- From: Kurosaka, Teruhiko <Teruhiko.Kurosaka@iona.com>
- Date: Sun, 3 Nov 2002 11:33:29 -0800
- To: <Peter_Constable@sil.org>, <www-international@w3.org>
> From: Peter_Constable@sil.org [mailto:Peter_Constable@sil.org] > I'm curious about this extract from an appendix to the XHTML spec: > character encoding explicitly must include both the XML > declaration an > encoding declaration and a meta http-equiv statement (e.g., > <meta http-equiv="Content-type" content="text/html; > charset=EUC-JP" />). I think there are two problems. (1) The use of the word "must" is probably a mistake. It should be "should". It would be a mistake to require something in a spec to work around a bug of one product. (2) If they indeed intend to be a requirement, I see a danger here. First, it is inefficient to carry the identical information in a document. Secondly, it incrases a chance of creating a wrong document where the first and second methods of encodinf declaration declare different encodings. The quoted statement is emphasizing portability concern of the document. They might also want to add a statement that requires changing the encoding in these embedded tags, if the received document is in a different encoding than declared in the embedded tags. It is possible because HTTP level declaration of charset is stronger than the embeded information per HTML and XML spec. By the way, despite the mandate in XML spec that reads: All XML processors must be able to read entities in both the UTF-8 and UTF-16 encodings. I recently learned that there are XML parser implementation on a "limited configuration" device that only recognizes Shift_JIS. Apparantly, because of the resource limitation, the implementator decided to ignore the mandate. Just another example of the reality that doesn't always match with the ideal world... -kuro
Received on Sunday, 3 November 2002 14:34:08 UTC