- From: Willy Tarreau <w@1wt.eu>
- Date: Wed, 16 Oct 2013 12:03:28 +0200
- To: Nicolas Mailhot <nicolas.mailhot@laposte.net>
- Cc: cowwoc <cowwoc@bbs.darktech.org>, Martin Thomson <martin.thomson@gmail.com>, Mark Nottingham <mnot@mnot.net>, Julian Reschke <julian.reschke@gmx.de>, HTTP Working Group <ietf-http-wg@w3.org>
On Wed, Oct 16, 2013 at 11:52:49AM +0200, Nicolas Mailhot wrote: > > Le Mer 16 octobre 2013 11:47, Nicolas Mailhot a écrit : > > > > Le Mer 16 octobre 2013 06:00, Willy Tarreau a écrit : > > > >> The main idea behind idempotence is to know whether or not a client can > >> safely resend a request that failed because of network errors. For > >> example, > >> sending a request over an established connection may result in an error > >> if > >> the server decided to close at the same time. But the client has no way > >> to > >> know if the request was considered or not, so it must resend it. If the > >> previous request was correctly processed, the second one may return an > >> error, but it's the client's job to handle this. I can give you an > >> example > >> here with unix commands : > > > > I'm not sure I agree with the example, I've seen cases where idempotence > > is a much stronger property. > > > > For example, since GET is idempotent, that means a malware checker service > > can replay (before or after the browser) get requests to check if the > > result is dangerous or not. And it does not affect the result the browser > > will receive on its own GET. > > > > If you say mv is an example of idempotence, that means all the broken > > websites that transform GETs into implicit POSTs are right, since the > > first GET destroys the result the following GET could expect. > > > > IMHO idempotence means a replay of the same command will give the same > > result, baring errors, website reorganisations, and data changes (when the > > request exposes some dynamic data store) > > Or to put it more succinctly an idempotent method must not trigger by > itself a state change server-side. Is that clear and short enough for > everyone? Yes it can trigger a state change. But it will not trigger another state change if done *again*. Willy
Received on Wednesday, 16 October 2013 10:03:57 UTC