Gather benchmarking stats

build_analysis_csvs.rb

# yes, it's a pile of hacks

require 'set'
require 'csv'
require 'pry'
require 'rugged'
require 'time'

filename = 'deployments.csv'
release_regex = /^release_[\d]+$/

staging_and_prod = []
CSV.foreach(filename, headers: true) do |row|
  if row['environment'] != 'preview'
    if row['domain'] == 'github.com'
      if release_regex =~ row['version']
        staging_and_prod << row
      end
    end
  end
end

repos = Set.new

repo_tags = {}

staging_and_prod.each do |release|
  repo = release['repo']
  project = repo.split('/')[1]
  tag = release['version']

  repos << repo

  if release['environment'] == 'production'
    repo_tags[repo] ||= {}
    repo_tags[repo][tag] ||= {}
    repo_tags[repo][tag]['deployed_at'] = Time.parse release['deployed_at']
  end
end

def resolve_reference_to_commit(rugged_ref)
  if rugged_ref.target.class == Rugged::Commit
    return rugged_ref.target
  else
    return resolve_reference_to_commit rugged_ref.target
  end
end

repos.each do |repo_name|
  folder = repo_name.split("/")[1]
  begin
    repo = Rugged::Repository.new("../#{folder}")
    repo.tags.each do |tag|
      if release_regex =~ tag.name
        time = resolve_reference_to_commit(tag).time
        repo_tags[repo_name] ||= {}
        repo_tags[repo_name][tag.name] ||= {}
        repo_tags[repo_name][tag.name]['merged_at'] = time
      end
    end
  rescue Rugged::OSError => e
    repo_tags.delete repo_name
  rescue Exception => e
    binding.pry
    next
  end
end

repo_tags.each do |repo,tags|
  repo_tags[repo] = tags
    .select { |tag, data| data.key?('merged_at') }
    .sort_by { |tag, data| data['merged_at'] }
end

def get_next_deploy_date(future_tag_data)
  future_tag_data.each do |tag_data|
    if tag_data[1].key? 'deployed_at'
      return tag_data[1]['deployed_at']
    end
  end
  nil
end

def hours(merged, deployed)
  if deployed < merged
    nil
  else
    ((deployed - merged) / (60 * 60)).round
  end
end

repo_tags.each do |repo, tags|
  tags.each_with_index do |tag_data, i|
    tag = tag_data[0]
    data = tag_data[1]

    if data.key?('deployed_at')
      repo_tags[repo][i][1]['delay'] = hours(data['merged_at'], data['deployed_at'])
    else
      deploy_date = get_next_deploy_date(tags.last(tags.size - i - 1))
      if deploy_date
        repo_tags[repo][i][1]['delay'] = hours(data['merged_at'], deploy_date)
      else
        repo_tags[repo][i][1]['delay'] = hours(data['merged_at'], Time.now - (60*60*24))
      end
    end

    repo_tags[repo][i][1]['year_merged'] = data['merged_at'].year
    repo_tags[repo][i][1]['month_merged'] = data['merged_at'].month
    repo_tags[repo][i][1]['day_merged'] = data['merged_at'].day
  end
end

out = {}
repo_tags.each do |repo, tags|
  tags.each do |tag_data|
    year = tag_data[1]['year_merged']
    month = tag_data[1]['month_merged']

    out["#{year}-#{month}"] ||= []
    out["#{year}-#{month}"] << tag_data[1]['delay']
  end
end

outfile = File.new('release_delay_averaged.csv', 'w+')
outfile << "year-month,average hours\n"
out.each do |month, data|
  average = data.reject(&:nil?).inject(0.0) { |sum, el| sum + el } / data.size
  outfile << "#{month},#{average.round}\n"
end
outfile.close

outfile = File.new('release_delay.csv', 'w+')
outfile << "repo,tag,merged_at,delay_in_hours,year_merged,month_merged,day_merged\n"
repo_tags.each do |repo, tags|
  tags.reject { |t| t[1]['delay'].nil? } each do |tag_data|
    outfile << "#{repo},#{tag_data[0]},#{tag_data[1]['merged_at']},#{tag_data[1]['delay']},#{tag_data[1]['year_merged']},#{tag_data[1]['month_merged']},#{tag_data[1]['day_merged']}\n"
  end
end
outfile.close

export_deployments.rake

# Upload this to the Release app in integration
# run it
# download the resulting CSV

require 'csv'

desc "Exports a CSV of historic deployments to /tmp/deployments.csv"
task export_deployments: :environment do
  deploys = Deployment.includes(:application).map do |deployment|
    if deployment.application && deployment.environment != 'preview'
      {
        version: deployment.version,
        environment: deployment.environment,
        application: deployment.application.name,
        repo: deployment.application.repo,
        domain: deployment.application.domain,
        deployed_at: deployment.created_at
      }
    end
  end.compact

  csv = CSV.generate do |csv|
    csv << deploys.first.keys

    deploys.each do |deploy|
      csv << deploy.values
    end
  end

  File.write('/tmp/deployments.csv', csv)
end