On Sun, 24 Jul 2011 22:56:29 +0200, [UTF-8?]SaaÅ¡ha Metsärantala wrote
> > think about increasingly precise balances...
> These questions are really interesting! There is no doubt about that! They
> may also probably be useful.
> If there are use cases, I would really like to include such precision
Use cases, I think, are a plenty. I think I have over the course of this
list presented quite a few....
> descriptions within the EDTF specification! No doubt about that!
> My point with this e-mail is the following:
> There is no way to express subtle semantics within a syntax which is too
> narrow for it. To express such semantic subtleties unambiguously, we need
But I think the semantics have long been implicitly expressed. Its our job
perhaps to lay them explicitly down in the text for our emerging "standard".
When I ask someone, for example, how much did the object weigh and they
respond "1 gram" what are they saying? They are implicitly saying that the
speciment with a precision of 1 gram--- perhaps, at best, 1/2 gram--- weighs
1 gram. When someone reports "1.200g" they are reporting that with a
precision of mg a sample tips the balance at "1200 mg". The
expression "1.2g" is not the same. Its expressing a measure with the
precision of decagram. While a 1200 mg sample weighs 12 decagrams, a 12
decagram sample might not weight 1200 mg to a precision of mg.
In weights and measures we typically have zeros (0) left of the decimal
point beyond the last place to express higher precision. A number such
as "1.000000" is expressing something beyong the number "1". The problem is
with 10s left of the decimal point. How can one distinguish
between "100" with a precision of 1, "100" with a precision of 10 or "100"
with a precision of 100?
The same "problem" exists with date expressions. We have well defined
expressions of year, month, day, hour, minute, second,.. perhaps even
century precision.. but not decade, whence the generalized "x" syntax.
The "stuff" is all there.. sure we could extend things to allow for a more
general means to express precision.. while I can see a few use cases I
do not think that they are needed just yet.. and would carry the danger of
complicating things.. expecially given the lack of widespread "precision
consciousness" at this time. Many people, I think, that we are addressing
with the standard don't have a background in natural sciences so some of
these basic concepts might, at first, seem stange. Its one step at a time!
What we are still missing then in the standard, I think, are:
- clear expression of what has been implicit:
o that "1994" has year precision, "1994-06" month precision etc..
A number of examples, I think, would be good.. down to highly precise
expressions such as !994-05-12T12:22.33Z" (decimal seconds following the
model I expressed for weights with the limitation to the Caesium
standard for time)
o the consequences of precision.
- how to compare things expressed in different precisions.
> a wider syntax helping people storing EDTF information to be aware of the
> (implied) semantics of the information they store. Other wise, there is a
> risk that some (many) people will store information containing implied
> "precision semantics" they are not aware of and some (many) people will
> read this information with an understanding of the implied "precision
> semantics" different from that of the person who stored this information.
We don't meed a wider syntax.. only to be more explicit and demanding in our
standard's text :-)
Edward C. Zimmermann, NONMONOTONIC LAB
Basis Systeme netzwerk, Munich Ges. des buergerl. Rechts