Hi.

I made an blob UDF called "blobsearch" that read all the blob (using
buffer, etc), sequencially, looking for a specific word.

Its performance looked good for me, although the bad algorithm
(sequencial).

After, I tried using the "like" operator, instead of my
function: "select .... from ... where xxx like '%searchword%'"
AND the performance was BETTER!

Any body knows a UDF (or another way to do that) that makes
a "smarter" search under blob (long text) fields?

I tried to do a binary search, but so I got another problem: how
to "seek" a blob (the reason of the other question posted in the
group).


thanks,
Lucio IP.