Update: RE: Caching dynamic content

I missed a bit. Doh1

> -----Original Message-----
> From: Jochen Dolze [mailto:dolze@epcnet.de]
> Sent: Sunday, 29 October 2000 4:53 AM
> To: squid-dev@squid-cache.org
> Subject: Caching dynamic content
>
>
> Hi,
>
> we are trying to cache dynamic content with squid-2.3stable. Our own
> scripting-language is able to produce Last-Modified and ETag-Headers.
>
> A reply from the server contains e.g.:
>
> Last-Modified: Fri, 18 Aug 2000 13:14:58 GMT
> ETag: "0-1cca-399d36d2"
> Content-Length: 8135
>
> It works fine with certain different URLs!
>
> Are these two contents different for squid, or is squid only
> taking the URL
> to identify different pages?:
>
> http://web.test.com/file.html?var=1
>
> Last-Modified: Fri, 18 Aug 2000 13:14:58 GMT
> Content-Encoding: gzip
> ETag: "0-2345-34232425"
> Content-Length: 3112
>
> http://web.test.com/file.html?var=1
>
> Last-Modified: Fri, 18 Aug 2000 13:14:58 GMT
> ETag: "0-2345-34232425"
> Content-Length: 7135
>
> In my case, MSIE5.5 got with squid the wrong cached object!!!
> Must i use
> different ETags for compressed/uncompressed content?

In the two above URL's you have the same URL
:http://web.test.com/file.html?var=1
The same Last-Modified
:Last-Modified: Fri, 18 Aug 2000 13:14:58 GMT
and the same ETag
:ETag: "0-2345-34232425"
and the content-length is DIFFERENT.

I believe you are breaking rfc 2616 with this behaviour

=== From RFC 2616 Section 3.11
   An entity tag MUST be unique across all versions of all entities
   associated with a particular resource. A given entity tag value MAY
   be used for entities obtained by requests on different URIs. The use
   of the same entity tag value in conjunction with entities obtained by
   requests on different URIs does not imply the equivalence of those
   entities.
===
You are serving different versions of the entity
http://web.test.com/file.html?var=1 with the same Etag. That violates
the MUST quoted above.

There is a similar constraint with Last-Modified.

If your URI was different it would not be an issue.

Rob
<snip>
Received on Sun Nov 05 2000 - 15:58:39 MST