- From: Joseph Kesselman <keshlam@us.ibm.com>
- Date: Wed, 10 Oct 2001 11:00:48 -0400
- To: www-dom@w3.org
The idea is that since you're calling XPath in the DOM world, you get a result in terms of the DOM tree. If the XPath returns text, it returns only first Text/CData node in any block of logically-contiguous text. (Which may involve multiple nodes, and may cross EntityReference boundaries.) A separate convenience function allows querying all the logically-contiguous text as a single string, ant it's the API user's responsibility to understand that when a Text node is returned they should use this approach rather than just asking that node for its own data/value. (Or, of course, the user can set up their own iterator/treewalker to examine the consecutive nodes, if they really care about the structure. We might want to include an appropriate NodeFilter in our documentation, though I think that would be considered to be on the wrong side of the declarative/descriptive line.) It's a compromise to match impedence between the XPath and DOM data models. I consider it a reasonable one, since anyone who wants only the XPath view can fairly easily derive that from these tools,, while those who want the DOM view still have the additional detail available. (If XPath and DOM hadn't diverged, this kluge wouldn't be necessary. But they did, and we're stuck with trying to come up with a reasonable interpretation and implementation despite that..) ______________________________________ Joe Kesselman / IBM Research
Received on Wednesday, 10 October 2001 11:01:49 UTC