Expressing 2- and 3-D coordinates

[martin hill]

Agreed, Doug.  If you have an array of primitives then space-separated encoding
is acceptable, often desirable.

Coordinates are not primitives, they have a variety of attributes depending on
their type, and forcing these attributes into arrays loses type/structure
description and validation, *and* obscures meaning from both human and machine.  

As you say, you can already define structures to do it, and that's what you
should be doing.  That's what the original complaint was about way way back when
some of us first saw array-based coordinates, not that array components should
be individually tagged.

Cheers

Martin

Quoting Doug Tody <dtody at nrao.edu>:

> I would argue that expressing an array with an explicit tag for every
> array element makes it significantly harder for a computer (or in many
> cases even a human) to process the information.  To do so implies that
> the values can be given in any order and that the reader must sort out the
> sequence to reconstruct the original array.  The tags have to be formally
> defined and recognized by the reader to correctly interpret the data; the
> reader has to check that all values are present and are not duplicated, etc.
> This is quite a lot of work to go to just to reconstruct a simple array.
> 
> If we are encoding an array then there is already a conceptual data
> model defined somewhere which defines what the elements of the array are.
> It is much easier for the software to just pull out and operate upon the
> entire array as an atomic object.
> 
> Of course, if we start to play tricks and encode complex things into arrays
> then it may be better to instead define a structured object with named
> elements, but we can already do that.     - Doug