I often think about performance of conditional expressions. I'm not just talking about the expense of whatever the test is, but the difference between the condition types used.
I’ve heard about it from a few sources, and Tony Andrews blogged about it recently, but never really been able to/had a chance quantify it. Roel Hartman has also blogged about this after reading my tweet. He shows some figures, but difficult to rinse, repeat & measure like a PL/SQL process.
For example, while I’ve always encouraged this type of expression for consistency (which runs dynamically), particularly with new developers learning the differences between item substitution syntax.
This performs better (declaratively):
I don't see this second debug statement when using the declartive method.
I welcome further discussion at this forum post.