Compatibility

Purpose

This page provides a comprehensive comparison of features, performance, and capabilities across all supported XML adapters.

Feature compatibility matrix

Feature/Operation Nokogiri Oga REXML LibXML Ox

Core operations

Parse XML string

✅ Full

✅ Full

✅ Full

✅ Full

✅ Full

Parse XML file/IO

✅ Full

✅ Full

✅ Full

✅ Full

✅ Full

Serialize to XML

✅ Full

✅ Full

✅ Full

✅ Full

✅ Full

Element operations

Create elements

✅ Full

✅ Full

✅ Full

✅ Full

✅ Full

Get/set attributes

✅ Full

✅ Full

✅ Full

✅ Full

✅ Full

Add/remove children

✅ Full

✅ Full

✅ Full

✅ Full

✅ Full

Replace nodes

✅ Full

✅ Full

✅ Full

✅ Full

⚠️ Limited1

Namespace operations

Add namespaces

✅ Full

✅ Full

✅ Full

✅ Full

✅ Full

Default namespaces

✅ Full

✅ Full

✅ Full

✅ Full

⚠️ Basic

Namespace inheritance

✅ Full

✅ Full

✅ Full

✅ Full

❌ None

Namespaced attributes

✅ Full

✅ Full

✅ Full

✅ Full

⚠️ Limited

XPath queries

Basic paths (//element)

✅ Full

✅ Full

✅ Full

✅ Full

✅ Full

Attribute predicates ([@id])

✅ Full

✅ Full

✅ Full

✅ Full

⚠️ Name only2

Attribute values ([@id='123'])

✅ Full

✅ Full

✅ Full

✅ Full

❌ None3

Logical operators

✅ Full

✅ Full

✅ Full

✅ Full

❌ None

Position predicates

✅ Full

✅ Full

✅ Full

✅ Full

❌ None

Text predicates

✅ Full

✅ Full

✅ Full

✅ Full

❌ None

Namespace XPath

✅ Full

✅ Full

❌ None

✅ Full

❌ None

Parent axis (..)

✅ Full

✅ Full

✅ Full

✅ Full

❌ None

Sibling axes

✅ Full

✅ Full

✅ Full

✅ Full

❌ None

XPath functions

✅ Full

✅ Full

✅ Full

✅ Full

❌ None

Special content

CDATA sections

✅ Full

✅ Full

✅ Full

✅ Full

✅ Full

Comments

✅ Full

✅ Full

✅ Full

✅ Full

✅ Full

Processing instructions

✅ Full

✅ Full

✅ Full

✅ Full

✅ Full

DOCTYPE declarations

✅ Full

✅ Full

✅ Full

⚠️ Limited4

✅ Full

Performance

Parse speed

Fast

Fast

Medium

Very Fast

Very Fast

Serialize speed

Fast

Fast

Medium

Very Fast

Fastest

Memory usage

Good

Medium

Medium

Excellent

Best

Thread safety

✅ Yes

✅ Yes

✅ Yes

✅ Yes

✅ Yes

Legend

✅ Full

Complete support with no limitations

⚠️ Limited

Partial support with workarounds available

❌ None

Not supported

Footnotes

1 Ox: Text node replacement may fail in some cases due to internal node structure

2 Ox: //book[@id] returns all book elements (doesn’t filter by attribute existence)

3 Ox: Use .find { |el| el["id"] == "123" } instead of XPath attribute value predicates

4 LibXML: DOCTYPE parsing works, serialization is limited (no perfect round-trip)

Performance comparison

Parsing performance (medium 10KB XML) 

Ox         ████████████████████████████████████████████████ 289 ips
LibXML     ███████████████████████████████████ 120 ips
Nokogiri   █████████████████ 76 ips
Oga        ███████████ 50 ips
REXML      ████ 15 ips

Serialization performance 

Ox         ████████████████████████████████████████████████ 39,203 ips
Nokogiri   █████████████████ 13,900 ips
LibXML     ███████ 5,600 ips
Oga        ████ 3,200 ips
REXML      ██ 1,500 ips

XPath performance (simple queries) 

Nokogiri   ████████████████████████████████████████████████ 64,958 ips
LibXML     ███████████████████████████████████████████████ 62,000 ips
Oga        ███████████████████████ 28,000 ips
REXML      ███████ 8,500 ips
Ox         ███████ 9,640 ips
Performance numbers are approximate and vary by system and document complexity.

Memory usage comparison

Adapter Memory delta Description

Ox

0.0 MB

Best - minimal allocation

Nokogiri

-0.1 MB

Excellent efficiency

LibXML

+0.1 MB

Very good

Oga

+0.3 MB

Good for pure Ruby

REXML

+0.5 MB

Medium overhead

XPath capability summary

Capability level Adapters

Full XPath 1.0

Nokogiri, LibXML, Oga

Basic XPath

REXML (no namespaces), Ox (no predicates)

Ruby workarounds needed

Ox, REXML (for complex queries)

Adapter selection decision tree 

Need pure Ruby?
├─ Yes
│  ├─ Need full XPath? → Oga
│  ├─ No external gems? → REXML
│  └─ Basic XPath OK? → REXML
└─ No (C extensions OK)
   ├─ Need maximum speed? → Ox
   ├─ Simple documents? → Ox
   ├─ Complex XPath? → Nokogiri or LibXML
   ├─ Industry standard? → Nokogiri
   └─ Alternative to Nokogiri? → LibXML

Recommendations by use case

Use case Primary choice Alternative

General XML processing

Nokogiri

LibXML

Maximum performance

Ox

LibXML

Pure Ruby requirement

Oga

REXML

No external dependencies

REXML

N/A

Complex XPath queries

Nokogiri

LibXML, Oga

High-throughput simple docs

Ox

LibXML

Namespace-heavy documents

Nokogiri

LibXML, Oga

Memory-constrained

Ox

Nokogiri

Testing across adapters

To ensure your code works across adapters:

# Test with multiple adapters
[:nokogiri, :libxml, :oga, :rexml, :ox].each do |adapter_name|
  next unless adapter_available?(adapter_name)

  context = Moxml.new
  context.config.adapter = adapter_name

  doc = context.parse(xml)

  # Test your operations
  results = doc.xpath('//book')
  # Add assertions
end

Migration guide

From Nokogiri to Ox:

  • Replace complex XPath with Ruby filtering

  • Test namespace operations carefully

  • Verify text node replacements work

From REXML to Nokogiri/LibXML:

  • No code changes needed

  • XPath queries will work better automatically

  • Performance will improve significantly

From Ox to Nokogiri:

  • Complex XPath will work automatically

  • No workarounds needed

  • Performance may decrease for parsing

See also