Using that approach I am able to build openconnect for linux with:
nix-build --option build-use-sandbox true --argstr system x86_64-linux -A openconnect
If I try the same using:
nix-shell -p nix-review nix-review pr --build-args="--builders ssh://root@nix-docker --argstr system x86_64-linux" 68780
nix-review just outputs:
No packages were successfully build, skip nix-shell.
Why is that the case? Is
nix-review automatically reading my OS settings and thus uses a darwin stdenv when running
nix-shell -p nix-review? Consequently it would then just not build anything in the second step because there is a mismatch between the host OS (darwin) and the derivation’s target OS (linux).
Regarding my approach: Is using the
--option build-use-sandbox true really taking any effect? I ask this because inside the nix-docker container’s nix.conf file the sandbox option is set to false. If I set this to true the build fails with a cloning permission error.
In general, what would be the best practice to review a linux pull request on darwin?