- From: Anne van Kesteren <annevk@opera.com>
- Date: Mon, 16 Apr 2007 07:14:54 +0200
- To: "Alexander Graf" <a.graf@aetherworld.org>, "Preston L. Bannister" <preston@bannister.us>
- Cc: "Dailey, David P." <david.dailey@sru.edu>, public-html@w3.org
On Mon, 16 Apr 2007 00:52:18 +0200, Alexander Graf <a.graf@aetherworld.org> wrote: >> After writing the above, I would like to suggest another principle: >> The HTML-next specification should be as short as possible. >> >> Strictly speaking, "Don't Break the Web" is a non-issue. Existing >> web pages will be interpreted just as they are today. The HTML- >> next specification only applies when the HTML-next version >> specifier is seen by the browser. (Just like XHTML did so [cough] >> successfully.) XHTML didn't do that at all. The only thing XHTML did was expressing the HTML language in XML. Since HTML was already a tree language where nodes in the tree have attributes this was fairly trivial. The only thing that has thus changed is the syntax, not the language. So the only thing that was fixed by XHTML (XML really) was the error handling of the syntax (there isn't any, you get some stupid error message in your browser). HTML5 fixed the syntax error handling for the HTML syntax in a manner believed to be compatible with the web. This still doesn't solve the problem for error handling with the language. What if an attribute has an incorrect value? What if an element is nested in an element within which it isn't allowed? For instance: <form action="a"><a href="b"><input type="button"/></a></form> XHTML didn't solve that. Browsers are handling the above case also identically in HTML and XHTML. It makes no sense to implement such logic twice in incompatible ways. Fixing the syntax of HTML is just a minor thing and has already been done to some degree (its even implemented and testsuites are available). -- Anne van Kesteren <http://annevankesteren.nl/> <http://www.opera.com/>
Received on Monday, 16 April 2007 05:15:35 UTC