Re: BUG #5716: Regression joining tables in UPDATE with composite types

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: "Andrew Tipton" <andrew(at)adioso(dot)com>
Cc: pgsql-bugs(at)postgresql(dot)org
Subject: Re: BUG #5716: Regression joining tables in UPDATE with composite types
Date: 2010-10-19 19:15:06
Message-ID: 7068.1287515706@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-bugs

"Andrew Tipton" <andrew(at)adioso(dot)com> writes:
> Attempting to execute an UPDATE that joins to another table where the join
> condition is comparing a composite type fails with the (presumably internal)
> error message "psql:testcase.sql:29: ERROR: could not find pathkey item to
> sort".

Fixed, thanks for the report!

BTW ... while this is unrelated to the cause of the problem, I think
this is quite an inefficient coding technique:

> CREATE TYPE price_key AS (
> id INTEGER
> );

> CREATE FUNCTION price_key_from_table(price) RETURNS price_key AS $$
> SELECT $1.id
> $$ LANGUAGE SQL IMMUTABLE;

> CREATE FUNCTION price_key_from_input(price_input) RETURNS price_key AS $$
> SELECT $1.id
> $$ LANGUAGE SQL IMMUTABLE;

> UPDATE price ...
> WHERE price_key_from_table(price.*) = price_key_from_input(input_prices.*);

Comparing composite types is probably a good two orders of magnitude
slower than comparing plain ints would be. I'm sure that coding
technique looks cute, but you're paying through the nose for it.
Consider making price_key a simple domain over int.

regards, tom lane

In response to

Responses

Browse pgsql-bugs by date

  From Date Subject
Next Message Jeff Davis 2010-10-19 19:40:14 Re: Recovery bug
Previous Message Jeff Davis 2010-10-19 16:51:35 Re: Recovery bug