"do people think it is sensible to include a feature in a specification when even the main proponents of the features are prone to disagreement?"

There aren't "proponents", there is one proponent, Ed  (see message that I just posted). So there isn't really any disagreement, rather there was misunderstanding, which I think is cleared up now .   --Ray

Sent: Tuesday, June 28, 2011 9:54 AM
Subject: Re: [DATETIME] On the usefulness of x

Maybe I could just offer the question, do people think it is sensible to include a feature in a specification when even the main proponents of the features are prone to disagreement? If confusion arises when people draft a specification, how much more when a specification is let loose on the people who haven't even heard the arguments either way?

The underlying point is that it may be counterproductive to specify something confusing, as people will use it in diverse, potentially incompatible or irreconcilable ways, while the veneer of using the same formalism will lead people to think that the same semantics in intended. This would be worse than not specifying that feature.

Simon

On Mon, 27 Jun 2011 16:57:19 -0400, Ray Denenberg wrote

> > {1950..1959} means the set (all of) dates 1959 through 1959. It is an
> > array of dates expressed with a precision of year.
> > The expression 195x, by contrast, is neither an array nor an expression
> > with a precision of year but a date with a precision of 10 years.
>
> Are you perhaps confusing 'x' with 'u'?

No. The expression 195u has a precision of year. It is used to express a
date in year precision where the year, however, is not (yet) known. The
expression, by contast, 195x is of decade precision (10 years) and does
not express any uncertainty. The expression, for example, "The journal was
published in 195u" I would take to imply that it was published in some year
in the 1950s but its uncertain, at this time, which. The expression, by
contrast, "The journal was published 195x" I would take to mean that the
said journal was published in at least one of the years in the 1950s.
195u -----------> 1 year in the range 1950 to 1959. The date is expressed
with precision of year.
195x -----------> 1 or MORE years in the range 1950 to 1959 expressed with

>
> 199u means "one of the years 1990, 1991, 1992, etc."
> 199x is supposed to mean "ALL of the years 1990, 1991, 1992, etc."

That is the difference in our views and why I think "x" is needed and can't
be  replaced by {..} expressions. {1950..1959) can mean ALL of the years
since its an expression of an array of dates with an implicit measurement
precision of year.
If one agrees that the expression "195x" has a precision of decade than it
can't be ALL since that would entail a readability (precision) of year.
The contruct 'x' (from right to left) I took to represent the
expressions "1950s", 18th Century etc.
The expression 195x (1950s) is, I think, like reading a short thermometer
with ticks every 10 degrees. Its hard to talk about where, for example, 33
degrees is..  its not readable to degree.
Its, again, like 1999. That expression does not mean "all" days, hours,
minutes or .. If I say that a journal was published in 1999 I mean that at
least one issue (perhaps more) was published in 1999 but not that an issue
was published every second. "1999" means one or more and not ALL. The
expression "1999" has a precision of year so I can't read precisely month,
day, hour, minute, second ...

>
> The example cited is someone listing years of publication. For example for
a
> book published in 1992, 1996, and 1998 you could list its years of
> publication as {1992, 1996, 1998}.
>
> But say it was published in 1988, 1990, 1991, 1992, 1993, 1994, 1995, 1996,
> 1997, 1998, 1999, 2002, and 2004.
>
> You could list its years of publication as {1988, 199x, 2002, 2004}.
>
> This is what we decided many months ago.

I did not decide this :-)
I was always, I think, speaking of precision. Wildcards never really
interested me except in queries. That is why I see 195x as being something
to implement (well its been implemented already by me as I have interpreted
it) but not 1x59. I can see some interesting questions expressable with
wildcards such as "turn of the century", "mid century" etc. but see them on
the side of query and less on the side of date representation for storage---
I would implement these, just as generalized 'u', in the query engine (state
machine) and not in my indexer (data storage model).

>
> --Ray

--

Edward C. Zimmermann, NONMONOTONIC LAB
http://www.nonmonotonic.net
Umsatz-St-ID: DE130492967

--
Simon Grant
+44 7710031657
http://www.simongrant.org/home.html