- From: Paul Leach <paulle@microsoft.com>
- Date: Wed, 30 Aug 95 10:30:24 PDT
- To: http-wg-request%cuckoo.hpl.hp.com@hplb.hpl.hp.com, http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
---------- Maurizio Codogno writes: ] ] % >Idempotent ] % ] % I second the request for a clear definition in the context of this spec. ] % Another good reason for this: idempotent is not in any dictionary ] % (including my Webster's unabridged) that I could find. ] ] Being a mathematician, my idea of idempotence is a function f such as ] f(f(x)) = f(x) . ] ] Now, I admit it is not something I'll put on the http specs, but what about ] ] "something which does not change if requested twice in rapid succession"? How about: "the results do not change in a way that matters (to the client or server) if repeated twice in a row (i.e., with no other intervening methods)" The reason that GET incrementing a counter can still be idempotent is that it doesn't matter to the client or server. (If it did, e.g., payment was based on hits, then this arguement wouldn't apply.) I don't have a position (yet) on whether GET should be required to be idempotent. However, if it isn't, then any non-idempotent GET *must* be marked non-cacheable. Paul
Received on Wednesday, 30 August 1995 10:36:36 UTC