I must admit, I only skimmed the article before posting the link. Having re-read the article, I agree, it's very thin on useful content, and doesn't give any insight into what MV actually have done.
The article seems to be a slight rehash of MV's own announcement here:
http://www.mvista.com/news/2005/realtime.htmlChanging the subject slightly, I read a long time ago how WinCE came about. Many people think they started with the Windows source code and stripped it down to make a new OS. In actual fact, they started from scratch, used the smallest subset of the Windows API that they could and still be useful, and only added new parts of WinAPI when people could persuade them. Unfortunately, this just led to more and more bloat in the full pocketPC version.. but apparently the minimal base WinCE OS and library is actually quite good (according to friends of mine who've been hardcore WinCE developers).
It's interesting to contrast this with the embedded linux projects which try and strip linux kernel down to make it lighter and leaner. Or with PalmOS which didn't have any legacy to support, but increasingly struggles to keep up with the highly directed monoculture of WinCE/PocketPC software and hardware.