intertwingly

I think using XHTML as a syndication format is a dumb idea. Then again I think XHTML was and still is a dumb idea but this idea seems even more ill-considered.

PS: Before anyone claims this is some Microsoft opinion. It isn't. Tantek is our HTML working group rep and he is all for this given my reading of his blog plus his opinion counts a whole lot more than mine in the grand scheme of things at least in that space.

I think Dare misses the point. To me, using "CSS class names for aggregation purposes" means embedding the structure of your content in your page, so that it can be parsed by an aggregator -- or indeed transformed into any alternative message/publishing/interaction format.
I use DIV ids to mark up the body, title and permalinks of my weblogs, and then I have a tool that parses them to create a) my RSS feed b) some Javascript indexes c) extracts to build my home page d) the HTML for updating my archive index page every week. All this multipurposed in a loosely coupled way from one XHTML source. The only pity is that it's a bit of a bodge at the moment, and the id="rssi0" labels are not very expressive. So I'd welcome any suggestions for progressing this concept to make it a bit more robust.

I was about to post a detailed response but noticed Mark Pilgrim has already beaten me to the punch with

http://diveintomark.org/archives/2002/11/26.html#syndication_is_not_publication

which eloquently points out most of the issues I saw with the XHTML-as-syndication-format proposal. The criticisms I have left that Mark doesn't bring up are probably not politically correct anyway given my employer. ;)


Posted by Dare Obasanjo

Re: That blog-that-is-also-valid-RSS (which isn't really valid RSS, it declares its version as 0.92 but uses namespaces, but the validator isn't that sophisticated). Anyway, the page doesn't display properly in Opera (latest stable version, appears unstyled), and doesn't display at all in Lynx (latest stable version, offers to download) or my copy of IE 5.5 (offers to download, although this may just be a symptom of an ongoing problem I have with my copy of IE). Works in Mozilla, though. Woohoo.

It's an interesting thought experiment, but I don't think it's proving the point you thought it was proving.

Mark, I guess that depends on what point you think I was trying to prove.

In the nirvana that is the distant future, there really is no need for XHTML. All that is needed is XML + CSS.

I don't believe that we can get there in one step, but meanwhile I believe in encouraging things that are helpful along the way (like supporting namespaces, and ignoring elements that one doesn't understand).

Sam, I wish I could believe in that future. I really do. My life would be much simpler. My day job is currently consumed by a project that has me debugging nested tables (of my own devising!) because our client has clients that are still using the dreaded Netscape 4 and our client is not in a position to force them to upgrade, AND they are of the mind that everything should look exactly the same in all browsers. (Ever wonder why my own site is getting more and more stripped down and "semanticly pure"? Classic overcompensation behavior.)

So we might get there eventually, but frankly I doubt it. Technologies don't evolve forever; they go a certain distance until they hit diminishing returns and lose the attention of a critical mass of end users (to push for innovations) and developers (to deliver them), and then they just stagnate. We in the blogging community are all in love with the web now, but most people just take it for granted, it more or less does what they want it to do, they've already been lowering their expectations for years, and future attention will be spent elsewhere.

How does pure XML + CSS benefit end users? I don't see a compelling reason for them to upgrade. Embedding other XML datatypes? MathML? SVG? RDF? Please. Reduced bandwidth? Minimal, again you're hitting diminishing returns, your limited time and resources are better spent elsewhere (like implementing mod_gzip).

My site is readable in Netscape 4 and Lynx. Hell, it's readable in Netscape 1 and Mosaic. It's unstyled but it's there. That text/xml blog is not. It defines a cutoff point for basic functionality (display), and sets it too high.

And once XHTML 2.0 comes down from on high, old browsers won't even understand links. LINKS! The single thing the web has going for it, and they changed the semantics of links. (In XHTML 2.0, anything can have an HREF attribute; you don't need a separate tag.) How long with it be before we can safely use *that* little nugget of a standard? In five years, will we all be arguing about supporting legacy browsers like IE 6 and Netscape 7? No, the entire world will have moved on and the field will have stagnated.

"HTML needs a RANT tag." --Alan Cox

HTML is the 3270 datastream of the late 20th century. There is no problem that can't be solved by a smart individual and a little screen scraping.

While XML + CSS would yield poor results in older browsers and would possibly yield bad effects in even recent browsers for certain purposes - wouldn't XML + XSLT provide a better answer as it would actually transform the data for its intended purpose - posting on the web, syndication, etc?

Look at http://www.terra.es/personal4/alsanan/xml/index.xml . I did it some months ago. I wanted RSS to be the place where my stories get placed. But the same XML has a reference to a XSL filter that makes the browser turn the tags into a beautiful webpage.

It's gonna be a grocely big redundant overhead if you start mixing a lot of different applications.
At this moment (2003-Jan) it's probably better to use server-side XSL Transformations, or to manually write seperate documents. *snuff*
Personally, we're still actively waiting for the user-side software revolution, that will catch up with the latest and most current standards.