What I don't really get from that thread is its purpose and motivations.
I am not feeling comfortable with the implicit and explicit extension to a whole amplifier design class.
@pma shown a use case (theoretical or real, that is still to determine) where a given specific class D implementation is not behaving well with some specific load testing. Fine with that.
Now, everybody jump and get crazy about it.
First, we still not have any evidence that this is a real issue in real life with real speakers.
Second, if proven that for a very specific category of speakers that represents an extremely small pourcentage of sold speakers, this specific amplifier has an issue, fine, future buyers are warmed.
Third, let's not generalize from one set of specific tests on a specific implementation to conclude that class D is bad (I am refering to some posts from the thread).
I don't really care in the end about class D.
I just want people to not be fooled by a misleading conclusion resulting from these specifics tests.