On 11/09/2010 11:34 AM, Ray Denenberg, Library of Congress wrote:
> We have to distinguish "the endpoint of an interval is another interval"
> from "the end point of an interval is a choice among several consecutive
> years";

The latter is not what I meant.  Rather the case is "the endpoint of an
interval is questionable/approximate/unknown etc" - and we have a framework
for times that are "questionable, approximate, unknown etc".  A "choice 
several consecutive years" is just one of many ways in which an endpoint
may be uncertain.

Is there some other meaning (or use case) of "the endpoint of an interval is
another interval" than the endpoint being questionable/approximate/unknown?

> an interval is a continuous period, vs. a discrete set of
> (consecutive years).
> So if we want to express, let's say "the interval beginning with one of the
> years 1670, 1671, 1672 and ending 1680"
> it would be
>      [1670, 1671, 1672]/1680
> or
>      [1670-1672]/1680
> On the other hand if you want to express "the interval beginning sometime
> within the interval beginning in 1670 and ending in 1672 and ending 1680"
> it would be
>      1670/1672//1680

I don't believe there is a meaningful difference.

> There may be preferable alternatives to the .xx. syntax currently proposed.
> I'm fairly sure however that we don't want to introduce angle brackets,
> because after all this spec is intended to be used heavily with XML.

That's certainly a valid concern - though only for people typing raw XML 
Probably most people benefiting from this spec wouldn't be typing and
reading raw XML - but I don't know the target "market" that well.
	--Per Bothner
[log in to unmask]