RE: URIs, content adaptation, DISelect and XSLT

Splitting text can be trivial, or complex, depending on the nature of
the text. Where there is much contextual dependency within the text,
splitting it can cause misinterpretation of the content by the end user.
Related text needs to be kept together, the order of consumption needs
to be maintained. Internal textual references need to be avoided (e.g.
"... as mentioned on the previous page ...") because the page flow has
been augmented by the internal pagination. Where forms are involved,
splitting of the forms needs to be done in a manner that does not break
the data model, nor separate closely bound controls (e.g. a field and
its label). So, while at first, splitting may seem trivial, I can
confirm from experience that it is not.

---Rotan.

PS I would prefer that we do not use the term "chunk" as this is already
an established term within the HTTP specifications, as in the "chunked
transfer encoding", which has nothing to do with automated pagination.



-----Original Message-----
From: public-uwa-request@w3.org [mailto:public-uwa-request@w3.org] On
Behalf Of Dave Raggett
Sent: 12 December 2007 15:03
To: Max Froumentin
Cc: Smith, Kevin, VF-Group; public-uwa@w3.org
Subject: Re: URIs, content adaptation, DISelect and XSLT


On Tue, 11 Dec 2007, Max Froumentin wrote:

> As for Dave's original question about pagination, I'm not sure how 
> an XPath function would work. Dave, could you send an example or 
> two? One way of doing it without a function is to enable 
> pagination URI parameters, like ?page=. Then when the browser 
> requests http://www.example.com, then the server which has decided 
> to paginate returns the first page of HTML, including a link to 
> http://www.example.com/?page=2. That seems to do the trick.

If the content were created with chunking in mind, it could include 
markup indicating good places for boundaries. For example, blogging 
tools let you indicate where to split large blogs, although this 
tends to use a comment, which is a bit of a hack. I can imagine 
using span, p or div element with a class, but it doesn't really 
matter just so long as the XSLT can pick it out.

When comes to splitting text programmatically, I would think that an 
XPath function would be feasible. Such a function could look for 
good places to split the text based upon criteria set through 
paramaters passed to the function. The function would return the 
n-th chunk of the text. I haven't written such a function but don't 
think it would be particularly hard to do.

You're right that you could put the chunk identifier as part of a 
query string. Another approach is to include it as part of the URI 
itself and use some URI rewrite rules to map the requested URI into 
what you need to pass to XSLT, e.g.

   http://www.example.com/chunk2843

where the string chunk2483 is generated by the chunking process.

Of course if the page is using CSS to reposition the content out of 
the markup order, then this will need to be taken into account when 
determining which content to include in each chunk. The adaptation 
process needs to act on both markup and styling ...

Cheers,

  Dave Raggett <dsr@w3.org> http://www.w3.org/People/Raggett

Received on Wednesday, 12 December 2007 15:10:40 UTC