Re: rfc 2616 query

----- Original Message -----
From: "Henrik Nordstrom" <hno@hem.passagen.se>
To: "Robert Collins" <robert.collins@itdomain.com.au>
Cc: <squid-dev@squid-cache.org>
Sent: Thursday, February 08, 2001 6:58 PM
Subject: Re: rfc 2616 query

> Robert Collins wrote:
>
> > Two cases: One is compression, the other is with operations on the data
> > which may (deterministicly) change the body - ie Joe & Moez's
> > acceleration url rewriting. While the second case is rare, I don't want
> > to confuse browsers because of it :]
>
> I would not dream of trying to support range requests in entity
> rewriting setups of any kind, except possibly cache hits on already
> rewritten content, but even for hits I have doubts. You must be 100%
> sure that the rewrite is identical to what would have been done if the
> filter worked on the whole entity, and that the requestor bases his
> request on a entity fragments with the exact same rewrites applied..

Thus the 'deterministic' word. I agree on the dificulties, and I don't plan to write such a beast myself. I do plan to make
allowances for things like virus scanners that may want to perform a full-body request upstream, check for viruses and then return
the requested range though. That's a slightly easier case to guarantee consistency on.

And don't even start thinking about users that request 50% of object A from cache B, and then the other 50% of object A from our
cache!

> > Also other things allowed but not recommended, such as content-type
> > transformations will need to interact cleanly with transfer-encoding.
>
> Hint: transfer-encoding is a hop-by-hop thing taking place in the
> communication between two or more hops. content-transformation is a
> entity operation taking place at one or more of the hops.

Yes. Content-transformation may affect entity type and size, and thus interacts with range requests... and if they interact with
transfer encoding then the issues arise. That was the tie-in with transfer-encoding in my comment :]
transfer-encoding only affects message size, not entity type or entiry size.

Thanks for the feedback...
Rob
Received on Thu Feb 08 2001 - 01:33:01 MST