It could make sense to allow tests for failing statuses: that some side-effects have already happened even when an error appears (i.e. writing some log files), that no incomplete modifications to the file system happened, non-applicable modifications are rolled-back, or that the failure of the command appears with a specific execution output.
It could make sense to allow tests for failing statuses: that some side-effects have already happened even when an error appears (i.e. writing some log files), that no incomplete modifications to the file system happened, non-applicable modifications are rolled-back, or that the failure of the command appears with a specific execution output.