[
Lists Home |
Date Index |
Thread Index
]
From: "Jeff Greif" <jgreif@alumni.princeton.edu>
> So that symbolic distances can be used:
> kilometer(length-of-Amazon-river-in-meters)
> --> length-of-Amazon-river-in-meters/1000
> ?
Could be. I'm not sure that all information about a type can or should be
crammed into a single identifier. (Or that astronauts prefer named types,
but that's another story. ;-)
I prefer the point of view that types have multiple attributes (along the
lines of what XML Schema datatypes calls facets) and that one of these is
unit-of-measure. But as long as it's clear that the domain of the function
as posed is a distance with an implicit unit of measure (which may be null,
i.e., a pure number) it makes sense.
Bob Foster
> Jeff
>
> ----- Original Message -----
> From: "Bob Foster" <bob@objfac.com>
> To: "Roger L. Costello" <costello@mitre.org>; <xml-dev@lists.xml.org>
> Sent: Thursday, July 24, 2003 11:27 AM
> Subject: Re: [xml-dev] Re: An approach to describing the relationships
> betweenunits-of-measure
>
>
> > It seems that kilometer, if it is to be viewed as a function, must
operate
> > only on the unit of measure property without affecting the type of the
> > value. If the domain is Distance, the range must be Distance; the
> difference
> > is that the argument distance is expressed in some (unspecified above)
> unit
> > of measure and the result distance is expressed in kilometers.
>
>
|
