Re: Performance for seq. scans

> > We were in a similar position and went for the 'Full Text Indexing" extra.
> > You'll find it in contrib/fulltextindex.
> > It creates a function which you call on a trigger to produce an index of
> > words for specified fields. These indexes do get _very_ large (one of ours
> > is ~800 MB), but it does work very well and speeds searches up enormously.
>
> If I understand you correctly, that's word-based? It's just splitting
> on whitespace and punctuation?  Unfortunately, that's not quite what
> we need --- our wildcard searches needn't have their '%' on word
> boundaries.

It is not word-based.  It slices each word into pieces and indexes it.

--
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026