RE: Satisfying IMS from HEAD or IMS would help where browser cach es

From: Nottingham, Mark (Australia) <mark_nottingham@dont-contact.us>
Date: Sat, 31 Oct 1998 15:13:02 +1100

I need to reread your message to really get what you're saying there,
but FYI Netscape servers do NOT include a LM header when the request is
a head. This violates the spirit of the HTTP, but not the letter...

Regards,

> -----Original Message-----
> From: David J Woolley [mailto:djw@bts.co.uk]
> Sent: Saturday, October 31, 1998 4:39 AM
> To: squid-users@ircache.net
> Subject: Satisfying IMS from HEAD or IMS would help where
> browser caches
>
>
> One thing I've noticed with CERN originally and now when testing
> squid2/patch2 is that browsers with internal caching tend to
> frustrate a corporate cache. The basic problem is that if an
> If-Modified-Since request for an object cached in the browser, misses
> on the corporate cache, one has to wait for a round trip time (and do
> a demand dial, even if the request could be satisfied from the
> browser cache) if the corporate cache never had the page (cache was
> built after the browser cached the page) or has lost the page
> (limited resources, CERN's absolute time limit, or a demand dial
> failure, etc., causing the the cache entry to be wiped by an error
> response).
>
> What I used to do on CERN, is occasionally trawl the logs for 304s
> that had been proxied (or zero lengths, which came from some earlier
> browsers) and refetch them at a cheap rate time. However that
> involves fetching the whole page, which might prove not to be useful,
> and, as a side effect, marking it recently accessed, so that a
> potentially large page might be kept longer than the true access
> pattern suggested.
>
> I found the following URL which suggests that someone (not sure how
> official this is) was considering caching IMS responses in squid:
>
> http://hem.passagen.se/hno/squid/historic/IMS.html
>
> However, it also occurs to me that HEAD ought to contain all the meta
> data. However, caching a page in IE4, bypassing squid, then manually
> doing a HEAD for the page through squid, then finally restarting IE4,
> configured to use squid, results in a TCP-MISS. Searching the code
> tends to confirm this, but I wasn't 100% certain at that stage.
>
> Is there anything in HTTP/1.1 or common misimplementations of HTTP
> that would make satisfying IMS from HEAD unsafe (assuming reasonable
> plausibility checks are done)? Would it be so unsafe as to make it
> unreasonable to have a leaf cache option to permit it? If not, doing
> TCP-IMS-HIT from HEAD for GET requests, would allow me to fetch just
> the HEADs for recent TCP-MISS/304 events.
>
> --
> David Woolley - Office: David Woolley <djw@bts.co.uk>
> BTS Home: <david@djwhome.demon.co.uk>
> Wallington TQ 2887 6421
> England 51 21' 44" N, 00 09' 01" W (WGS 84)
>
Received on Fri Oct 30 1998 - 20:47:18 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:42:52 MST