W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > January to April 1997

Re: Issues-list item "CACHING-CGI"

From: Andrew Daviel <andrew@andrew.triumf.ca>
Date: Wed, 16 Apr 1997 11:39:13 -0700 (PDT)
To: "Gregory J. Woodhouse" <gjw@wnetc.com>
Cc: http-wg@cuckoo.hpl.hp.com
Message-Id: <Pine.LNX.3.91.970416111611.20092F-100000@andrew.triumf.ca>
X-Mailing-List: <http-wg@cuckoo.hpl.hp.com> archive/latest/3078
On Wed, 16 Apr 1997, Gregory J. Woodhouse wrote:

> The more I think about it, the more I think the right thing to do is not
> to return the current time as Last-Modified: but (if this makes sense)
> return the last modified date fore the data used to generate the response.

At some point I wrote http://vancouver-webpages.com/proxy/log-tail.pl 
(Perl using LWP4) which does this kind of thing. It's a bit of a 
nuiscance compared to just doing

Content-type: text/html

<title>Here we go!</title>

and I confess I haven't moved to LWP5 yet - there may be a package to do 
it now.
In my search engine I attempt to handle IF_MODIFIED_SINCE, set 
Last-Modified to the modification date of the last key to change, and set 
Expires to the expected database update time. I haven't yet added any 
HTTP/1.1 headers.
Some notes on this are at 

In my copy of Squid; I have hierarchy_stoplist set to cgi-bin,? to match
the cache_stoplist of the parent, but cache_stoplist set to "map? gif?"
A scan of the access log shows several hits on DejaNews queries.

Server-side imagemap queries are clearly cacheable, but would have a very 
low hit rate under normal conditions. Hmm; I think the output is usually 
a redirect hence uncacheable; oh well ...  map? was from the xerox Parc 
map server

Andrew Daviel
TRIUMF; Vancouver Webpages
Received on Wednesday, 16 April 1997 11:41:35 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:40:19 UTC