W3C home > Mailing lists > Public > www-html@w3.org > November 2004

Issues with <dl> [continued from Re: Concerns about the "l" element name <l>]

From: Thomas O'Connor <me@ocoth.id.au>
Date: Fri, 5 Nov 2004 10:18:57 +1100
Message-ID: <016701c4c2c4$aa6aa1b0$4d607acb@dellwin>
To: <www-html@w3.org>

> Here we go again. A definition list is a list of definitions. If you don't
> know the definition of "definition", or more exactly don't care about it,
> you can use <dl> for whatever you like - for formatting.

I do know the definition of definition.  I do care about it.  I do see how there are certain problems with it, as you bring up in
your lengthy discussion[1], however you a taking a strict* view on the meaning of definition in the context of a "general-purpose
markup language designed for representing documents for a wide range of purposes across the World Wide Web" by supplying a
"generally useful set of elements" and that does not aim to provide "every possible markup idiom"[2].

And it isn't being used for formatting.  It is being used to associate terms with values, whether it be words with meanings,
speakers with speech, properties with values, labels with elements - which are all semantically valid under a loose* interpretation.

> And the W3C has
> shown the way in the grossly self-contradictory section on <dl> in the
> HTML 4 specifications.

Grossly self-contradictory? Surely it is a matter of interpretation, and is only self-contradictory from a strict* view point.

> There should be no <dl> element in XHTML 2.0, unless it is intentionally
> designed to be compatible with traditional abuse of <dl> in HTML.

Rather than "abuse", this alternative use is formed from a different interpretation of the purpose of <dl>, which the W3C themselves
has used in implementation[3].

> Shouldn't we _first_ have markup
> for a single definition, before considering a list of definitions (which
> would appear to be a simple compound idea, composed of "list" and
> "definition")?

We do have markup for a single definition, <dfn>[4], however you already are aware of this.  I gather by suggesting a "simple
compound idea" you mean that rather than having two separate methods each with their own structure like we have now, we have a
method that is simply repeated as the number of definitions increases.

> If you mean a set of name:value pairs, why do you call it a definition
> list? And what's the _meaning_ (semantics)? It sounds like pure abstract
> structure.

You are defining the value of certain properties.  It isn't pure structure.  The meaning is that it associates one term, with a
value, whether it be its meaning or value. It "defines" something, in a loose* manner.

The W3C in its HTML 4.0 recommendation[5], XHTML 2.0 draft [6] and validator implementation[3] of <dl> does not imply or act from a
strict view point, but in fact it implies that <dl> is actually loose*.

After discussing this with many members of a web development IRC channel[7], it has become apparent that people are generally split
on the issue, with some taking the strict* view of "definition", like Jukka has, and others following the apparent loose* W3C view.


* by loose in this context I mean not limited to words and their meanings, but any term and it's value, so therefore by strict I
refer to the limiting to words and their meanings,

[1] http://www.cs.tut.fi/~jkorpela/def.html
[2] http://w3.org/TR/xhtml2/#abstract
[3] http://validator.w3.org/check?uri=http%3A%2F%2Fw3.org - "Tip Of The Day" is marked up in a <dl>
[4] http://w3.org/TR/xhtml2/mod-text.html#edef_text_dfn
[5] http://w3.org/TR/html4/struct/lists.html#h-10.3 - Two examples, one strict* - the "dictionary" list, one loose* - the dialogue
[6] http://w3.org/TR/xhtml2/mod-list.html#sec_11.1. - Two examples, one strict* - the "dictionary" list, one loose* - the dialogue
[7] irc://irc.freenode.org/web


Thomas O'Connor, me@ocoth.id.au, http://ocoth.id.au/

----- Original Message ----- 
From: "Jukka K. Korpela" <jkorpela@cs.tut.fi>
To: <www-html@w3.org>
Sent: Thursday, November 04, 2004 10:35 PM
Subject: Re: Concerns about the "l" element name <l>


>
> On Thu, 4 Nov 2004, Thomas O'Connor wrote:
>
> > I believe that it is vital for semantic purposes that definition lists
> > remain how they are in HTML 4.0 through to XHTML 2.0 unless
> > another name:value element set is included.
>
> Here we go again. A definition list is a list of definitions. If you don't
> know the definition of "definition", or more exactly don't care about it,
> you can use <dl> for whatever you like - for formatting. And the W3C has
> shown the way in the grossly self-contradictory section on <dl> in the
> HTML 4 specifications.
>
> There should be no <dl> element in XHTML 2.0, unless it is intentionally
> designed to be compatible with traditional abuse of <dl> in HTML.
>
> If you want something that really means a list of definitions, you should
> at least name it differently, to avoid association with <dl> and to make
> the markup less cryptic. On the other hand, you might also consider how
> useful such a specialized construct is. Shouldn't we _first_ have markup
> for a single definition, before considering a list of definitions (which
> would appear to be a simple compound idea, composed of "list" and
> "definition")? This would get rather complicated, since it's a real
> problem and not just a made-up one, see
> http://www.cs.tut.fi/~jkorpela/def.html
>
> > Definition lists allow for a huge number of name:value pairs to be
> > semantically marked up in a way that cannot be reproduced through any
> > other element set.
>
> If you mean a set of name:value pairs, why do you call it a definition
> list? And what's the _meaning_ (semantics)? It sounds like pure abstract
> structure.
>
> > Of course there are always tables, but they are more
> > restrictive in their styling and positioning, plus are a bit of overkill.
>
> If you mean that a special case of a table, namely a table with two
> columns, needs specialized markup, then let's discuss _that_ (and not
> definition lists). And if you have problems in formatting, then that's a
> styling issue - and shouldn't it be solved in a manner that helps in
> formatting, say, 1-column and 3-column tables, too?
>
> -- 
> Jukka "Yucca" Korpela, http://www.cs.tut.fi/~jkorpela/
>
>
>
Received on Thursday, 4 November 2004 23:19:01 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 27 March 2012 18:16:01 GMT