Pavel Cisar wrote:

>
> Sorry, still can't see why it would be good to have it built in the
> system, when you can solve that easily at "application" level, i.e. use
> hashes and queries like:
>
> select textid from texttable where texthash=<valuehash>

1) Eliminate the question of index key size completely, regardless
of page size.

2) Recognize duplicate entries in public discussion groups
to avoid spammers - SlashDot currently does that.

>
>
> What many people would really want is quick search for value parts in
> long values (STARTING WITH, CONTAINING), not exact match for long value.
> And this type of index is not applicable for this operation.

That's a different issue- for that you want full text search and
there the question is whether it should be field/table based or
cross field and tables - also it needs proximity and some of the
other rules that search engines use to prioritize results.


Regards


Ann