class MessagePack::Factory::Pool::AbstractPool

def checkin(member)
  # If the pool is already full, we simply drop the extra member.
  # This is because contrary to a connection pool, creating an extra instance
  # is extremely unlikely to cause some kind of resource exhaustion.
  #
  # We could cycle the members (keep the newer one) but first It's more work and second
  # the older member might have been created pre-fork, so it might be at least partially
  # in shared memory.
  if member && @members.size < @size
    member.reset
    @members << member
  end
end