14 Apr
2008
14 Apr
'08
5:50 p.m.
AMDG Schrom, Brian T wrote:
Steven,
A couple of general comments:
1. I would appreciate having this type of error analysis in boost, so I encourage development in this area 2. I would like to have "pluggable" methods of error tracking. For example, most of the time I want partial derivatives with the dependencies tracked, but sometimes I want to know the absolute error of a calculation.
I definitely agree. The code I posted originally was just for proof of concept.
3. I've used the units library quite a bit since it was accepted and made my own class for handling errors, but a nice integration of these two ideas would be nice to see in boost too.
Hopefully they should work together without extra effort... In Christ, Steven Watanabe