Yes, there's been a few threads on this topic in the last couple of years. Most have dealt with it as CANPA, or some other term that essentially means the same thing. Some of these threads have horrified me, so I hope that the reference quoted by alf5071h sinks in with people.
The fact is that a NPA procedure is likely to always be a manoeuve based on a series of minimum altitudes all the way to the MDA. Therefore, how the pilot elects to execute the approach is only optional to a certain extent. The quoted reference is saying that you still cannot bust any of the procedure altitudes, especially not the MDA.
So, for the benefit of anyone who still thinks that they can fly CANPA all the way to MDA and THEN initiate a missed approach... you're playing with fire and, sooner or later, you WILL get burned. The problem with this is that it unnecessarily spoils the day for the poor bugga who designed the procedure. He (or she) still has to front the Magistrate to explain that the design of the procedure complies with Pans Ops (or TERPS, as applicable) criteria.
Even after the designer is exonerated - months or YEARS later - that person has, understandably, a crisis of under confidence. And it's all because some idiot thought he knew better.
I'm not pointing the bone at anyone here, merely trying to give the procedure designer's perspective. Pilots need good approach procedures and, therefore, you need procedure designers who not only know and understand the criteria, but have the confidence to apply it in the trickiest of situations. After all, anyone can design a procedure to a place when there's no obstacles to worry about.
Pans Ops provides for NPA procedures that employ a glide slope and, therefore, also account for "sink thru" at MDA. If you're using one of these procedures, that's fine. If not, please don't expect to be able to safely "sink thru" a MDA on any other type of NPA.
Sorry if this sounds a bit strong in tone, but I've stayed silent on similar threads when I probably should've piped up.