Performance

Benchmark stdlib

Guesses about speed are often wrong, so measure with the benchmark library. The bm helper runs each report block and prints timings.

Compare two approaches side by side and keep the faster one only when the difference survives repeated runs.

require "benchmark" Benchmark.bm do |x| x.report("plus:") { 1000.times { "a" + "b" } } end

Magic frozen_string_literal comment

Every string literal normally allocates a fresh mutable object. The magic comment at the very first line freezes all literals in the file instead.

Frozen strings can be shared safely, which trims both allocation count and garbage collector pressure.

# frozen_string_literal: true name = "Ada" puts name.frozen?

Avoid garbage

Intermediate arrays and strings cost time twice, once to build and once to collect. Bang methods transform in place and skip the throwaway copy.

Prefer map with a bang over plain map when the original array is no longer needed elsewhere.

names = ["ada", "bob"] names.map! { |n| n.upcase } puts names.inspect

Measure first

Optimize the hot path, not a random line. A monotonic clock around the suspect code shows elapsed seconds without wall clock jumps.

When the number refuses to shrink, the bottleneck lives elsewhere, so profile a wider scope before rewriting more code.

start = Process.clock_gettime(Process::CLOCK_MONOTONIC) 1000.times { "x" * 10 } done = Process.clock_gettime(Process::CLOCK_MONOTONIC) puts(done - start)

Next: Modern Ruby tour

Article author: Arthur Isaev