<thomasd>
bsima: I have a pkg.nix generated w/ cabal2nix, and I have a shell.nix that is generated automatically from the default.nix file, I think
<bsima>
so what I do is add a "compiler" arg to the nix expression
<clever>
thomasd: instead of using pkgs.haskellPackages on line 5, use haskell.packages.ghc844, and rename the part on the left of the =, so it wont conflict with things
<clever>
thomasd: you can also move parallel-julia up into the haskell overrides
<thomasd>
parallel-julia is the package I am building
<thomasd>
clever: what do you mean by "rename the part on the left of the ="?
<thomasd>
clever: "error: undefined variable 'haskell'" on the very last line :/
<thomasd>
clever: thanks for helping so far!
<clever>
thomasd: oops, thats pkgs.haskell
<thomasd>
clever: error: Package ‘cudatoolkit-9.2.148.1’ in /nix/store/v6wv39y6ahw65g1nyr9pxjkzw4pq8j4g-source/pkgs/development/compilers/cudatoolkit/default.nix:149 has an unfree license (‘unfree’), refusing to evaluate.
<thomasd>
woah
<thomasd>
i am not having anything to do with cuda, lol
<clever>
thomasd: you forgot the -A param, so its trying to build everything
<clever>
thomasd: nix-build -A parallel-julia
jluttine has quit [Ping timeout: 246 seconds]
<thomasd>
clever: nice, something is building now :)
<thomasd>
clever: what does `-A parallel-julia` tell nix, or should I look it up?
<clever>
thomasd: it tells it to lookup that attribute, in the set default.nix returns
<thomasd>
right. synonymous with --attr
<thomasd>
clever: if I returned `parallel-julia` as a result of the nix expr in `default.nix`, it would work without -A, right?
<clever>
yep
<clever>
but then its harder to debug the other packages
<thomasd>
clever: thank you, let's see if this works!
jluttine has joined #nixos
goibhniu has quit [Ping timeout: 250 seconds]
ddellacosta has quit [Ping timeout: 240 seconds]
ddellacosta has joined #nixos
leothrix has joined #nixos
dbmikus_ has quit [Ping timeout: 245 seconds]
eadwu has joined #nixos
pie_ has joined #nixos
sevanspowell has quit [Ping timeout: 256 seconds]
Ariakenom_ has quit [Quit: Leaving]
rcshm has joined #nixos
<{^_^}>
[nixpkgs] @Infinisil merged pull request #53767 → ssh: put custom options before generated options → https://git.io/fhZMH
<clever>
thomasd: looks like a build failure within ghc 8.4.4 itself
<clever>
thomasd: you may need to use a different nixpkgs version
rcshm has quit [Remote host closed the connection]
rcshm has joined #nixos
<gchristensen>
anyone have udev + systemd user service experience? I'm having a real hoot of a time.
<gchristensen>
I have this udev rule, for handling when my thunderbolt-connected display is attached: «ACTION=="change", SUBSYSTEM=="drm", TAG+="systemd", ENV{SYSTEMD_USER_WANTS}="hello.service"» watching the output of «udevadm monitor --environment -s drm --udev --environment» I see events I think should match: https://gist.github.com/grahamc/a824fbc60e09b7066b5b4bdd6b73e5e0 but my hello.service isn't being reliably
<{^_^}>
To get a sha256 hash of a new source, you can use the Trust On First Use model: use probably-wrong hash (for example: 0000000000000000000000000000000000000000000000000000) then replace it with the correct hash Nix expected.
shibboleth has joined #nixos
rcshm_ has quit [Remote host closed the connection]
rcshm has joined #nixos
joepie91 has quit [Excess Flood]
joepie91 has joined #nixos
<{^_^}>
[nixpkgs] @matthewbauer pushed 7 commits to priorities: https://git.io/fhdGi
<{^_^}>
[nixpkgs] @matthewbauer pushed 0 commits to priorities: https://git.io/fhdGX
<{^_^}>
[nixpkgs] @matthewbauer opened pull request #56031 → Add some more priorities → https://git.io/fhdGM
<{^_^}>
[nixpkgs] @uri-canva opened pull request #56033 → buildBazelPackage: autodetect nix toolchain instead of Xcode one on macOS → https://git.io/fhdGh
fresheyeball has quit [Quit: WeeChat 2.2]
Supersonic has quit [Disconnected by services]
Supersonic112 has joined #nixos
Supersonic112 is now known as Supersonic
<{^_^}>
[nixpkgs] @kalbasit closed pull request #55582 → bazel: set BAZEL_USE_CPP_ONLY_TOOLCHAIN=1 on Darwin → https://git.io/fhQBa
<yl[m]>
infinisil: w.r.t https://github.com/NixOS/nixpkgs/pull/53120 the tests are not working and on master VirtualBox is not working with the latest master. We need your feedback on it
<yl[m]>
gchristensen: sorry I missed your message earlier. Yes, it does work without the latest kernel on master. Like yourself, I had to pull this PR into my configuration to make it work. I have not had any time to take a stab at the test, but maybe infinisil would be able to.
o1lo01ol1o has quit [Remote host closed the connection]
zopsi has joined #nixos
daniele- has joined #nixos
andreabedini has joined #nixos
Rusty1 has quit [Quit: Konversation terminated!]
zopsi has quit [Quit: Oops]
zopsi has joined #nixos
o1lo01ol1o has joined #nixos
zopsi has quit [Client Quit]
zopsi has joined #nixos
o1lo01ol1o has quit [Ping timeout: 250 seconds]
ddellacosta has quit [Ping timeout: 250 seconds]
vk3wtf has quit [Ping timeout: 258 seconds]
slack1256 has joined #nixos
ddellacosta has joined #nixos
agander has joined #nixos
reinhardt has quit [Read error: Connection reset by peer]
rawreraw has joined #nixos
drakonis has quit [Quit: WeeChat 2.3]
rawreraw has quit [Quit: WeeChat 2.3]
jackdk has quit [Ping timeout: 272 seconds]
rcshm has quit []
endformationage has quit [Ping timeout: 246 seconds]
simukis has joined #nixos
<srhb>
Hmm. How is autorandr supposed to work? I don't understand how I configure it to understand my various layouts,
dermetfan has joined #nixos
<srhb>
Alternatively, is there some other service that will do this for me? Understanding docking, undocking and picking a layout accordingly.
ddellacosta has quit [Ping timeout: 244 seconds]
agander has quit [Ping timeout: 268 seconds]
LnL has quit [Ping timeout: 255 seconds]
<srhb>
Looks like home-manager might be more fully fledged
DrLambda has quit [Quit: Leaving.]
LnL has joined #nixos
LnL is now known as Guest5095
<{^_^}>
[nixpkgs] @FRidh merged pull request #55372 → buildEnv: break with a proper error if one path is actually a file → https://git.io/fh979
<{^_^}>
[nixpkgs] @FRidh pushed commit from @Ma27 to staging « buildEnv: break with a proper error if one path is actually a file »: https://git.io/fhdcz
hyper_ch2 has joined #nixos
<Myrl-saki>
Great. I have no idea what I'm doing now. Losetup worked a while ago without sudo, now I'm getting operation not permitted.
palo1 has joined #nixos
<Myrl-saki>
Oh, I'm stupid lol. The file itself has root permissions.
slack1256 has quit [Remote host closed the connection]
Jetien_ has joined #nixos
rauno has quit [Remote host closed the connection]
phreedom has quit [Remote host closed the connection]
phreedom has joined #nixos
ost has quit [Remote host closed the connection]
MayeulC has joined #nixos
<tilpner>
fuzzy-id: I don't know what you mean by "haskell exec". That could mean "I want to execute Haskell code", or "I want to execute code compiled from Haskell", and those two interpretation require vastly different closures.
<MayeulC>
Currently trying to install NixOS on a raspberrypi 3 (aarch64), I can't get the serial working (works until "Starting kernel ..."). Any ideas?
<tilpner>
fuzzy-id: If you already have an executable compiled from Haskell, then yes, you probably just want to keep ghc out
<fuzzy-id>
yeah, it's the latter. i have an executable compiled from haskell…
<MayeulC>
I also tried appending this to the kernel command line from the boot partition. Same result.
vidbina has quit [Ping timeout: 255 seconds]
ost has joined #nixos
ThatDocsLady_ has joined #nixos
<makefu>
MayeulC: which image are you using? the one from hydra? because that should just work(tm) for the raspi3
<MayeulC>
makefu: Yes. Unfortunately, I can't plug in a monitor (my VGA screen+adapter doesn't support the mode set by the pi -- maybe I could edit it in the config, though)
<MayeulC>
So I am going for a headless installation. I get U-boot on the console, but nothing from "starting kernel..." onwards :/
<MayeulC>
(By console, I mean trough a serial adapter, of course)
sigmundv__ has quit [Ping timeout: 272 seconds]
<{^_^}>
[nixpkgs] @vcunat pushed to staging-next « go_1_10: fix tests after binutils update »: https://git.io/fhd8a
vidbina has joined #nixos
wfranzini has joined #nixos
Ariakenom has quit [Read error: Connection reset by peer]
Ariakenom has joined #nixos
init_6 has joined #nixos
init_6 has quit [Client Quit]
init_6 has joined #nixos
Guanin has joined #nixos
<fuzzy-id>
found it! shrinked the image from >4gb to 53mb! :D
jasongrossman has quit [Read error: Connection reset by peer]
sigmundv__ has joined #nixos
<{^_^}>
[nixpkgs] @groodt opened pull request #56044 → td: init at 0.16.4 → https://git.io/fhd8N
<infinisil>
Oh and btw, yesterday/two days ago, I asked how you could have all the options in all of your NixOS modules appear in `man configuration.nix`, the answer is it's not possible currently, but there's #56020 and the previous #47177 that allow for that
<rnhmjoj>
is there a way to prevent systemd conflicting with openresolv? both executable and man page for resolvconf is replaced by the dummy command provided by systemd-resolved and it's interfering with dhcpcd hooks
ost_ has joined #nixos
ost has quit [Read error: Connection reset by peer]
<edef>
rnhmjoj: services.resolved.enable = false?
<{^_^}>
[nixpkgs] @vcunat pushed 22 commits to staging-next: https://git.io/fhd4E
<edef>
rnhmjoj: idk if you actually rely on systemd-resolved
<noonien>
hello
<noonien>
has anyone managed to get nixos running on the steamlink by any chance
ost_ has quit [Remote host closed the connection]
<rnhmjoj>
edef: yes, it's disabled
<noonien>
oops, forgot the: ?
ost has joined #nixos
<edef>
resolvconf(1) in PATH very much not the systemd one here, but the manpage is
<ottidmes>
siers: the last command seems to imply you have those channels in both the user and root. You should only have them in the root user and then update nix-channel as root
<siers>
I'm on osx and I heard multi-user mode is broken or something, so I'll stick to single user mode
<siers>
ottidmes, Hm, but how did you know I also have it in root? What implied it?
<siers>
but but but "nixpkgs=~/.…/unstable" wouldn't that change this?
<ottidmes>
siers: I think that depends on how your NIX_PATH is setup, although I agree, I would have assumed that this should not have mattered since things set with -I should have prio over all
<__monty__>
env NIX_PATH=nixpkgs=... would.
<ottidmes>
siers: but I wonder, how did `nix-env -f $HOME/.nix-defexpr/channels/unstable -iA syncthing` not work for you, you did not answer my question what did not work, did you use nixpkgs.syncthing again, rather than syncthing?
<siers>
ottidmes, a second
<siers>
ottidmes, ah, I must've missed the -f
<siers>
that then should've worked
<siers>
and indeed it does
<ottidmes>
siers: good to know, because thats what I generally use when I use nix-env
<ottidmes>
siers: could you maybe print your NIX_PATH? I want to test something, but to be able to reproduce it, I have the best chance with your paths
<siers>
pm'd
<Taneb>
haskellPackages.proto-lens-protoc_0_5_0_0 doesn't build in master because it uses proto-lens rather than proto-lens_0_5_0_0
<Taneb>
Not sure where to fix that as hackage-packages.nix is autogenerated
dejanr has quit [Quit: WeeChat 2.3]
<musicmatze>
Hi, I am packaging an application that needs a c++17 compatible compiler and the `<filesystem>` header - what do I have to bring in scope for that? I already tried `gcc8`, but it does not help with the missing header
iyzsong has quit [Ping timeout: 258 seconds]
hyper_ch2 has quit [Ping timeout: 256 seconds]
<musicmatze>
Ah, gcc8Stdenv seems to work
peterhoeg has joined #nixos
<{^_^}>
[nixpkgs] @matthiasbeyer opened pull request #56048 → fltrdr: init at 0.1.0 → https://git.io/fhdB1
peterhoeg has quit [Client Quit]
<gchristensen>
srhb: did you sort it out? I spent time on it yesterday and it was horrible, but I think I found a nicer solution
<{^_^}>
[nixpkgs] @fpletz pushed commit from @r-ryantm to master « xmr-stak: 2.7.1 -> 2.8.2 (#55804) »: https://git.io/fhdzL
eadwu has quit [Quit: WeeChat 2.3]
Mateon3 has joined #nixos
Mateon1 has quit [Ping timeout: 246 seconds]
Mateon3 is now known as Mateon1
eadwu has joined #nixos
dramforever has quit [Quit: Leaving]
DrLambda has quit [Ping timeout: 250 seconds]
<gchristensen>
srhb: have you tried grobi? gilligan tells me it works better, I didn't use it though -- I did something else (on a call, though, so can't explain yet)
<srhb>
gchristensen: I did not, I'll check it out. No rush :D
o1lo01ol1o has joined #nixos
trevorriles has joined #nixos
<catern>
how does one configure plugins in Hydra?
<srhb>
catern: via hydra.conf. They each have a section like <pluginName>...</pluginName>
<catern>
I can't quite unravel this Perl magic... is there some kind of command line argument?
<catern>
srhb: aha, thanks
<srhb>
catern: The source code is probably your best bet for documentation for each of them. Alternatively git history
<catern>
Yeah that's what I'm reading
<kreisys>
Is anyone running the buildkite-agent successfully on nix-darwin?
<srhb>
catern: If you look for something like $self->{config}->{foo}, foo is the section you care about in each of them, you can sort of decipher what options they take.
<srhb>
catern: Right.. Most (all?) plugins work by calling a perl script on each build.
<catern>
Yes, I figured that much out
<catern>
But populating the list of plugins to use, I can't figure out
<srhb>
gchristensen: That seems like a good idea, thanks!
najibpuchong has left #nixos ["Leaving"]
<catern>
The Perl doesn't seem to use HYDRA_CONFIG/hydra.conf for anything besides getStoreUri and getGCRootsDir, AFAICT
<{^_^}>
[nix] @tollb opened pull request #2687 → Fix build --repeat with -K to use real paths → https://git.io/fhdzx
<lassulus>
would it be sane to autoenable xserver if a display-manager was enabled?
<samueldr>
isn't gdm waylandish?
<srhb>
catern: hydra-notify gets it via getHydraConfig
<lassulus>
hmm, so If I enable gdm I don't have to enable xserver?
<lassulus>
lets test
<srhb>
catern: I don't know the specific syntax for email notifications, you'll have to parse the specific plugin perl (since every plugin can implement its own syntax)
<samueldr>
I'm talking out of my behind right now :) I wouldn't know for sure, but IIRC gnome tries to nudge itself definitely ont the wayland side of the line
<{^_^}>
[nix] @tollb opened pull request #2688 → Fix nix-build --check -K in sandbox w/o root → https://git.io/fhdgJ
ddellacosta has joined #nixos
<catern>
srhb: no no, I just mean the generic syntax
agander has joined #nixos
<srhb>
catern: The generic syntax is <pluginname>...</pluginname>
<srhb>
catern: Each plugin can implement (...) basically however it sees fit.
<catern>
srhb: you said <pluginName>...</pluginName> - is that just inserted on its own line? the rest of the file is key = value\n so I'm a little perplexed
<srhb>
catern: Oh, sorry, each of those bracketed bits are on their own lines
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/fhdgm
<furrycatherder>
i tried using placeholder and i got --prefix=/1rz4g4znpzjwh1xymhjpm42vipw92pr73vdgl6xs1hycac8kf2n9, what does this mean?
<furrycatherder>
shouldn't it be /nix/store/foo?
knupfer has joined #nixos
<srhb>
furrycatherder: That is the special sauce hash that is eventually replaced.
<furrycatherder>
srhb: ahh i thought so :)
<srhb>
furrycatherder: (it really only depends on the string "out")
<catern>
srhb: okay, thank you, this is very helpful, I'll give this all a try
<srhb>
So you get a different hash from eg. placeholder "lib" which is replaced with the contents of $lib at build time, etc..
<srhb>
(Well, approximately)
<disasm>
would it make sense if forwardPorts is set in any container to set host net.ipv4.ip_forward = true?
wedens has quit [Quit: Connection closed for inactivity]
<furrycatherder>
srhb: is this part of patchelf?
<furrycatherder>
srhb: oh wait, no
<disasm>
I just spent a few minutes debugging why my ports weren't forwarded to my container, which wasn't a big deal since I knew what pieces to check for, but I could see someone with less experience not being so happy.
<srhb>
furrycatherder: No, it happens in libstore.cc
<srhb>
furrycatherder: So in nix itself.
<furrycatherder>
srhb: how do i use it in a nix-shell then? i can't just do ./configure $configureFlags anymore, can i?
<srhb>
furrycatherder: Er um. :) I mean, from a nix shell you can't really register the store path regardless of whether you're using placeholder or not.
<srhb>
furrycatherder: What are you trying to acheve?
<srhb>
achieve*
<furrycatherder>
srhb: ah sorry, i think i'm having the x-y problem :p
<srhb>
furrycatherder: Might be, yes!
<furrycatherder>
srhb: i need to pass just --prefix and --with-sqlite, but not --disable-static
xkapastel has quit [Quit: Connection closed for inactivity]
<furrycatherder>
srhb: i'm trying to build the yocto project pseudo package with isn't *really* an autoconf package
<greymalkin>
Hmmm... how are substitutions in `installPhase` being handled now? I have a package that uses substituteAll on a *bunch* of shell scripts which is no longer working.
<furrycatherder>
srhb: do i just give up using mkDerivation and write my own derivation?
<srhb>
furrycatherder: I don't think so. It seems to me you should be fine at least if you write your own configurePhase?
<srhb>
furrycatherder: Sorry, I'm not sure how we got from A to B :-)
zupo has joined #nixos
johanot has quit [Quit: WeeChat 2.2]
<furrycatherder>
srhb: i want to write configureFlags = [ "--prefix=foo" "--with-sqlite=${sqlite.dev}" ]; and not have the extra --disable-static, which their configure script complains about
<furrycatherder>
srhb: doing the "--prefix=foo" part is how i started messing with placeholder
<furrycatherder>
srhb: but you're right, i should just write the whole configurePhase manually
<furrycatherder>
srhb: thus, x-y problem
<furrycatherder>
srhb: :P
<srhb>
furrycatherder: There might even be an attr that removes --disable-static
<srhb>
like, enableStatic or something like that..
<symphorien>
dontDisableStatic iirc
<srhb>
dontDontEnableStatic
<srhb>
I hate the dos and donts. x-P
<furrycatherder>
lol
<srhb>
furrycatherder: symphorien is right, it's in pkgs/stdenv/generic/setup.sh
<furrycatherder>
srhb: awesome thanks
shibboleth has joined #nixos
<furrycatherder>
yay!
rcshm has joined #nixos
<furrycatherder>
and now i get ./makewrappers: /usr/bin/env: bad interpreter: No such file or directory e_e
<symphorien>
patchShebangs .
<symphorien>
this will... patch the shebangs of scripts in .
<srhb>
See, this is a really good name. :-P
<symphorien>
:)
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
rcshm_ has joined #nixos
drakonis has joined #nixos
eadwu has quit [Quit: WeeChat 2.3]
schneid3306 has joined #nixos
rydnr has quit [Ping timeout: 256 seconds]
trevorriles has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
zupo has joined #nixos
eadwu has joined #nixos
gagbo has joined #nixos
benwaffle has joined #nixos
dbmikus_ has joined #nixos
<furrycatherder>
yay!!
<benwaffle>
When do packages from the github repo show up in the package repo?
<furrycatherder>
benwaffle: hydra builds the channels if that's what you're asking
<benwaffle>
specifically, gitAndTools.hub 2.9.0 is in the nixpkgs repo, but when I do nix-env -u I don't get updates for it
<gchristensen>
sorry about nixpkgs-unstable, we had trouble with darwin for a bit there -- but I thin kthat has been resolved
pie__ has quit [Ping timeout: 240 seconds]
trevorriles has joined #nixos
jasongrossman has joined #nixos
<catern>
how can it be that there are no examples of anyone configuring Hydra for email notifications on the internet?
<catern>
am I missing something? does Hydra actually have email notifications turned on by default?
<manveru>
why did zoom-us break? i get `This application failed to start because it could not find or load the Qt platform plugin "xcb"` :|
<sphalerite>
manveru: I think there was an issue about it
<manveru>
trying to build master to verify...
<manveru>
but the issue seems to be about something else
<benwaffle[m]>
how do I add a `buildInput` only for osx?
fresheyeball has joined #nixos
<gchristensen>
buildInputs = [ your normal ones ] ++ (pkgs.stdenv.lib.optionals pkgs.stdenv.isDarwin [ these are only on darwin ])
samrose_ has joined #nixos
samrose has quit [Ping timeout: 245 seconds]
<benwaffle[m]>
why is there a nixpkgs.libiconv (libiconv-osx) and a nixpkgs.libiconvReal (libiconv)
daniele- has joined #nixos
<benwaffle[m]>
when i build the package 'luit' I get missing iconv.h, but libiconv is a dependency of luit
<clever>
benwaffle: i think one of them depends on glibc
benwaffle has quit [Quit: Page closed]
<benwaffle[m]>
clever: yep, i see it
vidbina has quit [Ping timeout: 255 seconds]
<eyjhb>
I am trying to setup a `sdb` external luks encrypted SSD, but when I use the configuration here - https://termbin.com/mn4n (the commented part with data is normally uncommented), then at boot I am dropped to a rescue shell.. Any ideas?
justanotheruser has joined #nixos
ddellacosta has quit [Ping timeout: 246 seconds]
<catern>
at last, finally, I have made hydra send an email
daniele- has quit [Quit: daniele-]
reallymemorable has joined #nixos
<srhb>
catern: So what was it? :)
<reallymemorable>
what notepad-style software do people prefer on nixos?
daniele- has joined #nixos
<srhb>
reallymemorable: I'd guess people tend to use whatever their normal editor is..
<catern>
srhb: just need a line like "email_notification = 1", no <Plugin></Plugin>
<srhb>
catern: :| Sorry for leading you astray then.
<reallymemorable>
gchristensen: thanks!
<catern>
that's fine, it looks like that syntax is at least parsed by the Perl config stuff, so I may need it for other plugins
<reallymemorable>
that looks good
<catern>
srhb: and you did clue me in on HYDRA_CONFIG being used at all, so still, thanks
<{^_^}>
[nixos-hardware] @danbst opened pull request #101 → thinkpads: refactorings + modularization + some new options → https://git.io/fhd2H
<catern>
I suspect that all plugins are always loaded, perhaps
Ralith_ has joined #nixos
Ariakenom has quit [Ping timeout: 268 seconds]
<srhb>
catern: Yeah, iirc they are. And some of them are quite slow to load (and they're loaded on every hydra-notify invocation)
<srhb>
catern: especially the s3 ones are slow iirc.
<srhb>
(It becomes a problem if you're doing a lot of builds, since each notification can take upwards of a second)
<{^_^}>
[nixpkgs] @zimbatm merged pull request #56033 → buildBazelPackage: autodetect nix toolchain instead of Xcode one on macOS → https://git.io/fhdGh
<{^_^}>
[nixpkgs] @zimbatm pushed commit from @uri-canva to master « buildBazelPackage: autodetect nix toolchain instead of Xcode one on macOS (#56033) »: https://git.io/fhd2p
<reallymemorable>
but i dont understand what is the standard postgres package
<srhb>
reallymemorable: Why not services.postgresql.* ?
<reallymemorable>
i can add that as a package?
<reallymemorable>
in my environment.systemPackages?
<srhb>
reallymemorable: What are you trying to do exactly? Do you want to run a postgres db or do you just want the psql client?
ixxie has joined #nixos
<reallymemorable>
i need psql client
<srhb>
OK, then you're doing the right thing.
<reallymemorable>
`services.postgresql.*` so add that to my system packages
<srhb>
reallymemorable: No no.
<srhb>
reallymemorable: Just add pkgs.postgresql :)
<catern>
(ugh, the Perl part of Hydra supports using a SQLite database but the C++ part requires pgsql... frustrating being pgsql is so slow to start from scratch)
<srhb>
reallymemorable: Which is 9.6.11 on NixOS 18.09 currently.
<reallymemorable>
got it thank you
SyrupThinker has joined #nixos
sondr3 has joined #nixos
<reallymemorable>
and the same for SQLalchemy
<reallymemorable>
?
<reallymemorable>
I can just put that in systemPackages
<ivegotasthma>
question, I've been trying to update my version of youtube-dl for a long time, but it's still stuck at the old version
<ivegotasthma>
I ran `nix-channel --update && nix-env --upgrade` as root
<ivegotasthma>
I then ran `nixos-rebuild build && nixos-rebuild switch`
<ivegotasthma>
in my local user I did `nix-channel --update && nix-env --upgrade`
<ivegotasthma>
the version is still old, any idea what I might be missing?
<srhb>
reallymemorable: So in general, all "packages" are trees of directories like /bin /share /lib /...
justanotheruser has quit [Ping timeout: 268 seconds]
<srhb>
reallymemorable: Putting a package such at this in systemPackages simply symlinks it in place under /run/current-system/sw
<reallymemorable>
ah
<reallymemorable>
ok
<srhb>
reallymemorable: That is, once you added postgresql, /bin/psql was linked in place to /run/current-system/sw/bin/psql
<srhb>
reallymemorable: This goes for all packages.
<srhb>
ivegotasthma: What version did you expect, what version did you get, and what does nix-info say?
<sondr3>
ivegotasthma: you need to check the version in the channel you're installing from, not the nixpkgs repo on Github
<ivegotasthma>
so home-manager is out of date?
leotaku has joined #nixos
<samueldr>
catern: if you end up using it, please report issues or annoyances :)
skaapgif has quit [Ping timeout: 256 seconds]
<samueldr>
and do note that it offers only the poorest kind of isolation, but shouldn't dirty any "global" system state; once the hydra-in-a-bag folder deleted, nothing should be left (except for gc-able things)
<sondr3>
ivegotasthma: either that or the way that home-manager installs packages, installing it on my machine yields installing 'youtube-dl-2019.01.30.1'
<sondr3>
via nix-env -iA nixos.youtube-dl
<catern>
clever: samueldr: as it happens I'm making essentially the same thing, just in Python and with, well, a fair bit more sophistication I think
<{^_^}>
[nixpkgs] @vcunat pushed to staging-next « gcl*: fix with updated binutils »: https://git.io/fhdaP
<srhb>
ivegotasthma: I guess the first question really is how you installed it. It's possible you used home-manager, but you haven't used home-manager switch since updating your channels.
<catern>
samueldr: I don't see any tricks to get around Hydra's need for being a trusted-user, have you thought about that?
<ivegotasthma>
srhb: it's installed as a home-manager package
<catern>
that need is pretty annoying I feel...
<srhb>
ivegotasthma: Assuming your home.nix is using your regular channels, you should be able to just home-manager switch then
<samueldr>
catern: it's mostly geared towards hacking on the web parts, and I'm not a hydra buff; but since it runs under your own user, I guess if your own user is trusted it would be trusted
<catern>
yeah, but my own user isn't trusted :)
<samueldr>
trust issues? :)
<catern>
I don't trust myself!
<samueldr>
yeah, then uh *scrambles away*
<catern>
haha
<{^_^}>
[nixpkgs] @vcunat pushed to staging-next « filebench: fixup build by using older bison »: https://git.io/fhda9
<catern>
I'm thinking of doing some kinda hack with a nix-daemon running in a container for the use of Hydra
<srhb>
What is it that the queue-runner does that needs trust anyway?
<samueldr>
I guess it's like written on the tin: a convenience thing, the minimum viable way to run a hydra instance, but definitely not in any way like you would in prod :)
<clever>
catern: if you set NIX_REMOTE, it wont even need a daemon
<clever>
that would dump everything into /tmp/deletethis/nix/store/
<catern>
srhb: buildDerivation in store-api.hh
<clever>
and chroot automatically
<catern>
srhb: which sends over a derivation as text, and builds it, without putting it into the store
<clever>
(technically, it doesnt chroot, it just prefixes all paths automatically, it has some bugs too)
<srhb>
catern: Ah.
<catern>
not sure why that needs trust though ¯\_(ツ)_/¯
<ivegotasthma>
srhb: amazing, thanks!
<catern>
clever: hmm, interesting, I'll think about that... you say it doesn't chroot at all? the main thing I was concerned about with my container plan was, how to get the hydra binaries into the container while still being able to write to /nix/store in the container. I guess local?root=/tmp/foo is... just a different store prefix? that makes a lot of sense...
<clever>
catern: i think the sandboxing mechanics are the only time it does chroot like stuff
rcshm has quit [Remote host closed the connection]
<catern>
yeah, it makes sense. it's obvious in retrospect: I don't need a container or anything, I can just use a different store prefix
<clever>
ive also used local?root= with nix copy, to unpack a tarball of a store, then merge it into an existing store, that may not be at /
<{^_^}>
[nixpkgs] @vcunat pushed to staging-next « zeroc_ice: disable problematic parallel make »: https://git.io/fhdVJ
furrycatherder has quit [Quit: Lost terminal]
ma9e has quit [Quit: leaving]
<{^_^}>
[nixpkgs] @vcunat pushed to staging-next « stp: fixup build by using older bison »: https://git.io/fhdVY
xvnvx- has quit [Ping timeout: 258 seconds]
<{^_^}>
[nixpkgs] @costrouc opened pull request #56057 → pythonPackages.sparse: init at 0.6.0 → https://git.io/fhdVB
morr_ has quit [Ping timeout: 250 seconds]
wfranzini has quit [Remote host closed the connection]
<samueldr>
at least they didn't use my name (SEO is safe!) :)
ost has quit [Ping timeout: 240 seconds]
<samueldr>
my full name is, allegedly, unique, so it's annoying to see my zero work for being the first result for a search being annihilated by one weird bug :)
rgrau` has joined #nixos
<LnL>
hah
rgrau` has quit [Remote host closed the connection]
<samueldr>
(I say allegedly since it's unproven, but it's using a pair of names that's relatively uncommonly paired, and relatively local to the country and province, and confirmed to be unique in the province a few years back)
dbmikus_ has quit [Ping timeout: 252 seconds]
slack1256 has joined #nixos
otulp has joined #nixos
superherointj has joined #nixos
superherointj has quit [Client Quit]
<tokudan[m]>
samueldr: google claims 655 million results for my name, so i'm on the other end of the spectrum ;)
<clever>
tokudan[m]: my name pulls up 190 million hits...
<ottidmes>
mine 1.25 million, when I wanted to register my email adress it was already taken, so I added that email address to my chat program at the time, and had a nice chat :P he lives relatively close by too, but unrelated
kreisys has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
illegalprime has quit [Ping timeout: 255 seconds]
eadwu has joined #nixos
o1lo01ol1o has joined #nixos
tbenst[m] has joined #nixos
ubert has quit [Quit: Leaving]
Ariakenom has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
Jetien_ has quit [Ping timeout: 245 seconds]
Ariakenom has quit [Read error: Connection reset by peer]
<pie__>
im pretty sure googles result counter is bullshit :D
<tokudan[m]>
pie__: yes, it is. that should be common knowledge and why i wrote the word "claims"
neminis has joined #nixos
kisik21 has quit [Ping timeout: 252 seconds]
kisik21 has joined #nixos
<tbenst[m]>
Hi all! Curious if anyone has had luck installing printer drivers? For my printer, there are .deb and .rpm available, but unfortunately generic drivers are not working & my model is not available on gutenprint. Xerox helpfully provides files for a Unix install, but not sure how to install on nixos. Here's the directory tree & postinstall script: https://gist.github.com/tbenst/bc1155bfdc97f495452ca13c2f364b50
orivej has joined #nixos
rgrau has joined #nixos
<ottidmes>
gchristensen: I am getting a merge conflict label with my PR, it was outdated, but I redid the PR and force pushed a commit on top of the latest master. Could it be a delayed response to me reopening it when it was still outdated?
Ariakenom has joined #nixos
<{^_^}>
[nixpkgs] @dasJ opened pull request #56058 → exim: Fix build with LDAP → https://git.io/fhdrR
<das_j>
Is there a way to retrieve the nixpkgs version in an overlay?
phizzz has joined #nixos
<pie__>
tbenst[m], is there a ppd file somewhere?
<pie__>
ah
reinhardt has joined #nixos
* pie__
tries to look at the listing
<srhb>
das_j: It depends from whence your nixpkgs came. generally, lib.version tries its best.
<pie__>
oooook that looks like its not the trivial case, i dont know
<das_j>
srhb: So in an overlay it's self.lib.version?
<srhb>
das_j: or super.lib.version, yes.
<das_j>
Thanks :)
<srhb>
das_j: I would call this a little unreliable, depending on what you need it for.
<srhb>
(And how you get nixpkgs)
<das_j>
srhb: I need an overlayed package for 18.09, but not for 19.03. So I need some condition
hedning has quit [Remote host closed the connection]
zupo has joined #nixos
hedning has joined #nixos
zupo has quit [Client Quit]
<phizzz>
hi there, i'm trying to setup up kubernetes on nixos and the kubelet service is not starting https://pastebin.com/xmQt5Uam
<tbenst[m]>
pie__: yeah I'm sorta at a loss for even the general strategy here, I've been spoiled by installers for drivers on ubuntu. One approch would be to try and install the .deb (https://reflexivereflection.com/posts/2015-02-28-deb-installation-nixos.html) but idk if that is viable for a driver anyways
<reallymemorable>
I am trying to build a branch on my client and I got an error: `trace: Please set your nix-path such that ssh-config-file points to a file that will allow ssh to access private repositories. The builder will not be able to see any running ssh agent sessions unless ssh-auth-sick is also set in the nix-path.`
hedning has quit [Remote host closed the connection]
<pie__>
tbenst[m], i dont know mut maybe fhsuserenv
<pie__>
*but
<reallymemorable>
Does this mean I have to modify ssh-config-file so that it points to my /.ssh folder?
<pie__>
or rather, maybe someone else that has a better clue can help
hedning has joined #nixos
<pie__>
tbenst[m], the problem isnt the deb or rpm format, its that that printer driver is full of .so files
trevorriles has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<phizzz>
some errors about deprecated options like --cluster-dns --cluster-domain --hairpin-mode. i don't even know how to edit the kubelet.service (which is in the store, of course...)
reinhardt has quit [Quit: Leaving]
<tbenst[m]>
pie__: gotcha, yeah fhsuserenv makes sense, I wonder if/how that integrates with the rest of the OS so I can use ctrl-p..?
<ottidmes>
gchristensen: never mind, it has since been removed, sorry for the noise
reinhardt has joined #nixos
<ninjin>
I seem to constantly fail to hit the binary caches and have to build Julia all the time which is a royal pain (Hello LLVM…), is there a way to debug as to why? I am puzzled as to why Julia is the only package on my system that has this behaviour.
<pie__>
tbenst[m], no clue :I thats what i was thinking about too
<pie__>
tbenst[m], linux printing usually goes through CUPS in some way
<pie__>
florianjacob, do you maybe have some idea about the above?
<srhb>
ninjin: The usual culprits are config and overlays. What channel are you on? We can try to figure out which dependency (if any) is different from upstream.
kreisys has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<phizzz>
if no one has ideas about kubernetes, then how about a more general question. like, how do i modify a systemd service script?
<ninjin>
srhb: `nixos https://nixos.org/channels/nixos-18.09` is the only one as far as `nix-channel --list` is concerned. Should I dump by config somewhere as well?
<ekleog>
Do we have any docs on how to run update scripts defined by passthru.updateScripts?
illegalprime has joined #nixos
<srhb>
ninjin: What does nix-info say? (it'll tell us the exact revision you're on currently)
<ninjin>
I do have `autoUpgrade.enable` on, maybe that is my sin? Wasn’t sure how to get security packages in a sensible way otherwise.
<ninjin>
s/packages/patches/
<das_j>
Is there a way to conditionally set variables? Not something like optionalString (because this results in "" if the condition is false), but something that won't run at all if the condition is false
<ekleog>
1
<srhb>
ninjin: can you try: nix-shell -p nix-diff --command "nix-diff $(nix-instantiate '<nixpkgs>' -A julia) $(nix-instantiate -E 'with import <nixpkgs> { config = {}; overlays = []; }; julia')"
<ekleog>
oops
<srhb>
ninjin: Oh, hm. Maybe it's just plain broken at that revision.
<srhb>
ninjin: Did your build ever succeed?
<ninjin>
srhb: I have managed to get `jul
<ninjin>
Meh.
<ninjin>
I have managed to get julia_10 to build cleanly.
<infinisil>
What's the safest/cleanest way to allow NixOS services to read the ACME generated certificates and keys in /var/lib/acme/${domain}/*.pem?
<ninjin>
But I am still puzzled why it builds at all.
<ninjin>
I have a full desktop system running and that single package can’t be fetch as a binary? Surely I have done something odd.
<das_j>
infinisil: I usually copy the pem files to a directory that is owned by root:$servicegroup and give them 640 permissions
<das_j>
This needs to be done in a acme hook of course
<ninjin>
srhb: What did we expect from the nix-diff command?
<pie__>
tbenst[m], https://nixos.wiki/wiki/Printing "Open the PPD as a text file, and check that it does not mention FHS paths like /usr/bin. If it does, this method is unlikely to work, as the PPD file depends on executables not present on your system."
ddellacosta has joined #nixos
<srhb>
ninjin: No output if you have no local modifications to it or its dependencies, but I think I know what the culprit is.
<ninjin>
From what I can tell it dumped nothing on stdout.
<infinisil>
das_j: Hmm I see, why not $serviceuser:$servicegroup?
<das_j>
infinisil: So the service cannot change the permissions of the file
<infinisil>
Oh smart
<das_j>
Shouldn't matter if you use serviceConfig.ReadOnlyPaths
<tbenst[m]>
pie__: thx for link, hopefully can get to work though.... Sure enough, '/nix/store/sajf50mga0gi9igdcv8inbldc0623kym-cups-progs/lib/cups/filter/' has no file XeroxQScrip. However, I do have that file in the .deb lib/cups/filter/XeroxQScript
<pie__>
tbenst[m], technically that could probably be patched to point to the correct paths but the binaries would still need to be made to work for nix
<pie__>
for nixos
<infinisil>
das_j: What directory do you use? /var/lib/$servicename?
<pie__>
see stuff about patchelf or something maybe
<das_j>
infinisil: /var/lib/secrets/$servicename. My deployment system also deploys other secrets (dhparam, ssh keys, ...) to that path. So I just need one directory that is owned by root:$service to manage all secrets (useful if the group changes)
Guanin has quit [Ping timeout: 259 seconds]
<das_j>
Also, some services might write to /var/lib/$service
<ninjin>
srhb: Got it, internal tests failing.
proml has joined #nixos
<infinisil>
das_j: But /var/lib/$service/secrets would also just be one directory
<infinisil>
das_j: Or do you mean that everything like that is in the single /var/lib/secrets directory?
<ninjin>
That is supremely bad, I may be able to help if I read up a bit on the NixOS part of things since I am very well familiar with the Julia codebase.
<srhb>
ninjin: Yes. I don't see any changes to it recently, so if yours builds correctly, it might be a transient test failure
<srhb>
ninjin: Which is very bad on Hydra.
<srhb>
ninjin: (exactly because you get to compile it again and again)
<das_j>
Yes, depends on what you prefer, I don't really see any reason why one way would be better. I just wanted one place to store all secrets, but splitting them might have some upsides I didn't notice yet
<infinisil>
das_j: Alright cool thanks
<srhb>
ninjin: If you need help understanding the Nix part of things, do poke.
<das_j>
It's just useful because for me, the deployment system puts everything there
<ninjin>
srhb: I have seen pretty much exactly the same error on 1.1, it appears to be some HTTP timeout.
<das_j>
you're welcome
dermetfan has quit [Ping timeout: 255 seconds]
equivrel has joined #nixos
proml has quit [Client Quit]
<srhb>
ninjin: uuuuh.
<ninjin>
srhb: Any recommendation where to start reading?
<srhb>
ninjin: The sandbox shouldn't have network access...
<srhb>
ninjin: Since that would be impure.
<srhb>
ninjin: Where do you see this?
<das_j>
srhb: Btw, is there a way to FORCE network access?
<srhb>
das_j: disabling sandbox. You should not.
zupo has joined #nixos
<das_j>
Damn it, okay
<ninjin>
srhb: Sure, but this is for testing. I assume that it may be testing some part of the HTTP library which would ned network access.
<srhb>
ninjin: Ah, as long as it's local that's fine.
rfold has joined #nixos
<srhb>
ninjin: Going out into the world is a nono...
zupo_ has joined #nixos
<srhb>
ninjin: I also see a random seed in there somewhere, that should be fixed in the nix build, hopefully.
<srhb>
I don't see it though, so that may also be an issue.
<equivrel>
Hi, does anyone know what I should do if I want to run two instances of a service on the same machine at the same time? I am trying to deploy some CI with nixops, and I want to run two instances of buildbot-worker.
tmaekawa has joined #nixos
<srhb>
equivrel: If it's not a nicely parameterized service, maybe you can do something with nixos-containers?
<infinisil>
das_j: You can do whatever you want with the network in a sandbox as long as it doesn't influence the output with a fixed-output derivation
zopsi has quit [Quit: Oops]
<equivrel>
srhb: how could I tell if it's nicely parametized? I presume it should be since the reason for buildbot's master-worker architecture is so that you can run multiple workers?
<infinisil>
Can download tons of stuff but just do `touch $out` in the end :P
<phizzz>
srhb: i get this error when i put that in configuration.nix: The option value `services.kubernetes.kubelet.extraOpts' in `/etc/nixos/configuration.nix' is not of type `string'
<das_j>
infinisil: I thought this might be an easy way to pass secrets. That's my main problem now
zupo has quit [Ping timeout: 245 seconds]
<srhb>
phizzz: Oh, sorry, apparently it should be just a single string, not a list of strings
<das_j>
I don't want them in the store
zopsi has joined #nixos
zopsi has quit [Excess Flood]
zopsi has joined #nixos
zopsi has quit [Excess Flood]
<srhb>
equivrel: Technically, that's a module. I can see it's not, because there's only the single systemd service, that is, buildbot-worker -- not something like "buildbot-worker-${someparam}"
<phizzz>
srhb: i'm not sure what you mean
<infinisil>
das_j: You could use impure env vars
<das_j>
infinisil: Is there documentation on them?
<infinisil>
Oh um, not sure if these don't end up in the store
<das_j>
> This attribute is only allowed in fixed-output derivations
<{^_^}>
error: syntax error, unexpected IN, expecting ')', at (string):218:32
<infinisil>
Oh yeah, darn
<equivrel>
srhb: do you think it is straightforward to just make the name of the system service depend on some option, to deduplicate them? Or does it make more sense to start looking into containers?
<phizzz>
thanks srhb that seems to work
zopsi has joined #nixos
zopsi has quit [Excess Flood]
<srhb>
equivrel: I don't know the service well enough to answer that.
matijja has quit [Ping timeout: 240 seconds]
<equivrel>
OK, thanks for the help though.
ninjin has quit [Ping timeout: 256 seconds]
zopsi has joined #nixos
zopsi has quit [Excess Flood]
<infinisil>
das_j: Oh, how about sandbox-paths, in `man nix.conf`
<srhb>
equivrel: (Questions that influence that answer are things like: Will they need different config files, params, working dirs, ...)
abrar has quit [Quit: WeeChat 2.3]
<sondr3>
quick question, I'm trying to access my NixOS machine on my local network when it's serving a folder with python http.server, but I cannot access it from my laptop
zopsi has joined #nixos
zopsi has quit [Excess Flood]
abrar has joined #nixos
<gchristensen>
srhb: sorry to say I had a failure for autorandr to do the right thing, a scenario where xrandr doesn't think my monitor can do the resolution it actually can.
<{^_^}>
[systemd] @andir opened pull request #26 → v239-security: fix of CVE-2019-6454 → https://git.io/fhdoL
<das_j>
Oh, that might be what I need
<sondr3>
running the same command on my laptop though I can access my NixOS machine, so there's something I need to configure I think on it
<das_j>
Well, it would be for *all* builds but better than nothing
zopsi has joined #nixos
zopsi has quit [Excess Flood]
abrar has quit [Client Quit]
<srhb>
gchristensen: Grah! I still think it's better than what I have though.
<das_j>
Silly question, but does Hydra support multiple nix daemons?
ninjin has joined #nixos
<gchristensen>
also a funny thing is redshift had made my screen quite red when I captured the setups, and now changing profiles causes it to turn red before redshift says "hey, what're you doing?"
zopsi has joined #nixos
zopsi has quit [Excess Flood]
<ninjin>
srhb: Sorry, I think I hit the swap a bit too hard on this old desktop and rebooted. Checking out the Julia build script now to see what is up in terms of the seed etc.
<sondr3>
oh, I forgot that the firewall is on by default, opening the port worked fine
<srhb>
ninjin: Woops :) Welcome back.
<srhb>
ninjin: If you figure out how to set the seed, we can try to reproduce the failure on Hydra with the seed it emitted. I guess that would imply we found a Julia bug (?) but we really should not be using random seeds.
<ninjin>
srhb: My guess is that some test is unstable, it has happened from time to time.
philippD has joined #nixos
<das_j>
What happens if I set the speed factor of a machine to 0?
zopsi has joined #nixos
zopsi has quit [Excess Flood]
<das_j>
Will Hydra just not schedule anything, then?
<srhb>
ninjin: Sounds likely.
zopsi has joined #nixos
zopsi has quit [Excess Flood]
koray has joined #nixos
zopsi has joined #nixos
zopsi has quit [Excess Flood]
<das_j>
save
<das_j>
whoops
hiroshi has quit [Ping timeout: 246 seconds]
koray has quit [Client Quit]
<bsima>
how do I use fetchFromGitHub to download a pinned nixpkgs?
trevorriles has joined #nixos
<bsima>
anyone have an example offhand i could see?
alex`` has quit [Quit: WeeChat 2.3]
<clever>
bsima: fetchTarball is better for fetching nixpkgs itself, since you need a nixpkgs to use fetchFromGitHub
<bsima>
ok
<bsima>
i was using (import <nixpkgs>{}).fetchFromGitHub
<clever>
i prefer fetchTarball, because i can then unset NIX_PATH entirely
<clever>
and then any <nixpkgs> mistakes will fail, rather then cause problems
rydnr has joined #nixos
rauno has joined #nixos
zopsi has joined #nixos
zopsi has quit [Excess Flood]
slack1256 has quit [Quit: sleep]
drakonis has quit [Quit: WeeChat 2.3]
itscaleb has left #nixos ["ERC (IRC client for Emacs 26.1)"]
<{^_^}>
[nixpkgs] @elseym opened pull request #56061 → radarr module: add more options and refactor → https://git.io/fhdoX
<ninjin>
srhb: Yep, and we are not using it as far as I can see.
rauno has quit [Ping timeout: 250 seconds]
kreisys has joined #nixos
<srhb>
ninjin: Indeed.
<ninjin>
Perhaps a sensible thing to do would be to set it to hash of the release?
<srhb>
ninjin: So maybe something like preCheck = "sed magic to insert --seed=someseed" there?
<srhb>
ninjin: That or just a static seed..
eadwu has joined #nixos
<ninjin>
srhb: And trust Julia to catch issues upstream?
<ninjin>
srhb: Let’s see if I can code this and make it my first patch.
<srhb>
ninjan: Regardless of which seed we choose, when it succeeds, we'll only be running it once anyway (afterwards it will be cached) for any given dependency set and version..
<srhb>
ninjin: I'm not sure there's a huge value in ensuring that the seed varies in this case.
<srhb>
(between versions/dependency sets, I mean)
<srhb>
ninjin++
<{^_^}>
ninjin's karma got increased to 1
Ariakenom has joined #nixos
<ninjin>
srhb: Sounds far to me, I have not done software packaging/porting since I was on FreeBSD about ten years ago.
<ninjin>
So I am a bit outdated in terms of sensible standards.
<ninjin>
I don’t even think we did integration.
<srhb>
ninjin: At any rate, we're possibly a bit "weird" in wanting deterministic builds first and foremost.
<srhb>
(meaning, definitely no random seed influencing success/failure)
johanot has joined #nixos
<ninjin>
srhb: I can see the arguments for it, over in Julia upstream we probably want to explore as large of a surface as possible at all times.
eadwu has quit [Client Quit]
<ninjin>
srhb: Is there a handy way to check the build once I modify the `default.nix` expression? Sorry if this is in the manual.
ThatDocsLady_ has quit [Ping timeout: 259 seconds]
<srhb>
ninjin: `nix-build path-to-your-checkout -A julia` to a first approximation
<ninjin>
And this is why I bloody love NixOS. ♡
<srhb>
ninjin: Don't you know it. :-)
Soo_Slow has quit [Remote host closed the connection]
<srhb>
Are you trying to reproduce it with the seed from the failure first?
<ninjin>
srhb: Yeah, will take a while though so I don’t expect a patch for 24 hours… LLVM is a beast. Perhaps if I check things out on the Julia build cluster.
slidercrank has joined #nixos
<srhb>
ninjin: :3 Sounds great. Feel free to ping me if you put up a PR, since I have some context now.
<ninjin>
srhb: Roger!
eadwu has joined #nixos
ddellacosta has quit [Quit: WeeChat 2.2]
ddellacosta has joined #nixos
<{^_^}>
[nixpkgs] @elseym opened pull request #56062 → jackett module: add more options and refactor → https://git.io/fhdKL
ost has joined #nixos
<fpletz>
johanot: either you rebase or I'll do it before merging, as you prefer :)
superherointj has joined #nixos
matijja has joined #nixos
<fpletz>
johanot: do you have time in the next few days to look at calbrechts changes? he also fixed lots of service dependencies so might see less errors in the logs :)
<{^_^}>
[nixpkgs] @worldofpeace pushed 3 commits to master: https://git.io/fhdKW
zopsi has joined #nixos
zopsi has quit [Excess Flood]
agander has quit [Ping timeout: 258 seconds]
zopsi has joined #nixos
zopsi has quit [Excess Flood]
zopsi has joined #nixos
zopsi has quit [Excess Flood]
ddellacosta has quit [Client Quit]
ddellacosta has joined #nixos
ninjin has quit [Remote host closed the connection]
ddellacosta has quit [Client Quit]
ddellacosta has joined #nixos
ninjin has joined #nixos
<reallymemorable>
I created a user in configuration.nix and did not specify a password. When I tried to `sudo chown -R ...` a folder, it prompts me for a password. When I just try to hit enter, it says wrong password.
khh7778 has joined #nixos
<reallymemorable>
Am I required to define a password?
<khh7778>
hello
<manveru>
reallymemorable: sudo needs the password of root
<khh7778>
im interesting in install nixos or guixsd
reinhardt has quit [Quit: Leaving]
<reallymemorable>
its nott accepting root's password
<srhb>
manveru: password of the user* with rights to sudo
illegalprime has quit [Ping timeout: 255 seconds]
<{^_^}>
[nixpkgs] @romildo opened pull request #56063 → pygmentex: add tlType attribute; install latex style and demo → https://git.io/fhdKR
<manveru>
hmm, true :)
<manveru>
but if the user has no password?
<srhb>
Then you can't sudo, by default.
<manveru>
can't say i ever tried that
<alex_giusi_tiri>
khh7778: hello
<JosW>
reallymemorable: you need to add a password to the user, passwd "youruser"
<alex_giusi_tiri>
reallymemorable: you would need to add this: "security.pam.services.sudo.allowNullPassword = true;"
<khh7778>
i want to add this channel to my irc android client, howto?
<srhb>
khh7778: Set up access to the freenode network and join the #nixos channel. How that's done depends on the client in question.
<alex_giusi_tiri>
if you want to allow to sudo if the user has no password
<reallymemorable>
I created SSH keys for my root user in NixOS and added them to GitLab and GitHub. I was able to clone the repo. But I needed to create another user on my NixOS machine. I copied the same keys over to that user’s directory. But when I try to clone the same repo with the same keys from that user, it says permission denied. Does anyone know what this is happening?
sondr3 has quit [Ping timeout: 244 seconds]
<alex_giusi_tiri>
make sure that the key files are now owned by that user
<reallymemorable>
with chown?
<alex_giusi_tiri>
yes
<alex_giusi_tiri>
that's something I sometimes forget to do myself
<reallymemorable>
I need to run tthat from root
<reallymemorable>
or from within that user
<alex_giusi_tiri>
as root
<manveru>
ssh-copy-id ftw :)
ixxie has quit [Ping timeout: 245 seconds]
zupo has joined #nixos
matijja has quit [Ping timeout: 244 seconds]
zupo_ has quit [Ping timeout: 272 seconds]
<alex_giusi_tiri>
so, I've been trying to compile ceph for some time now, but I can't seem to succeed
<manveru>
garbas: are you around?
<srhb>
alex_giusi_tiri: lejonet might be able to help.
agander has joined #nixos
<alex_giusi_tiri>
thank you
<srhb>
alex_giusi_tiri: What's the problem? I'm probably a few versions out of date, but I remember the cmake build being pain and horror.
ATuin has joined #nixos
<alex_giusi_tiri>
should I chat with him directly?
<alex_giusi_tiri>
well, depending on the version, the kind of error changes
<alex_giusi_tiri>
I should probably show what I've been working with...
* lejonet
hides
<srhb>
alex_giusi_tiri: I the name to summon. Might have better luck during earlier (european) hours :)
<srhb>
aha!
<alex_giusi_tiri>
oh
anderslundstedt has joined #nixos
<lejonet>
Ceph is ceph :P
khh7778 has quit [Ping timeout: 256 seconds]
<lejonet>
Which version you battling with?
<alex_giusi_tiri>
I was recommended 13.2.4
daniele- has quit [Quit: daniele-]
<alex_giusi_tiri>
what version would you have?
<srhb>
Speaking of which, we should deprecate ceph-dev..
<lejonet>
12.2.2 I think, let me double check :P
<srhb>
And libceph..
<lejonet>
ah, 12.2.0 even
<alex_giusi_tiri>
ok
<lejonet>
13.2.0 seems to be latest stable, 12 series is LTS
<alex_giusi_tiri>
I made some local modifications to nixpkgs to other packages too, along the way
<kreisys>
Does anyone know which nixpkg on darwin provides the command "ioreg" (if any?)
<lejonet>
srhb: sounds like a sane way to do it, that way I can fetch just that branch if memory serves me right
<srhb>
lejonet: You can, among other things, yeah :)
pie__ has quit [Ping timeout: 246 seconds]
<clever>
argent0: its added to an env variable that cc-wrapper will force gcc to look into
<clever>
env | grep zlib --color
<clever>
argent0: its in both NIX_CFLAGS_COMPILE and NIX_LDFLAGS
<argent0>
clever: thanks
<lejonet>
srhb: its nix-build /path/to/pkg.nix to do a local build only, with a symlink to the results in the current dir?
<clever>
lejonet: -o can also be used to put the result elsewhere
<srhb>
lejonet nix-build path-to-checkout -A ceph
<lejonet>
clever: ah, figures :)
khh7778 has joined #nixos
<lejonet>
sweet, that way I can test the compile locally before I try and deploy it :)
<srhb>
lejonet: I think there's a few rebase issues, not sure what yet exactly..
<srhb>
Probably minor.
<lejonet>
srhb: well the conflict seems to be with a file that has been removed in krav1's branch (generic.nix doesn't exist in his pkgs/tools/filesystems/ceph dir)
<khh7778>
Does GuixSD have built-in KDE support?
<srhb>
lejonet: Yeah I got rid of that, but we also got rid of python2Packages.argparse in the meantime. I think it needs to just go, should be included anyway.
<srhb>
khh7778: try the #guix channel? :) They probably know better than us.
_kwstas has joined #nixos
_kwstas has quit [Remote host closed the connection]
<lejonet>
srhb: I really don't see why the package was divided up into default.nix / generic.nix in the first place, was it because there was a plan on having more than one? (maybe thats where ceph-dev came into play)
dermetfan has joined #nixos
<srhb>
lejonet: Back when I started working on it I think we have several versions and also the boost split was weirder still..
<lejonet>
srhb: makes sense then
Glider_IRC_ has joined #nixos
<srhb>
lejonet: Around the same time they overhauled the build system (I think that's when they started using cmake?) and generic got really useless...
<srhb>
Or at least, that's how I vaguely imagine it was. It was a while ago.
<lejonet>
srhb: yeah, the overhaul wasn't THAT long ago, but still like around 10 I think?
<srhb>
I can barely remember what happened a week ago. :-P
<lejonet>
xD glad I'm not alone there :P
<lejonet>
I do not look forward to compile this on my odroid-hc2's...
<betawaffle>
where in the manual does it tell me what attrs are available to me in a module's arguments?
<betawaffle>
(ie. config, lib, pkgs, etc)
<clever>
betawaffle: if you remove the ... element, then you will stop accepting args you didnt name yourself
<ottidmes>
I know of hnix, but do people know of other formatters? considering the bullet "Formatting Nix code", I was hoping one already exists based on rnix, but I cannot find any
<clever>
betawaffle: then it will error out, and reveal the names of things you didnt claim to accept
<betawaffle>
clever: i get that, i mean what other stuff is passed?
_kwstas has joined #nixos
<betawaffle>
for example, my configuration.nix has `{ config, pkgs, ... }:`
<betawaffle>
but i know `lib` is another one i could specify if i wanted
<clever>
betawaffle: nixops will add resources, nodes, and deploymentName
<ottidmes>
betawaffle: you can easily test this by adding a catch all { ... }@args: and then do attrNames on args
Glider_IRC__ has quit [Ping timeout: 246 seconds]
<betawaffle>
basically what i'm asking is where are those arguments defined/passed to my module
<srhb>
bsima: You get all the build information as a json file.
<srhb>
So your "script" can do all the logic you care about.
<bsima>
ok, so i could get the path from that json and deploy it with 'nix-copy-closure'
<lejonet>
alex_giusi_tiri: 13.2.2 compiled just fine with that PR for me, so 13.2.4 should work just as well (I'm going to try with 13.2.4 after I've tried some of the binaries)
<srhb>
lejonet: How big is your machine x_x
<bsima>
thanks srhb
<gchristensen>
srhb: I don't suppose you would want to make a run command plug-in to dump jobs to rabbitmq, would you? :)
<lejonet>
srhb: I've only got a 1700X :P
<srhb>
gchristensen: Um. Probably.
<srhb>
Why rabbitmq? ofborg uses that?
<srhb>
lejonet: Mine says 76% :-P I jealous.
<gchristensen>
*shrug* it is generally what I use as a bus
<lejonet>
srhb: xD I just put a 16c/32t epyc in my rack, could try with that and see how long it takes on it ^^
<srhb>
gchristensen: I don't have any opinion on that, was just curious :)
<srhb>
lejonet: Nice!
<alex_giusi_tiri>
ok, that's good news; thank you guys :-)
<lejonet>
and at least the ceph binary could fetch info from my 12.2.0 cluster without any problem
<srhb>
lejonet: Didn't you write, like, a NixOS test? :-P
<lejonet>
srhb: yes I did, but always nice to test on a "real" cluster too ^^
<srhb>
Yarr.
<srhb>
lejonet: I'm heading off for tonight, and I just got told that krav is on vacation, but do you want to team up on ensuring this gets in before the 19.03 freeze? :)
<lejonet>
its really nice that krav took the time to figure out ceph-volume too
<srhb>
Definitely!
<lejonet>
srhb: Sure!
<srhb>
Great :)
<lejonet>
I'm gonna trigger the test on 13.2.2, then compile 13.2.4 and do the test on it too, it "should" just be a version bump and current build process Just Works(TM)
<srhb>
Awesome. :)
<lejonet>
because then we could just ask krav to bump the version to 13.2.4 so when it gets commited, we have latest stable :)
<alex_giusi_tiri>
:-)
drakonis has joined #nixos
<manveru>
FRidh[m]: you around?
<srhb>
lejonet: Yep, makes perfect sense! Seeya. :)
<lejonet>
srhb: have fun!
* lejonet
is now at loss how to trigger the test in nixos/tests and make it use the checkout ceph pkg instead of whats in NIX_PATH
<callahad>
Can someone explain *why* nixos-unstable is 8 days old on https://howoldis.herokuapp.com/? It looks like the Hydra nixos/trunk-combined/tested job succeeded, so are we just waiting for everything to build?
<symphorien>
reallymemorable: are you using bash or another shell ?
<reallymemorable>
bash
<__monty__>
reallymemorable: The other commands work?
<Denommus`>
symphorien: ... I really need to stop using standalone nix and moving to NixOS
Denommus has quit [Ping timeout: 252 seconds]
<symphorien>
Denommus`: ah then just create a gc-root for your package anywhere
hedning has quit [Quit: hedning]
iqubic has joined #nixos
<symphorien>
nix-build "<nixpkgs>" -A sl -o ~/gc-root-for-sl will prevent sl from being removed as long as the symlink exists
<symphorien>
sh /Users/reallymemorable/.nix-profile/etc/profile.d/nix.sh
<makefu>
iqubic: buy a bigger ssd? i mean they got super cheap these days
vk3wtf has joined #nixos
<symphorien>
which is not the same :)
<Denommus`>
symphorien: how do I create a gc-root? Just symlink the package anywhere?
<iqubic>
My Nixos partition is formatted with ZFS. It's gotten too full, even with compression.
<symphorien>
Denommus`: nix-build has a -o option for example
<symphorien>
for the details I don't know
<reallymemorable>
wow
<reallymemorable>
thanks
<reallymemorable>
i am an idiot
<reallymemorable>
i thought i had copy-pasted that...i guess not
<iqubic>
At this point whenever I boot into Nixos, several things fail, and I'm dropped into tty1. I can no longer access the graphical login manager on tty7.
<clever>
iqubic: nix-collect-garbage
<ivegotasthma>
-d
<iqubic>
I would do that, but I can't log it.
<iqubic>
*in
<clever>
iqubic: text mode login fails?
<iqubic>
That is correct.
<clever>
iqubic: try adding ` single` to the end of the kernel command line in grub
<manveru>
does anyone know the proper way of using toPythonApplication for a python3 application that depends on a python2 application? seems like it can't build because of conflicting PYTHONPATH
<iqubic>
clever: I'm going to wait for my phone to have enough charge so that I can use it to access IRC. I don't want to be flying solo here.
jpo-joyent[m] has quit [Quit: issued !quit command]
aanderse has joined #nixos
aanderse_ has quit [Ping timeout: 250 seconds]
<{^_^}>
[nixpkgs] @fpletz merged pull request #55872 → kdevelop: added ability to include kdevelop plugins, kdev-php: init at 5.3.1, kdev-python: init at 5.3.1 → https://git.io/fh5Oz
<iqubic>
At this point I'm seriously considering nuking windows from my hard drive.
<ottidmes>
iqubic: I am curous how much space did you give NixOS with ZFS?
<iqubic>
98.12GB
knupfer has quit [Ping timeout: 250 seconds]
wfranzini has quit [Remote host closed the connection]
<ottidmes>
iqubic: potential downsides of nuking Windows are the obvious ones, you dual booted for a reason I presume, so if you dont have another drive to install Windows on, you might have some trouble getting those things you used on Windows to work properly on NixOS (i.e. time consuming, success not guaranteed)
wfranzini has joined #nixos
<iqubic>
The main reason I dual boot is because I use windows for printing.
<ottidmes>
iqubic: that is twice what I have for my NixOS on btrfs
<iqubic>
I just can't get printing to work on Nixos.
<ottidmes>
me neither, well I got it to work, but the results were far from ideal
<iqubic>
ottidmes: Yeah, and I think I know what is taking up all the space.
<iqubic>
I really really really need to prune my IRC logs every now and then.
dermetfan has quit [Ping timeout: 272 seconds]
<symphorien>
iqubic: (shameless plug) if you store is too big, you can use nix-du to know what to remove
<ottidmes>
iqubic: also for general disk space usage, I use ncdu from time to time to cleanup my drive
<clever>
iqubic: ncdu can also be used
<das_j>
What is the nix 2.0 alternative to nix-instantiate?
<clever>
i recently found that $HOME on a server i help manage, was twice the size of /nix/store/
<ottidmes>
das_j: nix eval
<iqubic>
I have been logging 4 IRC channels basically 24/7 for the past few months and storing the logs in plain text, in 4 gigantic files. And I have never looked at these logs ever.
<iqubic>
I should just delete these logs, and turn off logging.
<betaboon>
sphalerite: you around ?
<clever>
iqubic: several years worth of irc logs are only taking up 5gig here
jpo-joyent[m] has joined #nixos
<sphalerite>
betaboon: yep, not for much longer though
<iqubic>
Oh, well... IDK what the issue is then.
<das_j>
ottidmes: How do I instantiate a system this way?
jpo-joyent[m] is now known as Guest69324
<betaboon>
sphalerite: we had a talk some time ago about urxvt and mouse-pointer size.
<ottidmes>
das_j: but they are not completely the same feature wise (nix eval can instantiate during eval, while nix-instantiate has working --arg/argstr and a strict evaluation flag without requiring going to string/json, which nix eval is limited to)
<reallymemorable>
how do i change group membership on directories and files in nixos?
<betaboon>
sphalerite: just wanted to tell you that i just switched from urxvt to alacritty. mouse-pointer size works there out of the box
Guest69324 has quit [Client Quit]
<ottidmes>
reallymemorable: chown?
<__monty__>
iqubic: Almost 20 channels over a couple years: 194MB
<iqubic>
So how can I log into Nixos when I don't get put into the graphical system, and text login fails.
enick_372 has joined #nixos
<clever>
iqubic: add ` single` to the kernel cmdline in grub
<ottidmes>
das_j: if you print the derivation path, it will instantiate it on the fly
<iqubic>
I'll try that now that I can get my phone to work and I can access IRC there.
<das_j>
ottidmes: Yes, but how do I print the derivation path? :/
<das_j>
ah, --raw
enick_372 is now known as jpo-joyent
gagbo has joined #nixos
<das_j>
Wait, that's not the drv path, that seems to be the full system
Denommus` has quit [Remote host closed the connection]
<reallymemorable>
ottidmes: sorry meant to ping you on that -- see above
<sphalerite>
betaboon: good to know!
<ottidmes>
reallymemorable: yep, thats how its usually done
<betaboon>
sphalerite: and I'm quite happy with alacritty. tho the config is a little weird (its yaml but i cant load it with builtins.fromJSON)
<reallymemorable>
ottidmes: how can I specify a group without a user
<reallymemorable>
i.e. can i just change the group ownership
<ottidmes>
reallymemorable: chgroup
<ottidmes>
reallymemorable: chgrp
<reallymemorable>
chgroup -R groupname /directory
<ottidmes>
(never use that one, hence the typo)
iqubic has quit [Ping timeout: 256 seconds]
jpo-joyent has left #nixos ["User left"]
<__monty__>
Can't you just use chown?
<ottidmes>
reallymemorable: yep, but chgrp instead of chgroup, my bad
jpo-joyent has joined #nixos
<reallymemorable>
ok noted thank you
<{^_^}>
[nixpkgs] @delroth opened pull request #56069 → git: build diff-highlight if Perl support is enabled → https://git.io/fhdir
<ottidmes>
__monty__: you are right, its actually mentioned in the manpage, you can just omit the username in front of the colon and than it acts like chgrp
<ottidmes>
__monty__: reallymemorable: "If the colon and group are given, but the owner is omitted, only the group of the files is changed; in this case, chown performs the same function as chgrp."
<{^_^}>
[nixpkgs] @jacereda opened pull request #56070 → xavs: build on darwin → https://git.io/fhdiK
<das_j>
Is there a way I can tell nix "This is my drv, please build it"?
<ottidmes>
das_j: nix store --realize
<ottidmes>
das_j: nix-store --realize
<das_j>
Heck, that's why nix-build didn't work
<das_j>
ottidmes++
<{^_^}>
ottidmes's karma got increased to 15
<lejonet>
srhb: purely fyi, 13.2.2 and 13.2.4 compiled just fine, but the test whines at the package override I do, so I'll need your help to debug that :)
<__monty__>
Though tbh my rational was "I don't wanna remember all these specific commands." : )
<ottidmes>
das_j: "nix-build is essentially a wrapper around nix-instantiate (to translate a high-level Nix expression to a low-level store derivation) and nix-store --realise (to build the store derivation)."
<das_j>
ottidmes: Is there a way to specify the out link for --realise?
<ottidmes>
das_j: its described under the common options section of the manpage in nix-store
<das_j>
Ahh, that's why I didn't find it
<ottidmes>
clever++ (you helped me too many times)
<{^_^}>
clever's karma got increased to 93
gagbo has quit [Ping timeout: 272 seconds]
iqubic has joined #nixos
das_j has quit [Remote host closed the connection]
<iqubic>
Alright. I'm in a recovery shell. What now?
das_j has joined #nixos
<phizzz>
any one have experience with kubernetes on nixos? i can ping the pods on the nixos host (which is on virtualbox), but not the virtualbox host...
<__monty__>
Pretty sure I still owe some karma clever++
<{^_^}>
clever's karma got increased to 94
<reallymemorable>
ottidmes: is there a flag to produce some output log for chown or chgrp
<johanot>
phizzz: you probably need routes to your cluster network on the host
<reallymemorable>
im interested to follow what happens
<reallymemorable>
but no text gets outputted
<infinisil>
Follow up on my earlier acme question. When I initally set security.acme.${domain}.postRun, is there an nice way to have this run when the certificate already exists? I'd have to wait until the first refresh otherwise (cc das_j)
<iqubic>
I just tried running "ping www.google.com" and got a system error
<phizzz>
johanot: any ideas how i would do that?
<das_j>
infinisil: Sorry, I'm not using the nixpkgs module, I have my own one which does things differently
<infinisil>
das_j: Oh I see, so it handles that correctly?
<infinisil>
das_j: Your module that is
<das_j>
But you can do something like me. The "hooks" option (type lines), which get concatenated to a script. The script is called *every time* after the acme unit ran. Wait, I'll gist you the utility function from the top
<das_j>
infinisil: It depends on PowerDNS :/
<clever>
iqubic: normal nix-collect-garbage
asymmetric has joined #nixos
<clever>
iqubic: that should free up enough space to make it run normally
<infinisil>
das_j: I think what I can do is just write a simple systemd service that does the `mkdir && cp && chown && chmod`, then I can order this unit after both the acme update and before the service starts
<asymmetric>
hey, how do i import a specific file after fetchFromGitHub?
<infinisil>
das_j: I mean, if I put it in preStart, and the service doesn't start, it doesn't matter, because only the service needs the commands ran in the preStart, right?
<manveru>
samueldr: you around?
<das_j>
That's right if you don't have your service running until the cert expires ;)
<das_j>
But that's very unlikely
<samueldr>
manveru: depends for how long and what concentration level
<infinisil>
das_j: Oh right..
<das_j>
Btw, some bug I found by gisting it (annoyed me for weeks): sha256sum also outputs the file name, meaning that the two strings are never the same m(
iqubic has joined #nixos
<betawaffle>
what's the best way to structure a git repo that holds nixos modules i want to reuse?
<manveru>
samueldr: i'm just trying to package gridsync, it's a python app that depends on qt... in the build_ext step it fails with `qt.qpa.plugin: Could not find the Qt platform plugin "xcb" in ""`
<johanot>
phizzz: I'm sort of guessing here, but.. you run a kubernetes node on a nixos virtualbox guest? Depending on how networking is setup in virtualbox _and_ in kubernetes, you might be able to do something like: "ip route add <pod-cidr> via <virtualbox-guest-ip>"
<iqubic>
Hello
<infinisil>
das_j: Yeah so I think a systemd service that gets run either when the cert refreshes or before the service starts would be perfect
<manveru>
samueldr: you got any hints that might help?
<das_j>
infinisil: That should do the trick, yes
<samueldr>
manveru: during build... that's definitely different than expected
argent0 has left #nixos [#nixos]
<{^_^}>
[nixpkgs] @bgamari opened pull request #56072 → nixos/gitlab: Introduce database pool size option → https://git.io/fhdPI
<infinisil>
das_j: I fear messing up the Wants/WantedBy/Requires/After/Before/whathaveyounot of systemd though lol
<asymmetric>
clever: i updated the gist to provide more context
<iqubic>
So I can't do much of anything in this rescue shell
<iqubic>
I need help here.
<manveru>
samueldr: that's the output, i'm using buildPythonApplication, which seems to use some additional steps somehow? i can run /nix/store/y41d0y4218qjm0cq06p8ka5b98778sy9-gridsync-0.4.0/bin/gridsync directly, but that fails to find the tahoe executable https://www.irccloud.com/pastebin/lLU6NF6T/
<clever>
asymmetric: line 12 needs a + and the /overlay.nix has to be quoted, and maybe some more () so import applies last
<das_j>
infinisil: It's WantedBy = [ your-service.service ]; Before = [ your-service.service ]; After = [ acme-whatever.service ]
<samueldr>
manveru: try setting something like QT_PLUGIN_PATH="${qt5.qtbase}/${qt5.qtbase.qtPluginPrefix}" in the environment to see whether it is due to purity or something else
<iqubic>
Everything is freaking out because I am out of space.
<samueldr>
or uh, look at the other QT_PLUGIN_PATH invocations
<das_j>
But that doesn't start the service after the acme service, so you need to put systemcl restart into postRun
<clever>
iqubic: what happens when you run nix-collect-garbage?
<infinisil>
das_j: But it also needs something for it to run every time acme-whatever finishes
<samueldr>
but... that's maybe not a "solution", but a workaround
<infinisil>
das_j: systemctl restart is a bit ugly..
<das_j>
If you find something better, let me know
* infinisil
thinks
<das_j>
infinisil: systemd.directives(7) is probably what you need now
<infinisil>
das_j: Yeah that manpage is a godsend
<infinisil>
Using it all the time
<das_j>
Also probably the longest one on my system
<das_j>
(apart from configuration.nix)
<infinisil>
Heh
<iqubic>
I can't collect garbage. There isn't enough space to make the file '/nix/var/nix/temproots'
<clever>
iqubic: are you able to manually delete a file anywhere, a non-nix file?
<iqubic>
So I'm screwed
<iqubic>
Clever: yes
<manveru>
samueldr: thx, i know shit about python and qt :(
<clever>
iqubic: try deleting some normal files, until nix-collect-garbage does work
<samueldr>
though manveru, if futzint with QT_PLUGIN_PATH solves it, it's something to note in the "ultimate fix" test suite
<iqubic>
Freaking hell. My home folder has had its content deleted.
<ottidmes>
infinisil: I have that problem on one of my servers, a systemd loop that I for now can work around, but which will be hell to properly fix (a loop between samba, zfs, and others), on case by case the way they are configured make complete sense, but taken into a complete picture it won't work
<clever>
iqubic: deleted, or just not mounted? is /home on its own dataset?
<asymmetric>
clever: updated again, still gets into infinite recursion - because the repo that's fetched has `import <nixpkgs>` in its default.nix
<asymmetric>
why is default.nix loaded at all though?
<infinisil>
ottidmes: Oh I think I remember you talking about this before, yeah seems like a paiyn
<iqubic>
I have the folder /home/iqubic but that is empty.
<clever>
asymmetric: you need to use a second copy of nixpkgs, from `import <nixpkgs> {}` to get your fetchFromGitHub
Serus has quit [Ping timeout: 250 seconds]
<clever>
iqubic: what does `zfs list` say?
<iqubic>
It is on a separate ZFS dataset though
<asymmetric>
clever: yeah, thought so :(
<clever>
iqubic: then its just not mounted, you need to mount it to see the files
<iqubic>
Ah.
elgoosy has joined #nixos
<iqubic>
zfs list says I have used all the space up.
<ottidmes>
infinisil: thats correct, and I thought I had fixed it, until my brother complained his backups were unreachable, ouch... and every attempt so far to try and fix it, has broken something else in the process, so since every works, despite some warnings in the logs, I keep my hands off for the moment, until I have the time (and courage :P) to properly take a look at it
<clever>
iqubic: did it list your datasets? or fail before that?
<iqubic>
In the avail column everything is set to 0B
<clever>
iqubic: thats to be expected
<elgoosy>
hi, nix is supposed to be a pure a language. Why/how does it allow an expression (getEnv "HOME"), which is not pure
<infinisil>
ottidmes: Did you open an issue in the systemd repo? Maybe they could help with that
<clever>
elgoosy: there is a special flag you can turn on to make it more pure, hydra enables it by default
<manveru>
samueldr: i think it was trying to run tests with a running X server, which wasn't there, even with plugins it doesn't find the display, so i disabled tests and it builds
<clever>
elgoosy: either restricteval, or pureeval
<iqubic>
Which is way way way too much
<clever>
elgoosy: which are set via `--option restrict-eval` or `--option pure-eval`
Serus has joined #nixos
<clever>
iqubic: then mount it, delete one or 2 files, get nix-collect-garbage working, then install and run ncdu on home
<iqubic>
So, what do I do now?
<iqubic>
How do I mount it?
<ottidmes>
infinisil: I might end up doing that, because right now it looks like a limitation of how services/mounts are configured, but I love to be told that I am wrong on that :P that there is just some option that I missed or some misconfiguration on my part
<asymmetric>
clever: what do i pass to a function that takes `self: super:` as arguments? how do i pass something that makes sense for self??
<elgoosy>
clever: thanks. However, if I assume that a package and downloading/installing depends purely on its inputs, now i can't be sure, cause maybe the package expression contains some impure expressions like (getEnv "HOME")?
<iqubic>
I have no clue how to mount a ZFS dataset
ost has joined #nixos
<clever>
asymmetric: the overlays attr of nixpkgs will handle that for you, and you usually dont want to be calling it yourself
<ottidmes>
elgoosy: thats why those things are generally banned from nixpkgs
<das_j>
Is there a way to provide defaults for all systemd units unless overridden?
<iqubic>
I have all the things
<das_j>
So I can do something like InaccessibleDirectories = /var/lib/acme?
skratl0x1C has left #nixos ["WeeChat 2.3"]
<asymmetric>
clever: error: attempt to call something which is not a function but a string with context, at /nix/store/pcb948jb0126d103118vapl88ip7ijk2-nixos-18.09.2203.9bd45dddf81/nixos/lib/fixed-points.nix:44:67
_kwstas has quit [Quit: _kwstas]
<clever>
asymmetric: oh, i forgot to import
aanderse_ has joined #nixos
<clever>
asymmetric: gist fork updated
kreisys has joined #nixos
aanderse has quit [Ping timeout: 255 seconds]
<iqubic>
Well, I have a ton of snapshot of home
Soo_Slow has joined #nixos
das_j has quit [Remote host closed the connection]
Soo_Slow has quit [Remote host closed the connection]
<clever>
iqubic: the `used` column from `zfs list -t snapshot -r -o name,used,refer,written amd/home` tells you how much space youll get back if you destroy that 1 snapshot
das_j has joined #nixos
<clever>
iqubic: you can likely get enough back to GC, from destroying 1 snapshot
o1lo01ol1o has quit [Remote host closed the connection]
<iqubic>
It's like a few megs each
<iqubic>
Except for the few that use a whole gig
<clever>
iqubic: then destroy a single one of those gig large snapshots
schneid3306 has quit [Quit: ZZZzzz…]
<iqubic>
How?
<dmj`>
how do I generate the sha256 hash of an object in s3 so I can use fetchs3 ?
<clever>
iqubic: 2019-02-19 18:46:58 < clever> and this will delete a single snapshot
wfranzini has quit [Remote host closed the connection]
<infinisil>
das_j: I've been wanting to make a PR for this to nixpkgs, enabling all secure defaults for systemd, such that every service needs to explicitly allow what it needs
<clever>
dmj`: if you are using flat hashing, then its just the plain old sha256 of the file
zupo has quit [Ping timeout: 258 seconds]
<das_j>
infinisil: Sounds like a huge thing to do
sigmundv__ has joined #nixos
<das_j>
(just like the icingaweb2 thing I didn't fix yet :x)
<infinisil>
das_j: Well it would be in stages, first add an option secureDefaults, false by default. Then transition all services over by setting it to true. Then make the default true
<das_j>
Oh
ferdek has joined #nixos
<das_j>
I could also think of something like systemd.servicesDefaults.serviceConfig where each service is //ed on
<infinisil>
and Warning: It might delete all your snapshots
<iqubic>
I want to change how often I snapshot.
<reallymemorable>
how can i check which users belong to a specific group?
<das_j>
reallymemorable: getent group $GROUPNAME
<clever>
iqubic: after you fix things, you can edit the nixos config for the automatic snapshots
<iqubic>
I will.
zupo has joined #nixos
xkapastel has quit [Quit: Connection closed for inactivity]
<reallymemorable>
thanks
<pie_>
does anyone use nixos + spacemacs + haskell + dante
rfold has quit [Quit: leaving]
<pie_>
not really sure how to set this stuff up
<iqubic>
The xargs thing is telling me I have too many arguments.
<iqubic>
For destroy that is
Ralith_ has quit [Ping timeout: 246 seconds]
__monty__ has quit [Quit: leaving]
<reallymemorable>
is there a way to turn off sandbox mode for all users?
<asymmetric>
clever: thanks, that worked! still would like if i didn't have to require nixpkgs twice, but i can live with this :)
<reallymemorable>
I have root and another user and im having some ssh issue that i can't solve
<reallymemorable>
and i want to just turn off sandbox mode globally
Ralith_ has joined #nixos
mek42_laptop has joined #nixos
<ferdek>
Hello fellow nix'ers! I was reading a lot of Nix documentation recently and decided to try it out. Remarkable piece of engineering. I have one question I cannot find answer to, before I run nixos-install command: what do I have to put in configuration.nix in order to set libc as musl?
<symphorien>
iirc systemd is not compatible with musl so you can't
<iqubic>
How do I modify configuration.nix to tell zfs to only snapshot every hour?
<iqubic>
How do I modify configuration.nix to tell zfs to only snapshot every hour?
<iqubic>
How do I modify configuration.nix to tell zfs to only snapshot every hour?
<iqubic>
Sorry about that.
<mek42_laptop>
I installed nix (just the package manager) onto a Debian WSL over the weekend and I do not have a ~/.nix-profile/etc/profile How do I obtain this file?
<{^_^}>
[nixpkgs] @orivej-nixos pushed commit from @orivej to master « aws-sdk-cpp: 1.7.49 -> 1.7.53 »: https://git.io/fhdPi
Guanin has joined #nixos
justanotheruser has joined #nixos
<iqubic>
das_j: that xargs part of the comand failed. ZFS tells me I passed too many args to destroy.
<das_j>
iqubic: Use xargs -n1
<ferdek>
symphorien: that explains it ;) so if I understand correctly, having the Nix itself (running anywhere, say debian for instance) I theoretically can build my own rootfs without systemd and with musl (something like NixWrt project I stumbled upon), but NixOs itself does not support this due to systemd?
<symphorien>
Yes
<symphorien>
See also not-os
<iqubic>
Now, how do I snapshot only on the hour?
das_j has quit [Remote host closed the connection]
das_j has joined #nixos
<iqubic>
And, how do I check if compression is on?
<ferdek>
symphorien: thank you, that looks like buildroot on steroids!
<manveru>
yay, i got gridsync to work, it's syncing now with Least Authority :D
fusion809 has quit [Ping timeout: 268 seconds]
wfranzini has joined #nixos
simukis has quit [Quit: simukis]
<iqubic>
So, how do I snapshot only on the hour in ZFS?
<simpson>
manveru: Nice!
das_j has quit [Quit: "Bye!";]
das_j has joined #nixos
<iqubic>
clever: can you help me get the snapshotting configured?
Ariakenom has quit [Quit: Leaving]
<ottidmes>
reallymemorable: going down that route you likely only increase your issues eventually. Its like just forcing something with root, while it ought to be done with the user, but because you are tired of trying to get it right, you just opt to sudo it, forcing it along. It sometimes works, but more often than not it will cause other problems in turn
slack1256 has joined #nixos
<reallymemorable>
ottidmes: understood
<reallymemorable>
Ive been battling to build this branch all day and am a bit exasperated, but will keep at it
<ferdek>
and just out of curiosity, as no documentation will tell me that, were there any decisions in NixOs community regarding systemd and it's future in NixOS? I don't want to start flame war, just asking for general direction the distribution has decided to move
<ottidmes>
reallymemorable: if you paste some of your errors and what your are working on, maybe others can help
kreisys has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<iqubic>
So nobody knows how to get snapshots only on the hour in ZFS on NixOS?
<mek42_laptop>
iqubic, how often are they happening now?
dbmikus_ has quit [Ping timeout: 250 seconds]
<reallymemorable>
I have root and a user both with the same ssh keys in their respective .ssh folders. I am able to clone the repo, but when I try to locally compile it, I always hit this error: https://paste.ofcode.org/s3sNpJzbkP3r9TZDc6jd3J
<ottidmes>
ferdek: its unlikely to go away any time soon, since its so ingrained in NixOS at this point
<reallymemorable>
From what I understand, the nixbld-ers dont have access to the .ssh keys
<reallymemorable>
so they dont get authenticated
<reallymemorable>
i could be wrong though
<iqubic>
I have every 15 minutes, every hour, every day, every week, every month, and every year.
<symphorien>
ferdek: there is a secret fork called slnos which may go in this direction
<iqubic>
But nixos manages that for me.
<iqubic>
I think.
<iqubic>
I only want to snapshot hourly.
<reallymemorable>
also every time i run tthe build command from the user
<reallymemorable>
clever: yeah i added the user to nixbld in the extraGroups part of configuration.nix
<clever>
reallymemorable: thats the problem, nix will use the members of nixbld as build users
<clever>
reallymemorable: when it selects a user, it kills everything in that user
das_j has quit [Client Quit]
das_j has joined #nixos
<clever>
reallymemorable: so, the user you login as, must never be in the nixbld group
<reallymemorable>
so i remove nixbld from extraGroups?
<clever>
yeah
erasmas has quit [Quit: leaving]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<iqubic>
So how do I change how frequently snapshot happen?
eadwu has quit [Quit: WeeChat 2.3]
<clever>
iqubic: you cant change how often they happen (as far as i know), but you can limit how many daily it keeps, and the same for every other category
<iqubic>
I see
<clever>
you can also set the zfs property `com.sun:auto-snapshot:weekly=false` to just stop weekly entirely, read the description of .enable for more examples
shibboleth has quit [Quit: shibboleth]
<ferdek>
ottidmes: symphorien: Thank you guys for the information. I will check the secret sect fork ;), not-os (for eventual embedded use with musl library) and also Guix with it's Shepherd (but honestly I rather like nix language than lisp-like Scheme). I totally understand why systemd got it's popularity, why some hate it and why this might be the best choice for NixOs as of now. Thank you once again :)
<reallymemorable>
clever: so thats a marginal improvement -- thanks -- it doesn't kick me out, but I am still getting an error about a nix-path variable that i dont know how to set: `trace: Please set your nix-path such that ssh-config-file points to a file that will allow ssh to access private repositories. The builder will not be able to see any running ssh agent sessions unless ssh-auth-sock is also set in the
<reallymemorable>
nix-path.`
<clever>
reallymemorable: you want builtins.fetchGit
<iqubic>
Now I run 'ndcu'?
<clever>
iqubic: yep
<reallymemorable>
is that something i add to configuration.nix?
<clever>
reallymemorable: instead of using pkgs.fetchgitPrivate, you use builtins.fetchGit
<reallymemorable>
ok
<iqubic>
I don't have ndcu installed
<reallymemorable>
so i add that to my systemPackages
<iqubic>
What package is it in?
<clever>
iqubic: ncdu
das_j has quit [Quit: "Bye!";]
das_j has joined #nixos
<reallymemorable>
clever: how do I use builtins.fetchGit? it doesnt appear to be a system package
<clever>
reallymemorable: its a function, used to fetch git repos
<reallymemorable>
i have already cloned the repo though
<reallymemorable>
im just trying to build it
<clever>
reallymemorable: what nix function are you calling, that results in the error your getting?
<reallymemorable>
nix-shell i believe
<clever>
reallymemorable: what is in your shell.nix?
<reallymemorable>
where is that
<iqubic>
I just saved 2806.06 MiBs by deleting old generations
<clever>
reallymemorable: in the directory you ran nix-shell in
hiroshi has joined #nixos
<clever>
reallymemorable: it could also be called default.nix