[Performance] Suggestion: Use `pygixml` as a high-performance XML backend

[MohammadRaziei]

Hi @dapper91,

First off, thank you for building pydantic-xml. It’s an elegant library that bridges the gap between Pydantic and XML seamlessly. I’m a big fan of the project.

I am the creator of pygixml and I believe it could significantly boost parsing performance for pydantic-xml, especially for large payloads.

Why consider pygixml?

• Massive Speedup: It is up to 14x faster than Python’s standard ElementTree and significantly faster than lxml in many scenarios (up to 8-14x vs stdlib, while lxml is only ~2x vs stdlib).
• Zero Runtime Dependencies: Unlike lxml, it requires no system libraries (libxml2). It bundles the C++ parser directly.
• Tiny Footprint: 0.45 MB package size vs. 5.5 MB for lxml.
• Low Memory: Uses ~7x less memory than standard parsers.

Performance Comparison (5,000 elements):

Library	Parse Time	Memory	Size	Dependencies
pygixml	0.0009 s	0.67 MB	0.45 MB	None
lxml	0.0041 s	0.67 MB	5.48 MB	libxml2
stdlib (ET)	0.0083 s	4.84 MB	0 MB	None

Impact on pydantic-xml:
Since pydantic-xml relies on XML parsing to validate and instantiate models, switching to or supporting pygixml as a backend could reduce latency and memory usage for applications processing large XML documents.

If you are open to this, I’d be happy to help integrate pygixml as an optional backend.

Best regards,

Mohammad Raziei
Creator of pygixml