RE: Liaison with CSS WG to provide a mechanism for expressing the style of document semantics

Robert -

Looking over this again, I still do not see how this fits in with any of our
goals for HTML 5. Since I am a relative newcomer to this group, I may not
understand those goals, but from my perspective, this proposal creates a
mechanism to put content into a style document.

Regarding the concern of legend reuse, I put forth that anyone using the
same legend that often may want to reconsider their usage; they most
certainly will want to not use plain, static HTML for such repetitive
content. I believe that while it may be a valid use case, it is not one that
we should work hard to *discourage* by making it more difficult, not
encourage by not only making it easier, but codifying it in the HTML 5
specification.

Take the following example of a logically equivalent (but bogus) proposal:

	Add a "value" property to CSS. Any HTML tag with the "value"
property set will be 	assumed to have either its "value" attribute set to
that value, or its CDATA value. 	An actual @value or CDATA content of
the tag itself will override any value property 	from the CSS.

I truly hope that this example illustrates the horror that I have at the
prospect of this proposal going anywhere. If we want authors to start
putting legend *text* in the style sheet, why not the contents of, say, the
menu bar? After all, it is pretty repetitive to type in those menus. Why not
the footer? There is another boring piece of text, ripe for inclusion in the
style sheet.

* And I am "positive" (sarcasm) that the "bad guys" out there would not use
this mechanism to cloak content or otherwise abuse the mechanism.* Right off
of the top of my head, I can think of a few deceptive practices which would
easily be modified to use this mechanism.

I still think this one should be shelved, it does not have any of our goals
in mind, and causes nothing but problems. The problems it *does* address
seem to be problems that are not the concern of HTML as a document format.

J.Ja

-----Original Message-----
From: Robert J Burns [mailto:rob@robburns.com] 
Sent: Thursday, May 29, 2008 5:06 AM
To: Justin James
Cc: 'HTML Issue Tracking WG'
Subject: Re: Liaison with CSS WG to provide a mechanism for expressing the
style of document semantics

HI Justin,

One more thing I meant to add to this reply. The use of CSS for the  
legend also permits reuse of legend information. Imagine the site-wide  
or community-wide use of a class selector.  For example a community of  
juridical authors might use the class "parties" to markup documents. A  
a set of alternate reusable stylesheets might be used to style all of  
the documents in this authoring community. There should be no reason  
to place the legend description within thousands upon thousand of  
document when a few stylesheets can simply have the declaration:

.parties {
	legend: "parties to the suit whether: plaintiff, defendant,  
petitioner or respondent";  /* or
         */ legend: uri(http:www.example.com/legend#parties)
	font-weight: bold;
         text-transform: uppercase;
}

To require the placement of the legend in the document itself would  
require it be repeated in every document. This does not prevent  
authors from referencing the same legend from their documents as well.

<link rel='legend' href='http:www.example.com/legend' >

or perhaps more precisely through RDF.

Take care,
Rob


>
> On May 29, 2008, at 4:33 AM, Justin James wrote:
>
>>
>> I can't figure out this wiki thing enough to put my comments on  
>> there, so I
>> will have to do it on the list instead (I suspect that the wiki is  
>> the
>> "right" place for it, but without a way of using it...).
>
> No, discussion here is certainly appropriate. I will try to add  
> links to the email discussion as it happens.
>
>> Looking at this, I am curious as to why in the world, after 10  
>> years of
>> begging people to separate their styling from their semantics, we  
>> would then
>> turn around and make a mechanism that allows people to embed  
>> content and
>> semantics (in this case, putting a string with a legend text is  
>> certainly a
>> form of content) into the style sheet. This really looks like a  
>> massive step
>> backwards. In this case, people should be using a tag in HTML with  
>> a *role*
>> of "legend" (and another attribute indicating the ID of the tag  
>> that it is
>> the legend of), with a stylesheet to style the legend itself. The  
>> legend
>> text does not belong in a *style* definition.
>
> I understand your point here. The URI value for the property does  
> permit the author to separate concerns in just the way you suggest.  
> That is the legend text can reside in the document itself, while the  
> presentation description gets automatically generated or handled  
> through the legend-presentation property. Perhaps I should include  
> just such an example with the legend in the original document.
>
> One of the reasons I turned to CSS for this is that I wanted it to  
> participate in the cascade. UAs might also assist users in  
> identifying duplicate presentation idioms resulting from the cascade  
> and correcting those. Imagine for example, the UAs default  
> stylesheet.  It might include:
>
> em {    legend: "emphasis"
>             font-style: italic;}
>
> Yes this legend is redundant with the specification of HTML.  
> However, it places the legend in the UA stylesheet where it can be  
> handled in the same manner with other legends. If the legend had to  
> reside in the HTML document, then the UA stylesheet could not also  
> provide the information. Likewise a user could not provide any  
> legend information, though they might also be interested in  
> selecting and styling a particular semantic. I'm a big advocate of  
> the separation of concerns, but there are times when CSS with its  
> cascade provides valuable abstraction and flexibility that cannot be  
> matched otherwise. Also it is important to provide authors multiple  
> ways to achieve the results they seek. For example, a stylesheet  
> author may identify particular semantics she wants to style with a  
> stylesheet, but not have write access to the original document so  
> this separate of concerns provides the needed flexibility to place  
> the legend in either place.
>
> Take care,
> Rob
>
>
>>
>> Here is another issue that needs to be introduced here for  
>> discussion,
>> as it will be added to the issue-tracker in time. This idea has been
>> discussed briefly within the WG and more off-list. I welcome
>> additional feedback now. As this involves nearly entirely only CSS,  
>> it
>> would merely be a rough draft proposed to the CSS WG from the HTML WG
>> if the WG decides to do so.
>>
>> Liaison with CSS WG to provide a mechanism for expressing the
>> semantics of styling.[1]
>>
>> [1]: <http://esw.w3.org/topic/HTML/SemanticPresentationLegendCSS>
>>
>
>

Received on Sunday, 1 June 2008 06:31:21 UTC