X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~mdw/git/mup/blobdiff_plain/cdb3c0882392596f814cf939cbfbd38adc6f2bfe..ddf6330b56bcfb657e0186b24b9b1422c51d3424:/mup/mup/rational.h

diff --git a/mup/mup/rational.h b/mup/mup/rational.h
new file mode 100644
index 0000000..8530dca
--- /dev/null
+++ b/mup/mup/rational.h
@@ -0,0 +1,108 @@
+/* Copyright (c) 1995, 1996, 1997, 2001 by Arkkra Enterprises */
+/* All rights reserved */
+/*
+ *	rational.h	definitions for using rational numbers
+ */
+#ifndef _RATIONAL
+#define _RATIONAL
+
+
+/*
+ * The code in rational.h/rational.c is written in a way that depends on
+ * certain variables being 32-bit numbers stored in two's complement form.  For
+ * most compilers, "long" works for this, but for some compilers, "long" is
+ * 64 bits, or maybe something else.  C standards after (and not including) the
+ * 1994 version guarantee the existence of a header file stdint.h which defines
+ * typedefs int32_t and uint32_t.  So for those standards we will use those
+ * typedefs from that header when defining our INT32B and UINT32B.  Otherwise
+ * we will just use long and unsigned long.  If these aren't 32-bit, two's
+ * complement, you will have to change this code.
+ */
+#if __STDC_VERSION__ > 199409L
+#include <stdint.h>
+#define	INT32B	int32_t
+#define	UINT32B	uint32_t
+#else
+#define	INT32B	long
+#define	UINT32B	unsigned long
+#endif
+
+/*
+ * Define the structure that represents a rational number.  The number
+ * should be kept in standard form; that is, in lowest terms with a positive
+ * denominator.  Zero should be stored as 0/1.
+ */
+struct rational {
+	INT32B n;	/* numerator */
+	INT32B d;	/* denominator */
+};
+
+typedef struct rational RATIONAL;
+
+
+/* macros for comparing rational numbers */
+/* numbers must be in standard form */
+
+#define EQ(a, b)	((a).n == (b).n && (a).d == (b).d)
+#define NE(a, b)	((a).n != (b).n || (a).d != (b).d)
+
+/* must call function for inequalities, because overflow possibilities */
+
+#define GT(a, b)	gtrat(a, b)
+#define LT(a, b)	gtrat(b, a)
+#define GE(a, b)	( ! gtrat(b, a) )
+#define LE(a, b)	( ! gtrat(a, b) )
+
+
+/* macros for testing a rational number */
+/* numbers must be in standard form, or at least have positive denominator */
+
+#define ZE(a)		((a).n == 0)
+#define PL(a)		((a).n >  0)
+#define MI(a)		((a).n <  0)
+#define NZ(a)		((a).n != 0)
+#define NM(a)		((a).n >= 0)
+#define NP(a)		((a).n <= 0)
+
+
+/* definitions for error handling */
+
+#define	RATNOERR	0
+#define RATOVER		1
+#define RATDIV0		2
+#define RATPARM		3
+
+#ifndef MAXLONG
+#define MAXLONG		0x7fffffff
+#endif
+
+extern int raterrno;			/* holds error code */
+#ifdef __STDC__
+extern void (*raterrfuncp)(int);	/* pointer to user error handler */
+#else
+extern void (*raterrfuncp)();		/* pointer to user error handler */
+#endif
+
+
+/* functions to do operations on rational numbers */
+
+#ifdef __STDC__
+extern RATIONAL radd(RATIONAL x, RATIONAL y);
+extern RATIONAL rsub(RATIONAL x, RATIONAL y);
+extern RATIONAL rmul(RATIONAL x, RATIONAL y);
+extern RATIONAL rdiv(RATIONAL x, RATIONAL y);
+extern RATIONAL rneg(RATIONAL x);
+extern RATIONAL rinv(RATIONAL x);
+extern RATIONAL rrai(RATIONAL x, int n);
+extern void rred(RATIONAL *ap);
+extern char *ator(RATIONAL *rp, char s[]);
+extern char *rtoa(char s[], RATIONAL *rp);
+extern int gtrat(RATIONAL x, RATIONAL y);
+#else
+extern RATIONAL radd(), rsub(), rmul(), rdiv(), rneg(), rinv(), rrai();
+extern void rred();
+extern char *ator(), *rtoa();
+extern int gtrat();
+#endif
+
+#endif