- From: Risto Kankkunen <risto.kankkunen@iki.fi>
- Date: Tue, 19 Oct 2004 22:54:32 +0300
- To: Edward Lass <elass@goer.state.ny.us>
- CC: www-html@w3.org
>>> It's great that you pointed out this real-life problem you are >>> having, because it's just one of those things that "<alt>" could >>> fix. > > I'm a bit worried by your tone, Risto, because it comes across as if > you're trying to "disprove" me. I was making an observation that > definitely does agree with that part of your premise (whether or not > it agrees with your solution). I'm sorry, if it sounded like that, but I was really grateful that you pointed out that incident. I think it is far more interesting to look at real world scenarios than to look at my artificial examples. > Well, I am suggesting caution against a "Christmas tree" attitude, > i.e., cautioning against hanging every possible ornament on XHTML. A > markup language can't be all things to all people. I understand that. But it's not that I'm proposing something new and complicated here. I'm proposing for a simpler and more powerful alternative to the current proposal. The current approach seems to revolve around the concept whether some extrenal resource is accessible (since the content negotiation in the current form can be handled with HTTP Content negotiation like you and the XHTML FAQ point out). Is this the #1 problem in WWW? If we go along the current proposal and then find out we need someting more versatile, we are at dead end. We have to make up new syntax and then we have partially overlapping functionality for doing the same thing. Whereas using "<alt>" now we get much more flexibility and room for improvement adding some controlling attributes to "<alt>" later on, if needed. I understand that people might disagree which notation is more elegant. But I hope that it is accepted that the "<alt>" syntax lends itself naturally to express constructs that cannot be done with the draft proposal. And this versatility comes for free. Here's a JavaScript implementation of the "<alt>" element processing: // Input: DOM <alt> element // Returns: Render tree function handleALT(element) { // Try out all the children in succession for (var e = element.firstChild(); e; e = e.nextSibling()) { var render = handleElement(e); // Could we render the child element without error? if (render) return render; } render null; } Implementing the draft mechanism is not easier (I'd say it's harder), since any element handler needs to check for "src" attribute, (at least for "<object>") peek for possible "<param>" elements and discard other content, and on failure, skip those "<param>" elements and render the rest. And the end result is not as flexible. > I suspect by the time XHTML2 is matured, non-Latin fonts will be much > more commonplace. I think I'm not going to download 36 MB of CJK fonts (that are freely available to anyone who wants to use them right now) before I can read those characters. I'd like to show my preference by not installing the fonts and give the browser the ability to pick an alternative spelling that I might be actually able to read, maybe even pronounce. > (But does device independence preclude us from assuming this?) I think the need for for customized and alternative content only grows, not just because better support for Unicode, but because more varying devices have user agents, and because of the modularization of XHTML/XML in the form of specialized sub-languages (e.g. MathML). >> It's also interesting to see, that you don't think multilanguage >> documents are common... > > I never said that, nor do I agree with it. OK, I read too much into your comment "Is this problem common enough[...]". I apologize. Regards, Risto Kankkunen
Received on Tuesday, 19 October 2004 19:54:29 UTC