>My personal belief is that entity version is much more
>useful than server version to solve the problems
>that exist today.

I still don't understand why you think that there exists such a thing
as an "entity version".  There are HTTP/0.9 messages and HTTP/1.x
messages and (sometime next century, apparently) HTTP/2.x messages.
The entity has no version because it is payload and is completely
described by the entity header fields (which, BTW, can be extended
independently of the protocol version numbers).  Placing a version
number on those fields would be even less useful than the MIME-Version
header field of MIME.

For proxies, I think the best design is for them to store each response
in their own internal format (whatever is easiest/smallest/quickest for
that particular application).  One thing I would never expect would be
for it to store the response with the chunked transfer encoding still
intact -- it is, in all cases, more efficient and effective to dechunk
the response as it is being stored and save the Content-Length (and
trailer) of the result as header fields.


Received on Monday, 11 August 1997 11:00:16 UTC