<{^_^}>
[nixpkgs] @Ericson2314 pushed 2 commits to release-20.03: https://git.io/JJa6U
afreak has quit [Ping timeout: 240 seconds]
<ToxicFrog>
Hrm
<ToxicFrog>
I have a patch for the borgbackup module.
<ToxicFrog>
I want to, if I can, copy the important parts of it into my /etc/nixos so that my local nixpkgs checkout is just for submitting patches and not what I'm routinely building the system against.
<ToxicFrog>
However, I can't just copy the module into /etc/nixos and import it because then it collides with the existing one in nixpkgs.
<ToxicFrog>
Is there an equivalent to `overlays` for nixos modules that lets me patch the existing module? Failing that, is there a way to undeclare the existing module so I can replace it outright?
alp has quit [Ping timeout: 256 seconds]
<evanjs>
Also, lol what is up with GH actions right now? Every time I comment on an issue I have open I get (two?) emails about failed jobs
<iwq>
is there any alternative for generating Nix code than just printing lines of text without any parser?
tobiasBora2 has joined #nixos
tobiasBora2 has quit [Ping timeout: 240 seconds]
Rusty1 has joined #nixos
<infinisil>
iwq: Generate JSON, then import the json from Nix and transform it to what you need
philr_ has joined #nixos
<energizer>
i need my derivation to have access to the `gatttool` binary. how do i provide that? i tried the new dependencyEnv, but `The command was not found or was not executable: gatttool.`. https://bpa.st/ZVAQ
<ToxicFrog>
Ok, new question: how do I change the arguments a package is called with?
<ToxicFrog>
recoll is broken in unstable (it's missing a Qt dependency) but the dependency is for the GUI, which I don't need; building of the GUI is gated on the `withGui` argument to the package, but I can't see any way to change that short of modifying all-packages.nix
<infinisil>
siraben: Left a small comment, but it looks good after that :)
<energizer>
infinisil: oh i think dependencyEnv isnt what i want, but something's still wrong https://bpa.st/C5WQ
<infinisil>
energizer: What if you remove your custom override
<siraben>
infinisil: After rebasing I did a git pull and git push, and merged it, does that sound right?
<infinisil>
energizer: Does the wrapper then have NIX_PYTHONPATH set somewhere?
<infinisil>
siraben: Ah no, you need to force push, no merging
Emantor has joined #nixos
<energizer>
infinisil: then it does the pythonusersite thing so btscale module exists but i lose gatttool
<infinisil>
siraben: So I see you rebased to commit 75efe4ce065, but that still contains 3 commits in total, 2 of which are for adding the package
<siraben>
Hm, I'm unsure of what to do now.
<siraben>
git rebase -i 5f5188706366c49ba956ac8d7cb777fda29acc98 ?
<infinisil>
siraben: So I suggest doing: `git reset --hard 75efe4ce065` to undo the merge. Then `git rebase -i HEAD~3` to change history again. Then squash the last commit into the second last one (read instructions there)
<siraben>
Ok
<infinisil>
Then `git push -f`
<infinisil>
When you `git log` there should only be 2 commits by you then
_deepfire has quit [Remote host closed the connection]
bbennett371 has quit [Ping timeout: 244 seconds]
<infinisil>
siraben: Yeah, but let's wait until the first PR is merged, otherwise you'll have to rebase again because the maintainer commit will give a conflict
<siraben>
infinisil: Thanks for the merge, so what do I do about the other PR?
<infinisil>
siraben: Alright, first PR merged, now ideally you'd rebase the second PR on top of latest master, such that you have the maintainer commit in your branch
<infinisil>
siraben: I usually do it like this: git log to find out how many commits you added in your PR branch (say N), then `git fetch upstream` (or however your remote is named) to fetch new commits. Then rebase ontop of master with `git rebase --onto upstream/master HEAD~<N>` where N again is the number of commits you have
<infinisil>
After that, do the rebase -i again to squash everything related into a single commit
stephaneyfx has quit [Read error: Connection reset by peer]
<WRMilling>
Hello all, I am attempting to add a deviceTree.overlay to my nix configuration, but it keeps giving me a file not found (with a path that I can verify actually exists), just a bit confused why nixos-rebuild can't seem to find it. I am hoping I am doing something silly and maybe someone can put me on the right track.
simba1 has quit [Ping timeout: 272 seconds]
orivej has joined #nixos
sjkelly1 has quit [Remote host closed the connection]
<WRMilling>
Let me know if you would like to see anything else
<infinisil>
WRMilling: Ah I see the problem, replace ./dtb-overlays/overclock.dtbo with "${./dtb-overlays/overclock.dtbo}"
<infinisil>
This is a bug in nixpkgs
<WRMilling>
Whats the difference? (or should there be a difference?)
<infinisil>
The problem is that device-tree runs toString on the overlay paths
<infinisil>
> toString ./some-path
<{^_^}>
"/var/lib/nixbot/state/nixpkgs/some-path"
<WRMilling>
Got a more relevant error (at least something I can dig farther one, FDT_ERR_NOT_FOUND)
<ToxicFrog>
gchristensen: in the meantime, will setting `networking.domain` to the tail have a similar result?
<infinisil>
And this turns the path into a path where the file directly is
<gchristensen>
ToxicFrog: I'm not sure to be honest
<gchristensen>
ToxicFrog: it depends on your use case
<ToxicFrog>
Diffing implies that it probably will, but....
<infinisil>
WRMilling: However since the path is needed during the build, which runs in a sandbox, it can't access that
<WRMilling>
That makes sense.
<infinisil>
WRMilling: By using "${./...}", this explicitly copies the file into the Nix store, which is accessible in the sandbox
<infinisil>
iwq: No
mrosenbe_ has joined #nixos
mrosenbe has quit [Read error: Connection reset by peer]
<WRMilling>
Thank you for the quick help infinisil
kraeXen has quit [Remote host closed the connection]
<infinisil>
:)
<ToxicFrog>
Hrm. /etc/hostname is going to end up as `ancilla` rather than `ancilla.ancilla.ca`, though. Not sure what that will break but it's probably going to be something.
gustavderdrache has joined #nixos
kraeXen has joined #nixos
<gchristensen>
ToxicFrog: yes, the change is very heavy handed for a property which has no requirement to be so strict
orivej has quit [Quit: No Ping reply in 180 seconds.]
<ToxicFrog>
Actually, it might be ok
proofofkeags has quit [Remote host closed the connection]
<ToxicFrog>
Appropriate entries in the hosts file are still written, so gethostbyname should still do the right thing
<WRMilling>
It looks like the overlay is trying to apply it to every dtb that exists in the kernel's dtbs directory, is that correct? (Which would be why it is failing with NOT_FOUND error)
o1lo01ol1o has joined #nixos
bbennett371 has quit [Ping timeout: 244 seconds]
user12345 has joined #nixos
o1lo01ol1o has quit [Ping timeout: 256 seconds]
orivej has quit [Quit: No Ping reply in 180 seconds.]
kraeXen has quit [Ping timeout: 260 seconds]
orivej has joined #nixos
bbennett371 has joined #nixos
stephaneyfx has quit [Remote host closed the connection]
stephaneyfx has joined #nixos
matthewcroughan has quit [Ping timeout: 240 seconds]
sigmundv has quit [Ping timeout: 265 seconds]
<infinisil>
WRMilling: I'm just seeing https://github.com/NixOS/nixpkgs/pull/79370 which does some changes around device-tree, one of which is to add a filter, maybe that would work for you?
<energizer>
can i have systemd delay starting my service again depending on its exit code?
fresheyeball has joined #nixos
lambda-11235 has joined #nixos
Supersonic112 has joined #nixos
sigmundv has joined #nixos
maier has joined #nixos
tobiasBora2 has joined #nixos
Supersonic has quit [Ping timeout: 260 seconds]
Supersonic112 is now known as Supersonic
<danderson>
no, systemd doesn't have sensible backoff built into it :(
<WRMilling>
I am starting to think that having nix run a script to apply the patch might be easier right now (since the file location relative to the kernel will be known).
<danderson>
that would do a static sleep, rather than backoff
<danderson>
if you want to make it work exponentially, you need to retain state between runs, which you can't do
<danderson>
either that, or add a wrapper program that runs your actual program and babysits it without ever exiting
<danderson>
which works, but then breaks integration with systemd in various ways (e.g. you can't tell if your daemon actually restarted, because the various lifecycle events never happen in systemd, because from systemd's perspective your process monitor hasn't exited)
<danderson>
another idea we had at tailscale was have the daemon keep its own backoff state on disk, and sleep during startup if it's been restarting a lot
<danderson>
which works, except if the filesystem is read-only, or other weird problems that coincidentally would probably also cause a crash-loop in the first place
<gchristensen>
nice idea for other less capable inits too
<danderson>
so we concluded that if writing the restart state fails, to actually work right you have to be super conservative and sleep for your maximum backoff time
<danderson>
... which then leads to problems on stateless-ish systems that deliberately wipe /run and whatnot between executions
<danderson>
because in those systems the filesystem looks like it's working, but then something external to the program is sabotaging you between runs
* gchristensen
kicks dirt
<danderson>
it just never ends, and would be solved trivially by having systemd just do it for us :/
<danderson>
(this whole discussion was kicked off by just such a broken machine somehow losing its entire on-disk state between runs *and* also crashlooping rapidly, which generated a ton of noise and load on our servers)
<gchristensen>
maybe there is a PR or issue about this
<danderson>
(so it's not a hypothetical scenario, we were actually trying to figure out how to defend against a completely byzantine case we were seeing for real)
<danderson>
there's a systemd issue, yes
<danderson>
with lots of people saying they want it, and nobody offering a patch
<danderson>
so, we get the init system we deserve :)
<gchristensen>
:)
<danderson>
(I expect it actually turns out to be kinda challenging to keep this state, for some reason, and that's why it's not been done)
<gchristensen>
it would take 15 years for rhel/debian to take it anyway
<danderson>
yup, and that's the other problem
<danderson>
fixing it now means we can actually rely on it some time in 2030
<gchristensen>
it'll be faster to wait for nixos to take ove
CodeSpelunker has joined #nixos
<danderson>
heh
<danderson>
another ridiculous idea we had was to ship a _second_ service
<danderson>
whose only job is to sit there and keep track of how often the main one restarts
<danderson>
and then tell it over a unix socket how long to sleep on startup
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to release-20.03: https://git.io/JJaDw
<danderson>
but at that point we're clearly 100% being ridiculous
<aaronjanse>
I'm doing it because Redox cross-compiling uses pkgsStatic
<aaronjanse>
Is there a straightforward way to work around this? I'm fine with xorg not working; the problem is that I think the large number of recursion errors is causing issues
CodeSpelunker has quit [Quit: CodeSpelunker]
proofofkeags has quit [Remote host closed the connection]
<infinisil>
WRMilling: If you have the time, it would be awesome if you could see if that PR works for you by applying it to your system, would help with getting that PR merged :)
<WRMilling>
I was actually just typing to ask how to apply the PR
<WRMilling>
The filter does look like it would solve my problem.
sentinal8473 has joined #nixos
<infinisil>
Oh welcome then!
<WRMilling>
Danke, trying to get a full config working before I install to internal storage (running on SD card right now)
proofofkeags has quit [Ping timeout: 240 seconds]
<infinisil>
To apply the PR you'll have to get a git checkout of nixpkgs, then check out the commit of your current system (nixos-version displays the git commit at the end of the version string), then fetch the PR branch and cherry-pick its commit on top. Finally you can use `nixos-rebuild switch -I nixpkgs=/path/to/nixpkgs` to make it use that nixpkgs
pjt_014 has joined #nixos
jb55 has quit [Remote host closed the connection]
jb55 has joined #nixos
<WRMilling>
What would be the proper branch for 20.09pre236091.28fce082c8c ?
<WRMilling>
Also, what a large repo
<infinisil>
aaronjanse: Don't use nix repl for debugging, it can be very misleading because of its odd behavior in regards to caching and laziness. You can use `nix-instantiate --eval default.nix -A pkgsStatic.<TAB>` or so instead
proofofkeags has joined #nixos
<infinisil>
WRMilling: The 28fce082c8c is the commit, branches are moving though so you can't really rely on those
<WRMilling>
Makes sense
<infinisil>
I guess you could just use the stable branch though
<infinisil>
Which might make your system update by a bit
<infinisil>
WRMilling: The stable branch would be nixos-20.03
<infinisil>
Oh you're using 20.09 though (unstable)
<infinisil>
No branch for that then :)
<infinisil>
Oh wait no there is duh, it's nixos-unstable
<WRMilling>
using a image overlay by samueldr for the pinebook pro, so far so good.
<WRMilling>
Uses unstable by default. (better aarch64 support I think)
<infinisil>
I see
<colemickens>
How do Flakes work in terms of channels for regular users?
<colemickens>
Is the idea that instead it's just flake urls that resolve via the registry?
<drakonis>
colemickens: have you been checking out the other side of the fence?
<drakonis>
oh my
<drakonis>
flakes got finally merged
<drakonis>
holy moly
<colemickens>
Maybe? I've been "on" flakes for a while, but i still maintain my own nixpkgs so I don't care about channels normally.
<colemickens>
But I'm trying to figure out how to do "cloud VM image" in a flakes world and an out-of-tree module.
<infinisil>
drakonis: Experimental still though
<drakonis>
i'm aware
oida has joined #nixos
<colemickens>
Is it even clear what is laid down for a flake-based install?
<colemickens>
Does it still just drop a configuration.nix ?
kalbasit has quit [Ping timeout: 240 seconds]
drakonis has quit [Quit: WeeChat 2.8]
<WRMilling>
The PR seems to have worked (if my overlay dtbo is going to work, is another question)
o1lo01ol1o has joined #nixos
<aaronjanse>
> holy moly
<aaronjanse>
Holy cow
drakonis has joined #nixos
<{^_^}>
undefined variable 'holy' at (string):318:1
<WRMilling>
Going to have to disconnect to test to update, wish me luck...
codygman has quit [Read error: Connection reset by peer]
codygman has joined #nixos
<hsngrmpf[m]>
I'm not sure right now if I remember correctly. But wasn't there a tool that shows me the version history of programs packaged in nixpkgs? Let's say I need an older version of some program and want to know which nixpkgs commit has this program version.
o1lo01ol1o has quit [Ping timeout: 240 seconds]
WRMilling has quit [Ping timeout: 245 seconds]
WRMilling has joined #nixos
<WRMilling>
It worked, my max freqency is higher now (dtbo applied), but now I have a problem with it not taking all the frequency changes. The PR works well.
justanotheruser has quit [Ping timeout: 260 seconds]
<aaronjanse>
infinisil: I think the evaluation error was caused by recursion in `pkgs`
o1lo01ol1o has quit [Ping timeout: 256 seconds]
<aaronjanse>
I'm now trying to figure out how Hydra jobsets are created so that `hydra-eval-jobs` doesn't split out `buildPackages.buildPackages.buildPackages.buildPackages...`
<aaronjanse>
*spit out
fresheyeball has quit [Quit: WeeChat 2.7.1]
pjt_tmp has joined #nixos
<{^_^}>
[nixpkgs] @kasimon opened pull request #94414 → Fix typo in linkerd package name → https://git.io/JJa7t
<raboof>
hmm. it seems tensorflow 2.3.0 specifically requests bazel 3.1.0 to build, but we have 3.3.0. what would be the way to resolve that? create a derivation for bazel 3.1.0 or patch tensorflow to build with 3.3.0?
buckley310 has quit [Quit: Connection closed for inactivity]
o1lo01ol1o has quit [Ping timeout: 265 seconds]
<c0c0>
I'm having trouble setting up NixOS with i3 on a new machine. After booting, I see LightDM, but after typing in my credentials, I see "Failed to start session". Here are my configuration.nix, hardware-configuration.nix and home.nix (they are absolutely minimal) and the X.0.log: https://gist.github.com/corsinpfister/70f0e7bff46e90345aab9305916d8646 I'm relatively new to Nix(OS) and appreciate any
<aaronjanse>
c0c0: maybe a window manager needs to be defined?
<aaronjanse>
Tomorrow afternoon I'll be able to try out that config in a VM. It's almost 3 am for me, so I don't have the energy to do it tonight
turlando has quit [Quit: Leaving]
turlando has joined #nixos
octowan[m] has joined #nixos
bbennett371 has joined #nixos
cyphase has joined #nixos
domogled has joined #nixos
doyougnu has joined #nixos
bbennett371 has quit [Ping timeout: 244 seconds]
nature has joined #nixos
<nature>
Hi from nixos :) (just finished installing it)
<aaronjanse>
Congratulations, nature!
<TheSirC[m]>
Glad you switched !
<nature>
I really love it so far
<nature>
I just have a little problem starting x (i am in the tty right now ^^')
__monty__ has joined #nixos
gthm has quit [Ping timeout: 240 seconds]
<aaronjanse>
Oh ha. What's the problem?
<nature>
Does any of you have some idea of what might be wrong ? I did setup a xinitrc, I enabled startx and xserver, bspwm (my wm) is setup correctly too but when I startx I have a black screen
<TheSirC[m]>
nature: Oh wow, I am also troubleshooting a Xorg problem
domogled1 has joined #nixos
<nature>
looking at the xorg.log I found that screen 0 is deleted because no matching config
<nature>
oh really TheSirC[m] ? What's up with you ?
bbennett371 has joined #nixos
<TheSirC[m]>
nature: SAME
<nature>
okok, let's figure it out then :)
<TheSirC[m]>
Fatal server error: [ 72.950] (EE) no screens found(EE)
domogled has quit [Ping timeout: 240 seconds]
domogled1 is now known as domogled
<nature>
I suppose that there might be some config missing in /etc/X11/xorg.conf
<c0c0>
aaronjanse: Thanks for the reply. I have 'xsession.windowManager.i3.enable = true;' in my home.nix (I'm using home-manager)
<nature>
Still not working, I really think I am missing a configuration in xorg.conf
<c0c0>
"intel" video driver should be fine for Intel UHD Graphics 630?
<MichaelRaskin>
As in X11 video driver?
<MichaelRaskin>
You might want to use modesetting
<nature>
what is that ?
<nature>
I am actually looking at my xorg.conf and I see a lot of "Device" section and one had the modesetting driver
orivej has quit [Ping timeout: 260 seconds]
orivej has joined #nixos
<c0c0>
MichaelRaskin: in services.xserver.videoDrivers
<c0c0>
What is modesetting?
bukkitgerman8608 has quit [Ping timeout: 256 seconds]
<MichaelRaskin>
Later kernels contain a more featureful i915 kernel video driver. Modesetting is an X11 video driver (and default for Intel GPUs for videoDrivers in the latest NixOS) that asks Kernel Mode Setting to do most of the configuration work, and packs less knowledge of GPU specifics
afreak has joined #nixos
<nature>
MichaelRaskin: So it's simply more generic ?
siers has left #nixos ["WeeChat 2.7.1"]
<MichaelRaskin>
More generic, and also more maintained
<MichaelRaskin>
(upstream)
<exarkun>
Having trouble getting virtualbox working. `VBoxManager start ...` fails with "VBoxManage: error: Details: code NS_ERROR_FAILURE (0x80004005), component MachineWrap, interface IMachine" and a dialog saying "Kernel driver not accessible"
<exarkun>
my config is `virtualisation.virtualbox.host.enable = true;` and `users.users.exarkun.extraGroups = [ "vboxusers" ];`
<exarkun>
and I rebooted after the new kernel/modules were built
<exarkun>
anyone know how to fix this?
tldr32 has joined #nixos
mrosenbe has joined #nixos
tldr32 has quit [Client Quit]
<nature>
OMG, it was my sxhkdrc file that was broken....
<nature>
c0c0: Make sure that your wm (in your case i3) is ok
<nature>
c0c0: in other words I had some commands in my xinitrc that would fail, make sure all the commands run in there are working
<c0c0>
nature: I just use home-manager's `xsession.windowManager.i3.enable = true;`, so I don't think anything's broken there
mrosenbe has quit [Ping timeout: 240 seconds]
<nature>
c0c0: what if you try another dm ?
<c0c0>
"Failed to start session" from LightDM is not helpful. How would I set up X but have LightDM disabled, so I can see more output?
tldr32 has joined #nixos
<nature>
c0c0: you have something like services.xserver.displayManager.sddn.enable = true
<nature>
c0c0: this should change your dm
<c0c0>
nature: I don't have lightdm in my configuration explicitly. Something sets it implicitly (maybe `xsession.enable = true;`?)
<nature>
yeah I think lightdm is default
bukkitgerman8608 has joined #nixos
<c0c0>
no, that option can't be it, as it's in my home.nix. Maybe it's `services.xserver.enable = true;` in configuration.nix
<nature>
if you wanna go without dm like me you have to set services.xserver.displayManager.startx.enable = true
<nature>
oh yeah I am speaking in configuration.nix
<nature>
try to setup i3 from there
<nature>
c0c0: are you familiar with xinitrc ?
<c0c0>
nature: I used it under Arch
pareidolia has quit [Ping timeout: 244 seconds]
<nature>
c0c0: haha a fellow ex-arch user, literally just switched yesterday
<nature>
c0c0: then do that, setup your xinitrc to launch i3 and in your system's configuration.nix you do displayManager.startx.enable
<nature>
you should then at least have some logs in ~/.xorg.log
nature has left #nixos ["brb"]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<c0c0>
nature: In which way did you "set up" xinitrc? Just placed a file, or setting some more NixOS-like way?
o1lo01ol1o has joined #nixos
<nature>
Just placed the file for now
<nature>
I think there is more advanced nix way for doing this but I was pointed by people in this IRC a few days ago to a prettu nice way of handling dotfiles
<nature>
The advised this over home-manager
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<c0c0>
nature: The idea I'm having for using NixOS and home-manager is to configure everything there, and not have additional configuration management systems aside
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
proofofkeags has joined #nixos
bbennett371 has joined #nixos
<nature>
c0c0: I think it would be ideal to have everything configured in one place and I love NixOS for allowing you to do that, though I just saw a repo of a guy configuring everything in nix and this is his vimrc: https://github.com/jagajaga/my_configs/blob/master/.nixpkgs/vimrc.nix
<nature>
c0c0: Like why not just have a normal vimrc then, so you at least have the syntax highlighting etc... ? And since you're probably gonna have a git with at least your nix file, why not just have a vimrc, xinitrc etc.. in there ?
<nature>
c0c0: At lease that's how I see it, leverage the power of nixos system-wise to easily configure, rollback etc.. and then still having a few dotfiles managed in a repo
<nature>
c0c0: Did you get i3 working ?
bbennett371 has quit [Ping timeout: 260 seconds]
<c0c0>
nature: For me the main point here is that I still want to manage generations of my whole config with NixOS / home-manager, and not use another tool to pull in files. If the files are more or less verbatim in the repo, that's fine, but I still want to be able to roll back everything in one shot
<nature>
c0c0: okok I see, fair point
maier has quit [Ping timeout: 240 seconds]
<c0c0>
nature: services.xserver.displayManager.startx.enable = true has the effect that X is started after login, but I haven't configured i3 yet
<nature>
c0c0: But I also imagined that once I get more comfortable with nix I'll be able to do something like "fetchGit" on my dotfile repo and actually have everything managed through nix, without home-manager
<nature>
c0c0: ok, so now X works at lease ?
<nature>
*least
<c0c0>
Yes, X seems to work
<nature>
k great :)
mallox has joined #nixos
<nature>
I really like that nix allows to adopt the nix way gradually, like you probably won't have too much trouble going from a startx setup to a more "nix way" of doing things
noonien has joined #nixos
markus1199 has quit [Ping timeout: 240 seconds]
<noonien>
hello folks, is there any documentation for lib.makeScope? or commented examples of it beign used? i want to learn what it's for, and what it does
<davidv7>
this is my current font in configuration.nix
<alekhine>
Hello everyone. I have a quick installation question: I'm trying to mount my filesystem, but I keep getting this output when running 'sudo mount /dev/sda2/NixOS/NixOS /mnt/nixos'
<alekhine>
mount: /mnt/nixos: special device /dev/sda2/NixOS/NixOS does not exist (a path prefix is not a directory).
<alekhine>
Sorry to be asking a basic question
<raboof>
does it ever make sense for a nix-packaged python program to look at the user-supplied PYTHONPATH environment variable?
<davidv7>
but there's some syntax issue or something with that and i can't find anywhere that helps me how to get a nerrd patched font overriden
<raboof>
I have a case where one python app calls another, but the latter fails because it takes into account the PYTHONPATH inherited from the former. I wonder if I should just clear the PYTHONPATH in the wrapper for the latter.
<ToxicFrog>
`mount -t vfat /dev/disk/by-label/BOOT /boot` completes without error but the filesystem is not mounted and there are no messages about it in dmesg.
<{^_^}>
[nixpkgs] @Atemu opened pull request #94442 → Introduce BuildFHSUserEnv with Bubblewrap as an alternative to chrootenv → https://git.io/JJVUj
knupfer has quit [Remote host closed the connection]
knupfer has joined #nixos
v0|d has quit [Ping timeout: 246 seconds]
proofofkeags has quit [Remote host closed the connection]
proofofkeags has joined #nixos
gnidorah has quit [Quit: Connection closed for inactivity]
h0m1 has quit [Quit: WeeChat 2.9]
proofofkeags has quit [Ping timeout: 246 seconds]
<nature>
Where do I find what are the overrides I can do to a packages, e.g. neovim
<gthm>
Is there a way of finding out what nix-daemon is up to? It's burning all my cpu on compiling a bunch of haskell, presumably during its daily system upgrade task, but I don't recall installing this much haskell and would like to find out which package it's compiling. journalctl is very quiet
davidv7 has quit [Ping timeout: 256 seconds]
<simpson>
gthm: nix-top, available in nixpkgs, might help.
<gthm>
thanks!
<gthm>
it finished for now, though my fans are still going strong. I'll try it next time
never_released_ has quit [Read error: Connection reset by peer]
<symphorien>
nature: did you check the resulting /etc/sudoers ?
<nature>
symphorien: do you know also know why my sudoers file is not overriden and just appended ?
<nature>
yes it contains the default config appended to the config in my configuration.nix
<symphorien>
because you are supposed to use the options to change the rest of the config ;)
<symphorien>
notably security.sudo.extraRules
<nature>
ooh so I override it with "[]" correct ?
<nature>
or I define my options in the extraRules instead of having a dumb file override ?
<symphorien>
the second option
alp has quit [Ping timeout: 240 seconds]
cmk_zzz_ has joined #nixos
<nature>
kk many thanks symphorien :)
<symphorien>
always prefer structured config because if another module has something to change in /etc/sudo it will merge config better
<nature>
got it, then may I ask you more general questions ?
<symphorien>
yes ?
<symphorien>
(don't ask to ask)
cmk_zzz has quit [Ping timeout: 264 seconds]
<nature>
I just installed nixos and I am figuring out how to configure everything, I am trying to understand when should I use ~/.nixpkgs/config.nix and when I should use configuration.nix
<nature>
like for instance, I want to configure neovim
<symphorien>
configuration.nix is for everything that is systemwide, and everything that does not run as your user
<symphorien>
everything that requires root privileges to be configured on traditional distros
<nature>
okok
<symphorien>
for user-wide configuration (that is, dotfiles mainly), I advise you to use home-manager
<nature>
so I should do my neovim config in .nixpkgs/config.nix for instance ?
<nature>
okok, what's the benefit of home-manager ?
<nature>
And another question, how can you use nixos-unstable instead of the stable pkgs for the systemPackages ?
mog has joined #nixos
Rusty1 has joined #nixos
<nature>
If I simply remove the stable channel will I achieve that ?
meh` has joined #nixos
<stigo>
nature: yep, replace the stable channel with the unstable channel.
<lejonet>
What is the easiest way to just append some commands to a service already existing ExecStartPre script? Just copy the script, do my additions and overwrite it with lib.mkForce or is there a way I can just append to it, so that if it gets changed by the module later, my additions will just be appended, instead of having to keep up with the changes in my conf?
<Thra11>
If a vim plugin relies on an external binary being in the path to work, should it depend on the binary, or is it the user's responsibility to install such dependencies?
<lejonet>
Essentially, ideally for me would be if there is some way to do += " extra string " to like preStart, and then be done with it
<Thra11>
e.g. coc-nvim doesn't work unless nodejs is installed.
<lejonet>
Thra11: logically, I would say that if you're installing it through a package manager, it should resolve all of its dependencies at install time, for example in this case I would say that the plugin should add nodejs to its path when being installed
<lejonet>
If you're installing it outside a package manager, its up to the user to ensure it can access what it needs
<lejonet>
But that is just my opinion
<nature>
Should you rather explicitly install a package via the environment.systemPackages or implicity with for instance programs.zsh.enable ?
Neo--- has joined #nixos
<numkem>
nature: if you have an option available, use the option. if there is a module for it, it means it requires more than just having the package in your PATH
<nature>
same with wireguard as well, if you configure it, is it still advised to explicitely install in ?
<balsoft>
lejonet: I think taking original contents, changing it and overriding is the best option. If the type were a concatenable one (e.g. lines), just assigning would concatenate, but that's not the case for ExecStart
<numkem>
nature: not sure I understand your question for wiregard
<balsoft>
nature: You shouldn't need to add stuff like that to environment.systemPackages
<lejonet>
balsoft: yeah, and trying to "be smart" and do like preStart = preStart + " things" will ofc result in infinite recursion
<nature>
okok, so the use of systempackage is just to put stuff in your PATH ?
lsix has quit [Quit: WeeChat 2.9]
<numkem>
nature: pretty much yeah. Technically it modifies your profile but in the end that's pretty much it
<balsoft>
lejonet: Hm, yes, true
<lejonet>
balsoft: I found that preStart exists outside of serviceConfig, and that is "just" strings concatenated with "\n", so maybe just assigning to <service>.preStart will work?
<balsoft>
But there's no guarantee that your changes are going to be appended _after_ what's already there for obvious reasons
<lejonet>
balsoft: Well that doesn't matter for me, I only said append because that is what I want to do, if its done in the middle, before or such doesn't matter
<balsoft>
type = types.lines;, so yes, it should work
<lejonet>
Thanks for the help, I'll try that then :)
<TheSirC[m]>
Would anyone packaged `amdgpu-pro` in the past ?
sigmundv has joined #nixos
<TheSirC[m]>
I am facing an issue while trying to package it. Reaching the buildPhase I get : `make[3]: *** No rule to make target '/build/amdgpu-pro-20.20-1098277-ubuntu-20.04/root/usr/src/amdgpu-5.6.0.15-1098277/amd/amdkcl/symbols.o', needed by '/build/amdgpu-pro-20.20-1098277-ubuntu-20.04/root/usr/src/amdgpu-5.6.0.15-1098277/amd/amdkcl/amdkcl.o'. Stop.`
simba1 has joined #nixos
domogled has joined #nixos
CodeSpelunker has joined #nixos
tarruda2 has joined #nixos
cole-h has joined #nixos
philr has quit [Ping timeout: 256 seconds]
<{^_^}>
[nixpkgs] @vbgl merged pull request #93814 → coqPackages.bignums: init at 8.12.0 for Coq 8.12 → https://git.io/JJBq9
<{^_^}>
[nixpkgs] @vbgl pushed to master « coqPackages.bignums: init at 8.12.0 for Coq 8.12 »: https://git.io/JJVso
<hsngrmpf[m]>
How can I check quickly if a certain store path is available from a binary cache?
<bbigras>
infinisil: anything I need to do to populate the vendor dir? It didn't exist when I did tests last night.
metasyntactic has joined #nixos
<infinisil>
bbigras: It should if you specified vendorSha256
<metasyntactic>
I'm trying to write a package for a new VR remote desktop package that support linux - I'm having trouble with autopatchelf applying to some dynamic libs that the package drags along - can anyone help?
<infinisil>
hsngrmpf[m]: For a path /nix/store/<hash>-<name>, do `curl -I https://<url>/<hash>.narinfo`
<TheSirC[m]>
But anyway, that's not the dragon I should be slaying because the output of building (read `buildPhase`) the derivation reads : `checking whether the C compiler works... no` which is way worse
<{^_^}>
[nixpkgs] @doronbehar opened pull request #94460 → restinio: init at 0.6.8.1 → https://git.io/JJVnW
<ogkloo>
I can't seem to get nixos aarch64 to talk to HDMI on a raspi 3B. Do I have to do anything special to get it to do that? I know the pi is fine because it boots NOOBS without complaint.
zebrag has quit [Quit: Konversation terminated!]
<infinisil>
yurb: `man configuration.nix`
zebrag has joined #nixos
alp has joined #nixos
cosimone has joined #nixos
<tldr32>
ogkloo: I had to use older kernel 4.19 on 3B+ to be able to see anything beyond initrd. On 5.6 (I think) it booted but without a display .. was able to login through ssh
fresheyeball has quit [Quit: WeeChat 2.7.1]
<ogkloo>
Oh, interesting. I'll try it without a display over ssh, since I don't really need a display for what I want.
<ogkloo>
Would I just edit the configuration.nix in the sd card or something?
<ogkloo>
If I did want to change the kernel versio
arjen-jonathan has quit [Ping timeout: 244 seconds]
<infinisil>
metasyntactic: Why is there a .so in the bin folder
<{^_^}>
[nixpkgs] @Mic92 opened pull request #94465 → opengrm-ngram: init at 1.3.11, phonetisaurus: init at 2020-07-31 → https://git.io/JJVCl
orivej_ has quit [Ping timeout: 264 seconds]
orivej has joined #nixos
<betaboon>
is anyone using docker-buildx and can tell me if it is using qemu when building for the same architecture of the host ? oO
julm has quit [Ping timeout: 264 seconds]
julm has joined #nixos
kraeXen has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
ezemtsov has joined #nixos
cosimone has quit [Quit: Quit.]
metasyntactic has quit [Ping timeout: 265 seconds]
<{^_^}>
[nixpkgs] @matthewmazzanti opened pull request #94466 → doc: Fix missing callout in go.xml → https://git.io/JJVWv
malook has left #nixos ["ERC (IRC client for Emacs 26.3)"]
kraeXen has quit [Ping timeout: 240 seconds]
kraeXen has joined #nixos
drakonis has quit [Quit: WeeChat 2.8]
Fare has quit [Ping timeout: 244 seconds]
drakonis has joined #nixos
julm has quit [Ping timeout: 256 seconds]
WRMilling has joined #nixos
<WRMilling>
Hello everyone! Quick syntax check. If I want to add a patch to the kernel being built, would this statement work? https://dpaste.com/GLWRNX7JX (patching an overlay I am using)
<aleph->
So question, got a nix module here I'm writing, I'm generating a yaml/json config file from nix expr lang and running into an issue with nested submodules. The submodules on lines 45, 115, 139 are causing me some issues. Setting them is supposed to give me a nested mapping underneath a higher mapping. However instead I just get a coercion error of the type "value is a string while a set was
<aleph->
expected" with a trace of: https://paste.rs/NMk So I'm wondering if I should be using a submodule type here given I expect the choices to be options, or maybe if there's something I should be using a attr set directly for? Module here: https://paste.rs/Vgh
<{^_^}>
[nixpkgs] @cust0dian opened pull request #94469 → navi: init at 2.7.1 → https://git.io/JJVlP
<duairc>
aleph-: Do you need the submodules there? I'm not sure about this myself but I've only ever found it useful to do that when you using with listOf or attrsOf. If you take away the submodule bit and the 'options =' bit inside of that, does that give you what you want?
<duairc>
aleph-: So basically you just define the authentication.type and authentication.second_factor modules directly without using submodules
LouisDK has joined #nixos
<metasyntactic>
I'm trying to use patchelf on a 3rd party binary, and it appears to work, but then I get a bunch of "no version information available (required by.." errors when I try to run the patched binary. google says that means the library version are too old, but they don't appear to be - what am I doing wrong?
euandreh` has quit [Ping timeout: 272 seconds]
ezemtsov has quit [Remote host closed the connection]
<aleph->
duairc: Hmm, I guess I don't need options. I can still set a default value with an attr set right?
<aleph->
Because that is something I'd like.
o1lo01ol1o has joined #nixos
BaD_CrC has quit [Remote host closed the connection]
BaD_CrC has joined #nixos
<balsoft>
exarkun: Are you running some other hypervisor? There can only be one.
<{^_^}>
[nixpkgs] @mweinelt merged pull request #94465 → opengrm-ngram: init at 1.3.11, phonetisaurus: init at 2020-07-31 → https://git.io/JJVCl
<aleph->
balsoft: To be clear I meant the fields of the attr set. Although I guess that would be the same as just setting a default for the whole attr set.
<balsoft>
E.g. if you set default to { foo = { bar = 10; }; goo = 20; } and then set smth like { foo.bar = 15; } in some other module, you'll get { foo.bar = 15; } as the final module (e.g. it overrides and not merges)
<{^_^}>
[nixpkgs] @mweinelt pushed 2 commits to release-20.03: https://git.io/JJV85
<balsoft>
metasyntactic: are you sure the RPATH is empty before you override it?
maier has quit [Ping timeout: 240 seconds]
eoli3n has joined #nixos
<balsoft>
metasyntactic: oh, it's a deb package. Maybe you need to add the /lib or /usr/lib or whatever of that package to RPATH too
<metasyntactic>
balsoft: I looked at the original binaries from the .deb with readelf -d and they didn't have a RPATH in them originally
shibboleth has joined #nixos
<duairc>
aleph-: Maybe I didn't explain myself very well, but try this out and see if it does what you want: https://paste.rs/Jzv
<balsoft>
metasyntactic: maybe there are some libraries that come with the package? If that's the case, you have to make sure the binary finds them by adding their location to RPATH
CodeSpelunker has quit [Quit: CodeSpelunker]
<aleph->
duairc: Ahhhh, got it. So with that patch I get "attribute 'name' missing, at /nix/var/nix/profiles/per-user/root/channels/nixos/lib/modules.nix:532:12"
andreas303 has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @primeos opened pull request #94471 → wayland-utils: init at 1.0.0 → https://git.io/JJV4U
<aleph->
duairc: Might just try setting the embedded submodules as attr sets only I think
<metasyntactic>
balsoft: I don't see the libraries that are included with the package mentioned in the ELF header, so I'm assuming that they are loaded dynamically?
eoli3n has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @wkral opened pull request #94474 → wayland-utils: init at 1.0.0 → https://git.io/JJVBf
<aleph->
balsoft: Yeah that worked. Guess I'll go without submodules when generating nested config file options
<balsoft>
metasyntactic: in order to load libraries dynamically, dynamic linker must know where to find them.
<balsoft>
On "traditional" distros, that's usually somewhere in LD_LIBRARY_PATH, which is /lib:/usr/lib or something like that.
eoli3n has joined #nixos
<balsoft>
metasyntactic: On NixOS, LD_LIBRARY_PATH is empty, and everything is hardcoded into RPATH. That might upset some Debian packages.
<metasyntactic>
it's relative to the executable? I tried modifying the patchelf line as follows but it didn't change the rpath in the resulting binary: patchelf --set-interpreter "$(cat $NIX_CC/nix-support/dynamic-linker)" --set-rpath "${libPath}:$out/bin/immersed-res/va1/:$out/bin/immersed-res/va2/" $out/bin/Immersed
<balsoft>
metasyntactic: Did that work?
<balsoft>
aleph-: nice
<metasyntactic>
balsoft: nope, didn't change the rpath line in the elf header either, which baffles me
<balsoft>
metasyntactic: Are you sure you're actually rebuilding the package? :P That happened to me a couple of times
<metasyntactic>
balsoft: it created a different dir in /nix/store?
<balsoft>
metasyntactic: ok, so it did get rebuilt. Interesting.
malik has quit [Remote host closed the connection]
CodeSpelunker has joined #nixos
<metasyntactic>
balsoft: I tried using autopatchelf too, but that fails to build because of the included .so files, which is really confusing as well - would you like to see that version?
CodeSpelunker has quit [Remote host closed the connection]
<balsoft>
metasyntactic: Ah, I see now. autopatchelf probably fails because it tries to patchelf so files themselves, which doesn't make sense here.
<metasyntactic>
well they do have unresolved dependencies, but as far as I can tell, the libs they're complaining about are libav, which is part of the specified dependencies
<lordcirth__>
In Nix, how can I have a function take an argument (a username) and return a config with users.users.$username = thing? That is, how do I use a variable as a set key?
<balsoft>
metasyntactic: please wait while I download this thing to play with it
<balsoft>
lordcirth_: You're pretty close
<metasyntactic>
balsoft: will do - I appreciate your help
<balsoft>
metasyntactic: Ok, so I got it to fail with ./result/bin/Immersed: error while loading shared libraries: libcurl-gnutls.so.4: cannot open shared object file: No such file or directory
<metasyntactic>
balsoft: weird, I haven't gotten that error before on my system
<balsoft>
nix-locate suggests that there is no derivation for that lib specifically
<metasyntactic>
balsoft: really? I have a bunch of them on my system. e.g.: /nix/store/hqhimaamycxwl7lsxssnbj24p6xibpld-curl-7.68.0/lib/libcurl-gnutls.so.4
<balsoft>
Huh, it's a bug in nix-locate
<metasyntactic>
balsoft: i tend to use ls /nix/store/*/lib/lib...
<balsoft>
Hmm
malik has joined #nixos
<balsoft>
I don't have libcurl-gnutls in my curl.out
<balsoft>
What nixpkgs version are you on?
<balsoft>
release-20.03?
<metasyntactic>
weird... unstable vs stable difference?
<balsoft>
Yeah, maybe
<metasyntactic>
I'm on 20.03 but I also install a handful of packages from unstable or master
<balsoft>
Whoops, forgot to change license back :P
alp has quit [Ping timeout: 260 seconds]
<metasyntactic>
balsoft: huh... nix/store/imr3zcvxcpzdcrd5jc48y8n9mwpppn4x-immersed-2.1-0/bin/Immersed: error while loading shared libraries: libcurl-gnutls.so.4: cannot open shared object file: No such file or directory
<balsoft>
Uh-huh
<balsoft>
Can you see if ls $(nix eval nixpkgs.curl.out)/lib contains libcurl-gnutls?
<balsoft>
(that's why nix-locate is better than ls /nix/store/... btw)
<clever>
which is googles library for collecting coredumps and shipping them home automatically
<LouisDK>
balsoft: The software needs libpthread.so.0 in order to be able to find tv tuners on the network.
maier has quit [Ping timeout: 265 seconds]
<balsoft>
LouisDK: Ok, so what happens if you just try patching it? What's the error?
<clever>
balsoft: if breakpad is setup correctly (by the original creator of the app), it should have already sent the core file back to them
<balsoft>
clever: that sounds scary
arjen-jonathan has quit [Ping timeout: 260 seconds]
<clever>
balsoft: its likely in the EULA you agreed to when downloading the app
<clever>
you did read that, right? :P
<balsoft>
clever: that makes me want to run everything in network-isolated containers even more
<balsoft>
I guess it's about time I continue that project
<LouisDK>
balsoft: If the lib is not found the program is unable to find any tuners.
<balsoft>
LouisDK: And you are implying that the lib is not found?
<evanjs>
anybody know if there are any skeleton files for e.g. nix modules for emacs or vim? On a related note.... is there an idiomatic way to add such files via my config? (NixOS _or_ home-manager)
<duairc>
infinisil: builtins.unsafeDiscardOutputDependency was what I needed the other day
<evanjs>
Or would I be adding them like I would add typical vi/emacs packages/plugins, etc?
<LouisDK>
balsoft: That stated when running the program with "strace".
<LouisDK>
*that's
<balsoft>
clever: BTW, would it work in this case? The program doesn't even reach the entrypoint
<infinisil>
duairc: Oh damn, didn't know this existed!
<balsoft>
LouisDK: strace is a bad way to debug such things. What does ldd ./your_program say?
<{^_^}>
[nixpkgs] @xeji merged pull request #94341 → enlightenment.efl: do not default to openjpeg 1 → https://git.io/JJaLF
<clever>
balsoft: if your seeing breakpad, then it must have reached main() i think, and setup the breakpad exception handlers
<infinisil>
duairc: What's the full code now?
proofofkeags has joined #nixos
<clever>
balsoft: but it may be failing too bad to phone home, and might be failing within breakpad's own coredump generation
<balsoft>
clever: I'm only seeing breakpad in the debug symbols, not in any output. If I gdb it, break on the entrypoint and start, it fails before the breakpoint.
<clever>
ah
<balsoft>
So I doubt it has time to phone home. If it does, that's some impressive levels of invasive telemetry
<clever>
whats the backtrace upon failure?
<{^_^}>
[nixpkgs] @Ma27 merged pull request #94377 → zfs: backport linux 5.8 compatibility patch to 0.8.4 release → https://git.io/JJauG
<duairc>
infinisil: Basically I make a systemd service that runs `nix-store --realise ${unsafeDiscardOutputDependency derivation.drvPath}`; the drv is included in the closure but not the out
<clever>
balsoft: i think on its own, breakpad just allows coredumping with metadata about when/how, and listing those coredumps
<clever>
infinisil: has that text gotten more corrupted since i last saw it? :D
<duairc>
infinisil: As far as I can see, yes, but I haven't tested it with any non-trivial derivations yet
<clever>
balsoft: not sure then
<balsoft>
clever: ok, thanks for your help anyways!
<infinisil>
clever: It spreads slowly!
<balsoft>
clever: also, maybe you could help exarkun with their VirtualBox issue?
<clever>
balsoft: got a pastebin with an error?
<balsoft>
22:11 <exarkun> Having trouble getting virtualbox working. VBoxManager start ... fails with "VBoxManage: error: Details: code NS_ERROR_FAILURE (0x80004005), component MachineWrap, interface IMachine" and a dialog saying "Kernel driver not accessible"
bbennett371 has joined #nixos
<clever>
balsoft: is the kernel module loaded? (check lsmod), how was virtualbox enabled and installed?
<balsoft>
clever: oh, also, disassemble did disassemble the function, I'll look into it
<LouisDK>
balsoft: It states that the lib is found, so the issue has to with some other error. hdhomerun_config should find a network-based tv tuner on the local network but finds nothing. Browsing the tuners ip address from the same machine works so the ip is reachable.
<clever>
balsoft: the disassembly should have an arrow in it, what is the opcodes around the arrow?
<balsoft>
clever: it's not my issue, it's exakrun 's :)
<{^_^}>
[nixpkgs] @mweinelt opened pull request #94484 → nixos/tests/postfix: migrate test to use tlsTrustedAuthorities → https://git.io/JJV0t
<LouisDK>
balsoft: The only relevant log I can think of posting is from running the program strace. It works as expected under Arch Linux? I don't know if any sort of comparison would help?
<bqv>
armin: it depends on a database that isn't guaranteed to exist, iirc, so ymmv
<balsoft>
LouisDK: yeah, that would be a good start
<infinisil>
armin: Yeah I investigated this with jlv recently, and it seems that the database that comes with channels doesn't contain everything it should anymore
<infinisil>
Didn't get to the cause
proofofkeags has quit [Ping timeout: 256 seconds]
<ogkloo>
clever it doesn't even seem to be connecting over an ethernet connection
titano has quit [Ping timeout: 240 seconds]
<clever>
ogkloo: do you have a serial adapter for the pi?
<ogkloo>
I don't
<clever>
ogkloo: no other pi's either?
<ogkloo>
unfortunately not, I bought it secondhand to play around with
<ogkloo>
it doesn't seem to be dead, but now that I think about it I haven't ruled out that part of the networking on it just doesn't work at all
<ogkloo>
the guy I bought it off of had an ancient NOOBS installation on it, so I figured it was that
<clever>
ogkloo: try mounting the rootfs on the sd card, do you see anything new under var/log/journal/ ?
arjen-jonathan has quit [Ping timeout: 272 seconds]
<ogkloo>
If I mount the SD card I just see /etc, /nix, and /boot, I assume I'm missing a step?
<LouisDK>
balsoft strace output: http://ix.io/2sU9 - hope it helps. Should I do the same thing under Arch where the application works?
<balsoft>
LouisDK: Arch version would probably be a good thing, yes
<clever>
okif /var is missing, then it likely didnt get very far in booting, which model of pi? what image did you flash to it?
<clever>
exarkun: line 90 puts the right version into systemPackages, and line 89 puts the kernel modules into extraModulePackages
<exarkun>
good to know
<clever>
exarkun: line 13-15 will grab the kernel virtualbox from your chosen kernel, and use .override to force it to use the virtualbox from line 8
<exarkun>
I had that turned on but I _also_ installed virtualbox, clobbering the working thing with a broken thing, yay :(
<clever>
exarkun: and 8-11, will respect virtualisation.virtualbox.host.package if you want to change the versions
Henson has joined #nixos
<exarkun>
got it, thank you very much
<Henson>
hi everyone, in a NixOS configuration, say I have a NixOS option that is a list and it's defined in multiple files. NixOS will merge those lists by concatenating them. I'm trying to figure out how to prevent the concatenation, and make one entry THE entry that is used. Will mkForce do this?
Fare has quit [Ping timeout: 244 seconds]
<clever>
Henson: if you use mkForce, then it will still merge, but it will only merge the values you used mkForce on
tobiasBora2 has quit [Ping timeout: 260 seconds]
<clever>
Henson: so if you set it 3 times, to [1], mkForce [2], and mkForce [3], you will get [2 3] out at the end
tobiasBora2 has joined #nixos
orivej_ has quit [Ping timeout: 240 seconds]
<Henson>
clever: ok, that makes sense. How about if I used mkOverride with different priorities. Will the end result be the merge of the highest priorities that are all equal?
<clever>
Henson: yep
<clever>
there is a default priority for when you dont set one, mkDefault/mkForce then just set things to be lower/higher then that main default