Improve performance of sorting milestone issues

This cuts down the time it takes to sort issues of a milestone by about
10x. In the previous setup the code would run a SQL query for every
issue that had to be sorted. The new setup instead runs a single SQL
query to update all the given issues at once.

The attached benchmark used to run at around 60 iterations per second,
using the new setup this hovers around 600 iterations per second. Timing
wise a request to update a milestone with 40-something issues would take
about 760 ms, in the new setup this only takes about 130 ms.

Fixes #3066

CHANGELOG

@@ -4,6 +4,7 @@ v 8.2.0 (unreleased)
   - Show last project commit to default branch on project home page
   - Highlight comment based on anchor in URL
   - Adds ability to remove the forked relationship from project settings screen. (Han Loong Liauw)
+  - Improved performance of sorting milestone issues
 
 v 8.1.0 (unreleased)
   - Fix bug preventing mentioned issued from being closed when MR is merged using fast-forward merge.

app/controllers/projects/milestones_controller.rb

@@ -75,11 +75,7 @@ class Projects::MilestonesController < Projects::ApplicationController
   end
 
   def sort_issues
-    @issues = @milestone.issues.where(id: params['sortable_issue'])
-    @issues.each do |issue|
-      issue.position = params['sortable_issue'].index(issue.id.to_s) + 1
-      issue.save
-    end
+    @milestone.sort_issues(params['sortable_issue'].map(&:to_i))
 
     render json: { saved: true }
   end

app/models/milestone.rb

@@ -105,4 +105,36 @@ class Milestone < ActiveRecord::Base
   def author_id
     nil
   end
+
+  # Sorts the issues for the given IDs.
+  #
+  # This method runs a single SQL query using a CASE statement to update the
+  # position of all issues in the current milestone (scoped to the list of IDs).
+  #
+  # Given the ids [10, 20, 30] this method produces a SQL query something like
+  # the following:
+  #
+  #     UPDATE issues
+  #     SET position = CASE
+  #       WHEN id = 10 THEN 1
+  #       WHEN id = 20 THEN 2
+  #       WHEN id = 30 THEN 3
+  #       ELSE position
+  #     END
+  #     WHERE id IN (10, 20, 30);
+  #
+  # This method expects that the IDs given in `ids` are already Fixnums.
+  def sort_issues(ids)
+    pairs = []
+
+    ids.each_with_index do |id, index|
+      pairs << id
+      pairs << index + 1
+    end
+
+    conditions = 'WHEN id = ? THEN ? ' * ids.length
+
+    issues.where(id: ids).
+      update_all(["position = CASE #{conditions} ELSE position END", *pairs])
+  end
 end

spec/benchmarks/models/milestone_spec.rb

+require 'spec_helper'
+
+describe Milestone, benchmark: true do
+  describe '#sort_issues' do
+    let(:milestone) { create(:milestone) }
+
+    let(:issue1) { create(:issue, milestone: milestone) }
+    let(:issue2) { create(:issue, milestone: milestone) }
+    let(:issue3) { create(:issue, milestone: milestone) }
+
+    let(:issue_ids) { [issue3.id, issue2.id, issue1.id] }
+
+    benchmark_subject { milestone.sort_issues(issue_ids) }
+
+    it { is_expected.to iterate_per_second(500) }
+  end
+end

spec/models/milestone_spec.rb

@@ -140,4 +140,32 @@ describe Milestone do
     end
   end
 
+  describe '#sort_issues' do
+    let(:milestone) { create(:milestone) }
+
+    let(:issue1) { create(:issue, milestone: milestone, position: 1) }
+    let(:issue2) { create(:issue, milestone: milestone, position: 2) }
+    let(:issue3) { create(:issue, milestone: milestone, position: 3) }
+    let(:issue4) { create(:issue, position: 42) }
+
+    it 'sorts the given issues' do
+      milestone.sort_issues([issue3.id, issue2.id, issue1.id])
+
+      issue1.reload
+      issue2.reload
+      issue3.reload
+
+      expect(issue1.position).to eq(3)
+      expect(issue2.position).to eq(2)
+      expect(issue3.position).to eq(1)
+    end
+
+    it 'ignores issues not part of the milestone' do
+      milestone.sort_issues([issue3.id, issue2.id, issue1.id, issue4.id])
+
+      issue4.reload
+
+      expect(issue4.position).to eq(42)
+    end
+  end
 end