- From: Charles Reitzel <creitzel@mediaone.net>
- Date: Wed, 7 Feb 2001 10:28:18 -0500 (EST)
- To: www-xml-schema-comments@w3.org
This topic came up in xml-dev recently, and I felt it important enough to post a comment here. Several features of XML Schema appear to infer DOM-like processing in that random access is required to the document contents and thus, the entire input document must be memory resident. From my non-expert reading of the spec, the Identity constraints and intra-document links (ID/IDREF) are the obvious culprits. There may be others. I like keys and keyrefs, don't get me wrong. But it would be useful to call out those features of XML Schema that are implementable using SAX. I.e. an element should be validatable by the time it's end element appears on the stream. Validation should require access only to children and ancestors, not siblings. As an application/schema designer I need rough predictability across implementations as to the processing costs of particular schema features. Charles Reitzel P.S. is it soup yet?
Received on Wednesday, 7 February 2001 10:28:20 UTC