Re: Proposal: restricting <LINK> to hyperlinks

From: (Murray Altheim)
Date: Fri, 25 Aug 95 17:14:01 EDT


I believe you are misunderstanding what I was responding to. You didn't
include the paragraph you were responding to, which is what I was

>Bert Bos <> writes:
>>This is a suggestion for a change of terminology in HTML and for a
>>slightly different interpretation of the LINK element. In short: the
>>term `hyperlink' should be used more in accordance with existing
>>hyperlink literature and the LINK element should consistently result
>>in a button in a toolbar, independently of any REL attributes.
>>>I don't think so.  Link is intended to define abstract *and* concrete
>>>relationships between resources on the Web.  Link is independent of
>>>HTML and does not in general define a button on a toolbar.  Instead,
>>>a toolbar widget seeks out links associated with the relations
>>>that are represented within the toolbar (i.e., the button is already
>>>defined, it just takes its reference value from a link href).

I wrote:
>>Wasn't the idea of a WG to work out these concepts? It is precisely the
>>current ambiguity surrounding META and LINK I believe Bert is trying to
You responded:
>No.  The purpose of this WG is to resolve the syntax and semantics
>of HTML and produce a specification that accurately describes them.

Bert has proposed to tighten the definition of LINK. Whether you believe
there is ambiguity over META and LINK is a matter of opinion. I believe
resolution of this matter falls within the purpose of the WG as you state:
we are trying to resolve the syntax and semantics of HTML in accurately
descriminating between LINK and META in a formal definition.

You state that LINK "does not in general" define a button on a toolbar. I
agree, in that LINK defines a link relationship, not a UA behaviour. I
don't believe it necessary or appropriate to specify UA behaviour with
regard to LINK, nor from reading Bert's message do I believe he does
either, merely the "toolbar test"[1] he describes is an informal way of
deciding what type of information would be appropriate to LINK as opposed
to META. We obviously would need this to be formalized into specification
language, but Bert's five criteria for LINK usage certainly doesn't
contradict current definition, only further tightens it.

>Finally, there is NO ambiguity surrounding META and LINK.
>Yes, their functionality does overlap, but that is not ambiguous.

I don't think Bert's proposal, my ideas and your ideas regarding the
definition of LINK and META are at odds. I disagree that there is no
ambiguity surrounding META and LINK; discussions within this WG attest to
this fact. If the specification for both is not explicit as to where one
leaves off and another begins, how is this not ambiguous?

But before you and I go off and further argue over ambiguity, remember that
I agree with your definitions completely:
>META assigns unstructured metainformation content to a name.
>LINK defines a hyperlink relationship between the source document
>and the resource defined by the parameters to the LINK element.
>LINK is a special-case of META -- this specialization is significant
>and useful to WWW applications.

especially the last key sentence, which I believe Bert is proposing to
tighten with a more specific language. For example, in current definition
it would be unclear as to how to establish the relationship between a
document and its stylesheet. META or LINK?

Joe English <> writes on Fri, 25 Aug 95 15:14:02 EDT: [2]
>To handle links to metainformation resources,
>perhaps an HREF attribute could be added to <META>:
>        <META NAME=stylesheet HREF="...">
>would be used instead of
>        <LINK REL=stylesheet HREF="...">

This would allow META to handle document meta-information (such as a
stylesheet) without introducing the contrary usage of trying to LINK
(definitional "hyperlink") to the stylesheet and display it in a UA window.
This preserves LINK for true hyperlinks to resources that can be viewed or
otherwise rendered.

Further discussion of stylesheet implementation should be upheld for future
versions of the language, but the 2.0 definition of LINK and META should be
clear enough to provide a strong direction for resolution, ie.,
specification of our semantic model should match our intention.

>NOTE TO THE CHAIR: If you want to expand the scope of the charter to
>include all Link Relationships, then please request the IESG to add it
>to the Charter.  However, I think it would be better for the WG to focus
>on its current work.

I would not support this proposition. I had no intention of continuing the
discussion of Link Relationships in general as a point for resolution by
this WG. I agree with you that it is outside the scope and have ended my
discussion of this issue within this forum at Dan's request. Murray
Maloney's paper and other discussions of LINK attributes belongs in a
different forum, such as www-html or others.


[1] Bert Bos message "Re: Proposal: restricting <LINK> to hyperlinks" to
HTML-WG of Fri, 25 Aug 95 09:31:54 EDT.

[2] As an aside, as per Joe's suggestion, I would highly welcome the
addition of the HREF attribute to META, though as part of 3.0, _not as part
of 2.0_. As part of my current paper on HTML as a container for database
information, I would like to be able to reference lookup tables without
having them show up as browseable documents. META with HREF would solve
this, while LINK with HREF would be a potential source of confusion. I
propose moving this discussion to www-html.

      Murray M. Altheim, Information Systems Analyst
      National Technology Transfer Center, Wheeling, West Virginia