# Re: how to deal with TABLE heartburn

All butterflies are insects.  But...

It is true that the subspaces of a multi-dimensional space form a
Boolean algebra.

It is true that it is often useful to hid the axes inside a
subspace at times and look on a subspace decomposition as a
low-order multidimensional space.  So using "axis" to refer to a
subspace is a reasonable and natural stretch.

On the other hand, it is only this Boolean algebra semantic that
this attribute shares with AXES.

Calling the current attribute AXES is like saying that all
boolean algebras are subspace decompositions of multidimensional
spaces.

The best definition I have come up with for the actual semantics
of this attribute are:

(for each remote element referenced in AXES)
The content of the current cell is a member of a set
identified with the content of the referenced element.

This definition includes sets satisfying either "various
attributes of one entity" or "the same attribute across multiple
entities" as is required to cover the significance of the row and
column heads respectively in the "Courses in Bath" example.

The parallel to AXIS is seen clearly if we state the semantics
of AXIS as

The content of the current cell is a member of a set
identified with the content of the AXIS attribute.

"Axes" is a very clear notion to us math geeks, and it is not
what we are dealing with here.  And the (more narrow) math
concept will be something we want to be able to use as a point of
reference or analogy in defining future language features for
HTML.

In order to make the name extend gracefully to MegaZone's design
for the OPTION classification scheme, I would like to replace
"keep-with" as a suggested replacement name for "axes" and
suggest "belongs-to" which is a little stronger in connotation.

-- Al

Received on Wednesday, 22 October 1997 13:30:18 UTC