Re: Temporal database - no end date

Marshall wrote:
> So? You've admitted we can speak meaningfully about 2.3 children,
> and children are indivisible. So what is the issue with speaking of
> 2.3 chronons?

   The fact that, by definition, there is no such concept at 2.3 chronons. Chronons (or 'time quanta') are "indivisible". I'm not stating that as a claim. I point out that DDL assume it. It's fundamental to their approach.

  Reading the quote I provide earlier you will see that DDL are at great pains to decouple the chronon and the unit. It's perfectly reasonably to talk about 2.3 of a "unit" of time, but NOT 2.3 "quanta". When we store and reason about children we are not storing and reasoning about actual children. Rather, we adopt as a unit the notion of a "child", and make the assumption that quantities in this "unit" can be real numbers.

> > > I work on Oracle and I can affirm that the DATE in Oracle is one second
> > > of granularity. You say that 0.8 of the quantum is impossible, 0.8 of
> > > the day is impossible but 0.8 of 24*3600 is possible. 1/7 of 1000 is not
> > > exact (142) but the error is < 0.1 per cent. Why one can not select the
> > > granularity of chronon to achieve the exactness ? You can say, go and
> > > use floating point, but floating point is not exact too.
> >
> > The DDL book makes no claims about implementation. Nor should it.
> > It's a book that describes a logical model.
>
> You didn't answer the question.

 No. But I tried to respond to it. The point is that, logically speaking, no matter what granularity of chronon you pick, the same difficulty arises. On the other hand, reject the chronon idea, assume time is a continuum which can be measured in varying units, and no such difficulty occurs.

> > > I read about Snodgrass, and Snodgrass, he uses the name of chronon too.
> > > I want to know why you think that the chronon is bad in mathematic sense.
> >
> > Quanta are not 'bad'. They are not 'good'. The argument I am making
> > is merely that 'time quanta' (as described in the DDL book) represent
> > an innappropriate model for temporal reasoning because modeling time as
> > a continuum and temporal phenomenon using intervals over the continuum
> > is more general.
>
> You didn't answer this question either.

   I responded to it. I don't think "chronons" aren 'bad' in a mathematical sense. You can build perfectly consistent mathemtical models using integers. But integers alone are insufficient to describe the world. I think the idea of using 'chronons' in a temporal data model is inferior to the alternative. Received on Sun Jan 21 2007 - 23:01:36 CET