sequential scans and the like operator

There is a discussion going on on the sql-ledger mailing list concerning
whether indexes will provide any performance improvements.  The one that
caught my eye was whether using LIKE in a statement would force a
sequential scan.

I tried checking the PG list archives but fts.postgresql.org is a little
slow - as many know - plus a quick search using google didn't turn up
much.

My guesstimation is a leading or mid-field wildcard (_, %) would force a
sequential scan but a trailing wildcard would not (should not?).

I think (seem to remember) that this was discussed either here or on
the HACKERS list.  A pointer to the appropriate message/thread would be
great.


TIA,
Rod
--
                      Let Accuracy Triumph Over Victory

                                                       Zetetic Institute
                                                        "David's Sling"
                                                         Marc Stiegler