Fix the assembler's floating point number parser so that it can correctly handle...