From: "Edward C. Zimmermann" <[log in to unmask]>
> 2. A PROX/element.unit=container/distance=0/ element.containerName=C B
> This should allow for anonymous containers.
> Example: find 'A' and 'B' within the same container element
> A PROX/element.unit=container/distance=0 B
Yes the intent is certainly that the element.containerName modifier would be
> (although I really hate to call these prox and worse still talk about
> distance since there is no well defined metric with generic information
> which to make matters more complicated might include overlapping
Ed, I still don't follow your "no well defined metric (for distance)"
argument even after all these years. Whenever we talk about distance there
is always a unit. If distance is '3' and unit is 'word', or unit is
'sentence', or 'paragraph', is that not well-defined? There is no case
where a distance is proposed to be specified, where there would be no unit.
(The unit might be implied so that if it is omitted there is a well
So, 'unit=container, distance=0' (and assuming the spec supplies clear
semantics) meaning "in the same container". Is that not well-defined? We
might not allow a construct such as 'unit=container, distance=1' meaning
separated by one container, or consecutive containers, because that might
not be well defined, and in that case, 'unit=container' would be constrained
to be used only with 'distance=0'