Highlight Words

Jukka K. Korpela wrote:

You refuted the principle of using logical markup instead of semantically
empty <span>,

If <span class="foo" > is semantically meaningless, then you implicitly
limit the only expressible semantics of a HTML document to those
expressed in the DTD itself.

Why are semantics important in HTML? They're not important to
browsers, they're not important to CSS, so if semantics are to assume
any importance, they can only do this through being targetted at some
hypothetical semantically-aware consumer. There is no reason why such a
device is limited to the DTD and can't make sensible use of class.

Dag Sunde wrote:

A real net-nanny...

No, not really. Jukka is also the most knowledgeable HTML / SGML
expert you're ever likely to want to meet. Sometimes though he manages
to combine extreme pedantry with extreme unhelpfulness (I practice
both, but I try to limit it to one at a time). The fact that he's right
(almost always) is no excuse for pretending that there's no possibility
of an alternative view that isn't entirely wrong, or for simply being a
jerk.

We should pity him. It's months since he saw daylight and they've
probably already got to the cannibalistic stage.

Scripsit Andy Dingley:

If <span class="foo" > is semantically meaningless, then you
implicitly limit the only expressible semantics of a HTML document to
those expressed in the DTD itself.

I'm disappointed. I thought, from a long experience about your Usenet
postings, that you understood WWW issues fairly well. Now you express
considerable ignorance about some essential aspects and, moreover, express
it using technobabble. In reality, a DTD has absolutely nothing to do with
semantics; it is pure syntax by definition.

--
Jukka K. Korpela ("Yucca")
http://www.cs.tut.fi/~jkorpela/

Jukka K. Korpela wrote:

Scripsit Andy Dingley:

If <span class="foo" > is semantically meaningless, then you
implicitly limit the only expressible semantics of a HTML document to
those expressed in the DTD itself.

In reality, a DTD has absolutely nothing to do with
semantics; it is pure syntax by definition.

Of course the DTD by and of itself is pure syntax, but what else are
you suggesting to communicate the shared semantics?

For one thing, your original comment was that <strong> was somehow
meaningful when <span> wasn't. This only holds if you accept that the
DTD (in combination with the HTML recommendation) conveys semantic
meaning.


Semantics aren't a publishing problem, they're a communication problem.
Publish what you want, how you want, but if no-one else can understand
it, then there's little point in doing it.

So how _do_ we communicate the semantic schema from producer to
consumer? One perfectly acceptable (and almost commonplace) way is to
use an attribute like class with values from a shared vocabulary.
Dublin Core's element list being an obvious simple example.

It's also possible to express this schema by publishing it as a binding
to structures apart from class, such as the set of elements expressed
in the DTD. They're only syntactic _in_ the DTD, they're a little more
than this if you include the HTML recommendation too (as you yourself
did in advocating <strong>), but they can be as sophisticated as you
wish if you merely use the DTD as a list of elements upon which to
attach some further meaning.

Your point was that <span> was meaningless, which seems to imply that
the use of class is somehow unworkable (it isn't, plenty of people have
been doing it successfully for years). If this were the case though,
what else is available to use? The only option I know of for HTML is by
implied attachment through the DTD and with some assumed context, such
as the HTML rec. This isn't large though - the DTD isn't expressive
enough to allow a precise semantics to be defined through it, no matter
how you attach or extend it. A list of a few dozen elements can't
usefully express many more than that number of slots.

Scripsit Andy Dingley:

Of course the DTD by and of itself is pure syntax,

Please stop babbling. It's pure syntax, period.

For one thing, your original comment was that <strong> was somehow
meaningful when <span> wasn't. This only holds if you accept that the
DTD (in combination with the HTML recommendation) conveys semantic
meaning.

Nonsense. Stop making DTDs something mystic. They are just a fairly trivial
though contrived formalism for defining some aspects of syntax rules. At
present, DTDs have actually no impact on how browsers behave. (The document
type declarations have an impact, but _not_ due to browsers actually using
or even opening a DTD.)

So they have _nothing_ to do with the semantic questions. The _meanings_ of
elements have been defined in the prose of the HTML 4.01 specification.

Semantics aren't a publishing problem, they're a communication
problem. Publish what you want, how you want, but if no-one else can
understand it, then there's little point in doing it.

You seem to have completely missed the fact that browsers have, in fact,
recognized <strong> as something that should be rendered in some highlighted
manner, even in the absence of CSS (and even before CSS was available). I
mentioned this as a particularly good practical reason to use semantic
markup and not a dummy element - used just as a container to which
attributes can be attached - when there is a choice.

--
Jukka K. Korpela ("Yucca")
http://www.cs.tut.fi/~jkorpela/

Jukka K. Korpela wrote:

Scripsit Andy Dingley:

Of course the DTD by and of itself is pure syntax,

Please stop babbling. It's pure syntax, period.

Then in that case, <strong> is no better than <span class="foo" >

You seem to have completely missed the fact that browsers have, in fact,
recognized <strong> as something that should be rendered in some highlighted
manner, even in the absence of CSS (and even before CSS was available). I
mentioned this as a particularly good practical reason to use semantic
markup and not a dummy element

So you see the best hope for the growth of a "Semantic Web" as being
the pure-presentation aspects of a 10-year old default rendering on a
tiny handful of elements. That's hardly pushing the envelope, is it?

John Dunlop wrote:

You _could_ mark up emphasised text with SPAN,

Jukka said that to do that would be "essentially wrong". It isn't.
Maybe it's not a best practice, but it's far from being so emphatically
wrong.

"John Dunlop" <usenet+2004 (AT) john (DOT) dunlop.name> writes:

According to the HTML4.01 specification, STRONG is emphasis
stronger than EM, and SPAN is a grouping element with no associated
"presentational idiom". In other words, while visual user-agents
generally render STRONG in bold text, it's reasonable to assume that
user-agents applying only their own stylesheets don't render SPAN
distinctly.

FWIW, as someone who's "wrong" answer started this debate, I should add
that this is precisely why I chose span instead of em or strong. For a
stylesheet-aware browser, you can suggest distinctive presentations for
both span-highlighted keywords and ordinary em or strong elements, so
that it's clear to the user which is which

Without stylesheets, the use of em or strong would make no distinction
between emphasised text in the original document vs. text that's empha-
sised by virtue of being a found keyword. That would be a potential source
of confusion for the end user.

So, I figured it would be better to highlight the keywords in a way that
would be less likely to cause confusion for users of non-stylesheet user
agents.

sherm--

--
Web Hosting by West Virginians, for West Virginians: http://wv-www.net
Cocoa programming in Perl: http://camelbones.sourceforge.net

Sherm Pendley escribió:

"John Dunlop" <usenet+2004 (AT) john (DOT) dunlop.name> writes:

According to the HTML4.01 specification, STRONG is emphasis
stronger than EM, and SPAN is a grouping element with no associated
"presentational idiom". In other words, while visual user-agents
generally render STRONG in bold text, it's reasonable to assume that
user-agents applying only their own stylesheets don't render SPAN
distinctly.

FWIW, as someone who's "wrong" answer started this debate, I should add
that this is precisely why I chose span instead of em or strong. For a
stylesheet-aware browser, you can suggest distinctive presentations for
both span-highlighted keywords and ordinary em or strong elements, so
that it's clear to the user which is which

Without stylesheets, the use of em or strong would make no distinction
between emphasised text in the original document vs. text that's empha-
sised by virtue of being a found keyword. That would be a potential source
of confusion for the end user.

So, I figured it would be better to highlight the keywords in a way that
would be less likely to cause confusion for users of non-stylesheet user
agents.

Perhaps it could be <strong class="keyword"> instead of <span
class="keyword">. At least this makes explicit the intention of being
emphasized.

Regards.
--
Manuel Collado - http://lml.ls.fi.upm.es/~mcollado

Manuel Collado <m.collado (AT) fi (DOT) upm.es> writes:

Sherm Pendley escribió:

Without stylesheets, the use of em or strong would make no distinction
between emphasised text in the original document vs. text that's empha-
sised by virtue of being a found keyword. That would be a potential source
of confusion for the end user.

So, I figured it would be better to highlight the keywords in a way that
would be less likely to cause confusion for users of non-stylesheet user
agents.

Perhaps it could be <strong class="keyword"> instead of <span
class="keyword">. At least this makes explicit the intention of being
emphasized.

Like I said - without the stylesheet, there would be no visual difference
between <strong class="keyword"> and <strong>. It's not explicit at all;
the user is left with no easy way to know what's an emphasized keyword and
what's emphasized in the document itself.

IMHO, <span class="keyword"> has a more graceful fallback. In the absence of
stylesheet support, I think it would be preferable to omit the highlighting
entirely, rather than highlight keywords in a way that may cause the end user
some confusion.

sherm--

--
Web Hosting by West Virginians, for West Virginians: http://wv-www.net
Cocoa programming in Perl: http://camelbones.sourceforge.net