W3C home > Mailing lists > Public > www-talk@w3.org > May to June 2004

Re: HTTP 1.1 pipelining

From: Ian Clelland <ian@veryfresh.com>
Date: Fri, 28 May 2004 23:59:32 -0700
To: www-talk@w3.org
Message-ID: <20040529065932.GB1059@compaqdeb.casa.cavewallarts.com>

On Fri, May 28, 2004 at 09:14:54PM +0100, A Bagi wrote:
> 
> Idempotence: the ability of a Document to be transmitted and accepted more
> than once with the same effect as being transmitted and accepted once. This
> somehow does not mean no side-effects (web applications, GET)!
> Only idempotent requests can be pipelined, such as GET and HEAD requests
> with maximum scucess.  POST and PUT are dodgy business!!
> Ahmed Bagi
> Manchester

If we're going by the way that things are *supposed* to be, then PUT
isn't dodgy at all -- you should be able to submit the same PUT request
a hundred times, and have the outcome be the same as submitting it just
once. From the RFC:

  9.1.2 Idempotent Methods

     Methods can also have the property of "idempotence" in that (aside 
     from error or expiration issues) the side-effects of N > 0 
     identical requests is the same as for a single request. The 
     methods GET, HEAD, PUT and DELETE share this property. Also, the 
     methods OPTIONS and TRACE SHOULD NOT have side effects, and so are 
     inherently idempotent.


Of course, in the real world, things may not be this clean. But, for that
matter, neither are GET requests -- web applications routinely take
parameters from URLs and do horrible, non-idempotent things with them.
So from that perspective, GET is kind of dodgy too. Heck, even HEAD
would have side effects in half of the server-side scripts that I've
seen.

I don't think it makes sense to leave poor PUT out of a pipelining
application just because some developers might be abusing the HTTP
protocol. There are already worse offences out there, that will break
things much harder, than a non-idempotent PUT or two.


Ian Clelland
<ian@veryfresh.com>
Received on Saturday, 29 May 2004 03:00:15 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 27 October 2010 18:14:28 GMT