- From: Larry Masinter <masinter@parc.xerox.com>
- Date: Thu, 11 Apr 1996 19:16:57 PDT
- To: koen@win.tue.nl
- Cc: mogul@pa.dec.com, fielding@avron.ics.uci.edu, http-caching@pa.dec.com
I think the only thing left is to come up with an acceptable terminology. I don't like "weak validator" and "strong validator" since while there is "weak validation" and "strong validation" the tokens that are used to perform this validation are not themselves respectively "weak" and "strong". Let me propose two new words: content-token octets-token A content-token is used for 'weak validation'. An octets-token is used for 'strong validation'. Two entities have the same octets-token only if they have exactly the same sequence of octets, and can be used for range retrieval. Two entities have the same content-token only if they differ in semantically inconsequential ways (as deemed by the origin server), and can be used interchangably in retrieval of entire entities. Last-modified is a content-token that uses date, and so is not opaque. I used the word 'token' instead of 'validator' or 'identifier' or 'id', since it doesn't, by itself, validate, and 'identifier' is overloaded, since it only identifies the entity with respect to the context of a single URL.
Received on Friday, 12 April 1996 02:34:30 UTC