Re: GET on a reference

At 03:49 PM 6/29/98 PDT, John Stracke wrote:
>Jim Davis wrote:

>> I don't think John proposed that a PUT do redirection.
>
>Mmm...<dig, dig>...yeah, actually, I did; I proposed it for GET, HEAD, PUT, &
>POST.  

Urk.  You're right.  Apologies.

>> Don't we already
>> say that PUT on a referential resource is an error,
>OK, so why is it an error? 

To prevent the client from converting a resource from 'plain' to
'referential' by doing a PUT.  This was more important when we had strong
referential integrity as a possibility.  Probably less important now.

>If it's just because we think that PUTting onto a
>reference instead of the target is a Bad Thing, then we can return 302
instead
>of an error code, and then the client will PUT onto its target.  Or do we
have
>some deeper reason?

Hmm, so does PUT onto a referential resource change the resource from
referential to ordinary, or does it return 302 so that the client can
redirect?  Both seem useful.  As you point out, if the client wants to
change the resource type it can do a DELETE, then a PUT.  So I guess 302 is
the right behavior.

Okay, I'm convinced.

Received on Monday, 29 June 1998 21:10:14 UTC