> that would cause lots of compiler
> warnings and implicit truncation on UCS2 builds

Unfortunately, there is no such warning, or the initial problem we are trying 
to solve would have been spotted by such a warning (unicode_repr() calls 
Py_UNICODE_ISPRINTABLE() with a Py_UCS4 argument).

gcc has a -Wconversion flag, (which I tried today on python) but it is far too 
verbose before version 4.3, and this newer version still has some false 
positives. http://gcc.gnu.org/wiki/NewWconversion

But the most important thing is that implicit truncation on UCS2 builds is what 
happens already! The patch does not solve it, but at least it yields sensible 
results to wary code.
Or can you imagine some (somewhat working) code which behavior will be worse 
after the change?