- From: Adam Barth <w3c@adambarth.com>
- Date: Wed, 22 Sep 2010 14:33:20 -0700
- To: Bjoern Hoehrmann <derhoermi@gmx.net>
- Cc: Julian Reschke <julian.reschke@gmx.de>, HTTP Working Group <ietf-http-wg@w3.org>
On Wed, Sep 22, 2010 at 1:22 PM, Bjoern Hoehrmann <derhoermi@gmx.net> wrote: > * Julian Reschke wrote: >>Agreed, but it makes the spec more complex, and it's not sure whether >>it's worth the special case (that's why statistical data would be useful). > > Earlier this week I had http://cutycapt.sf.net/ load a bunch of URLs > (random selection of pages linked from the german Wikipedia, at most > three per host) resulting in about 125,000 requests. There was not a > single response with both Transfer-Encoding and Content-Length, or two > or more Content-Length headers, or Content-Length headers with any- > thing but digits and white space (especially no commas). Neither does > this happen for the Alexa 1000. Obviously this does not include certain > things like responses resulting from user interaction (say, posting to > a forum). > > It's a lot harder these days to get webservers send out messages with > broken framing, thanks to improvements to script interpreters, web > servers, and the interfaces between them, than it was a few years ago. > > (The usual strange things are included, like headers that end up as > a reason phrase in the status line of the response ala "HTTP/1.0 200 > OKContent-Type: text/html; char..."; portugese appears to be the most > popular language second to english in the reason phrase by the way, > followed by german and serbian...) As a control, did you look for multiple Content-Type headers? I know that these occur frequently enough that it's important to use the last one if you're not IE (i.e., if you actually use the Content-Type header for something). Adam
Received on Wednesday, 22 September 2010 22:04:06 UTC