W3C home > Mailing lists > Public > ietf-http-wg@w3.org > January to March 2005

Re: HTTP PUT method variations

From: Scott Lawrence <scott@skrb.org>
Date: Thu, 17 Feb 2005 15:05:02 -0500
To: Michel Drescher <Michel.Drescher@web.de>
Cc: HTTP Working Group <ietf-http-wg@w3.org>
Message-Id: <1108670702.11773.36.camel@sukothai.pingtel.com>

On Thu, 2005-02-17 at 18:15 +0000, Michel Drescher wrote:
> 
> 
> Scott Lawrence wrote:
> > On Thu, 2005-02-17 at 14:26 +0000, Michel Drescher wrote:
> > 
> >>Dear working group,
> >>
> >>I am exploring the capabilities of the HTTP protocol for implementing 
> >>parallel data distribution using established protocols.
> > 
> > 
> > As you correctly illustrated, HTTP can (and is) used to 'pull' data in
> > parallel requests.
> > 
> > Trying to overload PUT in the way you describe is more or less sure to
> > fail as soon as you encounter a server that doesn't understand it.  That
> > being the case, you are better off defining a body type for what you
> > want to do and using POST.  
> 
> It is a controlled environment - we provide both clients and server, so 
> we can make the server understand these requests.

Then why use HTTP at all?

> All I (more or less) wanted to know is that if my interpretation of the 
> RFC is valid (so we would be fully compliant) or not.

No - you are assuming constraints in server implementations that the RFC
does not require them to observe.
Received on Thursday, 17 February 2005 20:05:04 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 06:49:39 GMT