Speaking practically, a judgment from the United States, which is a bit dated, states that there should be a legal framework of rules for things like obtaining approval for the conditions precedent.
This ties in with what Ms. Ouimet said when she referred to mandatory testing in specific police actions. So it would not really be random; it would be mandatory, but under conditions approved in advance—by a higher police authority, for example, as opposed to an officer in the field. Ideally, the approval would be given based on evidence that links the demand to circumstances where a certain incidence of impaired driving has been observed. So it would have to be tied to, say, statistical evidence.
The new Criminal Code provision, as currently presented, speaks of random testing, and could go as far as giving an officer complete discretion. I don't think that's ideal. So I recommend that you consider specifying conditions, or a framework, that would tie the requirement to scientific evidence, based on European models, or on that court decision from the United States.