definition of NCName too strict

[jensopetersen]

I would normally use ASCII characters only for xml:id's and suchlike, but for a specific project it makes sense to use Chinese characters as xml:id's. Here eXist appears to me to be too strict. While it accepts characters in the Basic Multilingual Plane (the second one below), it refuses later additions to Unicode.

<graphs>
    <graph cp="3400" xml:id="&#x3400;"/>
    <graph cp="4E00" xml:id="&#x4e00;"/>
    <graph cp="20000" xml:id="&#x20000;"/>
    <graph cp="f0000" xml:id="&#xf0000;"/>
    <graph cp="100000" xml:id="&#x100000;"/>
</graphs>

[jensopetersen]

<x xml:id="1"/>

is well-formed, but Xerces holds on to the original XML 1.0 definition of NCName which was quite quickly superseded in XML 1.0 by the definition made in XML 1.1. Since eXist uses Xerces, this means that xml:id’s beginning with digits are held to be malformed in eXist, though they are well-formed according to XML 1.0.

[adamretter]

@jensopetersen I wonder if the Xerces property http://xml.org/sax/properties/document-xml-version could help, or perhaps even the feature http://xml.org/sax/features/xml-1.1?

See:
https://xerces.apache.org/xerces2-j/properties.html
https://xerces.apache.org/xerces2-j/features.html

[tuurma]

In the same vein see discussion at relaxng/jing-trang#188

I have encountered the issue when wanting to use some polytonic Greek characters

ͷ 0377 GREEK SMALL LETTER PAMPHYLIAN DIGAMMA
ϝ 03DD GREEK SMALL LETTER DIGAMMA
Ͷ 0376 GREEK CAPITAL LETTER PAMPHYLIAN DIGAMMA

which are perfectly kosher NameChar under XML Spec fifth edition https://www.w3.org/TR/REC-xml/#d0e804, alas not a Letter according to fourth https://www.w3.org/TR/2006/REC-xml-20060816/#NT-Letter

[duncdrum]

this seems to have been fixed, using the OP examples in 4.5.0.
Please open a new issue if there are still problems with NCName handling.