AW: [HACKERS] SourceForge & Postgres

[Zeugswetter Andreas SB]

> > anyway? ;-))  If so, a search for artistid 100050450 definitely *should*
> > use a sequential scan.
> 
> I tested this statement against the database and you are right, about 14
> seconds with the index, 4 without.

Now I don't understand the problem any more. Are you complaining, that
the optimizer is choosing a faster path ? Or are you saying, that you also
get the seq scan for other very infrequent values ?

Andreas