Matlab's Loss is Nobody's Gain

[David G Hough at validgh]

I am grateful to Stuart McDonald for pointing out this paper, a recent 
addition to the website of Prof. W. Kahan, and interesting for re-raising
many of the issues of the Java numerics discussion in the context of a 
completely different interpreted language.    The URL is

http://www.cs.berkeley.edu/~wkahan/MxMulEps.pdf

A recent Java paper is at 

http://www.cs.berkeley.edu/~wkahan/JAVAhurt.pdf