Re: text/html fallback using the Same Bytes

On Jan 7, 2011, at 19:45, Sam Ruby wrote:

> Full of gotchas.  Oh my goodness and golly.  More dire claims!  We're doomed?  Whatever will we do?

In the standards circles, it's a common response to ridicule claims that something is hard. That doesn't mean things aren't actually hard.

When well-formedness was claimed to be hard, Tim Bray reacted by calling anyone who can't get it right a bozo (http://www.tbray.org/ongoing/When/200x/2004/01/11/PostelPilgrim). Yet, it seems that well-formedness is really, really hard (http://lists.w3.org/Archives/Public/www-archive/2009Mar/0060.html).

Polyglot has to be harder than well-formedness, because polyglot documents have to be well-formed *and* they have to adhere to other rules as well. I have a reason to believe you might even actually agree (http://krijnhoetmer.nl/irc-logs/whatwg/20090326#l-475), but let's go over this once again for the benefit of those on *this* mailing list who haven't read the logs of all of our IRC discussions.

On any given topic, most people don't do their own critical thinking. Instead, they imitate some people who they think have it already figured out. After all, no one has the time to do his/her own critical thinking on every topic. Since people aren't doing their own critical thinking on every topic, they can't really make informed choices when they don't do their own thinking. Therefore, it matters what the default attitude communicated to people who aren't doing their own thinking (on a particular thing) is.

I don't have a problem with an outcome where the default attitude is not to do polyglot but people who do their own thinking (the kind of people who hang out on mailing lists such as this one) and arrive at the conclusion that they really need polyglot and go ahead and use polyglot markup.

What I do have a problem with is flipping this around: Generating an appearance that in general one should want to use polyglot markup and only private critical thinking can free a person from inflicting this additional hardship on oneself. I think it's wrong to make people who rely on the W3C to have done serious thinking about HTML and XML so that they don't have to (cue Ze Frank) do useless work (i.e. jumping through the polyglot hoops when they don't need to).

My concern is based on precedent. The "Polyglot Markup" publication of the HTML WG is basically Appendix C done better. We can look back as see what happened with Appendix C.

The XHTML2 WG (then called the HTML WG) promoted the use of XHTML 1.0 as text/html and promoted the use of the Appendix C. This promotion lead people to think that serving XHTML as text/html is some kind of a good thing. (If it wasn't, why would the smart people at the W3C promote it?) The WaSP posse chose to make "XHTML" their buzzword for promoting CSS-oriented authoring. Since XHTML as application/xhtml+xml didn't work out in practice (where practice included supporting IE), they promoted XHTML as text/html. So the word from both the W3C and the advocates on the conference circuit was that you were supposed to use XHTML as text/html.

Yet, XHTML served as text/html offered no actual beneficial effect in browsers. Moreover, since getting well-formedness right is exceptionally hard if you don't keep checking for it, most "XHTML" served as text/html wouldn't work with an XML parser anyway, so the theoretical benefit of XML compatibility is bogus in practice.

So what's the harm? Well, the first harm is that a lot of people did a lot of useless work trying (and most often failing) to make their content be XML in addition to working as text/html.

The second harm is that strongly held but weakly founded beliefs of the masses of Web authors generate useless requirements on software that's already quite OK for practical purposes. What prompted me to write http://hsivonen.iki.fi/wannabe/ was that I recommended using a tool and someone else smugly argued back that the tool was no good because it wrote the HTML element and attribute names in upper case.

More recently, I've noticed that Daniel Glazman (an authoring tool developer) asks me questions about polyglot from time to time on irc.mozilla.org. Even though he realizes that polyglot is pain, he expects his users to expect polyglot. That is, the expectation about an expectation to support polyglot causes pain to an authoring tool developer.

So when I have a generally negative attitude towards polyglot, I don't have a problem with the people on this mailing list or on the TAG using polyglot markup silently for their own purposes. I have a problem with generalizing from the perceived needs of specialists and making the polyglot appears as something that's worthwhile in the general case. Simply publishing the "Polyglot Markup" document without dire warnings looks like an endorsement. Someone will pick up polyglot and hit the conference circuit advocating for it for all purposes. After all, there has to be some magic in it if the W3C bothers to write a publication about it! Then, if the W3C publication itself doesn't come with warnings saying that you shouldn't bother with polyglot unless you've done serious thinking and you are sure you need it (so that you can refute bogus claims about polyglot by pointing to the "Polyglot Markup" doc itself), we'll have a decade of Appendix C redux with Web authors jumping through useless hoops and asking authoring tool developers to jump through hoops in order to meet perceived but most often unreal needs.

-- 
Henri Sivonen
hsivonen@iki.fi
http://hsivonen.iki.fi/

Received on Monday, 10 January 2011 09:32:19 UTC