XSD could make me very happy

You can tell from my recent posts that I have a mixed relationship with XSD. Yesterday's post explained my approach to XSD design, which grew out of my investigation of RNG, and which I'm pretty happy with. (I've been criticizin XSD a lot lately, but I have to give it credit for supporting a wide range of alternative approaches, including the one the I like). The one big issue that remains for me is versioning, which I've mentioned before. Noah Mendelsohn posted this message to the W3C TAG about potential changes to XSD to facilitate versioning by altering the unique particle attribution rule so that wildcards are weaker than element declarations, which take precedence. This would be a huge step and is exactly what I want. He goes further, adding the notion of declaring a default implicit extensibility model that injects wildcards at the end of or in between each particle in a complex content model. That would be a very cool feature too. If you are interested in the XSD versioning problem, read his email. For me, this change can't come fast enough, at the W3C and in System.Xml. Given this and the XSD style I talked about yesterday, I would probably give up on RNG (which isn't that hard, since people don't think it'll happen ;-).