RE: GET, POST, and side-effects

On Thu, 4 Jan 1996, Paul Leach wrote:

> ] (...late breaking synaptic activity:...)
> ] I just got an idea that might be simpler than trying to control this
> ] using headers, and doesn't require convincing any HTML people of
> ] anything.  Just invent some new methods that implement
> ] side-effect-free versions of POST, etc.
> ] Then we can define the cache actions differently based on
> ] the method.

Yup, and with a new method the UAs of the world don't need to confuse
users with queries re. rePOST.

> If it turns out that the client side can know to use these, fine.  If 
> the server is really the only one that can certifiy that (e.g.) a POST 
> is side effect free, then this won't help.

Let us now lose sight of the fact that the CLIENT is told by the server's
provision of a FORM with method=POST or RETRIEVE or whatever in the
HTML document what the server expects is likely to be the result of
the request.  I'm looking for a counter example, but I would expect
that most of the cases where the application design resulted in a POST
which had a result which didn't vary based on the repeat the POST
would be the case of an error message where it is somewhat doubtful
that caching would be useful anyway.

Dave Morris

Received on Friday, 5 January 1996 05:58:02 UTC