- From: Jim Whitehead <ejw@ics.uci.edu>
- Date: Thu, 22 Jan 1998 17:14:21 -0800
- To: "'SKREDDY@us.oracle.com'" <SKREDDY@us.oracle.com>, "w3c-dist-auth@w3.org" <w3c-dist-auth@w3.org>
Surenda, Thank you for taking so much time to comment on the protocol draft! More comments below: On Thursday, January 22, 1998 2:19 PM, Surendra Reddy [SMTP:SKREDDY@us.oracle.com] wrote: > > Jim, > > Thanks for your pointers on Dublin Core references and other discussions > related to adding properties > Author etc. After dwelving into those pointers, I agree with your argument > and inclusion criteria for adding any other properties. > > One more issue: Why collections are unordered collections only. It is > useful to have ordered collections. I am not > sure whether this issue had already been discussed at length and concluded > not to support ordered collections? ( > I am still reading through years worth of mailing archives). If there no > consensus reached on this topic yet, > I would like to hear comments from other members of this group. > > Jim what is your judgement call on this? I have not yet seen consensus on this issue on the list. > Do you agree with me renaming PROPFIND and PROPPATCH to GETPROP and > SETPROP? I know you may not > agree with as this not critical for successful implementation of Protocol > but i see that GETPROP and SETPROP > coveys the purpose of these methods ( again it may go back to an argument > of HTTP methods are opaue token)/ I think it's a bad idea to change method names this late in the game because the discussion of the working group has centered around these names, and thus although they are just opaque identifiers to the silicon computer, the carbon-based brain computer isn't so easily reprogrammed. People get used to discussing functionality lumped under a given name, and changing the names of the methods causes problems when discussing the same functionality under the new, not so well known name. Roy Fielding had the best rationale for this when discussing another method name change recently: http://lists.w3.org/Archives/Public/w3c-dist-auth/1997OctDec/0186.html My summation of the arguments concerning this other name change can be found at: http://lists.w3.org/Archives/Public/w3c-dist-auth/1998JanMar/0003.html A second, less compelling reason not to change PROPFIND to GETPROP is that some HTTP servers apparently see a method name beginning with "GET" and assume it is the GET method. The rationale for the choice of these method names is as follows. Since the state of a resource is now divided into the body (retrieved with a GET) and the properties, we decided to have some symmetry with the name of the PATCH method (which was in previous drafts, recently removed, will reappear in future versioning drafts) which is used to modify the state of the body of a resource. So, we named the method which modifies the state of the properties of a resource PROPPATCH. The name PATCH is a holdover to older HTTP 1.1 drafts, which also had a definition of PATCH before it was removed. (PATCH has had a checkered history.) Since we couldn't use a method name that began with GET, we decided to stick with a name that began with PROP for the property retrieval method name. Although we could have named the method PROPGET, we didn't like that name, because it connoted, with its use of the name GET, that the entire property state of the resource is returned, rather than just selected properties based on a selection criteria. FIND was a good substitute, hence PROPFIND. - Jim
Received on Thursday, 22 January 1998 20:24:14 UTC