gchristensen changed the topic of #nixos-dev to: NixOS Development (#nixos for questions) | https://hydra.nixos.org/jobset/nixos/trunk-combined https://channels.nix.gsc.io/graph.html | 18.03 release managers: fpletz and vcunat | https://logs.nix.samueldr.com/nixos-dev
orivej has joined #nixos-dev
phreedom has quit [Remote host closed the connection]
phreedom_ has joined #nixos-dev
<angerman> domenkozar: still around?
<angerman> I seem to have some issues making my remote builders use cachix.
<angerman> ahh. so I need to set it on the remote builder. And can't pass it via `nix-build`
jtojnar has quit [Quit: jtojnar]
jtojnar has joined #nixos-dev
jtojnar_ has joined #nixos-dev
jtojnar has quit [Ping timeout: 276 seconds]
jtojnar_ is now known as jtojnar
<clever> angerman: some of the --option flags only work if you are flagged as a trusted user in nix.conf (or have +w to /nix/store)
<clever> angerman: and when doing remote builds, they only apply to the local machine
<angerman> I am a trusted user by now :-) at least I can specify builders explicitly.
<angerman> Guess I really want nixops support for packet right now. So I don’t have to change the configuration file and nixOS-rebuild, but have nixops do that for me.
jtojnar has quit [Read error: Connection reset by peer]
jtojnar has joined #nixos-dev
<angerman> clever: would you know where to start/look for adding packet.net support to nixops? Is there maybe some tutorial?
<clever> angerman: first, does packet.net have an api to allow you to create/destroy a VM?
<clever> angerman: looks like a yes
<clever> "POST /projects/{id}/devices
<clever> creates a device
<clever> this is the python function responsible for creating a new ec2 instance
<angerman> Alright I’ll look after breakfast
<clever> angerman: to start with, i would implement create and destroy, and tracking the device id and state in the nixops state file
<clever> angerman: this part took me a bit to figure out, but attr_property creates a magic value, with getters and setters on it
<clever> so if you then attempt to get or set something like self.instance_type, nixops will automatically get/set it from sqlite
<clever> that will allow you to save your own per-backend state
<clever> angerman: once create and destroy are working, you will need a get_physical_spec() function, for deploy to work sanely
<clever> angerman: get_physical_spec will generate a nixos module that configures the machine, so it can continue to boot under packet.net
<clever> some hosters require a static IP to be configured in nixos, and you would have to track that in the state and configure it correctly
<clever> boot device configuration is also done here
<clever> angerman: get_ssh_flags and get_ssh_private_key_file may come into play, depending on how packet.net handles ssh keys
<clever> https://github.com/NixOS/nixops/blob/master/nixops/backends/__init__.py contains the default implementation of all of the methods
<clever> and https://github.com/NixOS/nixops/blob/master/nixops/backends/none.py has the none backend, which i use to manage several machines on my LAN
<clever> none's create will just generate a random ssh key, and rely on your ssh agent for that first deploy
<clever> and none doesnt support destroy
<angerman> clever: thanks lots of stuff to read up on!
<clever> feel free to ask other questions if you get stuck
sir_guy_carleton has joined #nixos-dev
phreedom_ has quit [Remote host closed the connection]
phreedom has joined #nixos-dev
lassulus_ has joined #nixos-dev
lassulus has quit [Ping timeout: 240 seconds]
lassulus_ is now known as lassulus
orivej has quit [Ping timeout: 244 seconds]
<dtz[m]> so how bad should I feel about this: (apparently fetchurl can fetch and extract nar's, which is awfully convenient....) https://github.com/dtzWill/nur-packages/blob/master/lib/from-nar.nix -- example usage although it could be more ergonomic: https://github.com/dtzWill/nur-packages/blob/master/pkgs/allvm/tools-from-nar.nix
<clever> dtz[m]: nix 2.0's fetchTarball can fetch&unpack a tarball
<dtz[m]> that one's easier since it's relocatable, this one does a bit of unpack'ing like we do with boottrap tools: https://github.com/dtzWill/nur-packages/blob/7443e165745dd4d727a8d3af9e8735f4ba630a14/pkgs/nix-mux/default.nix#L5
srk has quit [Ping timeout: 240 seconds]
srk has joined #nixos-dev
sir_guy_carleton has quit [Quit: WeeChat 2.0]
orivej has joined #nixos-dev
FRidh has joined #nixos-dev
orivej has quit [Ping timeout: 260 seconds]
tilpner_ has joined #nixos-dev
tilpner has quit [Ping timeout: 264 seconds]
tilpner_ is now known as tilpner
<sphalerite> dtz[m]: lol nice
Linus[m] is now known as sphalerit
<domenkozar> btw who is cancelling staging jobs?
<domenkozar> LnL: ?
<LnL> I just woke up :p
<domenkozar> maybe FRidh
<domenkozar> LnL: morning :)
<domenkozar> if you'd like to include anything into Tuesday weekly, please write to https://github.com/NixOS/nixos-weekly/pull/54
<{^_^}> nixos-weekly#54 (by garbas, open): Issue 2018/04
<FRidh> domenkozar: nope, I don't have permissions
<FRidh> should get them
<FRidh> anyway, makes sense to cancel staging
<domenkozar> why? :)
<FRidh> staging-next should finish so we can merge
<domenkozar> well that's darwin
<FRidh> now we have staging-next and staging, but haven't reduced job size, so there's a lack of capacity
<domenkozar> I've bumped staging-next to the front
<FRidh> ok
<FRidh> staging would be a lot more useful if it was smaller and would rebuild more often
<LnL> yeah, that would be better
<domenkozar> right
<domenkozar> I think we need something like release-sanitycheck.nix
<domenkozar> which sits between small and full
<domenkozar> so it tries to builds all major languages, platforms, etc
<domenkozar> but doesn't go build plugins
<domenkozar> or maybe just a few :)
<domenkozar> FRidh: you're one step ahead :)
<LnL> hmm, what about a generated list of all attributed with eg. >1k referrers
<domenkozar> could be the next iteration :)
<FRidh> LnL: that's an "easy" way to obtain important packages, however, I think if a sub package set doesn't have as many packages we should still test its logic
<LnL> right
<FRidh> how to do this. nix-instantiate everything and then use nix store -q --referrers ?
<FRidh> or maybe with new nix show-derivation --recursive
<LnL> does that work for drv paths?
<LnL> I was thinking of dumping the references of all attributes and then search in there
ixxie has joined #nixos-dev
<FRidh> domenkozar: could you abort staging and trigger a new staging-next. I just merged all the branches.
orivej has joined #nixos-dev
<domenkozar> FRidh: so retrigger both?
<domenkozar> done
<domenkozar> will cancel once they evaluate
<FRidh> well at the moment they contain the same
<domenkozar> oh you merged staging into staging-next?
<FRidh> yes
<domenkozar> I thought staging should be tested before it goes to staging-netx
<domenkozar> wasn't that kind of the point of splitting them up?
<domenkozar> there are stdenv rebuilds in staging that even weren't built
<FRidh> Testing happens in both, its just that in staging-next no new things get added. Ideally staging is stable enough before going into staging-next, but now with the big staging job it's hard to see.
<FRidh> fixes can be applied to staging-next
<FRidh> (including reverts)
<domenkozar> error: undefined variable 'gnome2' at /nix/store/94gxwsf0g4irzjkk20bip0y53nfcf77g-source/pkgs/applications/altcoins/parity-ui/env.nix:12:27
<domenkozar> in staging-next and staging
<FRidh> hm, I suppose pre-commit hook doesn't cover merge commits
<FRidh> ah, it did not even catch that
<FRidh> any suggestion for catching this? Following did not: exec nix-env -f . -qaP --meta --json --drv-path --show-trace >/dev/null
<Dezgeg> try nix-build --keep-going pkgs/top-level/release.nix -A tarball
<Dezgeg> or look what it runs
<domenkozar> yeah running tarball is usually a good check for eval
<domenkozar> you have to evalute all platforms, etc
<domenkozar> it would be cool if we added https://pre-commit.com/ support
<domenkozar> then commit hooks can be maintained in centralized fashion
<FRidh> that's an interesting project
<domenkozar> it does some cool tricks
<domenkozar> like git stash
<domenkozar> run hook
<domenkozar> git pop
Lisanna has quit [Remote host closed the connection]
<FRidh> oh master is also broken because of this
<FRidh> domenkozar: could you retrigger
<domenkozar> shot
<FRidh> mass rebuild on master, that's....odd
jtojnar has quit [Ping timeout: 260 seconds]
* peti just earned his "first emoji" batch on discourse.nixos.org. :-)
<domenkozar> :D
<domenkozar> gchris`wednesday: hmm, maybe ofborg could have a CI check that would fail if commit on release/master rebuild more than X packages?
<domenkozar> then if we defined how mass rebuilds are merged from staging, it would skip that check
ixxie has quit [Ping timeout: 240 seconds]
<{^_^}> ofborg#201 (by domenkozar, open): CI check for number of rebuilds
<domenkozar> peti: do you have a special trick for building all haskell packages and then generting platforms based on what works?
<LnL> there's a list of platform overrides in the package list IIRC
jtojnar has joined #nixos-dev
FRidh has quit [Remote host closed the connection]
<Profpatsch> Does it make sense to have a "bin" output for non-static binaries?
<Profpatsch> Because that always references "out" then, no?
<Dezgeg> but you don't always need the binary, say if your program links to libcurl but doesn't call the command line programs
<Profpatsch> +1
<Profpatsch> I’ve written a simple utility called nman that fetches the manpage by attrname.
<Profpatsch> So I notice that a lot of packages don’t even split their manpages.
<Profpatsch> And I wonder if manually specifying outputs is the way to go.
<Profpatsch> So I have to download a 1GB tesseract output just to see the manpage. :)
<Profpatsch> Maybe the solution should be to give nixpkgs authors better linters.
pie_ has quit [Remote host closed the connection]
pie_ has joined #nixos-dev
FRidh has joined #nixos-dev
ixxie has joined #nixos-dev
<dtz[m]> although until the man page situation is fixed, every split man page is a man page many users can't access in any reasonable manner
<dtz[m]> ideally we would fix that and split them off everywhere and everything would be good :)
<Dezgeg> it is accessible in nixos
<Dezgeg> not in nix-env indeed though
<Profpatsch> dtz[m]: There is a man page situation?
<dtz[m]> not accessible if you nix-env things -- I think nix-shell usually (?) works but I forget
<Profpatsch> Oh, people still use nix-env?
<dtz[m]> but notably everyone not on NixOS can't use man pages for things installed, which is kinda :(
<Profpatsch> nix-shell works indeed.
<Profpatsch> dtz[m]: But you can also point people to https://github.com/openlab-aux/vuizvui/blob/master/pkgs/profpatsch/nman/nman.go
orivej has quit [Ping timeout: 264 seconds]
orivej has joined #nixos-dev
orivej has quit [Ping timeout: 276 seconds]
<manveru> the nixos cache seems down
<manveru> hm, no
sir_guy_carleton has joined #nixos-dev
pie_ has quit [Remote host closed the connection]
pie_ has joined #nixos-dev
Sonarpulse has quit [Ping timeout: 260 seconds]
<domenkozar> Dezgeg fewliner seems to do the trick
<domenkozar> I hope we can merge staging-next asap
<FRidh> did anyone notice this before with staging-next? https://github.com/NixOS/nixpkgs/commit/6c72782d364149c7cfea1ea69fdbc1133acc80b5
<peti> domenkozar: I drop all "broken" entries from configuration-hackage2nix.yaml, re-generate hackage-packages.nix, and let Hydra build all of it. Then I cut & paste the Hydra page that lists all broken builds and use "sed" to turn it onto appropriate entries for configuration-hackage2nix.yaml. Voila. :-)
<domenkozar> :D
<domenkozar> peti: ok, once flags issue is fixed, we need to do that
<domenkozar> FRidh: what's this? :)
<domenkozar> oh the comment?
<simpson> FRidh: Thanks for the round-trip on #43437; since you did #42603, I was able to finish packaging Cozy, and verified that it works on one of their examples.
<{^_^}> https://github.com/NixOS/nixpkgs/pull/43437 (by MostAwesomeDude, open): Cozy
<{^_^}> https://github.com/NixOS/nixpkgs/issues/42603 (by MostAwesomeDude, closed): Z3 Python bindings should have a Python 3 variant
<FRidh> simpson: glad to hear it works
<FRidh> domenkozar: ah yes, the comment
<domenkozar> FRidh: but good data point, staging should run a few nixos tests.
lopsided98 has quit [Quit: Disconnected]
lopsided98 has joined #nixos-dev
ixxie has quit [Ping timeout: 260 seconds]
<samueldr> I feel dumb for asking, but could some of our Qt PATH issues stem from qtbase.bin not having a /bin/ folder in its output, and the libxqxcb.so file is in the .bin output?
<samueldr> looks like there's a patch to check in PATH for probable inputs, along with checking QT_PLUGIN_PATH
pie_ has quit [Ping timeout: 240 seconds]
pie_ has joined #nixos-dev
<simpson> It's too bad that ofborg doesn't let me play minesweeper or cribbage while waiting for checks~
FRidh has quit [Quit: Konversation terminated!]
lopsided98 has quit [Quit: Disconnected]
lopsided98 has joined #nixos-dev
sir_guy_carleton has quit [Quit: WeeChat 2.0]
pie_ has quit [Ping timeout: 276 seconds]
pie_ has joined #nixos-dev
goibhniu has joined #nixos-dev
sir_guy_carleton has joined #nixos-dev
goibhniu has quit [Ping timeout: 276 seconds]
Lisanna has joined #nixos-dev