Multiarranging (Mostly) Globally Euclidean ND Spaces

[Brent Turcotte <brentt@xxxxxxxxxxxxx>]

Suppose someone has a solar system table of statistics.  A solar
system table typically includes things like mass and diameter for
objects such as planets and moons.  If someone tried to include
all possible parameters, things get difficult, leading to a possible
application of ZigZag space.

A solar system table in ZigZag would go beyond a two dimensions
global Euclidean grid.  For some parameters, there is no data for
most of the objects in the solar system.  There is a distinction
between original data and derived data.  There are many ways of
grouping data by class: the sun, planets, moons.  Some one may
want to separate less significant moons from major moons.  Some
may want to include pictures and video along with the plain numbers.

Suppose that objects (planets, moons or the Sun) are visually in
rows and parameters are in columns.  A designer may want to have
dozens of arrangements of the rows and/or columns.  What would be the
best to represent and edit these different arrangments?  

Alternative:  different versions.  The designer would probably not
want to map the arrangements to different versions.

When the designer creates this 'solar system' table, it might be
necessary to link every item (number) to it object and its parameter.
If this wasn't done, problems could occur when items are hopped.

If it was not made into a grid, say the downward links were only
included, rearragments of the rows would screw up the data.

Yet in a grid, a rearragement of a the header of a column or row
would demand a rearragement of most of the links, otherwise things
might look increasingly messy.  Clones would not suffice.

A new view and bindings are likely necessary to handle this 
situation elegantly.  It seems that beinng globally euclidean 
complicates things.