Comments are formatted differently on mri and jruby

[jlahtinen]

To Reproduce

#!/usr/bin/env ruby

require 'nokogiri'
require 'minitest/autorun'

# Test for reporting issue
class Test < MiniTest::Spec
  describe 'document#to_xml' do
    it 'formats comments' do
      expected_xml = <<~HEREDOC
        <!--test comment-->
        <root>test xml</root>
      HEREDOC

      doc = Nokogiri::XML::Builder.new do |xml|
        xml.comment 'test comment'
        xml.root 'test xml'
      end
      xml = doc.to_xml(save_with:
                       Nokogiri::XML::Node::SaveOptions::NO_DECLARATION)

      assert_equal expected_xml, xml
    end
  end
end

Expected behavior

with cruby and jruby

1 runs, 1 assertions, 0 failures, 0 errors, 0 skips

Actual behavior

with jruby

  1) Failure:
document#to_xml#test_0001_formats comments [./nokogiri-issue.rb:26]:
--- expected
+++ actual
@@ -1,3 +1 @@
-"<!--test comment-->
-<root>test xml</root>
-"
+"<!--test comment--><root>test xml</root>"

1 runs, 1 assertions, 1 failures, 0 errors, 0 skips

Additional context

# Nokogiri (1.10.7)
    ---
    warnings: []
    nokogiri: 1.10.7
    ruby:
      version: 2.6.5
      platform: x86_64-linux
      description: ruby 2.6.5p114 (2019-10-01 revision 67812) [x86_64-linux]
      engine: ruby
    libxml:
      binding: extension
      source: packaged
      libxml2_path: "/home/jlahtinen/.rvm/gems/ruby-2.6.5/gems/nokogiri-1.10.7/ports/x86_64-pc-linux-gnu/libxml2/2.9.10"
      libxslt_path: "/home/jlahtinen/.rvm/gems/ruby-2.6.5/gems/nokogiri-1.10.7/ports/x86_64-pc-linux-gnu/libxslt/1.1.34"
      libxml2_patches:
      - 0001-Revert-Do-not-URI-escape-in-server-side-includes.patch
      - 0002-Remove-script-macro-support.patch
      - 0003-Update-entities-to-remove-handling-of-ssi.patch
      - 0004-libxml2.la-is-in-top_builddir.patch
      libxslt_patches: []
      compiled: 2.9.10
      loaded: 2.9.10

# Nokogiri (1.10.7)
    ---
    warnings: []
    nokogiri: 1.10.7
    ruby:
      version: 2.5.3
      platform: java
      description: jruby 9.2.8.0 (2.5.3) 2019-10-11 a1ac7ff OpenJDK 64-Bit Server VM 25.232-b09
        on 1.8.0_232-b09 +jit [linux-x86_64]
      engine: jruby
      jruby: 9.2.8.0
    xerces: Xerces-J 2.12.0
    nekohtml: NekoHTML 1.9.21

[flavorjones]

Hi, @jlahtinen. Thanks for reporting this.

What you're seeing is a difference in how two different XML libraries parse and format documents. The JRuby implementation of Nokogiri uses Xerces, and the CRuby implementation uses libxml2. Nokogiri is a relatively thin wrapper on top of these libraries, and inherits all this behavior from them -- and so there's nothing we can easily do to change this behavior.

Sorry we can't do more here to help you.