- From: David Brownell <david-b@pacbell.net>
- Date: Sun, 26 Sep 1999 14:25:47 -0700
- To: www-dom@w3.org
I thought I'd send some comments in now that it's last call. Briefly, what I saw (I focussed on the core) looks good -- addresses many (but not all) of the issues I had with the level 1 work, and I'm glad to see that things went along smoothly even in my absence! :-) Of particular note: * The system and public IDs are really for the "document type's external subset", not for the "document type", as it's now written. DTDs often consist of many components, each with their own identifiers. As written, those IDs could refer to one of dozens of identifiers -- not useful. I'd like to see DomImplementation and DocumentType describe this correctly (a small editorial change). * I can't see how I'd write a chunk of code that'll know it can use Level 2 features ... e.g. it's portable and needs to work with both L1 and L2. I'd like to see a "DOM" feature (for DOMImplementation) be defined, which would report on which level is supported. * It's confusing to have feature definitions scattered all over the spec. I'd like to see them collected in one table at one place, so their overall scope is more readily understood. * I still don't see why there's no solution for getting access to a system default DOM implementation, at least for Java. I'm posting an implementation in a followup post. I'd like to see something like this in the org.w3c.dom Java package. Thanks for the good work! - Dave Look I like the report format, but I still feel like I need the table of contents to have clickable links to let me get to each of the documented classes. This structure (sans such TOC support) is relatively hard to navigate. 1.1.7 Namespaces ... it may be useful to note that the DOM has adopted an 'early binding' model of namespace URIs to elements. With early binding, the URIs are permanently bound to elements as they get created. Late binding is more like what one would get by moving text around in a text editor -- the URI bindings are context-dependant. (So there are significant ramifications...) Also, I'm glad to see the simplification of having the namespace support be in the core interfaces versus being separate. Life is much easier that way! 1.2 General re compatibility ... I don't see how I can write a chunk of code that'll know to act differently if it's using DOM Level 2 or DOM Level 1. SUGGESTION: have a DOMImplementation 'feature". "DOM" would have values "1.0" (for DOM L1) or "2.0" (L2). 1.2 DOMException ... ambiguous re whether the other numeric codes are reserved to W3C or not. They should be. Similar comment for "Node" ... nobody except W3C should be defining new numeric codes for "nodeType". 1.2 DomImplementation ... createDocumentType() seems to have no accomodation for an internal subset. I'd suggest at least correcting the text; it's not the "document type system identifier" (or public ID), but the "external subset system identifier" (or public identifier). Also, having the "feature" identifiers scattered throughout the document is suboptimal. At least list the features from each chapter up at the front of the chapter, in a table to visually highlight them. 1.2 Document ... refers to "invalid" characters. "Illegal" would be more correct (with reference under DOMException to XML, HTML, and related specs for what that means). Of course the INVALID_CHARACTER_ERR name is grandfathered... 1.2 Document.importNode ... I'm rather uncomfortable with that name "import" since that implies the same object is in use (e.g. if I import something from Canada). "copy" is the appropriate word, and is even used in the documentation more than once. "import" suggests the wrong thing. 1.3 DocumentType ... as above on DomImplementation, the text should say the IDs are for the "document type's external subset", 7 Traversal ... I'm still not a fan of this stuff, but at least it's optional. (Seems like it should integrate with the view mechanism in some places, but doesn't -- oops.) App C Java Binding ... I ** REALLY ** think it'd be useful to define a standard way to get access to a DOM implementation, so that only parser connectivity really needs to be left to nonstandard APIs.
Received on Sunday, 26 September 1999 17:26:33 UTC