in reply to precision problem ?

It is probably "more than you care to know", but the full answer with all the mathamatical details can be found in http://docs.oracle.com/cd/E19957-01/806-3568/ncg_goldberg.html. Values for floating point constants applicable to your system can be found in the perl module POSIX. These details could be important if you ever really need the smallest possible epsilon for a given comparison.
Bill