libxml2 not ignoring whitespace when validating xsd:int

This is a copy of the old bugzilla issue #726766 (see https://bugzilla.gnome.org/show_bug.cgi?id=726766) first reported by Markus Schöpflin (2014-03-20 08:42:48 UTC ). The problem is still present today. I tested and confirmed it using the xmllint tool as provided by libxml2-2.9.12 (as packaged for Fedora 33 and 34)

copy of old bugzilla text:

Versions tested and affected: 2.9.1-2 (FC), 2.7.6-14 (CentOS)

Given the following schema:

<xsd:schema xmlns:xsd="http://www.w3.org/2001/XMLSchema">
  <xsd:element name="a" type="xsd:int"/>
</xsd:schema>

And the following document:

<a> 42</a>

Xmllint fails to validate this:

$ xmllint --schema test_schema.xml test_doc.xml 
<?xml version="1.0"?>
<a> 42</a>
test_doc.xml:1: element a: Schemas validity error : Element 'a': ' 42' is not a valid value of the atomic type 'xs:int'.
test_doc.xml fails to validate

According to XML Schema Part 2: Datatypes Second Edition, section 4.3.6 all atomic data types other than 'string' have their 'whiteSpace' constraint set to 'collapse', so I think the element 'a' should be valid.

There is even a comment to this effect in the libxml2 sources (xmlschemas.c / xmlSchemaTypeFixupWhitespace()):

/*

• For all atomic datatypes other than string (and types derived
• by restriction from it) the value of whiteSpace is fixed to
• collapse */