Re: Squid enhancement thought

Allen Smith wrote:
>
> On Jun 2, 4:24am, Henrik Nordstrom (possibly) wrote:
> > Dax Kelson wrote:
> >
> > > When a user requests a html document, the commercial box scans that html
> > > document for additional objects (mostly images) and starts retrieving
> > > those immediately. Shortly after, when the client requests those objects,
> > > the cache box either has them already, or is already getting them.
> > >
> > > Could this be implemented in Squid?
> >
> > Could: yes, Is: no, Given priority: very low.
>
> A minor version of this that wouldn't be nearly as hard to implement
> (I've looked at doing it myself, although it's currently pretty far
> down on my ever-growing list of tasks) would be requesting and caching
> any cachable referred object (referred via HTTP responses, not HTML
> coding).

Sorry for the long delay.

I do not quite follow what your are suggesting here. HTTP referred
references are not available until the client makes the request, and
Squid already fetches objects when the client requests them..

If your idea is that Squid should build a large database of how objects
refer to each other then I would say that it isn't either a minor task
to do, or very practical thing to do, and it will only give a very
limited gain in latency reduction as it only applies to cached objects.

--
Henrik Nordstrom
Spare time Squid hacker