W3C home > Mailing lists > Public > ietf-http-wg@w3.org > April to June 2009

Re: NEW ISSUE: content sniffing

From: Daniel Stenberg <daniel@haxx.se>
Date: Thu, 2 Apr 2009 09:19:17 +0200 (CEST)
To: Adam Barth <w3c@adambarth.com>
cc: Mark Baker <mark@coactus.com>, Julian Reschke <julian.reschke@gmx.de>, "Roy T. Fielding" <fielding@gbiv.com>, HTTP Working Group <ietf-http-wg@w3.org>
Message-ID: <alpine.DEB.1.10.0904020903350.15382@yvahk2.pbagnpgbe.fr>
On Wed, 1 Apr 2009, Adam Barth wrote:

>> Right.  The information you're talking about is irrelevant to most 
>> non-browser HTTP agents and so doesn't belong in the HTTP specification.
> It's relevant to any HTTP user agent that wishes to interoperate with 
> existing Web content.  For example, Imageshop (described earlier in this 
> thread) is not a browser but is interested in knowing when its users expect 
> an HTTP response to be treated as an image.

I think this is an example of mixed layerings that leads to no good.

HTTP is a transport layer and a HTTP transport does not need to perform any 
guessing of contents at all. It provides data and meta-data. Fine.

Applications that get stuff over HTTP may want to guess the content and ignore 
the Content-Type. Sure, but that's not HTTP.

I consider that a little like an unzip program checking that the .zip file 
looks like an actual zip before it operates on it. If the zip program 
downloads the file over HTTP it doesn't mean that we need to teach the HTTP 
spec on how the zip client may want to guess if the content truly is a zip or 


  / daniel.haxx.se
Received on Thursday, 2 April 2009 07:19:50 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:13:39 UTC