I think it will be a very long time before most hosting providers upgrade to PHP 5.3, because serious incompatibles with previously installed software will be a nightmare. I run my own servers and mostly serve my own websites, but even so, upgrading to PHP 5.3 on any of my servers would be sure to take down several active websites and would have difficult-to-anticipate consequences for many other sites. I wrote the code used by most of those sites, but the labor involved to read through tens-of-thousands of lines of code looking for and correcting potential problems is all out of proportion to the advantages PHP 5.3 provides.
The situation is even worse for most public hosting providers, because it is common for them to have hundreds or sometimes even thousands of customers on single servers. A single configuration change in situations like that can, and often does, cause enormous messes and angry customers.
I learned many years go about the importance of naming classes with unique prefixes. It not only greatly reduces the risks of name conflicts, but makes my own names quickly distinguishable even in code I wrote long ago and have almost complete forgotten about. I understand your issues with backward compatibilities, but the sooner the underlying problem is fixed, the fewer old third-party products and users there will be to worry about. It is easy for me to say this as an outsider who doesn't have to deal with the problems, but if I headed your development team, my inclination would be to rename all the classes and deal once with all the resulting backward compatibilities, rather than to change names here and there as necessary to "put out fires."
