on Mon Jul 14 2008, Daryle Walker <darylew-AT-hotmail.com> wrote:
Thanks for diagnosing this. Please put a ticket about this at http://svn.boost.org. If you could check the boost trunk (using the Trac browser it should be easy) first to make sure that it's not fixed, that'd be a bonus.
When you suggested to check the trunk, I wondered how could I check if I couldn't find the code before. Then that and the floating point comparison policy reminded me to check how I handled this for the Boost.Rational tests. When I checked, I handled the problem the same way. I guess this isn't a bug; I forgot that this is a design decision, just like the floating point comparison case. Sorry.
Of course, one could strongly argue that this design decision is a bug itself....
Then open a ticket about that :-) -- Dave Abrahams BoostPro Computing http://www.boostpro.com