home *** CD-ROM | disk | FTP | other *** search
- /*
- * Code to scale dimensions. Takes two thirty-two bit integers, multiplies
- * them, divides them by 2^20, and returns the thirty-two bit result.
- * The first integer, the width in FIXes, can lie between -2^24 and 2^24-1.
- * The second integer, the scale factor, can lie between 0 and 2^27-1.
- *
- * Here, unlike in TeX, we do the arithmetic exactly. Since the error
- * in the TeX arithmetic is parts per million, and since dvips makes no
- * layout decisions, this has no effect. (TeX stipulates that any
- * implementation of *TeX* needs to do the arithmetic exactly as
- * specified in TeX, but drivers need not.)
- *
- * Since this math is special, we put it in its own file. It is the only
- * place in the program where such accuracy is required.
- */
- #include "dvips.h" /* The copyright notice in that file is included too! */
-
- integer
- scalewidth(a, b)
- register integer a, b ;
- {
- register integer al, bl ;
-
- if (a < 0)
- return -scalewidth(-a, b) ;
- if (b < 0)
- return -scalewidth(a, -b) ;
- al = a & 32767 ;
- bl = b & 32767 ;
- a >>= 15 ;
- b >>= 15 ;
- return ( ((al*bl/32768) + a*bl+al*b)/32 + a*b*1024) ;
- }
-