Re: Definition lists - prior discussion of DL content model from

Albert Lunde (
Mon, 28 Jul 1997 19:41:58 -0500

Message-Id: <v03010d00b002e999f9c8@[]>
In-Reply-To: <>
Date: Mon, 28 Jul 1997 19:41:58 -0500
From: Albert Lunde <>
Subject: Re: Definition lists - prior discussion of DL content model from

[following up on my earlier post about DL]
>> The HTML 2.0 specification contained the following in the
>> description of DL:
>> | The content of a <DL> element is a sequence of <DT> elements
>> | and/or <DD> elements, usually in pairs. Multiple <DT> may be
>> | paired with a single <DD> element. Documents should not
>> | contain multiple consecutive <DD> elements.
>> Nevertheless, the declaration in the DTD was
>>       <!ELEMENT DL  - -  (DT | DD)+>,
>> allowing an arbitrary sequence. This declaration has remained
>> unchanged in 3.2 and also in the 4.0 draft, whereas the
>> respective texts don't contain the precise remarks about
>> element repetitions any more: The 3.2 spec just gives an
>> example for such a list, nothing more, the draft says that
>> "list items consist of two parts: an initial label and a
>> description", but still fails to explain precisely which
>> sequence is recommended, allowed or required.
>> Does that mean that any order is now considered `proper', or
>> has that part just been overlooked when the specs were
>> written? If the latter, why isn't the "should not" from the
>> quote above made a "must not" by choosing (DT+, DD)+ as the
>> content (which would, in addition, eliminate the possibility
>> of a DD as the first element in such a list)?
>I think the form of this part the HTML 2.0 DTD was discussed on the IETF
>html-wg list. I don't recall the details but I think it
>was a deliberate decision at that time, perhaps to follow
>pre-existing practice.
>    Albert Lunde            

Below are URLs to prior discussions of the content model of DL from the
html-wg (who wrote the HTML 2.0 spec). These are roughly in thread order,
as per the archives:

A number of content models were suggested... Here's a few quotes out of
context to suggest what was on people's minds:
= = =
>We talked about this at the meeting in Chicago, and I was
>never happy with the result. It is undoubtedly true that
>a document can have multiple terms followed by a single
>definition, and a single term with multiple definitions.
>While it is possible to argue that a similar visual effect
>can be achieved by using <BR> or <P> elements within either
>a <DT> or <DD>, that is not the point.
>I think that the content model for DL should be:
><!ELEMENT DL - - (DT+,DD+)*>
>The verbiage could go like this:
>The content of a DL element is a sequence of DT elements
>and DD elements, usually in pairs. However, multiple
>DT elements may be paired with a single DD element, and
>a single DT may be paired with multiple DD elements.
>I have probably overlooked some obvious reason why someone
>would want to have a <DT>+ without any corresponding <DD>,
>or vice-versa. I'm sure that some kind soul will point out
>the error of my ways. Thanks in advance. :-)
= = =

>Basically, the DTD defines what is legal. One of the problems though
>is backward compatability with legacy documents. Many HTML documents
>use only DDs within DLs. I believe the specification text is trying to
>instruct on the intended usage of the DL element.
= = =
>> Ok... we're trying to specify current practice, but we have on many
>> occasions decided that some practices are bogus and won't be part of
>> the standard.
>I'm still not convinced we should specify current practice where it's
>as bogus as (DT|DD)+. Either DL is a discussion list (in which case we
>want something like (DT+|DD*)*) or it's not (in which case we allow
>almost anything).
>If users are random enough to have used DD to indent a paragraph when
>they could have used BLOCKQUOTE I don't think we should encourage them.
>Better we should plan for align= and margin= in the future.
= = =

    Albert Lunde