stepcut has quit [Remote host closed the connection]
stepcut has joined #nixos
stepcut has quit [Remote host closed the connection]
stepcut has joined #nixos
stepcut has quit [Remote host closed the connection]
stepcut has joined #nixos
stepcut has quit [Remote host closed the connection]
stepcut has joined #nixos
stepcut has quit [Remote host closed the connection]
stepcut has joined #nixos
stepcut has quit [Remote host closed the connection]
stepcut has joined #nixos
hyper_ch has joined #nixos
stepcut has quit [Remote host closed the connection]
matt` has quit [Remote host closed the connection]
stepcut has joined #nixos
stepcut has quit [Remote host closed the connection]
stepcut has joined #nixos
stepcut has quit [Remote host closed the connection]
stepcut has joined #nixos
xkapastel has quit [Quit: Connection closed for inactivity]
evanjs has joined #nixos
phreedom_ has quit [Quit: No Ping reply in 180 seconds.]
kvda has joined #nixos
phreedom has joined #nixos
hmpffff_ has joined #nixos
hyper_ch has quit [Read error: Connection reset by peer]
hmpffff has quit [Ping timeout: 250 seconds]
hyper_ch has joined #nixos
stepcut has quit [Remote host closed the connection]
stepcut has joined #nixos
<samueldr>
aleph-: not sure, if it's guaranteed to stay compatible, I guess it'd do
<samueldr>
but then I figure it might not be
<aleph->
No probably not
<aleph->
Just seems silly having a service with the only change being package type. Course I honestly think emby should just be deprecated given it's no longer available. But eh.
stepcut has quit [Remote host closed the connection]
stepcut has joined #nixos
<gchristensen>
you could mak ethe service support changing the package
stepcut has quit [Remote host closed the connection]
day|flip has quit [Quit: ERC (IRC client for Emacs 26.1)]
<jasongrossman>
kvda: fetchTarball is memoised - its output is cached.
abf_ has quit [Quit: Leaving]
<kvda>
thanks jasongrossman. Is it possible to add a channel with nixenv and then refer to it by name in shell.nix?
<KarlJoad>
Hey guys, I have a question that I can't seem to figure out myself. How do I go about asking it here?
<jasongrossman>
,ask KarlJoad
<{^_^}>
KarlJoad: Just ask your question. It's the best way to know if anybody can help. Or look at other ways of getting in touch with the Nix community here: https://nixos.wiki/wiki/Get_In_Touch
<KarlJoad>
I am trying to mount a SMB/CIFS share declaratively in configuration.nix. I got it mounted and I can see it in my file manager. However, it's permissions are for root. Since this is a shared eBook library, I need RW access. How do I change the user-owner and group-owner for that directory, declaratively?
<mdash>
KarlJoad: probably set `options = "uid=1000, gid=1000";` in the mount declaration
mexisme has joined #nixos
KarlJoad has quit [Quit: Konversation terminated!]
KarlJoad has joined #nixos
dansho has joined #nixos
mexisme has quit [Ping timeout: 245 seconds]
mexisme has joined #nixos
abathur has quit [Ping timeout: 248 seconds]
<ashkitten>
uh so
<ashkitten>
upgraded my cpu and motherboard
KarlJoad has quit [Ping timeout: 258 seconds]
<ashkitten>
i'm going through my system bootloader from super grub2 disk
<ashkitten>
the first thing it says is systemd-udevd-control.socket: Failed to set invocation ID for unit: File exists
<ashkitten>
and then everything else fails
<ashkitten>
just looked up the error, apparently an issue with rdrand on ryzen 3000
___laika has quit [Ping timeout: 248 seconds]
KarlJoad has joined #nixos
adamantium has joined #nixos
<KarlJoad>
mdash: I tried that, but my gid is 100 and it still isn't doing quite what I expect. It's probably because I'm declaring it wrong.
<{^_^}>
[nixpkgs] @andir pushed 3 commits to release-19.03: https://git.io/fjD9v
palo1 has joined #nixos
KarlJoad has quit [Quit: Konversation terminated!]
KarlJoad has joined #nixos
palo has quit [Ping timeout: 244 seconds]
palo1 is now known as palo
endformationage has quit [Ping timeout: 248 seconds]
o1lo01ol1o has joined #nixos
Drakonis has quit [Ping timeout: 250 seconds]
o1lo01ol1o has quit [Ping timeout: 246 seconds]
m0rphism has joined #nixos
KarlJoad has quit [Ping timeout: 248 seconds]
daurnimator has joined #nixos
* daurnimator
waves. trying out nixos for the first time
<daurnimator>
haven't really found a good "getting started" guide
iqubic has joined #nixos
<daurnimator>
what's the "base-devel" equivalent?
<iqubic>
What GUI should I use for telling CUPS to print stuff?
<srhb>
daurnimator: THere isn't one / stdenv depending on how you look at it.
<srhb>
,libraries daurnimator
<{^_^}>
daurnimator: Don't install libraries through nix-env or systemPackages, use nix-shell instead. See https://nixos.wiki/wiki/FAQ/Libraries for details.
<srhb>
daurnimator: That might clear things up a bit.
<no-n>
the manual is pretty good
<iqubic>
So, what GUI should I use for printing stuff?
<iqubic>
clever: `eog` worked for me. The `eog` print dialog let me set the size properly. I know it is the right size because I screwed up and printed a second copy that was 5 inches taller. That image wound up being 1 inch taller than the 4 inches I was aiming for.
<iqubic>
thank you for your help.
<iqubic>
I'm just so pleased that this printer is working and accepting various different print settings.
<srhb>
gwen: Usually a package comes with udev rules that when added to services.udev.packages ensures that the node in /sys/... is oned by the proper group and has the right mode bits
trebuh has joined #nixos
<srhb>
gwen: See for instance the "light" package for such a rule set.
<gwen>
srhb: Hmm, I'm trying out "light" right now (via programs.light.enable=true), and although I'm a member of the video group, I can't set the brightness
<gwen>
srhb: It works with sudo though
<srhb>
gwen: Do the device nodes look right?
<srhb>
gwen: As in, does it belong to video?
<gwen>
As far as I can tell, all entries of /sys/class/backlight belong to root/root
linarcx has quit [Client Quit]
linarcx has joined #nixos
<srhb>
gwen: Did you have the module enabled when you booted?
<gwen>
srhb: Ah, no. I did it just now and still am in the same session
<srhb>
I'm guessing ACTION=="add" only triggers when the nodes are created, or via some other trigger.
llb has joined #nixos
<gwen>
srhb: OK, will reboot. See ya in a minute.
gwen has quit [Quit: Leaving]
gwen has joined #nixos
<gwen>
srhb: It worked!
<srhb>
gwen: Marvellous :)
<gwen>
srhb: So cool, finally I can set my MacBook keyboard brightness. Thanks!
<srhb>
niso: I tend to prefer it anyway because of the impedance match.
o1lo01ol1o has quit [Remote host closed the connection]
orivej has joined #nixos
<niso>
alright
<niso>
by chance someone is willing to share his nixops config for a gitlab/gitea - hydra CI/CD pipeline?
Ariakenom has joined #nixos
<johanot>
we have a self-hosted gitlab runner as well, a single instance with no garbage collection evar :p but we are considering S3 as backend for our store, which should make it easier to share between runners.
zupo has joined #nixos
zupo has quit [Client Quit]
<srhb>
I do something like place a gc root for each branch, periodically delete everything but the ten latest..
<srhb>
Arbitrary, but..
<johanot>
srhb: right. sidenote: we
<johanot>
whoops
<johanot>
we've also considered simply nix-copying result paths to a central store after each sucessful build.
zupo has joined #nixos
<srhb>
niso: It's rather difficult to do properly with nixops. Hydra is very imperatively set up. The main pain points is eg. the declarative project/jobset setup, which you basically do by browser.
<srhb>
niso: (Of course, you could automate this, but...)
<srhb>
johanot: Yeah, that makes sense too.
<manveru>
hm, i should just try using cachix with it... still hoping that domenkozar[m] will someday invite me to hercules :P
<niso>
srhb: i'm kinda wondering why this is the case?
<srhb>
niso: Why the config of eg projects isn't declarative?
<srhb>
niso: It has the feel of something that was initially intended to be multi tenant somehow. But ultimately, I don't know. :)
<srhb>
niso: Anyway, there are guides out there for setting up the declarative jobsets and then you never have to mess with it again except through a nix expression under revision control, which makes it tolerable. :)
<niso>
srhb: exactly, i'd expect nix related projects to treat declarative configurations as first-class citizens
<srhb>
Indeed..
<srhb>
Funnily, that's not always the case :-P
eagle2com has joined #nixos
<niso>
what also somewhat bothering me is that it's remarkably hard to use nix as a replacement for docker-compose - (also with nearly no dummy-projects available)
<srhb>
niso: For spinning up actual docker containers?
<niso>
srhb: no, for moving my dev-environments from docker-containers to nixos-containers
<srhb>
I've only heard someone do that like once before. The usual way is nix shells.
bakakuna has quit [Quit: bakakuna]
anon has joined #nixos
<srhb>
For development I feel nix mostly obviates the need for containers entirely.
<niso>
srhb: nix shells is fine as long as you don't need e.g. a database on your project
bakakuna has joined #nixos
<srhb>
Even then it's pretty fine.
anon is now known as Guest70275
<srhb>
Though the lack of nixos options is obviously a bit of a shame.
<niso>
srhb: how? (maybe i'm taking a wrong approach)
<niso>
srhb: e.g. right now i'm trying to setup a dev environment for yesod+mongodb
<srhb>
niso: Yeah, in that case I'd probably want the nixos bits too..
<niso>
srhb: question: can i use the container.nix options in imperatively started nixos-containers?
<srhb>
niso: I think in general I'd like a more pleasing, faster and more configurable test framework.
<srhb>
niso: What is container.nix?
<srhb>
niso: Do you mean stuff like allowedDevices etc?
<srhb>
I think you only get to fiddle with the individual container configs.
<niso>
srhb: this includes e.g. the "bindMounts" option, i don't see how i could use it within an imperativly created container (however i'm quite new, so maybe there's a way)
Okinan has quit [Quit: Konversation terminated!]
adamt has joined #nixos
adamt has joined #nixos
adamt has quit [Changing host]
magnetophon has quit [Ping timeout: 245 seconds]
<srhb>
niso: No, I think you're right, imperative containers only get to touch (in spirit) config.containers.name.config, not name.*
<niso>
what i don't quite get is where there is a toplevel isContainer config, however ther are the container.<name>.<options> aswell
<niso>
by chance you are aware of a way to work arround that?
<srhb>
niso: Not short of modifying the settings globally anyway, for all containers.
<{^_^}>
[nixpkgs] @Profpatsch merged pull request #65308 → buildBazelPackage: autodetect nix toolchain instead of Xcode on Darwin → https://git.io/fjDZW
<{^_^}>
[nixpkgs] @Profpatsch pushed commit from @kalbasit to master « buildBazelPackage: autodetect nix toolchain instead of Xcode on Darwin (#65308) »: https://git.io/fjD5M
Drakonis has quit [Quit: WeeChat 2.4]
o1lo01ol1o has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
usr has joined #nixos
<{^_^}>
[nixpkgs] @FRidh pushed 248 commits to staging-next: https://git.io/fjD5d
shomodj has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<dminuoso>
Does this mean some package is depending on libtinfo.so.5 and patchelf magic doesnt know which file to pick?
owickstrom has quit [Quit: Connection closed for inactivity]
<srhb>
dminuoso: No, it's just that two filies with the exact same paths and priority have been asked to merge into the same profile
<srhb>
dminuoso: files*
Ariakenom has quit [Quit: Leaving]
<dminuoso>
srhb: What do you mean by "profile" exactly?
<srhb>
dminuoso: Say ~/.nix-profile
<dminuoso>
srhb: Ahh I see.
<srhb>
dminuoso: Two packed that are required provide /lib/libtinfo.so.5 and nix doesn't know which to pick for ~/.nix-profile/lib/libtinfo.so.5
<srhb>
Gah, two packages*.. I can't write today.
<dminuoso>
srhb: So a nix-profile compares to the root directory of a conventional linux distribution in which you'd have defined paths where to find shared objects and binaries?
<dminuoso>
Right.
<dminuoso>
Gotcha.
<srhb>
dminuoso: In a sense, yes. It's a way to (among other things) produce an unambiguous environment that looks sort of like a regular tree with bin, share, lib...
<srhb>
By merging together a bunch of nix store paths.
<dminuoso>
That makes sense.
<pie_>
elvishjerricco: did you file any issues anywhere
<dminuoso>
srhb: Wouldn't that be an artifact of incorrect packaging? I mean presumably the derivation for arduino should throw libtinfo.so.5 and use the one from ncurses instead, no?
<srhb>
dminuoso: I don't know that thing in particular, so it might be that both ncurses and arduino should rely on a third package for libtinfo.so.5 at runtime.
<srhb>
dminuoso: But you may be right, yes.
<srhb>
dminuoso: You can either fix that or change the priority of one or the other so that nix will merge them together in the env, preferring the version from one of the packages.
<dminuoso>
srhb: Such a setting would be profile specific, wouldn't it?
<dminuoso>
Since whether one package should have a higher priority might depend on the exact environment Im settingup
<dminuoso>
I see.
<dminuoso>
Ahhh.
<srhb>
dminuoso: If you need differences, I'd just create multiple attrs with different priorities in meta..
chaoflow has quit [Ping timeout: 276 seconds]
<manveru>
can we just deprecate nix-env already? :P
* manveru
hides
<mpickering>
How do I change the location of my nixos-config?
<mpickering>
I need to tell nixos-rebuild to not look in `/etc/nixos/configuration.nix`
<mpickering>
but look in the new place
<srhb>
mpickering: change the NIX_PATH-entry nixos-config to point to it
<srhb>
For one time things, you can use -I for that.
<mpickering>
How do I use `-I`?
<srhb>
mpickering: Frankly I prefer keeping it at the default location because there's no declarative way to change it for the next rebuild that I know of, which can be pretty confusing.
<no-n>
11:14 PM <lm-> no-n: you don't get memory leak, you get 'still reachable', maybe nixos allocates some static buffer, and on exit it is not freed
<pie_>
elvishjerricco: while i screw around with making more space: it might be interesting to test with the 0.7 line on a 0.8 and see if its just the feature flags that mess us up https://chungy.keybase.pub/zfs-features-table.html
<pie_>
elvishjerricco: alternatively, we arent accidentally using luks2 or something? (did you also rebuild the luks partition?)
<pie_>
elvishjerricco: because grub doesnt support luks2
<pie_>
> sudo grub-probe -d /dev/sdd2
<pie_>
grub-probe: error: unknown filesystem.
<{^_^}>
undefined variable 'grub-probe' at (string):255:6
<pie_>
ok that might explain things...
<pie_>
cryptsetup luksDump /dev/sdd2 says "version: 2" which im assuming means luks 2
<pie_>
clever: some ideas re: yesterday ^
<pie_>
it still bothers me that grub just somehow autoselects what drive to use. but lets see if i can fix this now
orivej has quit [Ping timeout: 245 seconds]
<pie_>
i think some distros recently switched to defaulting to luks2 and thats what might have happened
<pie_>
elvishjerricco: it bothers me that we didnt even really get to ssee any sort of failure or anyhting grub just generated the wrong book config -_-
<pie_>
boot
<elvishjerricco>
Agh yea ok. That sucks. Just confirmed my machine uses luks 1
<elvishjerricco>
clever's probably right: probably better to just avoid grub's fancy stuff altogether :P
<{^_^}>
[nixpkgs] @FRidh opened pull request #65372 → qt5.wrapQtAppsHook: wrap more than just ELF's → https://git.io/fjDbd
<pie_>
elvishjerricco: do you think we could get some sort of warning added to config generation that warns if cryptodisk is enabled for an old luks or something
<pie_>
not really sure what to do but
<pie_>
i mean in the nix modules
<elvishjerricco>
I doubt it. The Nix config doesn't know what version of luks is in use
<niso>
elvishjerricco: how did you check the luks version?
<pie_>
cryptsetup luksDump /dev/whatever
<elvishjerricco>
^
<niso>
thx
* niso
uses 1 too
<niso>
pie_: congratz on figuring it out!
<niso>
elvishjerricco: to you too
<pie_>
im very frustrated with this problem
<pie_>
there has to be some sort of workaround to making other people not have to bang their head at the wall about this
<pie_>
how likely is it that at least several people are going to run into this for the next nixos release
<pie_>
oh cool new issue templates on nixpkgs github
<elvishjerricco>
pie_: Grub's luks and zfs support are pretty rarely used. Almost everyone with encrypted / just leaves /boot on an unencrypted drive that grub trivially read
<elvishjerricco>
So I'd say it's unlikely that people encounter problems unless they specifically read my blog post :P
<elvishjerricco>
So yea I'll have to update it
mkoenig has quit [Ping timeout: 246 seconds]
logzet has quit [Ping timeout: 264 seconds]
<danderson>
depending on your threat model, the only complete solution for encrypted boot is to use Secure Boot, with a signed bootloader and signed kernel
Makaveli7 has quit [Quit: WeeChat 2.5]
<danderson>
that way having an unencrypted kernel is not an issue, you run trusted code all the way to the prompt asking you to decrypt stuff
<danderson>
but it's still quite hard to set up, because people still have old fears about secure boot
<elvishjerricco>
Yea, but it'd be nice if the kernel could remain confidential, if it weren't possible for the kernels for different generations to be deleted, and if the kernel could be stored on redundant storage via zfs
<ehmry>
I used to have commit access to nixpkgs, but left the github group because I wasn't commiting anymore, seeing that the backlog of PRs is so big, can I apply to be reinstated as a committer?
<elvishjerricco>
Which is why I want a boot loader for EFI systems based off a linux kernel with EFI STUB :P
jojo88 has quit [Ping timeout: 272 seconds]
<{^_^}>
[nixpkgs] @abbradar opened pull request #65373 → bazel-remote, bazel-deps: mark as broken → https://git.io/fjDNO
<evanjs>
Not sure if related to home-manager (ping @rycee) or what, but does anybody have an idea as to why oh-my-zsh (without any plugins, even), which is currently installed via home-manager, breaks nixos-* command completion? Instead, it just lists the current directory on TAB as if I was using ls or something. My current workaround --
<manveru>
evanjs: `for f in $fpath; { find $f -name _nix 2>/dev/null }`
adamt has quit [Ping timeout: 258 seconds]
<manveru>
that's how zsh finds its completions, might help
dm9 has quit [Quit: WeeChat 2.5]
<manveru>
mine is in `/run/current-system/sw/share/zsh/site-functions/_nix`, so not related to home-manager
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
orivej has joined #nixos
<manveru>
ah, i also have `nix-zsh-completions` installed
<manveru>
that's where this file comes from here
veske2 has quit [Quit: This computer has gone to sleep]
<evanjs>
Sure, but that should be the case here, too: "Setting programs.zsh.enable = true in /etc/nixos/configuration.nix will automatically install and enable nix-zsh-compeletions."
seanparsons has quit [Ping timeout: 245 seconds]
<manveru>
ok then
<manveru>
the question is, what does your $fpath look like
<manveru>
first few line of the zshrc should set that
<evanjs>
Alright. I'll see what everybody there thinks. Thanks!
veske2 has quit [Ping timeout: 272 seconds]
srid6390803 has quit [Ping timeout: 272 seconds]
seanparsons has quit [Ping timeout: 245 seconds]
<pie_>
clever: why isnt boot.debug1devices the default? if boot fails its stuck where it is anyway so you might as well have the option to debug it by default
seanparsons has joined #nixos
<clever>
pie_: dont know
virus_dave has joined #nixos
vmandela has joined #nixos
<pie_>
clever: i mean is there any reason it shouldnt be?
<clever>
pie_: security in situations where you dont want to give root shells
zupo has joined #nixos
<pie_>
oh
<pie_>
i guess that makes sense
<pie_>
( i guesss)
vmandela has quit [Remote host closed the connection]
<clever>
dminuoso: in this case, its a nix-shell for using those programs, and it auto-loads their auto-completes into the shell
jasongrossman has quit [Ping timeout: 245 seconds]
<dminuoso>
clever: Ah that could be helpful, cheers.
<clever>
dminuoso: if your putting it somewhere system wide, you could also make some optimizations
<pie_>
mog: so i _would_ want hibernation support, would your variant work if my root is on a zfs+luks?
<clever>
dminuoso: using pkgs.runCommand, you can run `foo --bash-completion-script` at nix build time, and drop it into a .sh file, that you then simply source later on when opening shells
srid6390803 has joined #nixos
zupo_ has quit [Ping timeout: 244 seconds]
<clever>
dminuoso: oh, it says bash right there, what if you change that to zsh? :D
<dminuoso>
clever: What would be an equivalent way to do this from inside a nix-shell when I want to work on my parsing?
<dminuoso>
Oh wow no idea what my brain did there.
<dminuoso>
I want to interactively develop on my commandline flags, such that I dont have to reload my nix-shell.
<clever>
dminuoso: i think the 2nd argument there, is the path to the binary, so that would be something under dist/, and you may need to re-run source repeatedly
<dminuoso>
Perhaps I should just start with the problem Im having
<mog>
pie_, id assume so i only use ext4 but my drive is just fat32 for efi, and encrypted lvm with ext4 and swap
<clever>
pie_: of note, zfs doesnt support hibernation
<clever>
pie_: so if you have a zfs root, give up already :P
<mog>
i dont understand the benefits of zfs...
<mog>
ext just works, dont know what more youd want a filesystem to do
<clever>
mog: snapshots, zfs send|recv
<pie_>
clever: wait what
<samueldr>
pie_: my swap is encrypted transitively by being a swapfile on my ext4fs
<{^_^}>
[hydra] @edolstra merged pull request #657 → Create `pg_trgm` extension in the NixOS module → https://git.io/fjwLL
erasmas has joined #nixos
matthuszagh has joined #nixos
<matthuszagh>
hello, I'm trying to build a package using ccache, but I'm getting the error: "Directory '/var/cache/ccache' does not exist". The directory does exist and it has the correct permissions set. I checked the nixpkgs source code where this error is triggered (in ccache.nix) and everything looks fine to me.
<clever>
matthuszagh: the nix sandbox wont allow access to /var/cache/ccache/
<evanjs>
matthuszagh: Looks like programs.ccache.packageNames can be used to compile stuff via ccache. Or are you already using that?
<matthuszagh>
clever: the solution is to set `nix.useSandbox = false;`?
<matthuszagh>
evanjs: yeah that's what I'm using
__Sander__ has quit [Quit: Konversation terminated!]
wfranzini has quit [Remote host closed the connection]
<clever>
matthuszagh: you still wont have write access to the directory, and turning off the sandbox has a lot of purity problems
gentauro has quit [Read error: Connection reset by peer]
<Orbstheorem>
Hi ! How can I see a closure size without downloading ? (ie. I wanna know how much space I'd need to `nix-shell -p dodo`)
<matthuszagh>
clever: oh ok. is there any good way to use ccache then?
gentauro has joined #nixos
<clever>
matthuszagh: i dont see any feasible way to make it work during nix based builds, but nix-shell could still work
<matthuszagh>
clever: i do everything from configuration.nix so i guess that's an issue then
<infinisil>
Orbstheorem: Not really no
<matthuszagh>
clever: is ccache fundamentally at odds with the way nix performs builds, or it would just take some extra work to make them more compatible? i really like keeping everything in a declarative model (i use home-manager in configuration.nix as well) but I have quite a few overlays at this point and small changes to certain dependencies can trigger a long chain of rebuilds. is there another way to achieve what I'm hoping for? I know bazel
<matthuszagh>
exists, but I like keeping everything within nix if I can
<infinisil>
Orbstheorem: Well, it could work with some scripting
<infinisil>
Orbstheorem: By first getting the store path of the thing you want to build (e.g. with `nix eval nixpkgs.dodo.outPath)
<clever>
matthuszagh: nix basically does the same thing as ccache, but at the package level
<clever>
matthuszagh: if all inputs to a package are unchanged, it will reuse the entire build product
<matthuszagh>
clever: but then if i make even a small change to a dependency, it can fully rebuild a ton of packages
<infinisil>
Orbstheorem: Which then lists all runtime dependencies that has. Then you'd have to do the same thing for each of them recursively, while accounting for outputs you already have locally
<infinisil>
Orbstheorem: I'd love if Nix had some functionality that did this built-in
<clever>
matthuszagh: yeah, not much can be done about that right now
bakakuna has joined #nixos
<matthuszagh>
clever: ok fair enough. thanks for the input
shomodj has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<joko>
Hey, is anyone using an FHS environment in shell.nix so that downloaded binaries work directly? If so, could you share your setup?
<Orbstheorem>
I think one could script something that scans nixpkgs top-level and recursively queries the cache on anything not on top level to calculate its size
rycwo has joined #nixos
bakakuna has quit [Read error: Connection reset by peer]
* Orbstheorem
considers pushing that to his todo-list
bakakuna has joined #nixos
<clever>
Orbstheorem: *looks*
<infinisil>
Not sure if that's anything nixpkgs-specific
<joko>
adisbladis: I was thinking more like buildFHSUserEnv, I'm running gradle tasks which download extra binaries in the build directory, which I don't have the luxury to patch them one by one
fusion809 has quit [Remote host closed the connection]
<matthuszagh>
since adding a large binary package to my configuration.nix, nixos-rebuild switch takes a long time. each time I see: "warning: dumping very large path (> 256 MiB); this may run out of memory", which seems to cause the hold up. the binary package does not change. is there any way around this slowdown?
chaoflow has joined #nixos
<clever>
matthuszagh: use pkgs.requireFile
magnetophon has joined #nixos
<matthuszagh>
clever: do I set that to the same value as src? i.e. the tar.gz?
<aleph->
Huh caused a kernel panic while setting up a new machine
<aleph->
Neat
<wrl>
aleph-: 5.2 has been really sketchy for me
<matthuszagh>
infinisil: now i'm getting a curl error that it can't download this from any mirror even though it's available locally
<aleph->
wrl: Oh hey!
<wrl>
!
<aleph->
wrl: Yeah I did get an issue with my memory alignment not being correct
<aleph->
Maybe that caused it
<wrl>
i've just had weird like... usb issues
<aleph->
Or helped.
<wrl>
some weird xorg stuff
<wrl>
idk
<aleph->
Fun
<pingveno>
adisbladis: The default.nix I linked to includes those in buildInputs. I think there's a chance this is a Pillow-specific thing. It has a lot of weirdness in its setup.py.
<wrl>
yeah
<aleph->
Just trying to set up a small headless server
<pingveno>
Or at least it included zlib. Not sure about the others.
<infinisil>
matthuszagh: Does it not output a sha256 to use?
<{^_^}>
[nixpkgs] @AtkinsChang opened pull request #65385 → buildRustPackage: avoid appending to source code's cargo config → https://git.io/fjDjX
<infinisil>
matthuszagh: This is with builtins.fetchurl { url = file:///path/to/file; sha256 = "000000.."; }?
<aleph->
Eh I can fiddle with it in a bit.
<matthuszagh>
infinisil: yeah that's what i did, gave it an initial sha of all 0s to find the right sha
<pingveno>
Ah, pkgconfig was the missing piece of the puzzle.
<pie_>
yorick: sorry i dont know about haskell (yet?)
<infinisil>
matthuszagh: Hmm.. well maybe using clever's initial suggestion of using pkgs.requireFile might be better for now since this doesn't seem to work
<infinisil>
And it also makes more sense to use pkgs.requireFile if you intend to distribute this nix expression
<infinisil>
matthuszagh: Can you paste the error you get with builtins.fetchurl?
<matthuszagh>
error: cannot download Xilinx_Vivado_SDK_2017.2_0616_1.tar.gz from any mirror
<matthuszagh>
curl: (37) Couldn't open file /home/matt/src/Xilinx_Vivado_SDK_2017.2_0616_1.tar.gz
magnetophon has quit [Ping timeout: 246 seconds]
<infinisil>
matthuszagh: Ohh, you're not using builtins.fetchurl
<infinisil>
You're using pkgs.fetchurl
<infinisil>
It won't work with that
runciter has joined #nixos
<matthuszagh>
infinisil: hm now i get error: out of memory
<yorick>
pie_: seems to be working
<ajs124>
matthuszagh: I have a fhsenv for vivado somewhere, if you don't have one already. I couldn't manage to get the hwserver to run, though.
orthros has quit [Ping timeout: 260 seconds]
<ajs124>
So synthesis, simulation, implementation and bitstream generation will work, but you probably won't be able to program your fpga.
<infinisil>
matthuszagh: Ah darn, okay just use pkgs.requireFile then
magnetophon has joined #nixos
linarcx has joined #nixos
<runciter>
i'm using the nix package manager on another distribution. i've done `nix-env -i git`, but clones fail because libkeyutils can't be found, e.g. "/nix/store/<hash>/.../git-remote-https: error while loading shared libraries: libkeyutils.so.1: cannot open shared object file: No such file or directory"
<runciter>
what did i do wrong?
<runciter>
LD_DEBUG=all tells me that libgssapi_krb5 wants libkeyutils, and that `git-remote-https` fails to find libkeyutils in promising looking places like `/nix/store/<hash>-keyutils-1.6/lib/libkeyutils.so.1`, but that's a dangling symlink
<matthuszagh>
ajs124: i've got something for vivado but i'm not sure it'll work.. would you mind sending me what you have? shame about hw_server... have you figured any workarounds, or for the time being you can only program the fpga from a different distro?
<matthuszagh>
infinisil: i'll try that. side not, i was able to avoid the memory error by using the lessram pr for nix. that doesn't seem to work for builtins.fetchurl though
<{^_^}>
[nixpkgs] @edolstra merged pull request #65083 → lib: allow sourceByRegex to be composed after cleanSourceWith → https://git.io/fjMTM
<ajs124>
matthuszagh: my workaround was using windows. my buildFHSUserEnv basically boils down to "multiPkgs = p: with p.xorg; [ libXext libX11 libXrender libXtst libXi ] ++ [ gcc glibc.dev libstdcxx5 ncurses5 minicom strace ];"
<infinisil>
matthuszagh: you got a different error instead?
<matthuszagh>
infinisil: nope the same one: "error: out of memory"
<infinisil>
matthuszagh: Buut you just said you were able to avoid that error?
<matthuszagh>
infinisil: i can when i use src as a local directory, but apparently not with fetchurl
<infinisil>
You got a memory error when you just used src = /home/...?
<infinisil>
Because you originally just said that took a long time
<matthuszagh>
i did before the experimental pr (binary is huge, like 20GB)
<infinisil>
Ah
<matthuszagh>
experimental pr fixed the error, but still took a long time each rebuild. does that make sense?
<matthuszagh>
ajs124: gotcha thanks. hopefully i can get the whole thing working without resorting to another distro or windows etc. by the way, what's the policy for contributing packages that require a manual download first? can those be contributed to nixpkgs?
<infinisil>
I think
<ajs124>
they definitely exist. e.g. displaylink and citrix receiver are in there.
<matthuszagh>
well i imagine it makes a whole lot more sense to you than it does it me :). gonna try with requireFile now and will report back
<matthuszagh>
ajs124: cool, if i get it working i'll contribute it back then
<evanjs>
Figured the oh-my-zsh issue thing out. The system option for enabling oh-my-zsh works fine, but home-manager's oh-my-zsh.enable doesn't seem to load nix-zsh-completions properly (and requires a call to compinit, etc)
logzet has joined #nixos
srid6390803 has quit [Ping timeout: 272 seconds]
drk_lrd_877_ has joined #nixos
davidak has joined #nixos
Shoubit has quit [Ping timeout: 245 seconds]
___laika has quit [Quit: WeeChat 2.4]
matthuszagh has quit [Remote host closed the connection]
englishm has quit [Excess Flood]
englishm has joined #nixos
ddellacosta has joined #nixos
orbekk has quit [Quit: WeeChat 2.4]
<ashkitten>
hm, i haven't been able to get boot.grub.memtest86.enable to work properly. the option appears, but if i choose it in grub i just get the grub background frozen
morgib has joined #nixos
fragamus has joined #nixos
<clever>
ashkitten: legacy or efi?
<ashkitten>
efi
<clever>
ashkitten: i dont think memtest works on efi currently
<ashkitten>
oh
srid6390803 has joined #nixos
<ashkitten>
i guess i'll drop the memtest86+ efi stub in there manually then
linarcx has quit [Quit: WeeChat 2.5]
linarcx has joined #nixos
linarcx has quit [Client Quit]
linarcx has joined #nixos
eagle2com has joined #nixos
evanjs has quit [Quit: WeeChat 2.5]
sellout- has joined #nixos
Yaniel has quit [Ping timeout: 245 seconds]
Chiliparrot has joined #nixos
magnetophon has joined #nixos
wildtrees has joined #nixos
linarcx has quit [Quit: WeeChat 2.5]
linarcx has joined #nixos
<runciter>
if something i've installed from the binary caches fails because it can't find a library, does that mean the binary is broken?
linarcx has quit [Client Quit]
<infinisil>
Is the Nix store readonly on non-NixOS linuxes?
<cransom>
it would depend on single/multi user, no?
<adelbertc>
has anyone experienced strange issues with `buildRustPackage`? I am getting a `no matching package named `getrandom_package` found (required by `rand`)` error but looking at my `Cargo.lock` file I am using `rand 0.6.5` which does not mention a `getrandom_package`
<exarkun>
Should I expect a system crash, network interruption, or other local problems to result in corrupted (eg partially written) /nix/store contents? Or is this something particularly weird to encounter?
davidak has quit [Quit: Leaving]
<alienpirate5>
i don't want it to get built or anything
ddellacosta has joined #nixos
alex`` has quit [Ping timeout: 258 seconds]
ris has joined #nixos
<{^_^}>
[nixpkgs] @eraserhd opened pull request #65388 → tmux-ctrlw: init at 0.1.0 → https://git.io/fjyfU
<inquisitiv3>
How do I set DNS for WireGuard? I can't find any options when I search on the website.
zupo has quit [Ping timeout: 268 seconds]
magnetophon has joined #nixos
adamt has joined #nixos
adamt has joined #nixos
adamt has quit [Changing host]
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
joehh has joined #nixos
Okinan has joined #nixos
tonic1 has joined #nixos
joehh has quit [Ping timeout: 245 seconds]
troydm has quit [Quit: What is Hope? That all of your wishes and all of your dreams come true? To turn back time because things were not supposed to happen like that (C) Rau Le Creuset]
<___laika>
what is the difference between a set of package derivations for a language, say `langPackages`, and `pkgs.recurseIntoAttrs langPackages`? im trying to see what benefit it brings in this post https://blog.jethro.dev/blog/ocaml_with_nix/
<{^_^}>
[nixpkgs] @mattmelling opened pull request #65390 → screamingfrogseospider: init at 11.0 → https://git.io/fjyf7
wfranzini has joined #nixos
<{^_^}>
[nixpkgs] @worldofpeace merged pull request #52859 → cmake: use multiple outputs for GNUInstallDirs → https://git.io/fhTex
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to staging: https://git.io/fjyfd
troydm has joined #nixos
<clever>
___laika: i think recurseIntoAttrs specially marks it so `nix-env -i` can search that whole set
<clever>
and recurseIntoAttrs then makes -i even slower :P
<worldofpeace>
eyJhb: looking now :)
<eyJhb>
Great :)
<___laika>
that makes sense to me, but im having a hard time seeing the difference it makes in the setup for the blog post i mentioned. will recurseIntoAttrs make a big difference for the purposes of nix-shell?
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/fjyfx
<___laika>
ok, thats helpful. just to be sure ive got it, the only place recurseIntoAttrs is particularly useful is in expressions that reside in nixpkgs, for consumption by nix-env commands?
<clever>
___laika: pretty much
<tilpner>
___laika: recurseIntoAttrs will set recurseForDerivations. release-lib.nix also pays attention to the presence and value of that key, so it can affect CI with e.g. Hydra
<{^_^}>
[systemd] @abbradar opened pull request #30 → Additional paths to look for executables for systemd-sleep and systemd-shutdown → https://git.io/fjyJI
fragamus has joined #nixos
<{^_^}>
[nixpkgs] @grahamc merged pull request #65358 → kitty: fix wayland support (EGL: Library not found error) on swaywm → https://git.io/fjD1b
<aleph->
Seems my little box can't find it's init...
<aleph->
That's odd?
<ashkitten>
worldofpeace: indeed tests appear to be impure
ehmry has left #nixos ["bbl"]
ddellacosta has quit [Ping timeout: 258 seconds]
<ashkitten>
it failed even on channel:nixos-19.03
<ashkitten>
worldofpeace: with your patch it gives: error: some outputs of '/nix/store/a300iy3jqd6pdp6z1i17pfzvy5g36x7f-xdg-desktop-portal-1.4.2.drv' are not valid, so checking is not possible
<ashkitten>
i should probably backport the patches to add ryzen 3xxx pci ids to kernel and k10temp, but i decided to just apply them cleanly to linux-latest
<ashkitten>
wish me luck tbh
<ashkitten>
worldofpeace: sounds good
MmeQuignon has quit [Remote host closed the connection]
<averell>
what is the preferred style of adding review fixes? add fixups commits, or amend / force push yourself? what about rebasing on master?
<ashkitten>
worldofpeace++
<{^_^}>
worldofpeace's karma got increased to 25
<Thra11>
How can I get nix to fetch and unpack the sources required to build a derivation, then stop so I can study them? Is it possible to e.g do that but stop post-patch phase instead?
<colemickens>
I'll send one for mesa to 19.1.3 too
ng0 has quit [Quit: Alexa, when is the end of world?]
<worldofpeace>
sounds good colemickens. do you discern that the libva bump should dissapate the issues with it and iris?
<worldofpeace>
(at least when using mesa on master)
<averell>
Thra11: nix-shell -p <package> and then you can manually call the phases such as unpackPhase
jaykru has joined #nixos
<worldofpeace>
averell: It really depends on your reviewer
<worldofpeace>
averell: but in general, whatever allows the reviewer see the changes you've made in response to them.
<colemickens>
worldofpeace: I have no reason to suspect that bumping libva will help. On the other hand, I have good reason to suspect bumping mesa and/or the iris driver will: https://bugzilla.freedesktop.org/show_bug.cgi?id=110666#c8
<worldofpeace>
colemickens: lol, I've read the same issue. I now get a segfault with master
<averell>
Thra11: do man nix-shell and look at the first one in the Examples section, it's probably excactly what you want.
<andrewrk>
samueldr, according to that arch linux forum, to try the suggestion, I would want to disable firmware version 48 so that it tries the next oldest version
bakakuna has quit [Ping timeout: 245 seconds]
MmeQuignon has quit [Ping timeout: 268 seconds]
MmeQuignon has joined #nixos
abathur has joined #nixos
KarlJoad has joined #nixos
jaykru` has joined #nixos
<colemickens>
worldofpeace: surely some of the problem could come from trying to use the iris driver at a newer version with a mesa at a different rev? maybe?
mexisme has joined #nixos
<andrewrk>
I'm trying this: postInstall = "rm $out/lib/firmware/iwlwifi-cc-a0-48.ucode"; and then nixos-rebuild switch -I nixpkgs=/home/andy/dev/nixpkgs
<jaykru`>
anybody here know much about the cross compilation scheme set up in nixpkgs? I'm trying to add a new system for riscv64 embedded systems, with the LLVM triple being riscv64-none-elf and the libc being newlib.
jaykru has quit [Ping timeout: 245 seconds]
adelbertc has quit [Quit: Connection closed for inactivity]
<andrewrk>
hmm, that didn't prevent the kernel from finding version 48 still
<andrewrk>
is this the correct command to use nixpkgs source as the channel? sudo nixos-rebuild switch -I nixpkgs=/home/andy/dev/nixpkgs
<andrewrk>
my postInstall script didn't seem to get picked up
Yakulu has left #nixos ["Error from remote client"]
<colemickens>
ugh, what even is libva-vdpau
<worldofpeace>
colemickens: that's possible, but you know, so many cpu cycles to find out :P
<andrewrk>
does "dontFixup = true;" prevent postInstall scripts from running?
<colemickens>
seriously, though, what? libva-vdpau looks like it hasn't been touched in years? and gentoo is patching it to keep it building? grumble.
troydm has quit [Ping timeout: 246 seconds]
KarlJoad has quit [Ping timeout: 268 seconds]
troydm has joined #nixos
<worldofpeace>
i really wish they switched those things to use a pronounceable name.
__monty__ has quit [Quit: leaving]
<andrewrk>
ok yeah this is weird, I did `nix-build . -A firmwareLinuxNonfree` and it didn't seem to run my postInstall script of rm $out/lib/firmware/iwlwifi-cc-a0-48.ucode
<andrewrk>
this is 19.09.git.b5f5c97 (Loris) + the postInstall addition
<clever>
andrewrk: if the outputHash is uncchanged, nix will never re-run the derivation
<clever>
andrewrk: you likely want to remove it entirely
<andrewrk>
oh I thought the output hash was being computed based on postInstall scripts and such
<andrewrk>
but I guess this nix expr is overriding it for some reason?
<clever>
andrewrk: outputHash is a promise to nix, that the output will always have the given hash
<clever>
andrewrk: and if an output with that hash exists, it wont bother running the rest of the drv to make it
<andrewrk>
I see, thanks
webster23 has quit [Ping timeout: 250 seconds]
noudle has quit []
doyougnu has joined #nixos
KarlJoad has quit [Quit: Konversation terminated!]
knupfer has quit [Ping timeout: 250 seconds]
o1lo01ol1o has quit [Remote host closed the connection]
<slabity>
Updated my system today and now thunderbird won't start and firefox won't allow uploading files. Libreoffice is failing too. Did something major change in unstable/master?
<gchristensen>
wow, you might try rebooting?
veske has joined #nixos
doyougnu has quit [Ping timeout: 248 seconds]
<slabity>
gchristensen: Yea, I did. Reverting to my previous profile fixed the issue, but my most recent profile always has those issues.
<slabity>
For some reason thunderbird and libreoffice just seg-faults, so no useful error information either :/
<wildtrees>
that sounds pretty broken
<slabity>
Strange thing is that those apps aren't even on my system-profile. Just my user environment
<gchristensen>
ouch.
<maralorn>
slabity: So maybe bump both? system-profile und user environment?
<slabity>
Firefox's file error is `No such interface “org.freedesktop.portal.FileChooser”` which brings up a lot of results in google that say to install xdg-desktop-portal... But that is already running
<slabity>
I've tried updating thunderbird and firefox and it didn't seem to work
<samueldr>
yes, there was a new change
gagbo has quit [Ping timeout: 272 seconds]
<samueldr>
portals were enabled
hio has quit [Quit: Connection closed for inactivity]
<samueldr>
cc worldofpeace
<maralorn>
Hm, so slabity. I am not skilled enough to help you. But I can ensure you, you are not alone with your bug. The complaint came up at least twice today already.
<samueldr>
I don't personally know what it means, especially with regards to system/user profiles "mismatch"
<samueldr>
(if any mismatch)
<slabity>
Okay. As long as it's a known issue. I can easily revert profiles until it's fixed
<slabity>
Let me know if there's any information I can provide that can help debug
<averell>
libreoffice is also known, there's #65289
<slabity>
Looks like a lot of builds are broken on my machine: `riot-desktop` `virt-manager` `freecad` `evince` `libreoffice` and `wireshark-qt` all included
MmeQuignon has quit [Ping timeout: 245 seconds]
doyougnu has joined #nixos
doyougnu has quit [Client Quit]
MmeQuignon has joined #nixos
liberiga has quit [Ping timeout: 260 seconds]
<{^_^}>
[nixpkgs] @mmilata opened pull request #65397 → sympa: init at 6.2.44 + NixOS module → https://git.io/fjyTi
<gchristensen>
anyone know if github user jollheef has ever been on IRC?
civodul has quit [Quit: ERC (IRC client for Emacs 26.2)]
MmeQuignon has quit [Ping timeout: 258 seconds]
jaykru` has quit [Quit: ERC (IRC client for Emacs 26.1)]
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
matthuszagh has joined #nixos
liberiga has joined #nixos
<colemickens>
riot-desktop, virt-manager, evince are working for me on nixos-unstable
<colemickens>
oh, I missed the bit about portals
<colemickens>
but all the same
<matthuszagh>
does nixpkgs have support for downloading octave packages (e.g. signal, etc.)? i didn't see anything, but wanted to make sure before working on it myself
wfranzini has quit [Remote host closed the connection]
wfranzini has joined #nixos
Soo_Slow has joined #nixos
ris has quit [Ping timeout: 248 seconds]
<worldofpeace>
I don't know why firefox expects there to be an implementation of the filechooser interface to exist if xdg-desktop-portal is there
kvda has joined #nixos
mbrgm_ has joined #nixos
mbrgm has quit [Ping timeout: 250 seconds]
mbrgm_ is now known as mbrgm
<worldofpeace>
slabity: I can't reproduce thunderbird not starting
<slabity>
thunderbird 60.8.0 - Correct?
selfsymmetric-mu has quit [Remote host closed the connection]
<slabity>
Looking at stdout it may be caused by the exquilla extension and my exchange server
<worldofpeace>
idk why this happens with firefox, I'd expect their filechooser to to use it only when org.freedesktop.portal.FileChooser portal is available on the session bus to call the out of process file-chooser. is anyone familar with traversing firefox source code?
Soo_Slow has quit [Remote host closed the connection]
veske has quit [Read error: Connection reset by peer]