- From: Julian Reschke <julian.reschke@gmx.de>
- Date: Thu, 20 Oct 2005 21:08:37 +0200
- To: Lisa Dusseault <lisa@osafoundation.org>
- CC: WebDav <w3c-dist-auth@w3.org>, Cullen Jennings <fluffy@cisco.com>
Lisa Dusseault wrote: > > Well, I suppose there are multiple ways of answering the question. > Since the question explicitly asked whether the server should fail > "returning 207 Multi-Status with some portions giving a 412 > Precondition Failed", I still find that saying that 412 can't appear in > Multi-Status is at least a partial answer to that question. ...but: 1) A statement about 412 in multistatus is very different from adding a whole new section saying lots of other things, and 2) from a procedural p.o.v.: somebody asked a question, and there was a single answer, and no further discussion. To me, cases like this one are a perfect example for no changes of the spec being required. We shouldn't change the spec each and every time somebody has a question unless there's a consensus that there's actually something wrong with the spec. If it ain't broken, don't fix it. > Another answer would be to say that unmet preconditions MUST cause the > entire request to fail atomically even if the request involves multiple > resources. Do you think we should add that to the draft? After re-reading the thread I actually think that RFC2518 currently defines a behaviour for "Depth" that is unlikely to be implemented. So what we need to do is to sit down, write test cases and find out what current implementations do. *Then*, wen can discuss potential changes. But please let's first find out what currently servers do. Volunteers? Best regards, Julian
Received on Thursday, 20 October 2005 19:09:19 UTC