Re: Explicit revocation

Koen sets out two options for what to do when there is no Expires:
or other cache control returned from the header.  If I may be so
bold as to summarize, these are:

    Option 1: use a heuristic to decide how long to cache the respone
    
    Option 2: if the server is 1.1 or later, assume the worst case,
    e.g., that the document expires immediately.
    If the server is older (1.0 or earlier), use the heuristic from 
    option #1.
    
Although Koen writes:
    Given Jeffrey Mogul's excellent, though rather discouraging,
    statistics about the current use of Expires headers, I prefer
    option 1.
he also wrote:
    In particular, encourage 1.1 server authors to make the automatic
    addition of Expires or Cache-control headers to every reply a
    configurable option that defaults to `on'.

I would look at this in a slightly different way.  If we use option
#2, and the final HTTP/1.1 makes this clear, then we are providing
strong encouragement for server writers to send explicit cache-control
information.   Any server that doesn't do this will be bombarded with
extra requests, and so there would be a strong self-enforcement
mechanism for this part of the spec.

It's clearly not a good idea to be so strict with 1.0 servers,
even though their current behavior probably leads to lots of
faulty caching (and probably lots of redundant "Reload"ing, which
may defeat the performance of many caches).  This is why the
heuristic fallback is necessary.  But the extra implementation
effort to deal with Option #2 seems quite minimal, and I think
it will quickly lead to more effective caching.

-Jeff

Received on Tuesday, 2 January 1996 23:18:48 UTC