- From: Eric Eldred <eldred@eldritchpress.org>
- Date: Tue, 18 Jan 2000 19:57:26 -0500 (EST)
- To: JOrendorff@ixl.com
- Cc: www-html@w3c.org
> XHTML will not make old documents obsolete. > If you have correct HTML 4 documents, they will still be correct > HTML 4. Browsers will not suddenly stop rendering them when > XHTML becomes a Recommendation. Some new browsers and special- > purpose clients may require XHTML 1.1, but if that's an issue, > then the lack of <A target> is the least of your problems. Yes, that's true. I could just serve documents as text/html, or I could serve them as text/xml. But was not the point of XHTML that documents validated "against" its DTD would be not only well-formed but also valid, and be capable of being served as text/html to older browsers? If you wish to require that users obtain new browsers in order to read these documents, you lose the point of XHTML, and might as well just serve XML documents with arbitrary DTDs. And in fact your last point underestimates the problem. We are likely to face a host of widely different DTDs based on XML or XHTML, all of which try to use HTML features, but tailored to different user agents, such as the Open eBook specification or other portable devices. With HTML one could be compatible by including elements that were simply ignored if the user agent did not have that feature; with XML or XHTML one can only use elements in a particular DTD. > > Nielsen's complaint is not > > valid--the BACK command does work in Lynx, so if > > a user agent wishes to respond that way it can-- > > it is not a problem in the HTML DTD itself. > > I love the logic here. You've now said that: > 1 - Lynx ignores target="_new" > 2 - Lynx does the right thing. > > A lot of people would conclude: > 3 - The right thing is to ignore target="_new", and > if browsers should just ignore it then there's no point > carrying it forward in future versions of HTML. > > But your conclusion is: > 3a- target="_new" is great and should be supported in all > future HTML specifications. I didn't say that "target" was great. Only that it has a function and is used, and that prohibiting its use via a new DTD is not necessary. I pointed out that an HTML user agent might simply ignore that element, as Lynx does--that is the proper place to handle it, not by arbitrary DTD authoring. In other words, Lynx does the "right thing" to ignore it--but so does Internet Explorer do the "right thing" to handle it the way it does. That is the way HTML works. But not XHTML. My point is that almost all of the benefits of XHTML could be obtained simply by requiring that the document be well-formed in XML terms. If there is an element "target" in the document the parser needs to figure out where in the structure of the document this is allowed. The parser doesn't need to get involved in the semantics of the element--whether or not "target" has a meaning to the browser depends on more than just the document structure as set forth in the DTD. Moving the logic to the style sheet improves the matter not one bit. Whether or not "target" is a "good thing" or something horrible ought to be irrelevant when writing a DTD for XHTML; it should only be a case of whether the structure ought to be outlined so the browser parser can know how to handle it. I can see the earlier problems with framesets in XHTML: they cause structural problems because of the "body" element in the DTD. But I can't see that the same argument applies to the "target" attribute when it is not used in a frameset. Really, what happened to the "eXtensible" part of XML? Here we are arguing once again whether or not "blink" and its cohorts ought to be allowed to the table. But you are right that most of us will hold back and not rewrite our documents for XHTML. I would like to use "accesskey" but don't even have a browser to figure out how it would work. When I don't have to use <center> to center a subhead in Lynx then I will consider doing away with that hack and writing to "pure" XHTML and CSS. Meanwhile, good luck, W3C, in your arduous labors!
Received on Tuesday, 18 January 2000 20:09:37 UTC