> 1. The DTD is always available. This is false for the huge majority of > existing XML apps
No problem. If you cannot specify in the DTD that whitespace is ignoreable, it gets reported. > 2. It's easy to figure out which white-space is significant. Trust me on > this, a lot of smart people have wasted a lot of time trying to write > down simple rules. It's totally application-specific. Never got a problem with that. The DTD specifies when an element can have element content only, everything else inside the element is passed as ignorableWhitespace, and is simply ignored. I can further ignore that space in the application code, but there's no need to. It never existed to begin with. I'm still waiting to see one example of where this break and this example should: 1. Depend on the DTD being available. No DTD, no whitespace stripping. 2. Use the information model of the document, not the textual representation (I can understand why an editor will need the whitespace) 3. Prove that if whitespaces are ignored the application behaves in an inconsistent manner The latter can be done easily if you consider the simple case when a document is prouced in memory, then serialized with and without pretty printing. You get two different textual representations, the same canonical document. Your SAX events stream (textual representation) will be different, but your information model (calls to characters, DOM Text nodes) must be the same. arkin > > -Tim