- From: David Brownell <david-b@pacbell.net>
- Date: Tue, 28 Sep 1999 09:37:33 -0700
- To: www-dom@w3.org
keshlam@us.ibm.com wrote: > > Ignorable whitespace is an open issue. Knowing what's ignorable and what isn't > requires that the document be processed in the context of a DTD, so you can > distinguish element or text content from mixed content. The same is true for the "isSpecified" property of an attribute, which is already part of DOM (and not part of the last Infoset draft I looked at, though that may have changed by now). > We could provide a flag > that parsers would set, Why just "parsers" ?? Such flags should either be settable by any component, or should not exist. Else the API isn't testable in its own right. The current DOM has at least two flags that act like they're settabble by a parser. One is "isSpecified". The other is a "readonly" flag used for DTD related things -- the NamedNodeList objects in a DocumentType, and children of Entity/EntityReference nodes. (That is implicit, and can't be tested except by trying to change a readonly node.) > but making that flag continue to work as the document is > altered -- as the Text node is moved into a new context, for example -- would > then require custom solutions. But they do this already (see above). It seems like what's showing here is not a consistent logical approach, but rather a set of inconsistent judgements. I think the package would be a lot better given a consistent logical approach ... in fact, that's critical to having it be testable. There's still time to fully expose those flags (have them both gettable and settable). - Dave
Received on Tuesday, 28 September 1999 12:38:10 UTC