Using TOD (Time Of Day) formats for durations you need to also consider that these formats are specified to round down (as opposed to "common rounding"). (In this case combined with a sexagesimal system). The floatingpoint processor charged with the subtraction uses IEEE 754 'Double' dyadic format which mostly cannot exactly represent fractional numbers seeming "precise" when represented in text form based on decimal notation. Experiences as mentioned are mostly effects caused by the well known "cancellation of leading digits" (German: "Numerische Auslöschung") and automatic rounding basically unavoidable in machine arithmetic.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |