Re: Feedback on CT Document

On 26 Nov 2008, at 16:22, Francois Daoust wrote:

> I guess we may remove "distributed" if that's not clear, the meaning  
> being that the user agent black box in that case is composed of a  
> local component (from the end-user's point of view) and a remote  
> one, and is thus distributed.

OK - that might be worth explaining. I googled around and couldn't  
find much reference to that term.

>> 2. JSON or AJAX requests might also fit the definition of "web  
>> browsing" but should probably avoid transcoding. I realise this has  
>> already been discussed...
> An AJAX HTTP request and responses cannot be easily, if at all,  
> identified.
> A JSON HTTP request can probably be if the "Accept" HTTP header is  
> restricted to (or would "contains" be enough?) "application/json". A  
> JSON HTTP response can be identified through the Content-Type HTTP  
> header.
> I would not mind adding a guideline that prevents transformation of  
> JSON requests/responses, but I am not sure it truly addresses an  
> existing problem. Are you aware of any existing problem with that?

I have seen transformations in the past which affected non-HTML  
content by appending advertising I think - it was in the distant past,  
though.

>> 3. In section 4.1.2, reference is made to no-transform headers  
>> being used in XmlHttpRequests... but no-transform isn't mentioned  
>> in the XHR spec that's referred to. Is this a suggestion that they  
>> be used as a matter of course in XHR? Seems a bit cheeky to insert  
>> this stuff into someone else's spec if so ;)
> I do not think that having Cache-Control: no-transform directives  
> used as a matter of course is such a good practice.
> The important stuff is that the XmlHttpRequest API allows developers  
> to add the directive.

OK, my reading of 4.1.2 was that this is standard practice in XHR,  
which is what made it a good example to quote - but the XHR spec  
doesn't mention it.

>> 4. Section 4.1.3 seemed to be a bit stronger previously, placing  
>> the onus on proxies to ensure they only transcoded web content. In  
>> the current version this requirement isn't there, and the doc says  
>> "careful, or you'll break something".
> The problem is that it only "seemed" stronger.
> We are trying to remove normative statements that look like wishful  
> thinking but cannot be enforced in practice. There is no normative  
> way to detect "Web content intended for regular browsing".

At the risk of retreading old ground (something I seem to have a habit  
of doing)... is HTML, XHTML and WML not browseable content (therefore  
ripe for transcoding), and therefore by definition everything else not  
browseable (and therefore not suitable)?

--
Future Platforms Ltd
e: Tom.Hume@futureplatforms.com
t: +44 (0) 1273 819038
m: +44 (0) 7971 781422
company: www.futureplatforms.com
personal: tomhume.org

Received on Wednesday, 26 November 2008 19:05:53 UTC