I agree,
However we could also consider using the minimalistic aproach instead.
Basicly it says don't cache unless there is already a direct reference
to the data prior to the request.
This assumes that a second request indicates the likeleyhood of a third
or more. Wich should be more common close to the epicentre.
This should mean the number of cached copies never surpasses the nmber
of requests for a document.
But if we do decde to go probabilistic, I would suggest to include
'always cache on a direct reference'.
Neil
On 27 May 2001 18:42:56 -0700, Ian Clarke wrote:
> So one thing we might consider is making 0.3.9.2 incorporate the
> probabilistic caching that we have been talking about. It is fairly
> unanimous that such an approach is a good idea, would be quick to
> implement, and it could be ages if we wait for someone to get around to
> simulating it before we implement it.
>
> The caching would obviously be configurable, but should default to on.
>
> This may seem wreckless, but it could hardly make matters any worse...
>
> Ian.
_______________________________________________
Devl mailing list
Devl at freenetproject.org
http://lists.freenetproject.org/mailman/listinfo/devl
