AF447 crew walked the narrow path through what is completely uncharted territory. We have no way of finding out whether the astoundingly stable behaviour of their 330 at extreme alpha is general rule or whether they have through sheer chance found the power, weight, CG and control input combination that made their aeroplane fall in parachute-like manner instead of departing into violent oscillatory spin. Also there is no point in finding it out through forcing passenger widebodies to AoA way beyond critical one in course of flight testing. It would inevitably lead to destruction of a few of them and idea (which is still valid) is that pilots will properly react to stall warning or at least start recovery as soon as aeroplane truly stalls. Most of them still do.

It depends upon how slow(high AOA) you're at. Add in low vs. high altitude, and the corrective pitch attitude is - whatever it takes, which might be significantly nose low.

Well, we know that the airframe design was intended to have relatively benign stall characteristics - meaning that at the stall boundary there would be a vibration aspect (which, in the AF447 case, may have been misdiagnosed as vibration due to overspeed), followed by a relatively stable "mushing" descent profile. Modern wing designs seem to have these aspects anyway - the Birgenair B757 was similarly stable in aerodynamic stall, but eventually entered a spin due to asymmetric thrust following an airflow-induced engine compressor stall and failure.

Flight testing on modern types involved an unprecedented ability to capture data that could be fed back into the computers, allowing for extrapolation of that data to determine airframe behaviour beyond what would be considered safe in terms of a physical test. Older types were tested up to and beyond the stall boundary, but there was no way to capture the physical data in the same manner, and thus no way to feed that into simulated behaviour.

Yes.
Yes.

No.

Benign stall characteristics regarding the certification testing mean there is pronounced stall buffet, there is natural tendency to pitch down at stall (it doesn't imply it has be such that it can not be overriden by application of controls), there is no violent roll associated with airflow separation and aeroplane can be unstalled using conventional technique. If such characteristics cannot be achieved by natural means, it is allowed to use synthetic stall warning and preventing devices (shaker and pusher). There is no requirement for aeroplane to gently mush if crew insists on keeping AoA high by pulling. As a side note: in West Carribean 708 disaster, captain was so obsessed with keeping the nose up he manual wounded the trim to full nose up position. So much about the evils of autotrim.

Yes.

Correct but could be misleading. New types are tested up to and beyond stall boundary too. For certification they are just pushed beyond the lift limit into stall and promptly recovered, there is no requirement (or reason) to pull them into extreme alpha.

Sure - what I was getting at was that the design should not, for example, suffer an aggressive wing-down tendency once stalled. Your description is more precise than mine.

My information regarding the flight testing regime is a little vague in respect to precisely how far stall testing went, and does not specify whether going beyond the stall boundary was performed or not. If you have better info, it's all good. :ok: