Re: GSoC - proposal - Materialized Views in PostgreSQL

On Sun, Apr 11, 2010 at 10:13 PM, Florian G. Pflug <fgp@phlo.org> wrote:
> If continuous updates prove to be too hard initially, you could instead
> update the view on select if it's outdated. Such a materialized view
> would be a kind of inter-session cache for subselects.
>
> The hard part would probably be to figure out how to decide whether the
> view is outdated or not, and to deal with two concurrent transactions
> trying to use an outdates view (and both trying to refresh it). What
> makes the second problem hard is that you wouldn't want one of the
> transactions to wait for the other to complete, because this is not how
> SELECTs traditionally behave.

Well, the proposed project is to create views that only get refreshed manually.

...Robert