Whenever someone complains to me that the standard library is slow or "bloated" (whatever that means), I ask if they've ever profiled their code vs. the standard library version. 9 times out of 10 they have not, and are operating out of mythology rather than measurement.