I'm going to volunteer my 2 centavos to the following question raised by Mr. Optimistic.
there is no reason to think the software was at fault in this instance is there? Necessarily limited perhaps, but isn't it more to do with the presentation of information which is at issue ?
From the events I have seen so far, there is no cause yet to question the performance of the flight control software with relation to it's specification. In other words, it appears to have worked as designed.
One of the problems appears to be that there are a few holes in the design. The concept of "it can't possibly stall" appears to have stopped the specification effort a few clicks short of completion.
The aircraft probably needs a manual lateral trim control switch that is active only in roll direct modes. Lets see what BEA says about that!