Re: storing info in XSL-FO: new issue? [was: Draft TAG Finding:...] wrote:
> What is the "meaning" of a headline (sic)?

Put it this way: is there any reason you put what you put in your email 
subjects instead of, say, your sig? Would you buy a newspaper that 
printed out in bold letters on its front page the text of any random 
paragraph instead of something that allows you to know in a nutshell 
what the content is?

> Is there, for example, a worthwhile semantic difference between the 
> following?
> <h1>"Formatting Objects Considered Harmful"</h1>
> <p>Often repeated</p>
> <p>Poorly argued</p>
> and
> <p>""Formatting Objects Considered Harmful"</p>
> <ul>
> <li>Often repeated</li>
> <li>Poorly argued</li>
> </ul>
> and
> <svg:text>"Formatting Objects Considered Harmful"</svg:text>
> <svg:tspan>Often repeated</svg:tspan>
> <svg:tspan>Poorly argued</svg:tspan>
> </svg:text>

Well yes, I think it's fairly obvious... If I want to extract headlines 
from a document in order to generate a summary or a ToC, I'll be able to 
do it with the first one but not with the two others. Similarly a 
list-collasper will only work on the second one.

The semantic difference is strong and worthwhile to the point of being 

> To a human reader the meaning is pretty clear but to pretend that the 
> HTML variants shown (and there are many others) contain some sort of 
> immutable semantics is, I suggest, illusory.

I'm not sure what you mean by "immutable". I agree that "headline" 
probably meant something different (if anything at all) in ancient 
Greece... Apart from that I don't think anyone needs a crash course in 
ontology to know the difference between a paragraph and a headline.

If you really believe what you're saying, why do you use anything other 
than span and div?

> Are you seriously suggesting that SVG too is "harmful" and should be 
> abandoned because it lacks the historical domain-specific idiosyncracies 
> of HTML?

I don't know what Håkon is suggesting and won't answer in his stead but 
I don't think that SVG is harmful. The fact that it is 
presentation-orientated doesn't make much of a difference. In fact it 
does have what could be construed as domain-specific idiosyncracies 
given that it has some semantics, however shallow (a rectangle is a 
rectangle is a rectangle no matter what Illustrator thinks). Besides 
your example above is somewhat flawed. The "G" in SVG stands for 
"Graphics", not for "Text".

Besides, imho the best practice for SVG is to use it to render semantic 
content graphically as has been done to a limited extent with "taglibs". 
I don't know enough about XFO to argument a strong case, but I will not 
consider it harmful *if* people use it by sending semantic content which 
  is rendered in an XFO renderer after some transformation. If all 
that's sent is the XFO output, then yes I think it is harmful.

Robin Berjon <>
Research Engineer, Expway

Received on Friday, 16 August 2002 10:53:06 UTC