orivej has quit [Ping timeout: 240 seconds]
cransom has quit [Ping timeout: 276 seconds]
orivej has joined #nixos-borg
Mic92 has quit [Quit: WeeChat 2.0.1]
jtojnar has quit [Quit: jtojnar]
jtojnar has joined #nixos-borg
cransom has joined #nixos-borg
<LnL> the closure-info assert is because of nix 1.11 right?
<Dezgeg> yes
FRidh has joined #nixos-borg
<LnL> MichaelRaskin: ^
<LnL> I think it was your builder
<MichaelRaskin> I recently forgot to GC, yes, but I definitely do not run Nix-1
<MichaelRaskin> ps shows: /nix/store/2xzhrn49hzcpkg72sgqh6qdx0klckssv-nix-2.0/bin/nix-daemon
<LnL> this is evaluation, so the client
<MichaelRaskin> nix-instantiate (Nix) 2.0
<LnL> https://logs.nix.ci/?key=nixos/nixpkgs.38837&attempt_id=5afb5f4b-23ca-4b9c-a85a-138d3d94fa9b
<MichaelRaskin> (as ofborg)
<MichaelRaskin> Maybe the test refers to Nix1 itself?
<LnL> uh oh
<LnL> no it uses nix from PATH
<LnL> how are you running it?
<MichaelRaskin> just by a script manually started from a separate ofborg user
<MichaelRaskin> nix-shell ../shell.nix -I /home/repos/ --run ' target/debug/builder ../config'"$1"'.json '
<LnL> ah, that's it
<MichaelRaskin> Oh, the shipped shell.nix enforces Nix1.
<LnL> the shell includes "nix" from a pre 2.0 checkout
<LnL> I have a pr for that :)
<MichaelRaskin> Maybe you should merge it…
<MichaelRaskin> After all the mistakes on my side (like /dev/kvm access control), I blindly trusted the code in ofborg/ repo as compatible with ofborg expectations.
<LnL> well, I feel like everything in nixpkgs should still work with 1.11 until 18.09
<MichaelRaskin> Well, Borg could assume latest Nix for some of its own purposes (like details of log handling or timing or something)
<LnL> not sure what you mean, instantiation and realisation would have to be separated to fix this
<LnL> like how hydra handles evaluation
FRidh has quit [Remote host closed the connection]