Lists Home |
Date Index |
Michael Kay wrote:
> > Thus, distance is poor interchange data...
> > Good interchange data is position.
> The only way I know of to identify spatial position is as a distance
> from some agreed origin.
Yes, but the point is that
<distance to="destination airport">590</distance>
Has an implied referent: from "me" to the destination airport. That
makes the utility of the data very limited (only to clients the
understand the implied referent, which is mostly just "me", the sender).
With position, on the other hand, all referents are explicit.
Consequently, the data is useful not just to one client, but to any
client. Practically speaking, what this means is that with position
data clients can do things like calculate the aircraft heading, compute
time to fly-over, insert the data into a map, etc. Position is "high
The issue that Tony and Tom have raised is that, despite position being
more generally useful to clients, sometimes it's necessary to send the
calculated data (distance) rather than the fundamental data (position)
because the recipient may take the position data and generate a distance
that is not quite consistent (due to rounding errors, for example) with
the distance that the sender believes to be the case. This is the issue
we are wrestling with now.
> All data interchange relies on the sender and recipient having
> some shared knowledge.
There is no question there must be shared knowledge (metadata). For
If the data being interchanged is position data then the metadata is the
coordinate reference system. I went on to argue that there should be
precisely one, unambiguous coordinate reference system. It becomes the
"lingua franca" coordinate reference system.
If the data being interchanged is distance data then the metadata may
be, say, the units (meters, miles, etc).