Re: Changing PUT's idempotency after the fact [was: WebDav methods and idempotency]

Mark Nottingham wrote:
> 
> Well, that's a potential workaround, if both clients and servers know to 
> use it. I don't know enough about WebDAV autoversioning; is that a 
> specified behaviour (e.g., submit an ETag and then do INM in a 
> subsequent retry)?

If you do two PUTs with If-Match request header specifying the same 
Etag, the second one will fail if the first one succeeded (that's HTTP, 
any RFC3253 server that doesn't behave that way would be non-compliant).

> My concern is more regarding the process; having one RFC change the 
> semantics of a protocol element in another, widely-deployed RFC seems 
> like a no-no, at least without explicitly updating the other spec.

Agreed. The question is: does it? Note that it only applies to a very 
specific auto-versioning behaviour and it's impact on the resource's 
live properties.

Best regards, Julian

-- 
<green/>bytes GmbH -- http://www.greenbytes.de -- tel:+492512807760

Received on Wednesday, 16 March 2005 23:59:08 UTC