Re: The new, the improved ... FTS Searching of Mailing List Archives

Lincoln Yeoh <lyeoh@pop.jaring.my>

From: Lincoln Yeoh <lyeoh@pop.jaring.my>
To: The Hermit Hacker <scrappy@hub.org>, Bruce Momjian <pgman@candle.pha.pa.us>
Cc: <pgsql-hackers@postgresql.org>, Vince Vielhaber <vev@michvhf.com>
Date: 2001-04-28T01:29:35Z
Lists: pgsql-hackers
At 03:44 PM 27-04-2001 -0300, The Hermit Hacker wrote:
>On Fri, 27 Apr 2001, Bruce Momjian wrote:
>
>> > On Fri, 27 Apr 2001, Bruce Momjian wrote:
>> >
>> > Huh? *raised eyebrow*  This is a standalone application that they've
>> > donated to the project ... nothing that can be added to any of our
>> > distributions ...
>>
>> Isn't the text indexing something that can go into the distribution?
>
>to the best of my knowledge, everything they had for public consumption
>was added to v7.1, but Oleg would be better for that ... to get
>fts.postgresql.org, there was nothing special I had to do as far as the
>backend was concerned *shrug*

<featurerequest>
Well if stuff like that ends up in Postgresql would it be possible to index
LIKE '%xxx%' searches? That way all people have to do is create the
relevant index and use a fts_ops or something, and voila LIKE '%xxx%'
searches become faster, with maybe some performance+disk space hit for
inserts.

Would something like that be difficult to implement? I'm not sure how
function+fts index would work either.

I hope FTS for postgresql doesn't start looking like Oracle's
Context/Intermedia... Proprietary interfaces == "lock in" == "ick".
</featurerequest>

Cheerio,
Link.