My theory (which may be bs - people who know the actual facts probably are not authorized to spill the beans) is that not *that* much money was spent, since PySide is reusing binding creation stuff developed for Jambi/QtScript.

It might be interesting to go full circle and generate sip files (instead of Boost::Python wrappers) from the PySide xml api descriptions. This would be a legal "clean room" reimplementation of PyQt...