Well... the code's more readable at least!

[wolfwings]

I took things a step further than originally planned, from simply 'removal of bad assumptions' to 'make the code more portable and readable' as well.

This has resulted in me learning something about C that I always took for granted under NASM and other Assemblers. There's no way to define a macro to spit out an unknown-until-invocation number of lines of text.

I.E. As far as I can tell, there's no way to make a macro that I can use to simplify the insanely-convoluted piles of 'case' statements down to 'If THESE bits have THIS value' instead of 'If the value is X' anywhere from 1 to 256 times each. =-.-=

If anyone out there knows a trick to do such under stock C, or at least under GNU C, lemme know? It'll make the remaining code much easier to understand.

Now all I have to do is merge the 2x, 3x, and 4x include files into one file, though I'm also investigating making a series of lookup tables, one for each pixel location. Right now the code-size is a rather insane 100+k in the 4x case, but MUCH of it is redundant code. If I can use jump-tables (NOT a call-table) to speed that code up, things would be much smaller. And if I can get all the code to fit in 16-odd-k along with the data, suddenly things get much nicer for the CPU. =^.^=

And for anyone curious about the new code compared to the old code:

#define Whole(x)	(x) /* Always safe */
#define Half(x)	(((x)&_Mask_)>>1)
#define Quarter(x)	(((x)&(_Mask_&(_Mask_<<1)))>>2)
#define Eighth(x)	(((x)&(_Mask_&(_Mask_<<1)&(_Mask_<<2)))>>3)
#define Sixteenth(x)	(((x)&(_Mask_&(_Mask_<<1)&(_Mask_<<2)&(_Mask_<<3)))>>4)
/* ((c1*1.0000)) */
#define Interp0(c1)	(Whole(c[c1]))
/* ((c1*0.7500)+(c2*0.2500)) */
#define Interp1(c1,c2)	(Whole(c[c1])-Quarter(c[c1])+Quarter(c[c2]))
/* ((c1*0.5000)+(c2*0.2500)+(c3*0.2500)) */
#define Interp2(c1,c2,c3)	(Half(c[c1])+Quarter(c[c2])+Quarter(c[c3]))
/* ((c1*0.8750)+(c2*0.1250)) */
#define Interp3(c1,c2)	(Whole(c[c1])-Eighth(c[c1])+Eighth(c[c2]))

Compare to the original code:

inline void Interp1(unsigned char * pc, int c1, int c2)
{
  *((int*)pc) = (c1*3+c2) >> 2;
}

inline void Interp2(unsigned char * pc, int c1, int c2, int c3)
{
  *((int*)pc) = (c1*2+c2+c3) >> 2;
}

inline void Interp3(unsigned char * pc, int c1, int c2)
{
  //*((int*)pc) = (c1*7+c2)/8;

  *((int*)pc) = ((((c1 & 0x00FF00)*7 + (c2 & 0x00FF00) ) & 0x0007F800) +
                 (((c1 & 0xFF00FF)*7 + (c2 & 0xFF00FF) ) & 0x07F807F8)) >> 3;
}