Re: Mobile content heuristics

Sean Patterson wrote:
> In section 4.1.2 of the FPWD of the CT Guidelines, there is no mention 
> of examination of URLs to determine whether or not to transform.  I’m 
> mainly thinking of URLs that can be identified as mobile.  Examples include:
> m.*
> mobile.*
> wap.*
> *.mobi
> <domain>/m/*
> <domain/mobile/*
> Probably a CT proxy wants to perform no or minimal transformation on 
> sites with URLs matching one of these patterns.  Do we want to say 
> anything about that in this section?  (I think we discussed this topic 
> at one point, but I can’t remember whether we decided specifically not 
> to say anything about this in the guidelines, or whether it just kind of 
> fell through the cracks.)

It neither was decided nor really fell through the cracks... It is 
indeed something we need to discuss, so I'm glad you start the 
discussion ;-)
Save ".mobi", which was typically created to flag content as mobile, 
nothing ensures that URIs in the above list actually reference mobile 
resources. And URIs are supposed to be opaque: a URI ending with a 
".html" does not imply that the response will be an HTML page... in 
theory, that is.

But then, what we want to do here is to err on the side of caution. So 
it could be both harmless and useful to say that the CT-proxy should 
examine the URIs to identify likely-mobile ones and not transform these 

Note this triggers another point: let's suppose I have an 
"" site that is not mobile-friendly. I'm not performing any 
kind of server-adaptation and CT would be useful, but as it starts with 
"m.", the CT-proxy leaves it untouched. I have no way to say: "You can 
transform". The only flag I can use is "do not transform!" (the 
no-transform directive). Well, yes, I know, this goes back to what we 
initially wanted to do with cache-control extensions. And there's no 
real other way to do that for the time being.

> In section 4.4, heuristic examples are given for determining whether a 
> CT proxy should transform.  The second heuristic mentions content type 
> and doc type.  Francois has said he plans to put an item on the agenda 
> for next week to discuss whether we should be more specific about what 
> content types should or shouldn’t be transformed.  We could also say 
> more about doc types.  For example, the following doc types are usually 
> only used on mobile sites:
> <!DOCTYPE html PUBLIC "-//OMA//DTD XHTML Mobile 1.2//EN" 
> "">
> <!DOCTYPE html PUBLIC "-//WAPFORUM//DTD XHTML Mobile 1.1//EN" 
> "">
> <!DOCTYPE html PUBLIC "-//WAPFORUM//DTD XHTML Mobile 1.0//EN" 
> "">
> <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML Basic 1.1//EN" 
> "">
> <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML Basic 1.0//EN" 
> "">

Indeed, content types and doc types were part of the same discusion in 
my mind.
What about WAP?

> Regarding the third heuristic in section 4.4, I think it would be 
> illuminating to the reader to include an example of “<link> elements 
> specifying alternatives according to presentation media type”.  I’m not 
> sure this phrase will be obvious to anyone not intimately familiar with 
> HTML/XHTML.  (The example would be something on the order of: <link 
> rel=”alternate” media=”handheld” href=”” />)

It would also go in 4.2 to replace the editorial note and clarify the 
otherwise obscure "the medium for which the presentation is intended 
SHOULD be indicated" sentence.

Received on Monday, 5 May 2008 09:35:01 UTC