- From: Allen, Michael B (RSCH) <Michael_B_Allen@ml.com>
- Date: Tue, 31 Jul 2001 21:09:19 -0400
- To: "'Watson, Christopher'" <CWatson@lightspan.com>, "'WWW-DOM List'" <www-dom@w3.org>
> -----Original Message----- > From: Watson, Christopher [SMTP:CWatson@lightspan.com] > > Where should I trundle off to in order to ask of the DOM implementations > community questions pertaining to how the "live" aspects of NodeLists are > dealt with? I realize that the DOM FAQ[1] has a section titled "NodeList > issues", and I have read that. It gives a top-level explanation of how > It has been suggested that a common technique is to enable all NodeList operations with the capacity to construct an updated list at any moment and then have a "dirty flag" on the tree (or perhaps on each node) that would indicate reconstruction is necessary. I and at least one other person I know of punted on this. I would be very happy to hear more about this as well because I don't fully understand the above description myself. The problem is that it assumes the action used to generate the original list can be parameterised. For example, if the getElementsByTagName is used to create the list, you could save this information and the name of the tag used to perform the query. If the list is a simple list of children however, these parameters are not required. Perhaps this is not what was intended at all however but instead that the NodeList simply be patched as needed. When the dirty flag indicates that the tree (or branch of) has been modified, each member will be examined and compared to the main tree. All of this applies to NamedNodeMap as well which is even more troubling because the Map is not immutable. Mike
Received on Tuesday, 31 July 2001 21:09:22 UTC