[Gsoc] ZODB-related project ideas

[Malthe Borch]

On 19/03/2008, Martijn Faassen <faassen at startifact.com> wrote:
>  >  = An alternative object caching strategy for ZODB: object
>
> A challenging project. Nice from a research perspective, pay-off might
>  be anything between dismal and spectacular. I think work in this area
>  would be good.

I'm familiar with most of the ZODB code and quite interested in this
subject. The status quo of object caching is that the current
implementation is well tailored to the boring use-case where most
objects that are in high use can fit into the cache.

There are other important scenarios and it would be important to look
at an adaptive implementation. I have an intuition that it might be
feasible to also look at a global object cache to complement the
per-connection cache.

The topic is a good mix of theory and practice, and will require a
student with good problem analysis skills and some experience in
low-level programming.

\malthe