Re: tsearch_core for inclusion

"Joshua D. Drake" <jd@commandprompt.com> writes:
> Tom Lane wrote:
>> Surely the CREATE INDEX syntax has got enough warts on it already.

> Do you mean something like:
> CREATE INDEX ftil ON t1 USING GIST|GIN(C1 FULLTEXT);
> Where FULLTEXT is like VARCHAR OPS?

Yeah, that one.  It might be more consistent to spell it as "fulltext_ops"
but I wouldn't insist on it.

Of course the issue not addressed here is where you specify all the
secondary configuration data (the stuff currently handled by config
tables in the contrib implementation).  Perhaps the WITH clause would
work for that, though in the current code WITH is targeted at the index
AM not individual opclasses.
        regards, tom lane