wfranzini has quit [Remote host closed the connection]
<ivan>
lunik1: it's false after I found that it results in unplayable video on radeon
wfranzini has joined #nixos
<jophish>
Squarism, right. Any extra packages go in extraenvpackages. Its important to note that these shouldn't be dependencies as ghc sees them, as in they won't appear in 'ghc-pkg list' or in the Hoogle database. I use this list for tools such as cabal-install, Vim, other binaries which are useful only in the nix-shell and required for building
Supersonic has quit [Disconnected by services]
Supersonic112 has joined #nixos
wtv_nick has joined #nixos
Supersonic112 is now known as Supersonic
wtv_nick has quit [Client Quit]
wtv_nick has joined #nixos
ebzzry has quit [Ping timeout: 268 seconds]
dansho has quit [Quit: Leaving]
slack1256 has quit [Remote host closed the connection]
<clever>
ivan: video playback works fine for me, but has some tearing
<iqubic>
Does NixOS correctly respect Daylight Savings? Like, when the USA's clocks change at the start of November, will I have to manually change my system's clock?
<Squarism>
So I seem to fail making a an executable to be available on the path for a certain step of the build. Im not really sure if its because of my PATH isnt correct or if the the package hasnt been added to the build (Nixos doesnt seem to fail if you specify non-existen packages)
<Squarism>
what means to I have to debug the build?
zupo has joined #nixos
toppler has joined #nixos
xkapastel has quit [Quit: Connection closed for inactivity]
fusion809 has quit [Remote host closed the connection]
<karetsu>
how should I deal with a LaTeX medium scheme installation missing a specific package? tlmgr doesn't appear to be included in the scheme and I'm loathe to have to download each .cls/bst individually and scheme-full is overkill
<alexarice[m]>
yorick: Ah, I was missing that, thanks
ottidmes has joined #nixos
schjetne has quit [Ping timeout: 245 seconds]
<{^_^}>
[nixpkgs] @vcunat pushed 2 commits to release-19.09: https://git.io/Je4tJ
drakonis has quit [Ping timeout: 268 seconds]
drakonis has joined #nixos
<etu>
karetsu: I think it depends on your usecase. In that case I linked we generate a bunch of latex/org-documents as PDF on deploy of hugo.
FRidh2 has quit [Remote host closed the connection]
ebzzry has quit [Remote host closed the connection]
<karetsu>
this is just for editing a colleagues paper and then done with tex and back to org/pandoc so I'll dump it local to the paper I think
drakonis_ has quit [Read error: Connection reset by peer]
<etu>
karetsu: But if you have very different documents, it may be worth it? But I think most often you won't have *that* big differences so the extra packages isn't much overhead compared to running a medium or full sized texlive install
toppler has joined #nixos
<ptotter[m]>
I'm having trouble upgrading to nixos 19.09
<ptotter[m]>
I'm getting level=error ts=2019-10-16T09:12:37.250Z caller=main.go:216 msg="Error loading config" err="error reading config file: open /nix/store/wvny1dxlchpr7fcnay382mm6zgwxb3w3-checked-blackbox-exporter.conf: no such file or directory"
<{^_^}>
[nixpkgs] @FRidh pushed 374 commits to staging-next: https://git.io/Je4t3
<ptotter[m]>
the path is of course a symlink
<ptotter[m]>
but the file the symlink points to exists
<ptotter[m]>
and seems to be readable
ebzzry has joined #nixos
<ptotter[m]>
is "building '/nix/store/3gjs4p0wcaldary882hbwl3x2bicb03g-checked-blackbox-exporter.conf.drv'..." executed in some namespace where it cannot see /var/lib/prometheus/blackbox/config.yaml ?
gila has quit [Quit: My Mac Pro has gone to sleep. ZZZzzz…]
<rot13>
Hi, a very basic question: I'm trying to understand how a function defined in nixpkgs works. Specifically attrsToSection defined in `nixos/modules/system/boot/systemd-lib.nix`. I want to explore it using `nix repl` t
<rot13>
and try out different argumtents
<rot13>
how can I import this function into nix repl?
<rot13>
ddima: Thx, a lot! callPackage requires a path to the lib (systemd-lib.nix). Is there a shortcut by which I can refer to it other than pasting the whole path from /nix/store?
eraserhd has quit [Ping timeout: 265 seconds]
schjetne has joined #nixos
<ddima>
rot13: I did it from my local nixpkgs fork. you can also do sth like `systemdlib = callPackage <nixpkgs/nixos/modules/system/boot/systemd-lib.nix> {}`
<ddima>
or add arbitrary paths to the nix search path via `-I` and then reference them in a similar fashion.
<rot13>
ddima: That was exactly what I was looking for. thanks again!
<teto>
is there a way to convert a submodule to an attrset ?
toppler has joined #nixos
<hyperfekt>
The wiki page for Docker mentions that dockerTools requires nix 'on the host system'. Does this mean the system where the containers are run or the one where they are built?
<hyperfekt>
I presume the latter but I wanted to make sure before I make a clarifying edit.
<ddima>
teto: how do you mean? they are still attribute sets which you can access through `options`, like `options.services.foo`
<teto>
ddima: I am trying to use generators.toINI on an attrofs submodule and toINI fails because the submodule adds fields in the seemingly attrset
<teto>
I know you can do ".value" in general but it doesn't seem to work with submodules
<ddima>
teto: I think toINI might just be a bit too limited wrt nesting levels.
<teto>
ddima: the problem is not about nesting level, just that my config { a = b; } becomes in fact due to the submodule { a = { value= b; check=false; _merge=...} in which case toINMI legitimately fails
fendor has quit [Ping timeout: 240 seconds]
<jophish>
Squarism: how goes it?
karetsu has quit [Quit: WeeChat 2.4]
<Squarism>
jophish, sorry for not getting back to you. At work doing work things. (this is for a private project). I tried different variants. This is the last one I tried https://paste.ofcode.org/cMrDLkqB2SZuupHYbj9WFy
toppler has quit [Ping timeout: 268 seconds]
<Squarism>
Below is the error that wont go away - that I THINK is that System.Process doesnt find "R"
o1lo01ol1o has joined #nixos
fendor has joined #nixos
alexherbo2 has quit [Read error: Connection reset by peer]
toppler has joined #nixos
o1lo01ol1o has quit [Ping timeout: 246 seconds]
alexherbo2 has joined #nixos
<jophish>
Squarism: no worries, I'll take a look
<Ariakenom>
is nix-env -i in a nix-shell limited to within the shell?
bahamas has joined #nixos
<immae>
ariakenom: no, nix-env has effect on your ~/.nix-profile/, so it stays when you leave the shell
rauno has quit [Ping timeout: 276 seconds]
<Ariakenom>
immae: thanks
<Ariakenom>
is there a way to add packages locally when in nix-shell?
toppler has quit [Ping timeout: 246 seconds]
<immae>
you mean while being in a nix-shell, you want to add more things than what you gave initially?
<schmittlauch[m]>
Hi, what's the easiest way to use stdenv shell functions like `subst-var` outside of derivations, like in modules and systemd preStart scripts?
<immae>
I don’t know of a standard way, but you could go into another nix-shell that completes what you already have. Or following your idea of nix-env, you could use it with a `NIX_PROFILE=/some/temp/folder` (it will override the default value of ~/.nix-profile) env var and add /some/temp/folder/bin to your PATH manually
<schmittlauch[m]>
I mean `substitute --subst-var`
eraserhd has joined #nixos
<Ariakenom>
immae: ok
<immae>
it may or may not work depending on what your use case is though
pbogdan has joined #nixos
<Squarism>
jophish, its running. R and dependencies seems to be building.
<Shoubit>
I'm trying to package an electron application. After launching the renderer process I get an error about a missing libxkbcommon-x11.so.0 which I've included in the rpath when I use patchelf on the electron binary, but the error is still there. What else do I need to do to make it find this shared object file?
drewr has quit [Quit: ERC (IRC client for Emacs 26.3)]
cosimone has quit [Remote host closed the connection]
toppler has joined #nixos
cosimone has joined #nixos
cosimone has quit [Client Quit]
lunik1 has joined #nixos
Havvy has joined #nixos
schjetne has quit [Ping timeout: 268 seconds]
isHavvy has quit [Ping timeout: 240 seconds]
fusion809 has joined #nixos
drakonis1 has quit [Quit: WeeChat 2.6]
Havvy has quit [Ping timeout: 240 seconds]
bahamas has joined #nixos
erasmas has joined #nixos
toppler has quit [Ping timeout: 240 seconds]
Havvy has joined #nixos
cosimone has joined #nixos
<Squarism>
in
<Squarism>
ops
byrn has joined #nixos
<Squarism>
In a "runCommand" block how do i refer to the location where default.nix is locate. Say I want to copy some resource files into the resulting build.
drakonis_ has joined #nixos
<jophish>
Squarism: you can use ${./some/file} to refer to a file relative to the .nix file this is written in
fendor has quit [Ping timeout: 250 seconds]
isHavvy has joined #nixos
asymptotically has joined #nixos
Havvy has quit [Ping timeout: 276 seconds]
toppler has joined #nixos
drakonis has quit [Ping timeout: 240 seconds]
Havvy has joined #nixos
<wedens>
if I specify multiple values of `environment.etc.XYZ.text`, will they be merged or overwritten?
<infinisil>
wedens: Depends on what the type of the option is, let's see..
<Squarism>
thanks, that worked well.
isHavvy has quit [Ping timeout: 276 seconds]
<infinisil>
wedens: It's `types.nullOr types.lines`, and types.lines does indeed merge multiple definitions, so yes it does
<tetdim>
nix-env while setting up the build environment: executing '/usr/bin/bash': No such file or directory
Havvy has quit [Ping timeout: 276 seconds]
freeman42xx has quit [Ping timeout: 276 seconds]
bahamas has quit [Quit: leaving]
<wedens>
infinisil: thanks. should I use `lib.mkForce` to set specific value without merging?
bvdw has joined #nixos
knupfer has joined #nixos
<infinisil>
wedens: All definitions of the highest priority get merged, so all values with mkForce get merged together, which is probably good enough in your case
<wedens>
so, if I have only one `lib.mkForce` for the value, it'll be that value, correct?
isHavvy has quit [Read error: Connection reset by peer]
<Taneb>
simpson: everything's a cause for excitement if you're excitable enough! \o/
<buckley310>
does anyone here know how to boot EFI off of a raid1 ESP partition? the internet says i can use mdadm with metadata=1.0, but the install always fails. the best i could get was `grub.mirroredBoots`
<Taneb>
simpson: but it does look like it's getting ready to be merged
<Squarism>
jophish, sorry to bother you again. As I said, the compilation now works. But now I get the same problem running the end results. So R didnt make it to the runtime environment so to speak.
wfranzini has quit [Remote host closed the connection]
<Squarism>
(If it wasnt I had my deadline tomorrow Id be searching myself )
<jophish>
Squarism:could you paste the code please?
<jophish>
I'll have to go to bed in 10 mins but I'll have a quick look
<Squarism>
so building works but I get : server: R: readCreateProcess: runInteractiveProcess: exec: does not exist (No such file or directory)
<Squarism>
when running
<Squarism>
the "result"
<jophish>
Squarism:use in runCommand "lambda-forms" {nativeBuildInputs = [r];} ''
<jophish>
the attribute set there is appended to the call to mkDerivation, so you can put anything in there that one would normall find in the calls to mkDerivation
avn has quit [Remote host closed the connection]
<jophish>
that will make sure that "${r}/bin" is in PATH when the command is run
toppler has joined #nixos
<Squarism>
nix-build works flawlessly. Its just that ./result/bin/server throws this error
justanotheruser has quit [Ping timeout: 240 seconds]
<Squarism>
I thought runCommand was just used to run some post processing
<FRidh2>
let's say you want several programs available. With nix you could instead these individually using nix-env -iA. Better, put them in a list and install the list. On NixOS you could add them to environment.systemPackages
<FRidh2>
if your program needs certain packages at runtime, then you need to refer to them in your recipe in such a way that they can be found during runtime (as mdash says)
<{^_^}>
#69559 (by wkral, 2 weeks ago, closed): CUPS Printers not found in GTK applications
<FRidh2>
Squarism: you'll probably want to use wrapProgram (in makeWrapper package) to create a wrapper so R can be found
<evils>
how do i make (at least the man pages of) a package available to the user if they're enabling a service that uses that package?
7JTAARH42 has joined #nixos
17SAALY3L has joined #nixos
<dminuoso>
evils: I wonder, isnt it enough to just put them into manpages of your derivation?
<dminuoso>
Or no, I suppose installShellFiles is that hook
<dminuoso>
(or just plain doc/ from fixup)
<evils>
dminuoso: not sure what you mean with that
noudle has joined #nixos
byrn has quit [Quit: WeeChat 2.6]
<tetdim>
clever, Ericson2314 waddup theres errors in nixpkgs/pkgs/stdenv/freebsd/default.nix
<{^_^}>
[nixpkgs] @d-goldin opened pull request #71243 → prometheus-blackbox-exporter: coercing cfg to path → https://git.io/Je4Ci
<tetdim>
im fixing em, should i pr those also
toppler has quit [Ping timeout: 240 seconds]
<Ericson2314>
tetdim: Sure, but until we get official machines up for hydra.nixos.org for FreeBSD it's likely to bit-rot again
<tetdim>
rite
gila has joined #nixos
is_null has joined #nixos
<gchristensen>
Ericson2314: what would you think about things like freebsd (or cross, or riscv, or armv7l) being a different sort of build, where each instantiation all build jobs of prior instantiations are killed? that way it never really finishes anything, but as much of the bootstrapping as possible can be pre-built
toppler has joined #nixos
<tetdim>
Ericson2314, ok so nixpkgs is royally fucked for freebsd
<tetdim>
it basically errors non stop, itll be an entire thing to get that working
<simpson>
tetdim: Communities don't exist in a vacuum, and neither does code. Regardless of whether there's CI, we still need people who know FreeBSD to actively maintain those sections of nixpkgs.
<{^_^}>
[nixos-hardware] @Mic92 pushed 2 commits to master: https://git.io/Je4C7
<tetdim>
simpson, i agree..?
<simpson>
tetdim: I guess what I'm saying is that we want your PRs, but also you shouldn't have to feel like one single person needs to heroically unbreak an entire port.
<tetdim>
ah :)
<aristid>
yay, rebooting fixed my gtk printing issue (as i had already built the fix, but the environment variables need to be updated)
ixxie has joined #nixos
gaisseml[m]1 has joined #nixos
ee194350 has quit [Ping timeout: 268 seconds]
ee194350 has joined #nixos
<betaboon>
does nixos.org block ping ? oO
<gchristensen>
I believe so
ris has joined #nixos
toppler has quit [Ping timeout: 265 seconds]
<betaboon>
god damn ... I'm just setting up a new network with vlans, firewalling etc and was wondering why i couldnt ping out to wan from the guest network even tho i had the firewall configured accordingly :/
gaisseml[m] has joined #nixos
* betaboon
is not blaming anyone but itself
reanimus has quit [Ping timeout: 268 seconds]
ddellacosta has quit [Ping timeout: 276 seconds]
<Ericson2314>
gchristensen: I'm fine for quoting it's resources
<Ericson2314>
I mean if it's all cancelled builds and nothing finishes, well that's not too great
<gchristensen>
Ericson2314: what do you mean, quoting?
<Ericson2314>
*quota-ing
<Ericson2314>
oops
<Ericson2314>
but I am not to concerned on the exact method of the quota
<{^_^}>
[nixpkgs] @dpaetzel opened pull request #71244 → pythonPackages.mesa: init at 0.8.6 → https://git.io/Je4Cj
toppler has joined #nixos
Lears has quit [Ping timeout: 240 seconds]
drakonis has joined #nixos
Jackneill has quit [Remote host closed the connection]
<ShaRose>
so, I seem to be having issues with dnsdist. When it starts via the generated systemd service, it fails to bind the socket for access denied, but starting it manually from root works.
lord| has quit [Read error: Connection reset by peer]
mexisme has quit [Ping timeout: 245 seconds]
lord| has joined #nixos
<freeman42x>
when I run nix-shell in https://github.com/razvan-flavius-panda/Electron-Miso-Example/tree/electron-miso-jsaddle-hie I am getting following error: error: anonymous function at /nix/store/v7wxw8hnwxciazl47sa6xpsji59k4517-cabal2nix-app/default.nix:1:1 called with unexpected argument 'miso', at /nix/store/9r1vhvri4ba913g4vyg5jqrhg6pb9amn-source/pkgs/development/haskell-modules/make-package-set.nix:87:27
<freeman42x>
I don't know what to do about it since apparently the same code works fine in another project
o1lo01ol1o has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
knupfer has quit [Quit: knupfer]
<{^_^}>
[nixpkgs] @basile-henry opened pull request #71245 → <!-- Nixpkgs has a lot of new incoming Pull Requests, but not enough people to review this constant stream. Even if you … → https://git.io/Je4W7
knupfer has joined #nixos
toppler has quit [Ping timeout: 265 seconds]
o1lo01ol1o has joined #nixos
<{^_^}>
[nixpkgs] @leo60228 opened pull request #71247 → atheme: init at 7.2.9 → https://git.io/Je4Wd
psyanticy has quit [Quit: Connection closed for inactivity]
<{^_^}>
[nixpkgs] @worldofpeace pushed 3 commits to master: https://git.io/Je4ln
mexisme has joined #nixos
toppler has quit [Ping timeout: 276 seconds]
selfsymmetric-mu has quit [Remote host closed the connection]
<wormwood>
whats best practice when installing software on a system? should nix-env type installs be avoided if reproducibility is important?
<wormwood>
or should everything go / be imported into configuration.nix?
<wormwood>
what about at the user level?
<gchristensen>
I use systemPackages for everything I want system-wide, and I use a shell.nix for every project. so, no project-specific stuff in my global config
<wormwood>
(non-root)
<wormwood>
hmm, ok
<wormwood>
interesting
<wormwood>
gchristensen: so you don't do any user-level "global" installs via nix-env -iA?
<tilpner>
wormwood: Another option, users.users.wormwood.packages = ...;
<Ericson2314>
I might take it over from the original author or whatever, but I don't know about those python tools or what the normal way is 😊
lord| has joined #nixos
fendor_ has quit [Ping timeout: 240 seconds]
<red[m]>
isn't the maintainers datastructure the closest we have to an AUTHORS file?
<red[m]>
oh - I have a build infrastructure question - and the answer may very well be "yes" already - I'm just not a maintainer on anything that's broken (yet)
<Ericson2314>
red: tetdim was asking about nix not nixpkgs
<red[m]>
... but CPAN had this great function by where if any of your published packages broke when the new rev of perl came out or one of your dependancies dropped through the floor - it would email the maintainers of those packages that started failing tests
<tetdim>
GIT DERPED OUT AND DELETED MY COMMIT WHEN I PUSHED IT
<red[m]>
in my estimation is was way underpriced for what it was
<tetdim>
grrrrrrrrrrrrrrrrrrrrrrrRRRRRRRRRRRR
<gchristensen>
red[m]: oh wow
<red[m]>
(and they accepted an offer on it too)
<red[m]>
yeah - that'[s LTO6, which is perfect for me since I have historical backups on LTO4
<symphorien>
tetdim: git reflog might still have it
<red[m]>
LTO rules. 2 revs back to read, 1 rev back to write
<gchristensen>
hmm that is a really very nice idea to build in
<red[m]>
unfortunately, both the bacula and the bareos packages are b0rked in nixos
<red[m]>
and I know that at some point I'm going to have to fix and maintain the bareos thing
<red[m]>
People wonder why I'm obsessed with tape - it's simple.
<red[m]>
I did computer forensics and employee investigations at work for a fair few years.
<red[m]>
I have lost count the number of grown men who have openly wept in my cube after I tried to recover pictures of their children off dead devices.
<red[m]>
I tried backing up to the cloud... used crashplan. Crashplan deleted 7.2Tb of my data with a 7 day notice
jnas has quit [Remote host closed the connection]
<gchristensen>
neat.
<tetdim>
symphorien, IT IS
<tetdim>
THANK THE STARS
<symphorien>
:)
<red[m]>
and, being brutally honest - if I want my kid's pictures to still exist in 30 years... am I going to remember to pay my cloud backup subscription every month for the next 30 years?
<red[m]>
so yeah
<tetdim>
god i went into an adhd trance the last 3 hours of coding i have no idea what changes ive made
<lunik1>
I'm trying to get docker to work with "lockKernelModules = true;" but I'm getting "Error response from daemon: failed to create endpoint [...] operation not supported." I assume there is some kernel module that it needs, anybody know what it could be?
<nilsirl[m]>
How should docker-credential-pass be installed on nixos?
doublex has quit [Ping timeout: 240 seconds]
toppler has quit [Ping timeout: 268 seconds]
Neo-- has quit [Ping timeout: 265 seconds]
wfranzini has quit [Remote host closed the connection]
<AmandaC>
any idea why makeWrapper would be spitting this out? ` /nix/store/grr20p5m5k2qw0p0rfx7hihrvm8gd86a-hook/nix-support/setup-hook: line 43: /nix/store/rj9vnnj0cr6344vy8h5jcx08y5g9q0hj-vscode-with-extensions-1.39.1/Applications/Visual Studio Code.app/Contents/MacOS/Electron: Permission denied`
<{^_^}>
[nix] @adamse closed pull request #3152 → Use `lib.version` instead of the deprecated `lib.nixpkgsVersion` → https://git.io/Je4ci
<duairc>
I obviously know that doing so defeats the whole point of NixOS, but is there at all to temporarily edit the /nix/store? I finally got a massive python package to build, but I'm getting an error (I think paths are getting messed up somewhere) and it would be really nice if I could just quickly throw in few print(__import__('sys').path) in the right places to help me untangle what's going on...
o1lo01ol1o has quit [Remote host closed the connection]
toppler has joined #nixos
o1lo01ol1o has joined #nixos
knupfer has quit [Ping timeout: 264 seconds]
<{^_^}>
[nixpkgs] @philandstuff opened pull request #71266 → gnupg-pkcs11-scd: init at 0.9.2 → https://git.io/Je44q
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
Ariakenom has quit [Quit: Leaving]
<red[m]>
so - is it fair to say that if I only make changes on my nixpkgs branch and re-run a nixos-rebuild against it... the "rebuilding the following..." at the top of the rebuild output is the list of downstream?
<red[m]>
oh wait - it won't include anything that was successfully evaluated in previous runs
<red[m]>
crap
<red[m]>
thanks
<red[m]>
... and why is it building two different lvm2.drv s
toppler has quit [Ping timeout: 268 seconds]
ottidmes has quit [Ping timeout: 268 seconds]
lunik10 has quit [Ping timeout: 240 seconds]
o1lo01ol1o has quit [Remote host closed the connection]
lunik10 has joined #nixos
<duairc>
I can't get past this problem I'm having. I'm not sure if I don't know enough about Nix, or enough about python, or both, but I can't tell whose fault this is
lunik10 has quit [Client Quit]
<duairc>
I know what sys.path is
<duairc>
But python modules, they seem to also have a __path__ attribute
Jackneill has joined #nixos
<duairc>
If I have a package foo-bar that defines a foo.bar module, and a foo-bar-baz that defines a foo.bar.baz module, even though I have both foo-bar and foo-bar-baz in my PYTHONPATH, foo-bar-baz doesn't seem to end up in foo.bar.__PATH__
<duairc>
*__path__
<duairc>
Does this make any sense to anybody?
o1lo01ol1o has joined #nixos
drakonis has quit [Ping timeout: 265 seconds]
<c00w>
Does foobar have a dep on foo.bar.baz?
sellout- has joined #nixos
<c00w>
Or vice versa?
ng0 has quit [Quit: Alexa, when is the end of world?]
toppler has joined #nixos
civodul has quit [Quit: ERC (IRC client for Emacs 26.3)]
<duairc>
No, as far as I can tell. Actually, it's more like I have a foo.bar.baz and a foo.bar.quuz, that are independent of each other (except lots of them), and foo.bar.__path__ only seems to have a subset of them
lunik10 has joined #nixos
<duairc>
c00w: ^
o1lo01ol1o has quit [Ping timeout: 265 seconds]
fendor has quit [Quit: Leaving]
asymptotically has quit [Quit: Leaving]
drakonis_ has quit [Ping timeout: 264 seconds]
<red[m]>
So -- nixos-rebuild is building more than one package at once so I can't actually tell which of the packages is failing to build.
<c00w>
My guess - is that foo.bar.__path__ is set by foo.bar, and if it doesn't know about something it won't set it correctly.
<red[m]>
How would I determine that? (the output to stdout appears to be from every package building concurrently)
<c00w>
But why do you need foo.bar.__path__ populated anyway?
<c00w>
red[m]: Does --show-trace help?
<duairc>
c00w: I don't actually know. All I know is the program that uses all of these libraries crashes on ModuleNotFoundError: No module named 'foo.bar.baz'. But the directory that contains foo/bar/baz.py is definitely in sys.path.
drakonis has joined #nixos
<c00w>
Huh - got it. You've hit the limits of my common debugging. I'd probably look up how __path__ and import work in python and see if I could get a minimal repro (i.e. 2 dummy packages).
<duairc>
The only reason I'm mentioning __path__ is because when I googled it, someone suggested doing import foo.bar; print(foo.bar.__path__) to debug, and when I do that the path that contains foo/bar/baz.py isn't there (though maybe that's a red herring)
dermetfan has joined #nixos
<red[m]>
c00w: I'll try that now. It didn't suggest it so I didn't try it
Jackneill has quit [Remote host closed the connection]
toppler has quit [Ping timeout: 268 seconds]
<duairc>
c00w: I guess a making a minimal repro is a good shout
<red[m]>
I suspect it's systemd, because it's never dns ;-)
<red[m]>
I'll give that a shot now.
<red[m]>
duairc: all my servers run on unstable-minimal ;-)
<evelyn>
red[m]: whether "it's never DNS" depends if your crappy wifi router pings some unfortunate individual's nameserver by default without their knowledge
<{^_^}>
[nixpkgs] @jsamsa opened pull request #71268 → Fix path for darwin downloads → https://git.io/Je44a
<duairc>
extend_path helps, maybe?
<red[m]>
builder for '/nix/store/ywb4gvm86idl0v0613msaq6si5jifmv7-lvm2-2.03.01.drv' failed with exit code 2
<red[m]>
yet when I nix-shell -p lvm2
<red[m]>
it comes from a different store entry
abathur has joined #nixos
toppler has joined #nixos
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<red[m]>
Can you configure nixos-rebuild to only build one package at a time?
<infinisil>
red[m]: Yeah, --max-jobs 1
gjabell has joined #nixos
<gjabell>
anyone know if there's a way to force nixops to rebuild? I am setting up a matrix server but it won't use the up-to-date package for some reason
<red[m]>
(I also have "nix.buildCores = 1" too in my configuration.nix)
<gjabell>
defaults to v1.3.1, even though 1.4.0 is available, and running nix-shell -p matrix-synapse on the server even brings down 1.4.0
<infinisil>
red[m]: --max-jobs 1
<infinisil>
gjabell: Soo you're using nixops?
toppler has quit [Ping timeout: 250 seconds]
<red[m]>
doh - thank you
<gjabell>
yarp
<infinisil>
gjabell: The packages nixops uses usually come from <nixpkgs> in your NIX_PATH on the deploy machine, so what does `nix-instantiate '<nixpkgs>' -A <your matrix server, synapse? attribute>` eval to?
ambro718 has quit [Quit: Konversation terminated!]
<hpfr[m]>
My laptop is trying to build qtbase again :(
<gjabell>
guess I'll just cherry-pick the patch locally then
<__red__>
after the reload it can no longer connect to the postgres databasee
<__red__>
specifically:
<__red__>
Oct 16 22:29:04 evil.red synapse[6175]: twisted: [] psycopg2.OperationalError: FATAL: role "matrix-synapse" does not exist
<__red__>
which is concerning...
<infinisil>
That is a bit concerning indeed
* infinisil
checks the changes made
<__red__>
where do postgresql database files live again?
toppler has quit [Ping timeout: 268 seconds]
<__red__>
I'm hoping it didn't rerun an installation script which nukes the database :-P
<clever>
__red__: /var/lib/postgres i think
<infinisil>
Hm there weren't any obviously related changes
<__red__>
it *MAY* be that postgres bumped
<__red__>
checkign that now
<__red__>
how do you revert the switch again? (without going back to the grub menu)
<gjabell>
might need to restart postgres, I've had issues with prosody losing connection to postgres after an update
<__red__>
This is why I ask...
<__red__>
drwx------ 19 postgres postgres 4096 Oct 16 22:31 11.1
<__red__>
drwx------ 20 postgres postgres 4096 Oct 16 22:22 9.6
<__red__>
<hpfr[m]>
Does nix-collect-garbage have an equivalent in the nix [subcommand] form
<__red__>
That kinda suggests that we went from 9.6 to 11.1
<infinisil>
__red__: gjabell: The postgres service should also have been restarted with the update anyways
<gjabell>
I have postgres and prosody on different servers so they don't always update together
<infinisil>
Ah yeah then it made sense
<infinisil>
Although
<__red__>
Nope
<infinisil>
Ohh I have a suspicion..
<__red__>
it's apparently gone
<infinisil>
What's gone?
<__red__>
lemme check my **TAPE BACKUPS**
toppler has joined #nixos
<__red__>
;-)
broccoli has joined #nixos
<__red__>
I just did a psql \l with the 9.6 client and then
<__red__>
OH WAIT
<__red__>
I have to restart the server
<__red__>
sorry
* __red__
facepalms
<infinisil>
Ah the init change thing?
drakonis_ has joined #nixos
<__red__>
rolling back
<__red__>
yup
<__red__>
confirmed
<__red__>
it bumped postgresql from 9.6 to 11.1
<__red__>
and my data didn't go with it
lunik10 has joined #nixos
<__red__>
that's one hell of a surprise
drakonis has quit [Ping timeout: 268 seconds]
<broccoli>
hi, do someone have any experience with cuda/bumblebee? I managed to configure it somehow: I use the intel gpu for regular display, and start "optirun python" for instance when I need cuda, but I need to manually load nvidia_uvm kernel module, any ideas?
<gjabell>
oof
<__red__>
so I guess I have to either migrate my data, or pin postgres at that version
<gjabell>
isn't the whole point of the stateVersion to prevent that from happening?
<__red__>
well, I'm on unstable...
dermetfan has quit [Quit: WeeChat 2.6]
<__red__>
and I don't have a system.stateVersion in my configuration.nix
<gjabell>
ah
<__red__>
so this may very well be self-inflicted
<__red__>
What should you set it to when your nix-version looks like this: 20.03.git.aaa1ba3 (Markhor)
<infinisil>
__red__: Set stateVersion!
<infinisil>
,stateVersion
<{^_^}>
Setting stateVersion to the latest release doesn't upgrade anything and can only break your setup at best. To actually upgrade NixOS see https://nixos.org/nixos/manual/#sec-upgrading. If you want to update the stateVersion option regardless, Ctrl-F for "stateVersion" in https://nixos.org/nixos/manual/release-notes.html to see things that need to be manually migrated with the new value.
dermetfan has joined #nixos
immae has joined #nixos
<infinisil>
Not setting state version and updating is like setting stateVersion to the latest version
toppler has quit [Ping timeout: 265 seconds]
<gjabell>
so since nixops packages are built on the deploy machine and not the server, is it generally a problem to have both machines tracking different channels?
oldandwise has joined #nixos
<gchristensen>
no
<__red__>
infinisil: what should I set it to on a system (rolled back and working) that displays 20.03.git.aaa1ba3 (Markhor) as a version?
<__red__>
20.03?
<infinisil>
gjabell: I personally set it up so whatever nixpkgs the deploy machine uses gets copied to the server
<oldandwise>
how to install latest java in nixos?
<infinisil>
So no need for nix-channel nastiness on the server
lunik1 has joined #nixos
<__red__>
did the stateVersion bump then for unstable-small?... and how would I see that commit? where would it be?
<lunik1>
Does anybody on here use mergerfs? I'm running into some issues.
jb55 has quit [Disconnected by services]
<gjabell>
infinisil: hmm ok, do the nix-channels on the server even do anything if you're deploying with nixops?
<infinisil>
__red__: Well it ideally would be set to the version of NixOS you initially installed
jb551 has joined #nixos
<infinisil>
But since it hasn't been set before this shouldn't be done
<infinisil>
__red__: So use whatever value works, anything really, just so it's not unset
oldandwise has quit [Client Quit]
<infinisil>
Well I was gonna answer oldandwise but I guess not then
lunik10 has quit [Read error: Connection reset by peer]
lunik10 has joined #nixos
<__red__>
okay - thanks
toppler has joined #nixos
broccoli has quit [Quit: leaving]
erasmas has quit [Quit: leaving]
<infinisil>
lunik1: Nope, but maybe somebody can help if you actually described what the issue is
<lunik1>
basically mergerfs crashes as soon as I try to write to the filesystem
lunik10 has quit [Read error: Connection reset by peer]
<lunik1>
nixos 19.09, default kernel
<__red__>
infinisil: it didn't stop the upgrade or copy the data. Not a big deal - I'll just copy my data over from 9.6 to 11.1 and get on with my life :-)
<__red__>
I'm on unstable, so it's not like it's unexpected
wildtrees has quit [Quit: Leaving]
m0rphism has quit [Ping timeout: 240 seconds]
<infinisil>
Hm wait, but why is it 11.1 anyways
<infinisil>
The default should be 9.6 I think
<infinisil>
__red__: Do you have meguca enabled perhaps?
freeman42x has quit [Remote host closed the connection]
<__red__>
and I installed it and got 11.1 installed
<lunik1>
ok I am seeing the mergerfs issue on two different systems now. could somebody try to reproduce on 19.09? Install pkgs.mergerfs, create three empty directories, run "mergerfs /path/to/dir1:/path/to/dir2 /path/to/dir3", and then try to create and rm a file in dir3?
<__red__>
then, because of something else - I set my stateVersion to 19.09
<vika_nezrimaya>
Hey. NixOps seems to have a container backend, but it's largely undocumented. I assume that `nixops ssh` will still work when deploying as a container? Also how to forward ports from container? Can I make it connect to a bridge of some sort so the container will get its own IP from DHCP?
<__red__>
it would then install 9.6 instead on my next rebuild right?
<vika_nezrimaya>
(the container host is obviously accessible via ssh already)
<infinisil>
Unfortunately not setting stateVersion is more common than I'd like
growpotkin has joined #nixos
<__red__>
for most packages like that they let you pin your version in your configuration.nix anyways so easy enough to override people get into too much of a tither.
<infinisil>
And because it looks like a version number people like to increase it, which defeats its whole purpose..
<__red__>
I removed it because I went from stable to unstable
<__red__>
and didn't know what I should set it to
<__red__>
so just removed it
<__red__>
(since it didn't seem to break anything)
mexisme has quit [Ping timeout: 246 seconds]
<__red__>
I also removed the warning and the comment above it because "You're not the boss of me now"[tm]
<__red__>
lesson learned :-)
<infinisil>
I see, yeah you could've kept it
<infinisil>
Hehe :P
<__red__>
I've always been blown away by how resiliant nixos actually is
<__red__>
every time I do a talk on it, I switch my production server from unstable to stable and back again
<__red__>
it blows minds
<gchristensen>
same, __red__ :)
<__red__>
since half the people using .rpm based distros etc can barely run regular patches without everything breaking from time to time
<Ashy>
it's amazing
<flokli>
We might just want to increase stateVersion to something not matching the release number, like 3000, and then just increment the "suggested value" on every state migration we have to do, or every release
<danderson>
Probably a stupid question, but: are new nixos modules eligible for backport into 19.09? Or is it bugfixes only?
philr has quit [Ping timeout: 265 seconds]
<flokli>
danderson: hi :-)
<Ashy>
i've been completely failing to explain nixos to colleagues and friends
<danderson>
I ask only because I'm planning how long I will have to keep my shiny, just-merged pppd module in my own lib/ folder :P
<__red__>
Ashy: I have a presentation format you can borrow
<Ashy>
the closest analogy i can think of is that it's like git in the early days
<danderson>
flokli: hi! Thanks for the code review!
<__red__>
Ashy: it's really just a map for a demo - but I converted about 30 people with it
<Ashy>
it is going to revolutionize linux distros, but it hasnt quite reached critical mass yet
<infinisil>
flokli: Yeah maybe, but as discussed in #65314, I think the best fix for the current mess is to have per-module stateVersions
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to release-19.09: https://git.io/Je4BH
<infinisil>
flokli: Visibility for the end user? This shouldn't be a problem, they'll get an error if they don't set this option. Visibility for the module authors? Should not be a big problem, because the global stateVersion won't exist anymore but be replaced with a message explaining how to use mkStateVersionOption
<worldofpeace>
flokli: now we have it backported 😺
toppler has quit [Ping timeout: 276 seconds]
<flokli>
worldofpeace: ah
<__red__>
Basically: I go: 1. This is nix - platform independent. We don't break your env like brew and its ilk (that gets the Mac developers interested)
<{^_^}>
[nixpkgs] @worldofpeace pushed 19 commits to master: https://git.io/Je4BQ
<Ashy>
__red__: haha nice
<flokli>
infinisil: The error message generation, comparison etc. will be provided by some module-system-wide tooling? ok
<__red__>
2. If only there was an OS that used these principles... oh yeah, there is... it's called NixOS. Here's me being so confident I'll move my production systems back and forward major revisions. "Enterprise Linux" can barely patch itself.
<flokli>
mkStateVersionOption, ok
lord| has joined #nixos
<__red__>
3. If I do mess up - here's my rollback (show rollbacks and grub screen) -- That's recruited the OS developers.
<flokli>
danderson: we `cherry-pick -x ` the individual commits. In your case, it'd be a PR against the release-19.09 branch.
<infinisil>
flokli: Yeah that
<__red__>
4. Then there's nixops -- That gets the infrastructure people
<danderson>
flokli: perfect, thank you!
<__red__>
Then lastly... I show them qmk_firmware's github which has a shell.nix file in their repo for building microcontroller firmware.
<__red__>
Then I run that on my server, installing all the cross-platform compilers and dependancies... then when I exit the shell they've all gone poof
<__red__>
and that's it. Room converted
phyfey[m] has joined #nixos
<Ashy>
nice
<Ashy>
got a youtube link of you doing your talk? :P
<__red__>
Ashy: you do things like that and I may be encouraged to do it
Soo_Slow has joined #nixos
<__red__>
I'm actually kinda disheartened because I posted a few slides of my previous talks on twitter and got zero interest. Compare to when I inform the American audience that people in the UK eat with the fork upside down at 13,000 responses. Life is weird,
<danderson>
Wheee! Okay, now, I *was* configuring a nixos router at some point in the past, before I started writing nixos modules...
<judson__>
Thanks, btw, gchristensen and infinisil
<gchristensen>
worked from here ...
<judson__>
Hm. I'd thought maybe because it was inside an existing nix-shell that might be a problem. So I tried running it from outside a direnv'd directory.
<judson__>
Possible having it in direnv-lorri env is the problem.
<judson__>
That's weird.
<gchristensen>
hrm...
<infinisil>
And it worked outside the dir or..?
toppler has quit [Ping timeout: 240 seconds]
<judson__>
cp pg_back /tmp; /tmp/pg_back works.
<judson__>
But ~/dev/proj/pg_back doesn't (where there's a nix/lorri .envrc in ~/dev/proj)
<infinisil>
Weird
<judson__>
Ok, no -- it's that I'd named the script pg_back.nix-shell.
<judson__>
Removing that suffix fixes it.
<judson__>
And that suffix anywhere breaks it.
<gchristensen>
wat
<infinisil>
judson__: Even /tmp/pg_back.nix-shell?
<infinisil>
That's very weird
<infinisil>
Damn, can reproduce
<infinisil>
It happens anywhere indeed
alexherbo2 is now known as alex```
<judson__>
Yep
<infinisil>
Doesn't happen with the extension .nix
<judson__>
Glad I'm not the only one saying "wat"
ris has quit [Ping timeout: 258 seconds]
<infinisil>
That *has* to be a bug
<gchristensen>
how cool
<freeman42x>
how can I ehm... stop this balloo / tags.so / kde thingie stop using CPU doing like... nothing probably without switching from KDE?