Hi, Ludovic + Ricardo! Ricardo Wurmus writes: > Building and running virtual machines as part of the tests > seems expensive. Would it be feasible to mock the remote > interactions? I agree 100%. I've decoupled it from my patch series for now. We can always add it back later when it's implemented in a less expensive way. As for mocking -- I do like that idea, but that would only really be testing that calls to 'deploy-machine' et al. don't fail rather than ensuring that the implementation of 'guix deploy' does what it's supposed to do. The current tests make assertions about changes to the virtual machine. Ludovic Courtès writes: > As Ricardo wrote, I think that’s too much work to do in “make check”. > Plus this would only run when a “host store” is available, as we can’t > reasonably build QEMU and everything in $builddir/test-tmp. > > So I feel that the system test suite is a better fit, but I don’t > fully understand the limitations you hit, Jakob. > > Do you still have a draft of a system test that you wrote and/or notes > about what went wrong? Yep, I have an unsquashed commit history on my personal branch with all renditions of the test suite. I can pull it out tomorrow and write a detailed report on the issues I ran into. Thanks for both of your comments!