#942662 libxml2: rejects valid NCName characters

Package:
libxml2
Source:
libxml2
Description:
GNOME XML library
Submitter:
"brian m. carlson"
Date:
2019-10-19 19:09:03 UTC
Severity:
normal
#942662#5
Date:
2019-10-19 19:06:27 UTC
From:
To:
XML 1.0 Fifth Edition dramatically increased the range of characters
allowed in an NCName, and therefore in an ID (such as an xml:id
attribute).  However, libxml2 is still using the obsolete fourth
edition, which means that it rejects well-formed documents.

For example, the following is a well-formed document which is rejected:
---
<a xml:id="_ⅰ"/>
---

Because the document is erroneously considered ill-formed, xsltproc
cannot process it, either.

The fifth edition came out in 2008, so this is not a new phenomenon.  In
fact, upstream has a bug open with fixed files at
https://bugzilla.gnome.org/show_bug.cgi?id=675373; a suitable patch
should be applied to the Debian package.  If desired, I can synthesize
the files in that bug report into an actual patch, should it be applied.