[vidarh]

You're right, he's comparing the cost for that very simple case. Do anything remotely complex and it quickly becomes just noise.

[timr]

As with all things, it depends on context. If you're trying to write a graphics or audio system in Ruby, this kind of thing can really matter. If you're writing a Rails app, it's rounding error on the time waiting for I/O.

Profiling is essential when figuring out how to improve your code. What we're doing here is explaining one weird benchmark result.

[voronoff]

Except that's not what's being said:

"Why blocks make Ruby methods 439% slower" implies something that isn't happening here, at least to me.