Hi.
After looking again at the other email I've sent earlier, I realized that it
goes on for far too long, so I'll try to summarize my question more briefly.
1) How can FTI be made to NOT break up words into sub-strings? Most of those
are likely to be useless in my particular application. In fact, this feature
is why my FTI queries run slower on certain strings than sequential ILIKE
scans.
1.1) How do I make FTI only separate the text string into words on
non-alphanumeric characters (i.e. [!a-z|!A-Z|!0-9] - I think...). Is that
what it already does?
2) How can I make inserts work when enforcing unique records in the FTI
table? There is hardly any need for duplicates in the FTI table...
At the moment, the search for a particular string can take signifficantly
longer with FTI than using the plain ILIKE. FTI also returns 30K matches in
the FTI table for only 10K of records in the master table, which is clearly
not very useful. Doing a DISTINCT on the FTI table with that string will
yield 10K records, indicating that there are about 3 times more records in
the FTI table which are just duplicates taking up space.
I am NOT prepared to capitulate and use something MS SQL instead. How can I
get this to work with PostgreSQL? Who maintains the FTI contrib?
Kind regards.
Gordan
