Article:		XML on the Web Has Failed
Subject:		Useful but Misleading
Date:		2004-07-24 17:40:55
From:		MURATA
Although the article provides useful information, it is misleading. It makes people believe that RFC 3023 hampers interoperability of XML, but this is not true. It is not RFC 3023 but MIME (RFC 2046) and HTTP (RFC 2616) that specifies the charset parameter of text/* as the only authoritative information. RFC 3023 introduces nothing new about text/xml. Then, should MIME and HTTP blamed? I do not think so. We first have to understand why they use the charset param for every protocol and every text/* media type. Do we continue to introduce an ad-hoc encoding declaration mechanism whenever a new format (e.g., XQuery and RNC) is invented? Apart from text/xml, RFC 3023 provides application/xml, which is often more appropriate than text/xml. One of the reasons is that the omission of the charset means "Use the encoding declaration or BOM". A new I-D for XML media types makes this point clearer and even deprecates text/xml. http://www.ietf.org/internet-drafts/draft-murata-kohn-lilley-xml-00.txt One weak point of the article: it does not show how many "text/xml and non-ASCII" documents provide correct encoding declarations. If quite a few documents fail to specify the encoding dcl correctly, the problems boils down to "I18N is hard". MURATA Makoto (co-author of RFC 3023)

Previous Message

Next Message