[GRASS5] Raster_int()

[GRASS]

Hi developers,

I have a question about Raster() functions.

src/display/devices/lib/Raster_chr.c:
-----
int
Raster_char (int num, int nrows, unsigned char *array, int withzeros, int
color_type)
{
	...
		i = num ;
		while(i--)
			*(iptr++) = *(cptr++) ;
	...
}
-----

in the above code segment, we can know that 'num' means the length of 'array'.
However,


src/display/devices/XDRIVER/XDRIVER24/Raster.c:
-----
int 
Raster_int(int num, int nrows, int *array, int withzeros, int color_type)
{
	...

        for (j = 0; j < num; j++) {
            if (*arr == 0) {
                if (width > 0) {
                    for (i = 0; i < nrows; i++) {
                        XPutImage(dpy, bkupmap, gc, grimage, 0, 0,
                                  cur_x + start_col, cur_y + i, width, 1);
                    }
                    width = 0;
                    start_col = j + 1;
                } else {
                    start_col++;
                }
            } else {            /* non-zero pixel, put into the image */
                if (get_table_type() == FLOAT) {
                    if ( color_type )
                        XPutPixel(grimage, width++, 0,
                                  (u_long) _get_color_index(*arr));
                    else
                        XPutPixel(grimage, width++, 0,
                                  (u_long) *arr);
                } else if (use_visual->class >= TrueColor) {
		...
		}
	    }
	    arr++;
	}
	...
}
-----

in XPutPixel() calls, we always put pixels in the same row 0.
And if some zero values found, XPutImage() duplicates this row 0 into row
1-(nrows-1).

Supposing this case:

/* image

   00100
   01010
   10001
*/
int *array = { 0, 0, 1, 0, 0,
               0, 1, 0, 1, 0,
               1, 0, 0, 0, 1};

CASE A)
'num' is ncols.
Raster_int(5, 3, array, 0, 1);	/* num is width */

produces the following image:

		00100
		00100
		00100

CASE B)
'num' is the number of array elements.
Now,
Raster_int(15, 3, array, 0, 1);	/* num is the number of elements */

produces:

		00100 01010 10001
		00100 01010 10001
		00100 01010 10001


CASE C)
So, we should use Raster_int() like the followings:

	for(i=0; i<3; i++){
		R_move_abs(x, y+i);
		R_raster(5, 1, 0, array+5*i);
			/* this func calls Raster_int(5, 1, array+5*i, 0, 1); */
	}

produces:
		00100
		01010
		10001

'nrows' should be always '1' to get a correct result. Why is this needed?

Is this intended?
It's so inconvenient and not easy to predict results.


All modules may be working on 'CASE C)'. I think that this behaviour might
be related to many G_* functions such as G_get_raster_row() which manipulate
raster maps row by row.  G_get_raster_row() function reads raster map row
by row. Then we draw this row on a monitor. Then read again, ...
So we have not experienced any trouble drawing raster data.

I think that 'num' should be the number of colums and CASE A) should work
correctly.

Any good idea?


Regards,
Huidae Cho