Err, what you say in your last post is quite against the logic of updating the preview image to reflect the edits. And in fact would be quite impossible to do properly in case of e.g. heavy transform, where it would even be possible to get more pixels than you had before (how could they keep the image the same size in this case?). And how about pano merges?
No, the fact that the preview IS the image after all edits, transforms, crops, etc. have been applied is exactly what you want! That way smart software (like PM) can make use of that to show what the image looks like, including the edits. It would completely break my workflow (and render PM useless to me) if this wasn’t the case...
It was a prime reason for me to switch to a DNG-based workflow.