[Bf-committers] Boolean tool precision

[Ken Hughes]

Thanks for the input:

Ed:  I looked at the papers on comparing floating point numbers and they 
bring up some good issues.  I don't know how portable some of the 
"optimized" code techniques would be considering little/big endian 
issues, but I'll keep the concepts in mind as I look through the rest of 
the code.

Chris:  I'll put some #ifdefs in the code to make it easier to switch my 
float comparisons off, but I'd prefer to make my the more accurate 
comparisons the default so we can at least get some testing feedback.  
I've been looking at the gprof profiling results to see if I can 
optimize elsewhere to at least offset the speed loss; I've already found 
a few things which I'll include in the commit later this week (hopefully).

Since the code is C++ and I have a passing understanding of what's going 
on (although I  have programmed in C++, I haven't done so extensively) I 
will be poking some people on IRC for input for some of the 
optimizations to make sure I don't make matters worse. 

Ken