madjar
False-positives for haskellPackages
The patch from https://github.com/NixOS/nixpkgs/pull/7288 was reported as testing successful, but in fact no test build was ever run -- and if it had been, it would have failed. I suppose the same problem exists for rPackages and everything else that doesn't use recurseIntoAttrset.
Originally reported at https://github.com/NixOS/nixpkgs/issues/7304.
Thanks for pinging me !
I have to confess I'm not sure how to tackle this one. It is indeed a problem with recurseIntoAttrset: nox basically does two nix-env -qa to determine which derivation have changed based on their hash.
In that situation, I don't know how to make nox figure out which packages to build.
As for a workaround, there are legitimate cases where nothing changed and all is well, so I can't make it a crash. Maybe travis should be yellow or grey in that situation?
