Appendix D to Goldberg paper

[David Hough validgh business]

Doug Priest has kindly permitted me to post an appendix D he wrote to
clarify some points in David Goldberg's classic Computing Surveys paper,
What Every Computer Scientist Should Know About Floating-Point Arithmetic.
You can review the appendix or the paper at http://www.validgh.com.

For several years,
Sun has reprinted the paper as part of its Numerical Computation Guide
accompanying unbundled compiler products.

The appendix reviews issues that have been raised again in the Java on x86
discussion, and suggests why some of us have come around to the point of
view that IEEE 754 granted more liberty to compiler implementers than most of
them could use responsibly - and thus why it doesn't bother us that Java tries
a different approach.

And relevant to that, persons interested in Java compilers and interpreters
may find interesting: 

http://www.sun.com/workshop/java/jit/test_suite/index.html.