Code review comment for lp:~raof/mir/probe-client-drivers

> I'm not entirely sure what you think is fragile about executable_path()? We're
> already using it elsewhere in the tests.

Having looked again I think I misunderstood how this solution works. It isn't as bad as I thought.