- From: Lisa Dusseault <lisa@osafoundation.org>
- Date: Tue, 15 Aug 2006 15:39:24 -0700
- To: edgar@edgarschwarz.de
- Cc: w3c-dist-auth@w3.org
On Aug 15, 2006, at 2:33 PM, edgar@edgarschwarz.de wrote: > After reading the old drafts and also Roys comments: > - I would definitly go with Content-Type to give the diff algorithm. > Please no additional header :-( > - Find a simple mandatory binary diff which is free of IPR. > I'm no lawyer, but could it help to use a binary diff I use > for years now in an esoteric system called Oberon from ETH Zuerich. > Nobody complained about it in all these years :-) Can you try to verify its licensing status? It would be great to have an unburdened generally-useful diff algorithm. There are also two possible XML diff algorithms: Jara Urpalainen's, and Adrian Mouat's. Both have been published as Internet-Drafts in the past. > Only joking, but can anybody tell me what the problem with gdiff > is ? The gdiff algorithm doesn't have a Content-Type. To register a content-type, I think we'd have to publish an Internet-Draft: see http://www.alvestrand.no/pipermail/ietf-types/2004-September/ 000410.html. I believe we'd be OK publishing an I-D with just the IANA form and a reference to the W3C note, but I haven't gotten around to that yet. Help welcome. (BTW in hunting down this reference I found the time-range that saw lots of discussion on the HTTP WG list on the PATCH stuff: http:// lists.w3.org/Archives/Public/ietf-http-wg/2004AprJun/ and http:// lists.w3.org/Archives/Public/ietf-http-wg/2004JulSep/) Another candidate is VCDIff, but it's unclear how broadly VCDIff may be used. See <http://www.ietf.org/ietf/IPR/ATT-draft-korn-vcdiff>, and note it's limited to HTTP1.1. Does that mean RFC2616 only or does it include specs that extend HTTP 1.1 while retaining that protocol version header? IANAL. > - NO APPEND. > Perhaps Lisa and Suma could collaborate and provide a new draft. > And if somebody decides to go to another list. Please tell me to > subscribe > to it :-) > OTOH I think that PATCH has a special importance in the context of > versioning. So perhaps it could be a good idea to find a rough > consensus > here before going to the HTTP jungle. Thanks for the comments. Lisa
Received on Tuesday, 15 August 2006 22:39:42 UTC