Represent real numbers and organize files

Separate range from precision

A floating-point representation stores a mantissa and an exponent. The exponent changes the scale; the mantissa carries significant information. With a fixed number of bits, increasing exponent space usually leaves fewer bits for precision.

A value can be within range but not exactly representable. Rounding creates a small representation error which may accumulate through calculations. Normalization gives a consistent form and makes effective use of the available mantissa bits.

Match storage to access patterns

A record groups related fields. Enumerations restrict a value to named alternatives, while other user-defined types express a domain's structure. Choosing the type is part of modeling the problem, not just a way to shorten declarations.

Sequential processing reads records in order. Direct access finds a location without reading every preceding record. A hash-based organization needs a collision policy because different keys can map to the same location. Consider searches, updates, and storage overhead together.

Check yourself