Fastest way to iterate over the keys in a hash

Question

I initialized a hash as in this question and ran some benchmarks.

Benchmark.measure { a = h.keys }
#=>  0.010000   0.000000   0.010000 (  0.019832)
Benchmark.measure { a.each { |k| } }
#=>  0.060000   0.000000   0.060000 (  0.057262)
Benchmark.measure { h.each { |k, v| } }
#=>  0.320000   0.000000   0.320000 (  0.319768)
Benchmark.measure { h.each_key { |k| } }
#=>  0.310000   0.000000   0.310000 (  0.312656)
Benchmark.measure { h.each_pair { |k, v| } }
#=>  0.330000   0.000000   0.330000 (  0.331452)

I have thought that turning the hash into an array of keys would be slower, but it's not, and each_key has similar performance to each. Why is this? What is the best approach?

What is the best approach? To not worry about it, because it will not be your bottleneck if you have actual code in your block. — Mark Thomas, Oct 08 '14 at 01:01
Note that `a.each { |k| }` does something very different from the rest: it does not return keys, but key-value pairs. — Amadan, Oct 08 '14 at 01:08

score 0 · Answer 1 · answered Oct 08 '14 at 01:41

Looking at some code on Github I can tell you 2 words: It depends. If it's not obvious that your variable name is a hash, then using .each_pair will make it obvious (since a hash is a collection of key-value PAIRs). If your variable name mentions 'hash', then this makes sense: hash.each |key, value|. It doesn't make sense to do something like hash.each_pair do |k| where you pass 1 block argument to `each_pair, it makes the code look confusing.

From analyzing several pages on Github, I'm tempted to conclude that the "best approach" is when using .each, you should name your block variables |key, value| to make it very obvious you're iterating over a hash (the block arguments will give you a clue it's a hash). When using .each_pair, you can be a bit more ambiguous about naming the block arguments like |my_key_1, my_key_2| since .each_pair is saying you're iterating over a hash. It's all about looking at your code 1 week (or month) later and being able to INSTANTLY tell what's going on. Unlike with Python, Ruby doesn't have "one right way" to do something.

Fastest way to iterate over the keys in a hash

1 Answers1