W3C home > Mailing lists > Public > www-html@w3.org > October 2002

Re: FW: OL needs the start attribute

From: Sampo Syreeni <decoy@iki.fi>
Date: Thu, 17 Oct 2002 23:21:39 +0300 (EEST)
To: Bertilo Wennergren <bertilow@gmx.net>
cc: "'www-html@w3.org'" <www-html@w3.org>
Message-ID: <Pine.SOL.4.30.0210172217030.2674-100000@kruuna.Helsinki.FI>

On 2002-10-17, Bertilo Wennergren uttered to 'www-html@w3.org':

>If the details of the numbering are essential, and the document would
>suffer if those details are not presented exactly as whished, then they
>are part of the content, and should be specified directly in the

If the details of the numbering are essential, then what we're talking
about isn't an HTML ordered list, but a DL or a TABLE, both of which have
the capability of actually associating fixed labels with the list entries.

At the same time we all know that using DL's and TABLE's isn't really
good, either. There is a simple reason for this. Namely, the idea of a
numbered list has been presentational from the very beginning.

Numbered list items on paper serve a double role. First, they distinguish
an ordered (numbered) list from an unordered (bulleted) one. Second, the
numbers work as a way to refer to the items, individually. Where we talk
about an unordered list it's usually sufficient to refer to "one of the
items, above". Where an ordered list is involved, we need to distinguish
the items based on their position in the order. Hence, we utilize the
numbering: "item 2 in the above list". At the same time the numbers
indicate that the "block progression direction" of the content coincides
with the order relation of the data being represented by the list.

In hypertext there's no need for such constructs. If references need to be
made, they can be made with links. If order needs to be discerned, that
can be accomplished through any of a *huge* number of ways. What remains
of the semantics of numbered lists is plain order. Not the reference to
numbers (i.e. item labelling), but the plain order. That's what HTML
encodes, because more is not needed in the context of hypertext. Getting a
list to look like the way it has to be printed on paper is presentation,
and so is strictly secondary to the structural semantics of HTML.

Of course, that's not the whole story. Once there's lots of any particular
representation going around, the representation itself is bound to be
extended. It will be used to stand in for things it originally didn't, and
adapted to needs it originally didn't serve. Thus the way ordered lists
are represented on paper lead people to conceive of lists as arbitrary
labels associated with specific content, instead of visual indications of
order merged with link endpoints.

The "3a" example is about representing revision history, while inversed
numbering is mostly about poetic value. The latter is used when presenting
a list in straight order for some reason doesn't sound as good. (E.g.
countdowns to something, acknowledgeing winners, and the like, or just
artistic effect.) If either of them is coding-worthy, they should acquire
their own, explicit representation in HTML.

That's simply because HTML does away with that sort of thing. Its lists
represent order, as a physicist would conceive it -- there either is an
order, or there isn't.  Definition lists seem rather out of place, here,
until you realize that they're meant to stand in for all of the other
stuff, all the ways the list syntax has been used to map generalized
labels to bits and pieces of content. In this framework, a totally ordered
relation will be represented by an OL, a bag by a UL, and a discrete
function by a DL.

If you want things like "3a", inverse numbering, or whatever, you'll use
DL until there's a dedicated semantic structure in HTML which encodes what
you mean. You don't use the representational aspects of normal lists to
mimic what you want to mean, however far they could theoretically be

>If it's in CSS then it's just eye-candy. Eye-candy is OK, but then you
>should be able to live with a totally different presentation and a
>totally different numbering scheme.

You should, indeed. That's pretty much the point, in its entirety.
Sampo Syreeni, aka decoy - mailto:decoy@iki.fi, tel:+358-50-5756111
student/math+cs/helsinki university, http://www.iki.fi/~decoy/front
openpgp: 050985C2/025E D175 ABE5 027C 9494 EEB0 E090 8BA9 0509 85C2
Received on Thursday, 17 October 2002 16:21:41 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:06:01 UTC