Iterable caches are configured to count each entry as the number of underlying items in the iterable, aka a list of [1, 2, 3] counts as 3 items in the cache (not 1). Currently an empty iterable ends up counting as 0 items in the cache, which could cause the cache to use much more memory than expected.

A potential solution could be to do len(d) or 1 which would ensure any item in the cache counts as at least 1 item toward the cache limit.