Re: [xml-dev] How many text nodes in this element with a CDATAsection?

[Michael Kay <mike@saxonica.com>]

I believe that setCoalescing() only affects the DOM created by an XML parser; it doesn't affect a DOM created programmatically.

Saxon's XPath-DOM interface attempts to provide a coalesced view of a DOM that contains multiple adjacent text nodes,inced  allowing for programmatically-created DOMs.

However, I'm not 100% convinced it's worth the effort, since when you update a DOM programmatically, it's certainly possible create all kinds of garbage which Saxon won't be able to handle, such as elements with invalid names or namespaces.

Michael Kay

Saxonica

On 8 Feb 2022, at 15:12, Mukul Gandhi <mukulg@softwarebytes.org> wrote:

actually, Xerces's latest XSD 1.1 implementation for xs:assert within the 2.12.2 release doesn't have this bug, and therefore that doesn't require any fix.

The XPath 2.0 engine that we use, outside of any host language (on which Xerces's XPath 2.0 support is based upon), required this fix for XML CDATA nodes (the method, https://docs.oracle.com/javase/7/docs/api/javax/xml/parsers/DocumentBuilderFactory.html#setCoalescing(boolean) did the magic).

On Tue, Feb 8, 2022 at 7:07 PM Mukul Gandhi <mukulg@softwarebytes.org> wrote:

Hi Mike,

On Tue, Feb 8, 2022 at 6:03 AM Michael Kay <mike@saxonica.com> wrote:

In the XDM model the content of the Test element is a single text node. XDM does not allow two adjacent text nodes.

There are certainly XPath implementations, especially those running against a DOM, that get this wrong. You seem to be running against an implementation that has "cdata-section" nodes, which do not exist in XDM.

Thanks, for the clarification.

This was a bug with Apache Xerces's XPath 2.0 engine as well, that's based on DOM. We just applied the fix, as you've stated.

--

Regards,

Mukul Gandhi