(with no specific relationships between)
SDL Perl now has an improved build system, in the sense that it's trivial to write out SDL::Config. That module can answer the question "Which subsystems are available in this build?" That means that the tests for SDL::TTFont can bail out if SDL_ttf isn't installed.
Speaking of SDL, I've been chasing down some thread-related mishaps in Parrot recently. I've been seeing strange hangs at various points. gdb backtraces make me think it's thread-related, as my post to p6i shows some pthreads mishaps. There's a non-trivial possibility that my version of glibc is weird, though.
I saw a similar situation with MySQL a couple of months ago. That one's still unresolved. Part of the problem may be that architecture assumptions of Linux on x86 processors don't all translate well to Linux on PPC processors.
Any suggestions are welcome.
Finally, someone really ought to write bindings for ODE, the Open Dynamics Engine. Combine that with SDL and use a nice dynamic language as the glue and you'll really have something.