integer sizes

[mwm at ucbtopaz.CC.Berkeley.ARPA]

In article <8871 at brl-tgr.ARPA> gwyn at Brl-Vld.ARPA (VLD/VMB) writes:
>ANSI C will guarantee minimum sizes for integer types.
>A short or an int will have 16 bits minimum and a long
>will have 32 bits minimum.  For other sizes, round up.
>For more than 32 bits, find another way.

You mean, if I've got a program that needs values around 5 billion, and a
compiler/machine that has 60 bit longs, I shouldn't use C????

Likewise, if I've got a machine that has 18 bit shorts and (expensive!)  36
bit longs, and I need values around 100,000, I should use the longs?

Nuts to that. I'll use "uint60" and "int17", and give you a copy of the
include file that defines them. If you refuse to fix the include file, the
code will die. But that's your problem, not mine.

	<mike

Here's a the first piece of that include file, just to show you how easy
this is to do:

/*
 * Size definition for a generic byte oriented machine. Compiler needs to
 * understand "signed/unsigned byte", and "unsigned short/long" types.
 */

/* signed ints, size must have extra bit for sign */

typedef signed char	int1 ;
typedef signed char	int2 ;
typedef signed char	int3 ;
typedef signed char	int4 ;
typedef signed char	int5 ;
typedef signed char	int6 ;
typedef signed char	int7 ;
typedef short		int8 ;
typedef short		int9 ;
typedef short		int10 ;
typedef short		int11 ;
typedef short		int12 ;
typedef short		int13 ;
typedef short		int14 ;
typedef short		int15 ;
typedef long		int16 ;
typedef long		int17 ;