Decimal Testcases, version 2.09
Copyright (c) IBM Corporation, 2002. All rights reserved. ©
3 Nov 2002
[contents | next]


This document describes testcases designed for testing implementations of the general purpose floating-point arithmetic defined in the General Decimal Arithmetic Specification.[1] 

These testcases are intended to be both language-independent and representation-independent. They comprise individually identified tests, each describing a single operation and its expected results. The tests are grouped into files to make it easier to test a new implementation incrementally, and are available at:
The testcase files should be considered experimental (or ‘beta’), and may contain errors. They are offered on an as-is basis. In particular, even passing all the tests does not guarantee that an implementation complies with any Standard or specification, because the tests are not exhaustive.

Comments on this draft and the testcases are encouraged. Please send any comments, suggestions, and corrections to the author, Mike Cowlishaw (

For further background details, including specifications in various formats and related decimal arithmetic links, please see the material at the associated web site:

Appendix A summarizes the changes to this document since the first public draft.


The author is indebted to Aahz and Tim Peters for their contributions to this document. In addition, many people have contributed directly or indirectly to the testcases themselves; special thanks are due to Brian Marks for his meticulous investigations into different implementations of the base arithmetic.
[1] See

[contents | next]