Обсуждение: Re: [HACKERS] \dt and disk access\

> > > Try something with a _lot_ of data, and try to have
> > > the data inserted randomly if possible. Another thing to check is the
> > > fundamental time it takes to actually do sorting (for example, put the
> > > same data into a little test program which calls qsort(), or find the
> > > Knuth algorithm and call that directly). Sorts aren't (usually) cheap!
> >
> > See above.
> >
> > --
> > Bruce Momjian
> > maillist@candle.pha.pa.us
>
>     And why not to implement several sort algorythms which
> may be used in queries.
>     I think, somethink like GROUP BY ... USING psort - will be very
> usefull. Is it possible?

My goal is to choose a good sort algorithm that will not have to be
tweaked in sql.

- --
Bruce Momjian
maillist@candle.pha.pa.us

------------------------------