Thanks John.Bear in mind that I am not, and never have been, an electrician but, for what it's worth, that is what I would usually do, at least in the first instance. Indeed, if it were a sockets circuit and I was absolutely certain that no loads were connected to the circuit, I might well test the RCD 'from a socket' (with the MCB 'on'). In either of those cases, I think the only situation in which one would get misleading results would be if there were an N-E fault on the circuit - and one should have detected that (by IR testing) long before one got to testing the RCD.
Some real electricians may well disagree with such practices.
Again, for what the view of a non-electrician may be worth, I do all I can to minimise the number of things that are disconnected and then re-connected in the name of testing - since every time one does that one invokes a finite (hopefully very small!) risk that the 'testing' may create new faults/problems!
Kind Regards, John
For completeness, can I also understand how you prefer testing RCBO's? Based on the above, do you disconnect the loads (L&N) from RCBO before testing?