- From: Fred L. Drake, Jr. <fdrake@acm.org>
- Date: Sun, 14 Jan 2001 10:27:04 -0500 (EST)
- To: Dieter Köhler <dieter.koehler@ppp.uni-bamberg.de>
- Cc: "www-dom@w3.org" <www-dom@w3.org>
Dieter Köhler writes: > The 'attributes' property of the 'Node' interface is defined as being > readonly. Does this apply only to the property as such or does it > extent into the functionality of the NamedNodeMap too, e.g. does I can't speak to the intent of the API designers, but I've always seen it interpreted as "there is no set operation for 'attributes'". This makes more sense to me; setting the list would be a fairly expensive operation to make available through what may be syntactically innocuous in some language bindings. For example, this would appear as someNode.attributes = someAttributesList in Python; most Python programmers will consider this a "cheap" operation. > calling myElement.attributes.setNamedItem (instead of using > myElement.setAttributeNode) raise a NO_MODIFICATION_ALLOWED_ERR > exception? Actually, I'm not sure how else to get ahold of the children of a node in a general fashion, so this behavior would make the DOM nearly read-only in practice. What I *would* like is a way to determine that a DOM object is read-only; this should be exposed at the Node level (to see if a particular Node is read-only, perhaps because it is a descendent of an EntityReference), at the Document level (some particular Document is read-only), or at the implementation level (the impl. only supports read-only operation). This would be useful at least for developing a test suite to determine if an arbitrary implementation is operating as expected. There are enough nuances that a thorough test would need to be conditionalized for that more introspective features would be useful for. A set of standard "feature" names would be useful for this. Another would be persistence beyond a process, as we're starting to see more DOMs built on top of relational databases or other non-transient data structures. -Fred -- Fred L. Drake, Jr. <fdrake at acm.org> PythonLabs at Digital Creations
Received on Sunday, 14 January 2001 10:32:02 UTC