- From: Masayasu Ishikawa <mimasa@w3.org>
- Date: Fri, 12 Apr 2002 16:44:53 +0900 (JST)
- To: www-html@w3.org
Henri Sivonen <hsivonen@niksula.hut.fi> wrote: > | For the record, it was a deliberate decision to make the usemap > | attribute to IDREF and XHTML-aware browsers like Mozilla and Netscape 6 > | does (happen to) work with such image map used in XHTML documents > | served as 'application/xhtml+xml', but nevertheless, an erratum > | is planned for "5.8. Client-side Image Map Module" of Modularization > | of XHTML to make it a URI reference. > > I think changing a deliberate decision as an erratum is unfair to the > browsers that adhered to the original spec. (This is similar to the > "no underscore in CSS class selectors" issue.) > > Is it necessary to be able to reference maps outside the document? If > the referenced map is inside the document, IDREF seems like a more > natural choice. That was the rationale behind the decision. We did test a bunch of existing HTML browsers, and only Lynx and Links supported external map reference and all others failed. And both Lynx and Links are text browsers, so they don't really render 'image' maps. On the other hand, a couple of browsers happened to work with pseudo-IDREF syntax. Since modular XHTML is not required to be backwards compatible with HTML browsers and there weren't enough support for external map reference, the HTML WG thought that IDREF would be a reasonable choice, and the cost of supporting the IDREF syntax in XHTML user agents seemed minimal compared to support external map reference. This change should have been mentioned more explicitly in the spec, though. Nevertheless, since many people screamed against this change and existing XHTML user agents that understand IDREF syntax also supports fragment identifier syntax anyway (as the latter is still necessary to support XHTML 1.0, which uses fragment identifier syntax for backwards compatibility with HTML 4) so this change won't really break existing XHTML user agents, therefore the HTML WG accepted to make it a URI reference. > Old browsers don't support application/xhtml+xml anyway, so the > behavior of old browsers is a non-issue. (I am assuming that XHTML > 1.1 shouldn't be served as text/html. Correct ('SHOULD' should be interpreted as described in RFC 2119 here). The HTML WG is working on a document to clarify media type issues for XHTML Family documents, which will be published in the near future. Regards, -- Masayasu Ishikawa / mimasa@w3.org W3C - World Wide Web Consortium
Received on Friday, 12 April 2002 03:44:58 UTC