On Fri, 9 Sep 2011 12:42:33 +0200, [UTF-8?]SaaÅ¡ha Metsärantala wrote
> > integers with a bunch of xxx(s)
> I am not even sure that the "x"-notation ALWAYS would be the semantically
> most accurate alternate to exponential notation. If we choose to keep the
> exponential notation in EDTF, I would suggest to clarify in the
> specification which semantics are (or are not) associated with the
> exponential notation.
In scientific notation its customary to record all the significant digits in
the number. One often adds one more place as an estimation.
There is also an extended scientific notation that includes information on
the precision of this "estimated" digit.
This "estimation" digit is useful but conceptualy more demanding...
I think we would need to hammer this down in the standard...
Last digit always an estimate or not?
> > 6.55e7 versus 655xxxxx
> I had previously suggested to avoid dots and thought we agreed about that.
> My aim is to make it easier to implement an exponential notation avoiding
> constructs such as:
> which would mean "year 1234.5" and therefore is a bit problematic.
OK.. I think avoiding decimal places makes sense (especially since they can
be dots or commas).. It gets rid of redundant expressions.. 655e5 instead of
65.5e6 (or 65,5e6 as it would be written in Germany) or 6.55e7, 0.655e8 etc..
It also makes it immediately clear what the implied precision is..
nnnemm is always emm precision, e.g. 192e1 is e1 or decade precision--- or
if we consider the final digit as an estimation then century precision.
Edward C. Zimmermann, NONMONOTONIC LAB