Scientific Precision

As is the practice in most experimental science, the framework pays attention to the number of digits in a floating point number. Comparisons of expected and actual results are made to a tolerance implied by the number of digits.

The math I had above is bogus. I've taken a new approach. I now tweak the character representation to discover an appropriate tolerance. -- WardCunningham

See the source.

Generated data for the CalculatorExample sometimes fails in the last digit. This may be due to unusual rounding in perl, or maybe this formula isn't right. (It is still way more convenient than specifying an absolute tolerance.)

 
Last edited November 4, 2002
Return to WelcomeVisitors