Simon Willison’s Weblog

Markup answer: yep, that's the most appropriate tag to use. HTML's not all that good at relating stuff together in a sequential way: you've got definition lists which relate their DT and DD, a two-column table, which is a special case of the general "relate these things together in a tuple, and I've got lots of tuples" definition of "tabular data" (but a two column table seems a bit silly, which is why DLs exist :)) and label for with its corresponding form element, which is obviously inappropriate.

sil - 22nd October 2003 06:52 - #

I don't feel "confortable" using a *definition list* to present a Q&A sequence.

It seems to me that a sequence os paragraphs of two classes (question, answer) would be preferable.

Sérgio Nunes - 22nd October 2003 10:00 - #

I would say that a definition list is certainly not the appropriate tag. Suppose you want a list of all the definitions you have declared in you blog archives. I personally think, something like this is more appropriate:

<p class="qanda"><span class="question">question</span> <span class="answer">answer</span></p>

Martijn - 22nd October 2003 10:27 - #

Of course dl is appropriate. It's perfect for dialogue, transcripts, Q&A, and renditions of instant-messaging transcripts. Read the spec:

Another application of DL, for example, is for marking up dialogues, with each DT naming a speaker, and each DD containing his or her words.

Keep in mind that you can have multiple dts or dds in a row if you need them.

Joe Clark - 22nd October 2003 11:57 - #

The specification contradicts itself:

Definition lists vary only slightly from other types of lists in that list items consist of two parts: a term and a description.

A dialogue is not a list of terms and descriptions.

I personally think that the name and description of the <dl> element supercede an "oh, by the way" comment at the very end. You can read a thread about it in the www-html mailing list archives.

At the very least, would you say that the Q&A is a list of definitions? If not, then why use a definition list?

Jim Dabell - 22nd October 2003 12:07 - #

PS: Your anti-spam redirect drops fragment identifiers, as you can see with the link Joe provided.

Jim Dabell - 22nd October 2003 12:10 - #

The specs are ambiguous for sure, but I agree with Joe. Let's hope God does too, or you'll burn in hell :) I had another question: is it okay to change the markup of quotations, as you did? Granted, CNET uses ugly markup (a paragraph with bold tags and a break). So I would say yes, but then I can see how some people would say no, and yet other people would say maybe, etc. Any thoughts?

Lars - 22nd October 2003 12:26 - #

Jim: blast, they do as well. I can't think of a particularly elegant way to fix that either, without performing some kind of mangling on the link to encode the fragment # in a way that will pass through the redirect script's query path.

Simon Willison - 22nd October 2003 15:20 - #

Lars :

... is it okay to change the markup of quotations, as you did?

I've never seen anyone mention that before. But in the print world publishers readily change the "presentation" of an author's material, and so do those who re-publish:

The only alterations are of a ... typographical nature ... expansion of contractions, the use of single quotation marks [is this so that they can be confused with apostrophes?] ... reduction of excess capitalization [,] and clarification [sic] of the original erratic punctuation.

Source: Folio Society edition of Jane Austen. Smarmy b*stards! (Those are my additions in square brackets.)

But it is an interesting point. Henri Cartier-Bresson didn't retouch, or even crop, photos. For him that was a form of honesty. That seems a little extreme, but it's a valid point-of-view. How about historical documents on the Web? Say I want to post a letter from the Great War on the Web? Do I show linebreaks, crossings-out, mis-spellings? Are these part of the letter or not? Maybe the last two are but not the first one - but it's a tough call.

Michael - 22nd October 2003 15:51 - #

The "definition list" would wrong in the context you have used, mainly because you are referencing a "citation" from another site rather than defining a term.

Robert Wellock - 22nd October 2003 16:15 - #

For my own markup, I've arrived at a policy of flexibility when it comes to structures which don't have an explicitly defined set of tags for markup. The rationale behind this is that XHTML is not a complete specification (it's the fatherless cousin of HTML), so it will naturally not be able to address every imaginable semantic relation (something which, on the other hand, XML can do -- within certain boundaries, of course; c.f. family trees). So I think the use of DL to markup an interview could be legitimate.

Legitimate, but not covering all possible forms of an interview. What to do for a round-table or debate with more than two participants? Should a more abstract markup strategy be formed, then applied backwards to the two-participant model? Or can we apply two markup strategies to different conversations, differentiating on the number of participants -- like a far-field model and a near-field model? (sorry, physics background coming out here!)

As for Robert's comment, I think Simon took care of that by enclosing the DL in a BLOCKQUOTE, which even includes a CITE attribute pointing to the source of the quote. Furthermore, if you look at the XHTML source of this page, you'll see that the Source link after the quoted interview snippet is actually done through a Javascript bit written by Simon, thereby avoiding redundant markup; I think it's neat!

Micah - 22nd October 2003 18:52 - #

And, as Mark Pilgrim would say, Mu. ;-)

Seriously, when there's a clear-cut answer to a markup question, great. Otherwise, don't suffer much angst over the "right" way. It's likely others are muddled on it too, and that means the de facto semantics of the markup in question are not strong.

When that's true, the only meaning you can count on in the mess is what meaning you confer to it. So if, to you, dt/dd mean q&a, great. The rest of the web won't get that meaning out of it, but they can't make the assumption that dl is just word/definition, either.

It's too bad the semantics aren't clearer to begin with.

Jeremy Dunck - 22nd October 2003 21:11 - #

You're stealing Simplebits' fire, Simon. But seriously...

Of course it's a definition list. The text of the response defines what the question means to us in the context of this article.

Even in a dialogue, the dd defines what was spoken by the dt that particular time. This could be debatable, but debatable is good enough for me when I decide it's right for a particular document.

I think the spec was written with a calculated degree of specificity. The meaning of 'semantic' tags has built-in leeway to support the author's intent in many different contexts. Otherwise, the spec would have to include 1000 more tags: q-a lists, dialogue lists, phonebook lists, clan-affiliation lists, web navigation lists, blog-rolls, ad infinitum.

Michael Z. - 23rd October 2003 01:23 - #

An HTML usage cannot be "wrong" if the spec itself says it's OK. End of story.

Joe Clark - 26th October 2003 21:12 - #