- From: Julian Reschke <julian.reschke@gmx.de>
- Date: Fri, 23 May 2008 16:08:30 +0200
- To: Henrik Holst <henrik@witsbits.com>
- CC: Simon Perreault <simon.perreault@viagenie.ca>, w3c-dist-auth@w3.org
Henrik Holst wrote: > fre 2008-05-23 klockan 09:50 -0400 skrev Simon Perreault: >> On Friday 23 May 2008 09:36:46 Henrik Holst wrote: >>> There is no MUST in RFC4918 for the MKCOL Method so class 1 does not >>> include MKCOL as far as I can see. >> Is it intentional? I mean, RFC2518 says "All DAV compliant resources MUST >> support the MKCOL method." And advertising class 3 would not imply support >> for MKCOL, if I read the definition of class 3 correctly. >> > Well that is interesting, class 1 would thus be slightly redefined in > 4918 and supporting class 3 tells the client that your server follows > this redefinition I guess. > > Hopefully we can get some info from the people involved with 4918 how > the ideas where when MKCOL was changed from MUST. It wasn't intentional. As far as I recall, we lost the first paragraph without considering compliance. In practice I think it doesn't make any difference at all. If a server vendor can't implement support for creating collections, he/she simply won't. A 403 always was possible, RFC4918 now also allows 405. What difference does this make in practice? BR, Julian
Received on Friday, 23 May 2008 14:09:14 UTC