On Tue, 2004-10-12 at 11:20, Matt Fowles wrote: > According to the rt messages and back forth, this is resolved... > > http://rt.perl.org/rt3//Ticket/Display.html?id=31026 > > Is it?
It's *more* efficient, yes, but it's not pretty. The interpreter initializes a couple of NCI subs before it creates the place where it can store the cache, so the hash builds twice. I didn't dig around more to try workarounds. -- c
