Re: [HACKERS] "internal error" triggered by EXISTS()

David Gould <dg@illustra.com>

From: dg@illustra.com (David Gould)
To: tgl@sss.pgh.pa.us (Tom Lane)
Cc: pgsql-hackers@hub.org
Date: 1998-07-14T22:50:19Z
Lists: pgsql-hackers
> 
> I tried the following to find out whether a table has any records
> with field1 < X (for a constant X):
> 
> tgl=> SELECT EXISTS(SELECT * FROM table WHERE field1 < X);
> ERROR:  internal error: do not know how to transform targetlist
> 
> Is this a bug?  (I'm using development sources from yesterday.)
> 
> Am I using EXISTS() incorrectly?  The examples I've been able to find
> only show it as a part of a WHERE clause.
> 
> If it did work, would it be any faster than a table scan?  The code
> I was hoping to replace is like this:
> 	SELECT COUNT(field1) WHERE field1 < X;
> 	// test whether result > 0
> Since aggregates aren't optimized very well, this ends up reading
> much or all of the table, even if there is an index for field1.
> I was hoping EXISTS() might be smarter...
> 
> 			regards, tom lane
> 

Should have given a syntax error probably. But you might try:

select 1 where exists (select...);

Should be faster if and only if we are doing the existential query
optimization trick (stop on the first qualifying row).

-dg


David Gould            dg@illustra.com           510.628.3783 or 510.305.9468 
Informix Software  (No, really)         300 Lakeside Drive  Oakland, CA 94612
 - If simplicity worked, the world would be overrun with insects. -