Rebuild the entire graphic set of applications in pclos? I hope this is glaring obvious here, but you guys have a lot of big things that need to be done that don't look to promising for the out look of pclos. I don't know if doing the x11 packages would be first on my to-do list, but you have some huge hurdles no matter what order you do it.
The toolchain: If you guys are going to do this, then I doubt a buildbot is going to help in any way. There is tons of patching needed to be done for gcc 4.3 and higher. I believe the newer binutils is required for newer kernels as well which is needed for much of the functionality in other core packages.
rpm 4.4.6 (Jbj's precursor to rpm5): It still works fine, and is pretty much the last version of rpm to work well with apt-rpm, but the older rpm and apt-rpm clearly have performance issues. The longer this is put off, the harder it will be to take changes and specs from upstream Mandriva. For the packages that there is no local expertise in the dev group, this makes the work very difficult.
python: The 2.5 series is pretty dated and if you touch this then everything python has to be redone and then some, but if you don't do this, your bound to run into compatibility problems where newer releases of packages no longer really do legacy testing.
The x11 set of pkg: If you start on this and you do successfully rebuild the some 8,000 packages in the pclos repo, then you just wasted a huge opportunity to at least do something about 2 of the previous 3.
So maybe you say, let's take the easy route and rebase (which is probably the best idea), but apt-rpm goes out the door as an option :-( and with that goes synaptic too. But if you do start in on rebuilding all the packages, you are basically forking pclos and rebuilding it from a lower level in the OS.