Re: Cache revalidations

On 6 Oct 2001, at 10:08, Henrik Nordstrom <hno@squid-cache.org> wrote:

> This is what is said in RFC 2616 on updating headers:
>
> 9.4 HEAD
..
> 10.3.5 304 Not Modified
>
> If a cache uses a received 304 response to update a cache entry, the
> cache MUST update the entry to reflect any new field values given in
> the response.
>
> [squid only updates the freshness of the object, we do not update the
> response]
>
> We do obviously have quite a bit to do here..

 We need to detect that headers changed compared to cached object,
 excluding time headers that we can handle separately. If changed,
 we'd better unlink and refetch the object. imo.
 If only time headers changed that we handle and stuff into saved
 disk object head, we can still try to replace just those bits.

> Andres Kroonmaa wrote:
>
> > Do we need to talk about all http headers, or only those that are
> > currently covered with storeentry?
> > a) seems ok for current state.
> > b) is easy for FS like coss. for others it might be bad overhead.
> > c) if we talk only of storeentry prepended to object, then we can
> > consider overwriting just those bits after we have serviced object.
> > Then we'd open file for r/w instead of just r. If we talk of full
> > headers, then probably full object rewrite is the way.
> >
> > I'd definitely try to avoid rewriting full object anywhere but on coss.
> > So I'd vote for 'c'.

------------------------------------
 Andres Kroonmaa <andre@online.ee>
 CTO, Microlink Online
 Tel: 6501 731, Fax: 6501 725
 Pärnu mnt. 158, Tallinn,
 11317 Estonia
Received on Mon Oct 08 2001 - 03:45:32 MDT