depending on how long before doubles are implemented in PB, you could write a preprocessor that would translate a pb program that's using double precision math into inline asm functions and then compile the translated program.
for pure fun (got to love that word ) to see if they did any OpenGL programming and I found converted header-files for OGL. In that header file (gl.inc) they declared datatypes like this:
So, if one use inline asm. that should be possible right? And therefore you should be able to use commands like glOrtho() in OGL that require double.
Just a thought...
Well, we have already the datatype double defined with the F64 lib or another DoubleLib.