Safe assumptions

• char is guaranteed to hold only 0 to 127, and can be either unsigned or signedchar; you cannot assume one or the other. Avoid char unless you do not care about sign extension.
• unsignedchar can hold from 0 to 255; it can hold more.
• signedchar can hold from -127 to +127; it can hold more.
• short can hold from -32,767 to 32,767 (signed) or 0 to 65,535 (unsigned).
• int can hold from -32,767 to 32,767 (signed) or 0 to 65,535 (unsigned). Ints cannot be counted on to hold any more than a short. If you need something larger than a short, use a long. If a short is big enough, use int instead to improve efficiency by taking advantage of a processor's natural word size.

Always read and write an int by casting it to a short, because that's all you're guaranteed to get.
The Taligent stream classes do not read or write int because int doesn't have a portable representation.

• long can hold from -2,147,483,647 to 2,147,483,647 (signed) or 0 to 4,294,967,295 (unsigned).
• float is a IEEE single-precision number and double is a IEEE double-precision number. This is because the Taligent Application Environment runs only on processors that support the IEEE floating point standard and that support the single- and double-precision types.

In general, watch your assumptions carefully, and use typedefs instead of C types. For examples, see "Avoid raw C types with dimensions" on page 66.