Re: [HACKERS] create operator problem

Tom Ivar Helbekkmo <tih+mail@hamartun.priv.no>

From: Tom Ivar Helbekkmo <tih+mail@Hamartun.Priv.NO>
To: "Vazsonyi Peter[ke]" <neko@kornel.szif.hu>
Cc: pgsql-hackers@postgresql.org
Date: 1998-04-24T15:43:47Z
Lists: pgsql-hackers
"Vazsonyi Peter[ke]" <neko@kornel.szif.hu> writes:

> drop operator ~~ (text,text);
> DROP
> create operator ~~ (leftarg=text,rightarg=text,procedure=mylike);
> CREATE
> select 'this will be true'::text ~~ 'true';
> ?column?
> --------
> f
> (1 row)
> -- it seems not ;(

I got bitten by this, too.  There's special handling of ~~ hardcoded
into the parser, which expects that it implements the vanilla flavor
of likeness testing.  It has to do with enabling the use of indices to
speed up the matching.  Bottom line: you can't redefine it.

-tih
-- 
Popularity is the hallmark of mediocrity.  --Niles Crane, "Frasier"