Re: further proceeding

From: Ryan Lee <>
Date: Thu, 26 May 2005 17:47:53 -0400

Emmanuel Pietriga wrote:
> Ryan Lee wrote:
>> Emmanuel Pietriga wrote:
>>> I think that for this specific point we should discuss requirements.
>>> What I'd like to have:
>>> - be able to associate CSS instructions with specific graphical
>>> regions/entities of the displayed document. By regions and entities,
>>> I mean things like "the object that represents the resource", "the
>>> object that represents the property name", "the object that
>>> represents the property value".
>> Agreed, though by 'associate' I don't mean including CSS fragments in
>> Fresnel but allowing custom hooks to identify those regions for later
>> CSS selection.
> Yes.
>>> - be able to apply styling instructions based on conditions at the
>>> source RDF model level, i.e. be able to say things like "color in red
>>> all ex:age properties that have a value > to 47".
>> This is not a core requirement, is it? I agree that being able to
>> distinguish part of the RDF model for later styling is important, but
>> complex conditions implies that it's outside of core's capabilities.
> No, this is not part of core. But I think it would be really useful. And
> I guess it has consequences on the overall design.

Perhaps if the style hook proposal is general enough it could be adapted
for this.

>>> I also would like to have a well-defined intermediate tree format.
>>> Ryan, I haven't had time to look at the code you wrote. Did you
>>> already make a choice w.r.t that issue?
>> This is what it looks like right now, based solely on the selection
>> phase:
>> <results>
>> <resource label="Eric Miller">
>> <property label="name" uri="">
>> <value label="Eric Miller"/>
>> </property>
>> </resource>
>> <resource label="Ryan Lee">
>> <property label="name" uri="">
>> <value label="Ryan Lee"/>
>> </property>
>> <property label="Surname" uri="">
>> <value label="Lee"/>
>> </property>
>> <property label="depiction" uri="">
>> <value
>> label=";size=sm"
>> uri=";size=sm"/>
>> </property>
>> </resource>
>> ...
>> </results>
> A few comments:
> - wouldn't the value's text (which I guess label represents) be better
> put in a #PCDATA node? I'm thinking about multiline values, and more
> generally about long values. This only applies to value, not property.

Yes, it does make sense. I think I was just making it slightly easier
for myself to make a working example; not a design decision.

> - do we want a namespace for results/resource/property/...?

Yes. I made one up for my purposes, but we should agree on one.

> - Isn't there some kind of XML serialization of RDF that looks a lot
> like this? If so, should we consider using (and extending) it?

I don't know, but it's going to be different when we put the
non-presentational styling in (see next) - I suspect that will impact
whether we can conform to an existing XMLSchema?

> A question:
> - in your view, how does the non-presentational part of styles fit into
> this?

Besides changing text labels, I do anticipate putting explicit content
elements in so as not to affect the value text (i.e., commas for lists
not included in values themselves but as separate elements). Also class
attributes on arbitrary elements, I think.

Ryan Lee       
W3C Research Engineer    +1.617.253.5327
Received on Thu May 26 2005 - 21:46:23 EDT

This archive was generated by hypermail 2.3.0 : Thu Aug 09 2012 - 16:40:51 EDT