Comments on: Beats a reference cursor http://blog.mclaughlinsoftware.com/2009/03/19/beats-a-reference-cursor/ Michael McLaughlin's Technical Blog Sat, 04 Sep 2010 17:08:47 +0000 hourly 1 http://wordpress.org/?v=3.0.1 By: maclochlainn http://blog.mclaughlinsoftware.com/2009/03/19/beats-a-reference-cursor/comment-page-1/#comment-1533 maclochlainn Thu, 19 Mar 2009 15:11:47 +0000 http://blog.mclaughlinsoftware.com/?p=2088#comment-1533 Great reference article, Thanks. It looks like the only serious downside exists for wide record sets, but I went through it pretty quickly. If I missed something, let me know. Great reference article, Thanks. It looks like the only serious downside exists for wide record sets, but I went through it pretty quickly. If I missed something, let me know.

]]> By: Chris Neumueller http://blog.mclaughlinsoftware.com/2009/03/19/beats-a-reference-cursor/comment-page-1/#comment-1528 Chris Neumueller Thu, 19 Mar 2009 11:15:31 +0000 http://blog.mclaughlinsoftware.com/?p=2088#comment-1528 I agree that pipelined table functions are a nice addition to the developer toolchest. There are some issues to remember, however: A typical pipelined table function requires 2 additional object types, the result table type and it's record type. You can create them by hand or let Oracle automatically build them if the function's return type is declared in a pl/sql package spec. I'd also bet that selecting from pipelined table functions is a little bit slower than using ref cursors, pl/sql functions which return pl/sql tables or using procedures with out parameters (which can be pl/sql tables, too). Adrian Billington <a href="http://www.oracle-developer.net/display.php?id=423" rel="nofollow">wrote</a> about pipelined function issues some time ago, where he analyzes potential problems in much more detail. I agree that pipelined table functions are a nice addition to the developer toolchest. There are some issues to remember, however:

A typical pipelined table function requires 2 additional object types, the result table type and it’s record type. You can create them by hand or let Oracle automatically build them if the function’s return type is declared in a pl/sql package spec.

I’d also bet that selecting from pipelined table functions is a little bit slower than using ref cursors, pl/sql functions which return pl/sql tables or using procedures with out parameters (which can be pl/sql tables, too).

Adrian Billington wrote about pipelined function issues some time ago, where he analyzes potential problems in much more detail.

]]>