On 26/01/07, Helen Borrie <helebor@...> wrote:

> At 02:09 AM 27/01/2007, you wrote:
> >If I use the SP below to rebuild all index statistics which process
> >actually does the work. I could see that it could be either the
> >process that runs the SP (which is what I actually want) or it could
> >be the next process to try and use an index (similiar to GC), or its
> >probably something else I haven't thought of.
>
> It recalculates the selectivity of the index and writes the result to
> RDB$INDICES (and RDB$INDEX_SEGMENTS in v.2.x). To ensure that cached
> index pages are not contaminated by the changes, the in-memory
> selectivity graph is built at the time the database file is opened by
> the server. The effect won't be felt by the current process, but
> next time the database file is opened. This means different things
> depending on the server model:

thanks now I understand it effects better but I still don't see which
process is the 'it' that does the recalculation