout: 272 seconds]
<kenran>
right now, I'm relying on an existing cargo/rustc installation at least. I could ofc just add this to my shell.nix, and I've done similar things in the past. I'm just never really sure with nix, as I'm still quite new to the whole idea :)
<{^_^}>
[nixpkgs] @OPNA2608 opened pull request #82666 → cheesecutter: init at unstable-2019-12-06 → https://git.io/Jvi5I
<{^_^}>
[nixpkgs] @Lassulus pushed commit from @bhipple to master « geckodriver: upgrade cargo fetcher and cargoSha256 »: https://git.io/Jvi5R
domogled has quit [Quit: domogled]
philr has quit [Ping timeout: 256 seconds]
organixpear has joined #nixos
Mateon1 has quit [Ping timeout: 255 seconds]
Mateon1 has joined #nixos
<organixpear>
i am using nix on a non nixos distro. i have installed "htmldoc" using nix-env i htmldoc and it works fine, but the nix version is compiled without gnutls support or *something* so i have need to compile my own version. i would like to do this using nix, but all the docs i have read seem to indicate you can only really build nixpkgs on nixos. can anyone give me info on building packages in a non nixos
<organixpear>
situation?
<infinisil>
organixpear: That should work without problems, Nix/nixpkgs doesn't require NixOS
<organixpear>
infinisil: where do i start? what can i read? the nixpks manual that i found is less than helpful in practically helping me with the process.
aleph- has quit [Ping timeout: 258 seconds]
aveltras has quit [Quit: Connection closed for inactivity]
<NobbZ[m]>
In the wiki there is a page about how to do your own ISO, there the base expression is mentioned
jtcs has quit [Remote host closed the connection]
<glittershark>
all my cache.nixos.org downloads are actually stalling entirely
<NobbZ[m]>
glittershark I don't have problems
<glittershark>
mmk NobbZ[m]
<cole-h>
NobbZ[m]: Yeah, I've found that. Just wondering what the releases actually use, so I can try to track down an issue I'm having with a self-built image
<NobbZ[m]>
I haven't tried anything in the last 30 minutes or so...
gustavderdrache has joined #nixos
glittershark has quit [Ping timeout: 265 seconds]
jtcs has joined #nixos
yuken has quit [Ping timeout: 256 seconds]
yuken has joined #nixos
glittershark has joined #nixos
glittershark has quit [Remote host closed the connection]
andromeda-galaxy has quit [Ping timeout: 240 seconds]
<ottidmes>
chrisaw: Like so (just to give an example): self: super: { openstack-nova = super.callPackage ./nova.nix { }; os-brick = super.callPackage ./os-brick.nix { inherit (self) openstack-nova; }; }
<ottidmes>
chrisaw: Just in case you are not familiar with the inherit syntax (should be in the Nix manual), `inherit (self) openstack-nova;` is the same as `openstack-nova = self.openstack-nova;`
<glittershark>
yeah this is weird - it appears setting max-jobs=auto causes all downloads from cache.nixos.org to stall indefinitely
<glittershark>
passing --max-jobs=1 works fine
<glittershark>
this just started happening this morning
page has quit [Quit: leaving]
<NobbZ[m]>
Okay, I've a max jobs of 1 by default, as my system often becomes OOM otherwise...
<NobbZ[m]>
Okay, played a bit around... My son has the default theme for KDE. I have breeze dark. After I forced `Gome/GTK settings` to also use breeze dark every where and logging out and in again, it works fine!
<Ericson2314>
niksnut: are you around by any chance?
<jakobrs>
Oh, changing it back to Breeze makes the icons still appear
jakobrs has left #nixos ["Rebooting to test stuff"]
morgrimm has joined #nixos
<morgrimm>
Hey folks - I just installed on Catalina (all the volumes and volume permissions are good, normal default profile works well), but when I try and switch profiles and create a new one, it can't read the profile lockfile (not writable I imagine, permissions are denied)
<morgrimm>
How do I set writeable permissions for a new profile when nix can't create the profile in the first place?
<lovesegfault>
error: operation 'ensurePath' is not supported by store 'ssh://myserver.com'
<lovesegfault>
wat
<lovesegfault>
while trying to do a nix-copy
<lovesegfault>
*nix copy
<tilpner>
lovesegfault: Nix version on client and on server?
<{^_^}>
[nixos-hardware] @Mic92 merged pull request #150 → thinkpad/t495: fix backlight save/load and and battery conditioning. → https://git.io/Jviod
<{^_^}>
[nixos-hardware] @Mic92 pushed 2 commits to master: https://git.io/JviFs
<LnL>
tilpner: lovesegfault: FYI you can check that with nix doctor --store ssh://host
statusbot has quit [*.net *.split]
softinio has quit [*.net *.split]
dev3 has quit [*.net *.split]
Luker has joined #nixos
taktoa[c] has joined #nixos
ornxka_ has joined #nixos
liff has joined #nixos
electrocat has joined #nixos
eon` has joined #nixos
fadenb has joined #nixos
kumikumi has joined #nixos
wirew0rm has joined #nixos
statusbot has joined #nixos
Swant_ has joined #nixos
softinio has joined #nixos
dev3 has joined #nixos
erhandsome has joined #nixos
mog has joined #nixos
skorpy2009 has joined #nixos
endgame has joined #nixos
afics has joined #nixos
azazel has joined #nixos
endgame is now known as Guest15196
gchristensen has joined #nixos
catern has joined #nixos
Swant_ is now known as Swant
delli3 has quit [Ping timeout: 256 seconds]
<judson_>
Hrm. Not sure why this is escaping me: I've got an expression I'm trying to add to home-manager. I'd like to evaluate with with callPackage without having to add it yet to my HM configuration.
reactormonk has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @Mic92 merged pull request #82511 → rust_1_38, rust_1_42: drop patch version from attr → https://git.io/JviTX
<ottidmes>
If I want to set the default umask to 002 rather than 022, where should I change this? Right now I have it set in environment.extraInit, is there any better place. I know I have to configure sudo to adher to this, that is the only time it goes wrong at the moment, but I am still testing
<akegalj>
Hey all. I am confused what "Hetzner Cloud has NixOS ISO images that can be mounted on a running instance. You cannot directly choose NixOS as initial distribution when creating a VM. " means from https://nixos.wiki/wiki/NixOS_friendly_hosters ? I am just planning to use nixos so how should I interpret this. What mounting meens in this context ? (sorry if the question is dumb)
<tilpner>
akegalj: You can install NixOS, but you have to do it yourself
shapr_ is now known as shapr
<tilpner>
akegalj: E.g. with Ubuntu, you can directly order a server with it preinstalled. For NixOS, you can only mount an ISO from the webinterface, and then use that to install it yourself
<akegalj>
tilpner: does this meen I can mount my own iso or I have to install ie ubuntu and then convert to nixos (is this what they tried to said?)
shapr has quit [Changing host]
shapr has joined #nixos
<tilpner>
You can do both
<LnL>
cole-h: just noticed buildRustPackage builds depedencies twice (separate for the tests) which is kind of unfortunate
<tilpner>
You can't mount custom ISOs, but they provide a NixOS ISO for you to mount
<akegalj>
tilpner: thats what I wanted to hear. That's great news. Thanks!
<pikajude>
LnL: yeah, why does it build them again for the tests?
<{^_^}>
[nixops-aws] @AmineChikhaoui pushed 2 commits to master: https://git.io/JviN2
<danderson>
adisbladis: the tablet thing is a reMarkable tablet (remarkable.com). It's basically an e-ink tablet with a stylus, but the surface is textured like paper
justanotheruser has joined #nixos
<danderson>
so it's much more comfortable to write notes on, compared to a normal tablet where you're sliding on a glass surface
<adisbladis>
Yeah, I was just complaining about that in another irc-channel this morning and ended up finding the reMarkable
blibberblob has quit [Quit: WeeChat 2.7.1]
<adisbladis>
danderson: Your notes made me want one even more =)
<danderson>
I know a lot of people who use it to read RFCs and compsci papers. So far I only use this for notes, but it's working great so far
<danderson>
I never take notes on a computer because it's harder to correct, add scribbles, arrows etc.
<danderson>
and on paper it was a nightmare to organize
myskran has quit [Ping timeout: 258 seconds]
<danderson>
I've had this for ~2 weeks, and I have 5 active notebooks now :)
<adisbladis>
danderson: Have you tried getting your own software on there ?
<etu>
I'm pretty sure that I've had a meeting with a guy using a device like this once
<etu>
It seemed pretty cool
<adisbladis>
Because ideally I'd like something that could be live-synced over the internet
<adisbladis>
Which I'm totally fine with making myself, I just want to know if it's even possible
<danderson>
I haven't tried yet, no. There's a pretty active hacker community on the device, but I haven't looked into it at all
<danderson>
there is a builtin cloud syncing service for docs, and people have developed some linux interfaces for it
<danderson>
but that's all I know
<adisbladis>
Ok :>
<bqv[m]>
<joko "Hey, any idea with overriding /e"> should just be able to define the file in that directory. see the `target` attr
<danderson>
all I know is that my docs are saved somewhere in the Cloud if I need them
<bqv[m]>
do matrix replies come through properly on irc, btw?
<danderson>
but mostly I take notes on it, and then use the export function to email myself a pdf if I need it
<danderson>
it also has OCR functionality to convert notes to text, but my handwriting is really terrible so the conversion is also terrible :)
<danderson>
also doesn't help that half my notes are about tech jargon :)
mexisme has quit [Ping timeout: 240 seconds]
<tilpner>
bqv[m]: They come through, but they repeat a part of what you're replying to, which looks weird
<tilpner>
It's not as bad as multi-line message though)
<{^_^}>
[nixops-aws] @bhipple opened pull request #27 → Add black formatting check to PR validation in CI → https://git.io/JviNP
<bqv[m]>
where /etc/tinc is a preexisting directory
blibberblob has joined #nixos
<{^_^}>
[nixops-aws] @grahamc merged pull request #27 → Add black formatting check to PR validation in CI; switch from Travis to Github Actions → https://git.io/JviNP
<blibberblob>
Hi all, my nixos is borked right now... D: it fails at boot and drops me into a recovery shell. The error scrolls by too quickly to see but when I run `journalctl -xe` I see a message "overlayfs: missing 'lowerdir'"
<blibberblob>
I think this is related to docker but whatever's failing is preventing the network from connecting
<blibberblob>
so I can't rebuild nixos
<evils>
blibberblob: can you get to the boot menu (GRUB for example), and select a previous build there?
<blibberblob>
I don't see any previous builds in the grub menu. There's "default" and one other which I don't remember the name of and I'd have to reboot to see it
<evils>
it should be a submenu of grub
<blibberblob>
ok I can reboot in a sec
<blibberblob>
in the meantime is it normal behavior for `nixos-rebuild --rollback` to just show the help page
<infinisil>
Yeah, you're missing the `switch` argument
<blibberblob>
ok good to know
<blibberblob>
alright I was able to get a previous version to boot
<blibberblob>
how do I diagnose what's wrong?
<blibberblob>
the error flies by so fast in the boot log that I can't see what it's saying and I'm not sure how to scrollback
<evils>
i think you still have the logs of the failed boot
<blibberblob>
where do I find those
<evils>
journalctl
<blibberblob>
journalctl -xe produces a metric ton of output, I don't know what the failure is
<evils>
maybe journalctl -b -1 -u NetworkManager
<evils>
that's the previous boot, and logs for the unit network manager
<evils>
blibberblob: journalctl -b -1 will at least limit the output to that of the previous boot, shift + g to get to the end, then scroll up and look for red or yellow lines
reallymemorable has quit [Quit: reallymemorable]
<evils>
though just going down until you see a failure could avoid seeing errors caused by the root failure
<akegalj>
I wanted to try to install nixos on a virtual machine with qemu. I am not even sure can this be done on qemu, but I have tried to download graphical live CD and tried to run it with ` qemu-system-x86_64 -cdrom nixos-graph.qcow2 -boot order=d -drive file=nix-empty.qcow2,format=raw -m 1G` , but it fails with some timouts and enters to emergency mode. Is this expected? Should I use virtual box OVA instead
<ericdallo>
Hello, I'm stuck on a PR trying to make a new derivation and I can't find a header lib to fix the problem, could somebody help me with this?
<Yaniel>
you'll want libglvnd in your dependencies I think
<evils>
,locate GL/glx.h
<{^_^}>
Found in packages: nxproxy, mesa_noglu.dev, libGLU_combined, driversi686Linux.mesa.dev
<ericdallo>
Uow, that's a useful command for this channel. Thank you! I gonna try it!
myskran has joined #nixos
<danderson>
so, I'm finishing up my new laptop setup on NixOS. I have the basics (hardware support etc) figured out, now I'm applying polish.
<danderson>
Currently I manage my home dotfiles in git, by hand. Anyone have a favorite nix-based tool for that?
<niso>
ericdallo: you can install nix-index (nix-locate is the command mentioned before)
<danderson>
And, similar question for per-project programs+shell configuration, a-la direnv. I have Lorri written down in my setup notes, anything else I should check out?
<Yaniel>
home-manager, but lorri might be the successor to that?
<akegalj>
cole-h: thanks for the suggestion above - that did fix my problem
cosimone has quit [Remote host closed the connection]
<ericdallo>
Thank you for the tip niso
cosimone has joined #nixos
<danderson>
Yaniel: lorri looks like "direnv+nix for per-project hacking setups", rather than dotfile/etc management
<danderson>
but maybe I haven't dug far enough yet
<judson_>
danderson: 100% translate handwritten configs to home-manager, and use lorri (+direnv nix-shell) for dev environments
civodul has quit [Quit: ERC (IRC client for Emacs 26.3)]
philr has joined #nixos
<judson_>
Lorri is not for dotfiles.
<Yaniel>
oh
<danderson>
judson_: cool, good to know I seem to have accidentally figured out the right combo of software :)
blibberblob has quit [Ping timeout: 246 seconds]
<myskran>
danderson: what do you count as a nix-based tool?
cosimone has quit [Quit: Terminated!]
myskran is now known as abathur
<abathur>
heh
<judson_>
abathur, I'm curious what you have in mind
reactormonk has quit [Ping timeout: 240 seconds]
<danderson>
abathur: good question, I'm not sure :) I guess the things I want are: ability to install/upgrade software separate from the base OS, declarative spec of my configs, and ability to roll back painlessly
<danderson>
which seems to add up to "something something nix plus other stuff"
<danderson>
but admittedly I don't have a clear idea.
<danderson>
The two problems I'm trying to solve are: proliferation of random OS-installed software I only use for a single abandoned project (looking at you, npm garbage)
<danderson>
and the fact that managing my dotfiles in plain git hasn't scaled out, my configs diverge over time because I forget to sync stuff up.
<abathur>
yeah, AFAIK that's still roughly the state of things
<danderson>
forcing myself to use nix everywhere forces me to update the config and push it as the only way of changing anything
<danderson>
which _should_ make it easier to avoid deviations
<danderson>
(famous last words etc.)
<abathur>
yes; I still tend to drift a bit
<abathur>
I basically use the git approach, though via yadm
<cole-h>
I think you should give home-manager a try, danderson. While it won't solve everything, I enjoy it a lot
<cole-h>
Been slowly moving over my Arch-based system to it so the jump to NixOS is relatively painless. So far, it's been a good experience
<abathur>
I wish there was something a little more declarative, but I'm also not all-in on the nixify-all-config approach; some config needs to be cross-platform anyways
<danderson>
Cool, I'll get on that. I need to figure out something by tomorrow, because I get back to work and I need this laptop to be ready. Yay hacking on a deadline :)
<cole-h>
:D
<danderson>
yeah, I'm on the fence about dotfiles because in theory, I might want to use them on non-nix systems
<abathur>
and I'm leery of snookered into a very long yak-shave rewriting config stuff into nix
<danderson>
colemickens: fancy. I got secureboot working on my arch machine, so I'm very familiar with the mechanics of it. The one thing I don't know how to do is inline kernel signing in nixos activation
<{^_^}>
[nixpkgs] @Ma27 opened pull request #82691 → pass: don't rebuild when adding extensions, pass-audit: 0.1 -> 1.0.1 → https://git.io/JvixV
<danderson>
seems problematic to have a synchronous step in the current nix world, but I'd need that for "insert yubikey and tap to sign"
<{^_^}>
[nixpkgs] @doronbehar opened pull request #82692 → sacd: init at 19.7.16.37 → https://git.io/Jvixw
<danderson>
(note I'm going full-on secure boot: signed bootloader, signed kernel+initrd blob, only shows the drive decryption prompt if booting from a signed image...)
blaggacao has joined #nixos
<colemickens>
Graham has other fun stuff like their non-nix, non-var filesystems get wiped on boot to prove the nix-y stateless nature of the box. not sure if that's active, but we talked about it many months back and produced a working example
<evils>
niso: try #nixos-aarch64
<danderson>
that exists in mainline, right? `touch /etc/NIXOS_LUSTRATE` wipes everything except nixos on boot
<niso>
evils: thanks
<danderson>
that's how nixos-infect et al. take over the system
<colemickens>
danderson: I don't remember which pieces you need for the "only show prompt if you trust TPM" but I think this hits the other points you mention: https://github.com/NixOS/nixpkgs/pull/53901
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<{^_^}>
#53901 (by grahamc, 1 year ago, open): WIP: Sign systemd boot EFI images for secure booting.
<danderson>
TPM trust requires a bunch of extra work, but you can at least have the initrd ask the firmware if it was secure-booted
<colemickens>
kind of wonderfully simple, other than worrying about key storage/etc
<danderson>
that way the threat surface drops to "you have to pwn the firmware to boot a hostile kernel"
<danderson>
you need some help from the OEMs to sanely verify the firmware via TPM attestation, but one step at a time :)
chloekek has quit [Ping timeout: 255 seconds]
<danderson>
The problem I have is that I want to have some interaction to sign new things. My Arch laptop doesn't do that, it does a similar thing to gchristensen's PR: signing keys on disk, just sign new kernels as needed.
<gchristensen>
danderson: my / filesystem is ZFS, and it rolls back to a blank snapshot on every boot. I'm not sure if LUSTRATE does that exactly, I think it does more complicated things
<danderson>
which is fine, depending on the threat model. As long as your userspace isn't compromised, and the disk is encrypted, naked signing keys are fine
<danderson>
But I'd like to go one step further and store my kernel signing key in my yubikey, to require a key tap to sign new things
<danderson>
(for no good reason, other than "I'm a huge nerd")
<Yaniel>
I keep hearing great things about ZFS
<Yaniel>
but also horror stories
<gchristensen>
horror stories?
<Yaniel>
catastrophic data loss etc
<danderson>
aren't you thinking of btrfs?
<gchristensen>
sounds more like btrfs, circe 2015
<danderson>
ZFS is famous for being quite hard to lose data
<Yaniel>
no, btrfs I *know* :P
<danderson>
early ZFS would panic a bunch, go "uuuh something's wrong I'm not touching this", but afaik those were all recoverable with expert help
blaggacao has quit [Read error: Connection reset by peer]
<Yaniel>
it was the timebomb that almost killed my phone a couple of times
<gchristensen>
I lit my server on fire last week and didn't manage to lose any data :)
blaggacao has joined #nixos
<danderson>
but ZFS these days is rock-solid. It's the only FS I'd trust for precious data
* gchristensen
is going to get a lot of mileage out of that one
<danderson>
gchristensen: uh, story time. WTF?
<Yaniel>
nice nice
<Yaniel>
guess I'll have a look at ZFS for my desktop upgrade then
<Yaniel>
are there instructions for setting it up on nixos?
<cole-h>
Nothing comprehensive, I've found, but there are some posts and a wiki entry
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « python27Packages.braintree: 3.59.0 -> 4.0.0 »: https://git.io/JvixM
<gchristensen>
danderson: I plugged a NVIDIA Telsa K80 in. it has special wiring needs, with some pigtail adapters and other uniquely annoying requirements. needless to say, i remembered wrongly. I plugged it in, powered it on, got a very nice lightbulb effect from several of the cables before they burst in to flame
<cole-h>
wtf
<danderson>
woops
<danderson>
that's some expensive magic smoke
<Yaniel>
ouch
<gchristensen>
yeah. ... yep :)
<Athas>
gchristensen: that would have been worse a few years ago. A K80 was pretty expensive card.
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « python27Packages.azure-mgmt-compute: 10.0.0 -> 11.1.0 »: https://git.io/JvixH
<gchristensen>
they have lived in my basement since, because they take too much power and are too expensive to run.
<Athas>
It is still that most honourable of GPUs: one with no monitor outputs.
<danderson>
reminds me of a story from a former employer. Someone got paged for a machine failure. While troubleshooting, another machine failed, then another, then another.
ddellacosta has joined #nixos
<gchristensen>
uh oh
<danderson>
From the hostnames, they figured out that the machines were all in one rack, and failing bottom-to-top
<Athas>
Water leak!
<danderson>
so they called the datacenter, and said "rack 48 is on fire."
<Athas>
Oh, sign error.
<gchristensen>
hah
<danderson>
"what? No it's not, we have fire suppression systems, we'd know"
<danderson>
"I promise you, it's on fire. Go check."
<danderson>
5 minutes pass
<danderson>
"Yup, it was on fire. Sorry about that. Fixed."
<gchristensen>
"Fixed." hoo boy
<Yaniel>
"Sorry about that. Fixed." xDDD
<danderson>
turns out the bottom machine underwent "sudden catastrophic shutdown". As the heat and flames moved up the rack, each machine in turn hit overtemp limit and shut off
<cole-h>
Maybe it's my inexperience, but I'm shocked that they were able to grasp that it was fire from machines failing one after another
<danderson>
gchristensen: this was in the before-times, in a fairly janky datacenter setup :)
<niso>
turns out mounting a copy of /nix on /nix kills the system :/
<danderson>
I'm sure it's way better now
<judson_>
I dumped all my dotfiles into a directory, and they get copied into place during home-manager switch, along with a dire warning that they're "in transition"
<gchristensen>
hehe, nice
<judson_>
I've slowly whittled away at either translating them or considering them permanently "just text"
<gchristensen>
yes back when these K80's were in use, they were in a low-budget datacenter and I would not be surprised if they had caught fire :P
<danderson>
judson_: nice, thanks for the tip
__monty__ has quit [Quit: leaving]
<danderson>
heh, I have considered building some kind of hermetic enclosure around my home rack and just flooding it with CO2
<gchristensen>
at any rate, it very luckily destroyed 1 HD in a raidz1, and another HD in another raidz1, and I have replacements on their way.
<danderson>
as the ultimate "thou shalt not catch on fire" solution
<danderson>
well, I guess nitrogen would be better. CO2 still has oxygen atoms in there, if you try hard enough :P
Fuzen is now known as fuzen
<danderson>
but then remembered that's a really stupid idea, and stopped.
<danderson>
that's quite lucky :)
<danderson>
but hey, good outcome :)
<gchristensen>
yep!
<gchristensen>
big green check mark: it worked, my data is alive
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « python37Packages.elementpath: 1.4.1 -> 1.4.2 »: https://git.io/Jvixb
<gchristensen>
one of the problems with gasses which displace oxygen is when you accidentally displace *your* oxygen, and it can be hard to know when you're in danger
<cole-h>
gchristensen: Somewhat off-topic, but what is the best way to familiarize with zfs? I plan to use it when I switch to NixOS, but since that's one of the first steps...
<danderson>
gchristensen: yeah, exactly. I think eventually I decided that this would be an excellent way to accidentally kill my cats, and I'd rather just let the rack burn
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « fly: 5.7.2 -> 5.8.0 »: https://git.io/Jvipv
<MichaelRaskin>
Make sure it is hermetic, because in principle there are conditions where mineral oil itself burns
<danderson>
downside: all parts will be greasy for ever (completely impossible to clean), and you need a couple hundred gallons of storage to pump out the rack for servicing
<danderson>
... or buy a $10 smoke detector and insurance :)
<gchristensen>
and offsite backups
siiky has quit [Quit: leaving]
<simpson>
danderson, MichaelRaskin: I have done small-scale mineral oil. Can confirm that it is greasy forever. Surprisingly, it *doesn't* ignite readily; its specific heat is pretty ridiculous, and usually won't ignite unless vaporized.
<tomberek>
simpson: any need for pumps?
<simpson>
By "small-scale" I mean that I helped dunk some disassembled boards into a fish tank, which we then filled with mineral oil. Absolutely disgusting, but we were teenagers.
siiky has joined #nixos
<simpson>
tomberek: We used standard fans to circulate the oil. Changing oil probably would have been easier with a pump, but we didn't have one.
<niso>
you could also just assume that some servers might burn, thus you make the rest of the datacenter non-flameable
<danderson>
if this was in a datacenter I'd be less worried. They have plenty of fire detection and suppression gear
<danderson>
my basement doesn't :)
<niso>
surround it with buckets of water? :P
<danderson>
someday, I'll redo the electrical supply and have a bunch of emergency shutoff mushroom switches
<danderson>
there's nothing easily flammable in there, so it's mostly about "shut off electricity to make the electrical fire stop"
<aleph->
gchristensen: Thanks for that zfs article. Definitely can't wait to spin it up on my NAS come this summer
organixpear has joined #nixos
<organixpear>
i am trying to get xorg work on a non nixos distro using xorg from nixpkgs. i can get xorg to load and display a wm but i have no input. my distro is non-systemd. i have dbus installed plus xf86-input-libinput,-evdev,keyboard,mouse. what confounds my situation even more is X stopped producing logs...
ericdallo has left #nixos ["ERC (IRC client for Emacs 28.0.50)"]
<Yaniel>
you really love a challenge don't you
<evils>
aleph-: do you have any uncommited stuff towards getting kali's tools on nixos?
<aleph->
I do.. need to find the drive they're on
gustavderdrache has joined #nixos
nuncanada2 has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
orivej has joined #nixos
cosimone has quit [Remote host closed the connection]
cosimone has joined #nixos
o1lo01ol1o has quit [Ping timeout: 246 seconds]
<MichaelRaskin>
organixpear: just in case — eudev is running, right?
<{^_^}>
[nixpkgs] @flokli pushed 3 commits to release-20.03: https://git.io/Jvipu
Shouou has quit [Ping timeout: 272 seconds]
<organixpear>
I'm getting logs again. the logs show that the /nix/store paths for the xorg modules are being picked up but modules that i _know_ are install like intel the log reports that the module does not exist
akegalj has quit [Quit: leaving]
<hexa->
how can I get opencl with amd navi running when using wayland?
<{^_^}>
#62933 (by athas, 39 weeks ago, open): Restoring OpenCL support in Mesa
<organixpear>
the mesa-loader is looking in /run/opengl-driver/lib/dri. my real question is: is the nixpkgs xorg configured only to use systemd or something?
<Yaniel>
or by figuring out how to set up rocm 3.2 which is not out yet
<Yaniel>
that said, if you do manage it let me know, I'm very interested in getting that working as well
<hexa->
Yaniel: thanks, reading up
<hexa->
killercup[m]: might also be interested according to discourse ^ :D
<hexa->
lovesegfault: I'm having issues with folding@home, it's trying to run gromacs, but fails: https://bpaste.net/FG4A
reallymemorable has joined #nixos
<lovesegfault>
hexa-: hey
<lovesegfault>
hexa-: Are you running the right version?
<lovesegfault>
f@h 7.5.1 landed in master recently and it's what you should be using
<{^_^}>
[nixpkgs] @veprbl pushed commit from @symphorien to master « clementine: fix broken symlink (#82689) »: https://git.io/JvihU
dycan has joined #nixos
Soo_Slow has quit [Remote host closed the connection]
<hexa->
htop
<hexa->
oops
Soo_Slow has joined #nixos
<evils>
hexa-: there's amdgpu-pro, which didn't obviously fail last time i tried it, but still doesn't seem to work; there's also nixos-rocm: https://git.io/Jvihk that also doesn't die, but causes blender to crash for me...
<Yaniel>
evils: rocm only supports cards up to vega though :(
fenedor has quit [Quit: Leaving]
<Yaniel>
3.1 silently adds some navi stuff but not full support
<evils>
ah :(
<hexa->
evils: the only way to enable amdgpu-pro seems to be through xserver.videoDrivers, but I don't have an xserver x(
<lovesegfault>
hexa-: opengl.extraPackages?
<hexa->
possibly
<dycan>
hi, how to bring a file into a mkderivation and (in build phase) use ghc --include -optP(FILE_PATH)? I pwd and ls. the pwd is in /build/extra-files. ls shows it is there. But ghc says cc1: fatal error: -optP/build/extra-files/cabal_macros.h: No such file or directory
<dycan>
okay, I may just have typed something wrong...
troydm has quit [Ping timeout: 260 seconds]
maddo has quit [Quit: See ya]
reallymemorable has quit [Quit: reallymemorable]
Rusty1 has quit [Remote host closed the connection]
cosimone has quit [Quit: Quit.]
klntsky has quit [Remote host closed the connection]