- From: Shel Kaphan <sjk@amazon.com>
- Date: Mon, 13 Nov 1995 18:54:21 -0800
- To: Jeffrey Mogul <mogul@pa.dec.com>
- Cc: Larry Masinter <masinter@parc.xerox.com>, http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
Jeffrey Mogul writes: ... > > Which implies that the byte-range header should be named > > Request-byte-range-if-cache-valid-else-transmit-whole-file: 1-3 > > when presented along with a Cache-validator: header. I am with you in spirit, as I *much* prefer this kind of validation to date comparisons, but there's something wrong with these names for this purpose. (Something besides the length, that is). There is no "cache" to be valid, so perhaps what you really mean is Request-byte-range-if-cache-validator-would-still-apply-to-the-requested-document-if-it-were-cached: But seriously, I doubt http-wg is going to get consensus on this, given the lack of discussion last time these issues were brought up, which is why I wanted just the *option* of a "checksum" somewhere in this mechanism. But it's true that calling it a checksum means it should be ... a checksum ... so calling it something more general like "validator" would be better. I like the parameterized version of this: Request-range: bytes=X-Y; validator=<xxxxx>; other-parameter=<yyyy> I propose, then, that we choose a set of parameters that can be in this header, and (1) that each parameter name be the same as the name of some header actually received from the server when the initial document fragment was fetched (without byterange request), and (2) that these parameters be allowed to include (without limitation) last-modified, content-MD5, and content-length, the intent being that any parameter in this header is intended to help the server assess the validity of the request w.r.t. the available document, and (optional) (3) we extend the set of headers to include the generalization "validator" which might be used to contain (e.g.) a date, filesize, hash, checksum, or random number chosen by the server as pertaining to a unique version of the document, and allow "validator=<xxxxx>" to appear as a header in server responses as well as in this header. If we can agree to (3) then maybe later we can also think about using this in other appropriate places. --Shel
Received on Monday, 13 November 1995 19:05:49 UTC