Does this mean that there was a patch in the 5.8.1 branch? That's odd because the Changes for 5.8.1 don't mention a fix for that particular issue. I'm not complaining - I have code that ends up using that kind of construct and it's nice to know that 5.8.1+ doesn't have that memory leak. I'd just like to have some confirmation from p5p that this was indeed solved . . . on purpose. :-)