Issues when porting:

    This is an estimation of the porting issues; I haven't
    tried porting it, so there may be more.

SIZEOF:

    The code is rife with these assumptions:

        sizeof(int) == sizeof(long) == 4
        sizeof(short) == 2

    You'll have to fix a lot of stuff, including the bsp file
    formats in bspfile.h, if this isn't true.

    The "VERTEX" macro assumes that sizeof(dvertex_t) == 12.

    render.c has a macro which works if IEEE floating point,
    sizeof(float) == sizeof(int).

ENDIANNESS:

    There are several little-endian specific pieces of code:
       FLOAT_TO_INT()
       FLOAT_TO_FIX()

    You also have to enable the byteswapping code in bspfile.c

OS specific:

    Obviously all of the I/O things have to be worked on.
    The "I" is all in main.c, and the "O" is all in mode.c.
    Support for > 8-bit displays will require changing lots
    of code, since the surface cache would want to output
    16-bit...
