XML 1.0 Fifth Edition Specification Errata

E01

Section 2.2 Characters

Add a Note at the very end of the section as follows:

Note:

[Unicode] (conformance clause C06) says that canonically equivalent sequences of characters ought to be treated as identical. However, XML parsed entities (including document entities) that are canonically equivalent according to Unicode but which use distinct code point (character) sequences are considered distinct by XML processors. Therefore, all XML parsed entities SHOULD be created in a "fully normalized" form per [CharMod-Norm]. Otherwise the user might unknowingly create canonically equivalent but unequal sequences that appear identical to the user but which are treated as distinct by XML processors.

A document can still be well-formed, even if it is not in a normalized form. XML processors MAY verify that the document being processed is in a fully-normalized form and report to the application whether it is or not.

Section A.2 Other References

Add a reference to CharMod-Norm:

CharMod-Norm

W3C Working Draft. Character Model for the World Wide Web 1.0: Normalization. François Yergeau, Martin J. Dürst, Richard Ishida, Addison Phillips, Misha Wolf, Tex Texin. (See http://www.w3.org/TR/charmod-norm/.)

Rationale

The ill effects of (the lack of) Unicode normalization are noteworthy, but advice in a note is the best that can be done as part of an erratum, i.e. without changing the spec normatively.