Base datatypes in eml-attribute BoundsGroup preclude scientific notation
The 'minimum' and 'maximum' elements defined in the eml-attribute BoundsGroup
are currently based on the xml-schema data type 'xs:decimal'; however, this data
type does not support scientific notation so metadata creators are forced to
encode very large (or small) bounds in expanded decimal notation. This
represents an unnecessary burden and could introduce problems if authors are
required to display more digits than are known to be significant. Changing the
base data type to 'xs:float' would accomodate both decimal and scientific
notations so would not introduce backward compatibility issues.
Updated by Margaret O'Brien over 14 years ago
The advantage of using xs:decimal is that it's precise, and xs:float is not (and I presume, the original intent in making this choice?). We could switch this type now to ease the burden on document-authors, but it seems that float is not really the best representation of these values, and maybe we should not suggest so.
There might be another alternative forthcoming -- XML Schema 1.1 (still a working draft) is planned to have another data type "precisionDecimal" which is corresponds to a IEEE type, "floating-point decimal" with features of both - ie, the values and notation of float, but treated as decimal in arithmetic and storage
What should we do? relax the type now to xs:float, and consider xs:precisionDecimal when it is available? or just wait?