Re: Idempotent partial updates

Le 28 févr. 2012 à 01:28, Adrien de Croy a écrit :

> I've always had an issue with this concept of "Idempotence".  I really wonder whether it actually exists in the real web.
> 
> Many servers can and do have side effects for any particular method.

From the latest httpbis draft : "Request methods can also have the property of "idempotence" in that, aside from error or expiration issues, the intended effect of multiple identical requests is the same as for a single request. PUT, DELETE, and all safe request methods are idempotent.  It is important to note that idempotence refers only to changes requested by the client: a server is free to change its state due to multiple requests for the purpose of tracking those requests, versioning of results, etc."

> So one cannot claim that any method is in fact Idempotent?  Not in all cases anyway.  One could write a script for which POST was idempotent.  Likewise there are plenty of URIs where GET is not.
> 
> I think the concept is not well understood in the web authoring space, and no-one really takes much care to make website implementations honour the intent of HTTP wrt methods which are supposed to be Idempotent.
> 
> My point is, why persist with this language?

I'd say because it allows us to create distributed systems that benefit from it. Regardless of how well it is understood by some people, I am glad I can benefit from it in the systems I develop.

Philippe Mougin

Received on Tuesday, 28 February 2012 01:36:41 UTC