Unfortunately the threads aren't really preserved correctly in the archive. This query gets back most of the key discussions from Oct 2001
Not all of the following software is a library. Some other programs are mentioned, vaguely suspecting that people looking here might be interested in them, and to give possible developers of a Boost XML library an overview of related technologies (For example both a XML parser and Boost.Serialization could interface to a XML-Database).
Arabica is a XML toolkit, consisting of SAX, DOM and Xpath APIs. Is is written in a relatively modern C++ Style. It can handle UTF-8 and UCS-2 encodings and works with custom string types. The library currently does not contain an own parser implementation but rather wraps Expat, Xerces, Libxml and MSXML. It is BSD-Style licensed.
Tinyxml is a XML parser which currently only exposes a DOM API. It is advertised as beeing lean and easy to integrate into existing projects.
Xerces-C is a feature-rich XML Toolkit. It delivers SAX and DOM APIs. It features XML Schema and Namespaces support, too. The Apache Software Foundation also hosts a C++ XSLT implementation called Xalan and a webservices toolkit (Axis) available in C++. A XML/C++ databinding facility called Xmlbeans is developed in conjunction with Roguewave.
XSD is a code generator for binding XML-Schema types to C++. Both GPL and licensing for proprietary projects is available.
Berkely DB XML is a database for XML documents, featuring XQuery, XML Schemas and XML Namespaces among other things. It uses the C-written Berkely DB underneath. Both XML and non XML-objects can be stored in one database.
Sablotron is another XSLT and Xpath processor. It uses the Expat parser.