I guess the real question is:
what is more efficient - a join on a LIKE or a PLPGSQL function that
has 2 loops, 1 for each bottom level child and one that takes the
substring of that child one character at a time to get each parent?
It sounds like you are saying that the join will actaully use the index
and therefore it is optimal, not just possible.
mlartz@gmail.com wrote:
> If you use a btree index on the column, you should be able to use it in
> the like:
> http://www.postgresql.org/docs/8.1/interactive/indexes-types.html.
>
