Back in July I made the rather silly proposal to 'fork' KDE 3 to using Qt4. I made that proposal based on my negative experiences with KDE 4.0, before actually checking out the KDE 3 code and performing even a cursory review of the code base. I eventually performed those actions and came to really understand how complex and convoluted the KDE 3 codebase had grown over the years. And I came to appreciate, as others had warned, how difficult a task it would be.
It was only a short matter of time after making that announcement before KDE 4.1 was released. And with that release have come new releases of distributions that use it. An excellent example of a KDE 4.1 DE/distribution release is, in my opinion, Mandriva 2009. The more I work with the latest version of Mandriva and KDE 4.1.2 the more comfortable I am with the entire distribution. I honestly haven't felt this good about a Linux distribution using KDE since openSUSE 10.2 and KDE 3.
I believe the best course of action now is to work within the KDE 4 environment using Qt 4.4 and later, and add back in those features that are missing or adding new features that would have been difficult to add to KDE 3. Yes, I like the direction that Mandriva and KDE 4.1 is taking. I like it a lot.