- From: Kenji Takahashi <kt@nttlabs.com>
- Date: Tue, 12 Nov 1996 13:03:05 -0800
- To: Larry Masinter <masinter@parc.xerox.com>
- cc: w3c-dist-auth@w3.org
Larry, >Suppose that a versioning-aware server just returns a LINK header, e.g., > GET http://www.cern.ch/TheBook/chapter2 HTTP/1.1 >returns headers that include > Link: <http://www.cern.ch/meta-repository/TheBook/chapter2>; rel="MetaData" > >and then the client will know that it can do GETs against >the linked URL to get and modify the attributes of the included >URL. This is dangerous because attributes or something may change during two transactions. For example, the content may be changed after the GET of the content and before the GET of the attributes. This results in the inconsistency between the contents and attributes. So, I thingk the content and its attributes should be gotten in one transaction. How about the following (we are implementing this now). The attributes are sent as HTTP headers with the content with options. To specify options, I would like to propose "request-attributes" header for GET. If this header is not included in the GET request, attributes are not sent. For example, if you would like to get the "author" and "lock-state" attributes: GET http://foo.bar.content request-attributes: author,lock-state, etc... If you need all the attributes: GET http://foo.bar.content request-attributes: * If you need only the attributes. HEAD http://foo.bar.content request-attributes: * Also I would like to propose something like SET_ATTR to set attributes. Good symmetry, hah? GET <-> PUT HEAD <-> SET_ATTR Best regards, Kenji
Received on Tuesday, 12 November 1996 16:03:10 UTC