yaml_json_benchmark.rb

require 'benchmark'
require 'typhoeus'
require 'yajl'
require 'json'
require 'yaml'
require 'ruby-debug'
require 'uri'
 require 'yajl/http_stream'

hash = {:a => 1}
base_uri = "http://api.devver.net/projects/git%3A%2F%2Fgithub.com%2Fsinatra%2Fsinatra.git/revisions/"
shas = ["186a5bc9b66ac163e4f74a88c3406a1587d97eae", "e49365f1cd7299a3a63fea7594943acd1f289322", "79a99939167b039bc740d693eb25a4ed09b37b52", "1b8fda825e9cda5f44b26684119b19a49c2e416b", "f942ad76589d649b66458efe61648f13a08ef8a0", "e153415527c903191ae64cb5965fc33cb30b1a7a", "3cfd06ad86ca8febf339037e37c6765572c89f42", "849a5a95fc33e6ddd3633863c4bc3ee0065c503a", "15651d9ee14270048b511ab535287d3ddc965ad7", "30c2ef839649dc3b124672854d32a34b88b52501", "0915274b6959a4130517dd1c1b5790df4f200e9e", "171bf43341cd0f71c87299740876a25a9a58266b", "3ea36c94cf5f65d5d3244df07c384a2e70647853", "ceac46f0bc129a6e994a06100aa854f606fe5992", "035fe4705d54590b5477faeee1c9cd96a63fcc8d", "2723d7ec3b192d720fc4ae2d53c48fea4dd71dce", "ae34a6fde5e15e9ba3ca40cf800d0366e44eec1f", "19efdf869984071b801d02795ed70d5e31aa3e62", "f5efc370c8450a8b1308a3a87f743fe8b25e5097", "b6701db3faf1fba0bb5675e8a78385b618280fb2", "ccb66b0f3096773599fcff93925865c788d98d34", "8a7dd92e1c1b94ac1beeafda5af8898384a02f00", "e1638a43ad3a5c5ea70db56944a5230cc86e537a", "c496254f2be3d27bba238bf78c1c7376976eb924", "794437e921c8bbafd106aa99819af69293e61574", "e2c73bf1b036adb31853b2c92909eb447876d233", "11e1f8a971ce2e40d544815d359ee19dace48f4c", "7526dacf33d4411d7f0e98bceeadc91b46dc56bb", "7882efdf0bf24055f11be52a326f25efababc67b", "37eb3a767f6697241fa15c8f4edfd0f9a63d2f70"]
# change this to 5 to get an idea of perf difference for SHAs just on a dashboard graph
shas = shas[0,30] 
URLS = shas.map { |sha| base_uri+sha+"/raw_metrics"}

def make_typhoeus_requests(accept_type, &block)
  hydra = Typhoeus::Hydra.new(:max_concurrency => 5)
  URLS.each do |url|
    request = Typhoeus::Request.new(url, :headers => {:Accept => accept_type})
    request.on_complete(&block)
    hydra.queue request
  end
  hydra.run
end

metrics = []

Benchmark.bmbm(25) do |bm|
  bm.report('typhoeus yaml') do
    metrics[0] = []
    make_typhoeus_requests('text/x-yaml') do |response|
      metrics[0] << YAML.load(response.body)
    end
  end

  bm.report('typhoeus json') do
    metrics[1] = []
    make_typhoeus_requests('application/json') do |response|
      metrics[1] << JSON.parse(response.body)
    end
  end

  bm.report('typhoeus yajl') do
    metrics[2] = []
    make_typhoeus_requests('application/json') do |response|
      metrics[2] << Yajl::Parser.parse(response.body)
    end
  end

  # bm.report('streaming yajl') do
#     metrics[3] = []
#     URLS.each do |url|
#       metrics[3] << Yajl::HttpStream.get(URI.parse(url))
#     end
#   end
end

def noop
end


# This is intended to make sure we're not comparing the case where YAML is getting real data
# but JSON is getting an empty hash for some reason
# A few discrepancies may show up because some cached data in the real API has 'rails_best_practices'
# (for the YAML representation) while the JSON does not
metrics.each_index do |i|
  next if i+1 == metrics.length
  metrics[0].length.times do |j|
    if metrics[i][j].keys.map{|x|x.to_s}.sort == metrics[i+i][j].keys.map{|x|x.to_s}.sort
      #puts 'same'
    else
      puts "difference between #{i} and #{i+i} on #{shas[j]}"
      noop
    end
  end
end
puts 'done'