I'm the author of the original mint cache and completely agree that this implementation is a very nice improvement. ( the original was written under duress! ) Kudos, guys!
Andrew B. Thanks, Glenn! Sharing code is a beautiful thing. In case you're interested, one other feature we are considering adding is a "delete" method that merely lowers the timeout on the cached value (instead of truly deleting it), thereby triggering the mint refresh. That way one could avoid the dogpile effect when manually invalidating.