wrapped TTL can return nil from lookup-or-miss
This is a known issue for the TTL cache since it can invalid an entry on lookup and so the double lookup performed by through-cache and lookup-or-miss can cause a "hit" immediately followed by an invalidating "lookup" resulting in nil.
core.memoize already has spin logic to account for this. Similar logic should be added to lookup-or-miss (and the possibility of invalidation be noted on some other API functions, although the spin will not be added to those).
In 0.8.1 release.