Compaq Portable Mathematics Library

This document, dated April 2001, provides reference and exception information for the Compaq Portable Mathematics Library (CPML), version 7.1. CPML is a set of mathematical routines designed for high-level languages like Fortran and C, aiming for portability and consistency across various operating systems (e.g., OpenVMS Alpha, Compaq Tru64 UNIX Alpha) and hardware architectures.

Key aspects covered include:

• Mathematical Functions: A wide array of functions covering floating-point trigonometric, exponentiation, logarithmic, power, hyperbolic, algebraic, and complex evaluations.
• Data Types: Support for integer, various floating-point formats (VAX FFLOAT, GFLOAT; IEEE SFLOAT, TFLOAT, X_FLOAT), and complex floating-point types.
• Exception Handling: Detailed information on how the library handles "exceptional arguments" (mathematically undefined, out of range, or causing overflow/underflow). It defines specific exception conditions (overflow, underflow, invalid argument) and their default actions (e.g., trap, continue quietly) and return values (e.g., HUGERESULT, INVRESULT).
• Standards Conformance: Discussion of IEEE Std 754 considerations (e.g., NaN, infinity) and a list of routines conforming to X/Open Portability Guide (XPG4) standards.
• Routine Descriptions: Each CPML routine is documented with its interface, description, and specific exception behaviors.
• Technical Details: Appendices include critical floating-point boundary values and platform-specific entry-point names for direct calls.

The document is intended for compiler writers, system programmers, and application programmers who use CPML routines.