Comments on `Tradeoffs between mean-square and worst-case performances in adaptive filtering’