Re: XPath calculation

["Simon St.Laurent" <simonstl@simonstl.com>]

At 08:10 AM 2/10/01 +0800, Rick Jelliffe wrote:
>I think the problem comes down to seeing the document as a graph,
>with the root element, each element with an ID, and each unique set of
>vertices as potential anchor points. We want to find the shortest distance 
>to a potential anchor point, then the simplest way to express the path 
>between the anchor and the element.

Interesting how similar that is to XPointer child sequences, and their 
combination with bare names for IDs:
http://www.w3.org/TR/xptr#child-seqs

In general, this seems like a good way to do it.  The more IDs, the more 
stability, and the algorithms aren't too ugly.