- From: Claude Zervas <claude@utlco.com>
- Date: Tue, 11 Aug 1998 00:06:22 -0700
- To: www-dom@w3.org
DocumentFragment appears to be a placeholder for nodes or node trees that can be in the process of being cut/pasted from one document to another. During this phase it seems like the nodes might not be attached to a particular document. This would be very useful for creating templatized documents, for example, where some commonly used document fragments are cached and pasted into the document or rendered serially into an enclosing document when converted to ascii in a web server. However, the spec for Document clearly states that nodes may not exist outside the context of a particular Document. This appears to contradict the idea of DocumentFragments. I apologize if I'm completely missing something here, but any help clearing this up would be much appreciated! I've built an HTML40 parser that builds up a DOM tree from either a full document or valid document %block fragments. It would be nice to not have to always have to create a full Document object just for the fragments. My template engine takes a top-level document and a hashtable of named document fragments that are substituted into the output stream whenever a SUBST tag is encountered. This allows me to share document fragments with multiple master template documents since the merging is done at serialization time. Also, just a comment on a possible future server-DOM, it would be nice to be able to deprecate attributes like Node.nextSibling, Node.previousSibling and Node.ownerDocument that make the nodes completely encrusted into the collection they reside in. This makes it very inefficient if some of the nodes are remote objects and some are local. And please bring iterators back! Being able to mix and match nodes from different DOM implementations would be very handy, especially for distributed documents. As I understand it, the current DOM forbids this. Thanks, - Claude Zervas
Received on Tuesday, 11 August 1998 02:57:35 UTC