> Well, they are listed in http://www.unicode.org/Public/UNIDATA/DerivedAge.txt > If you search for "noncharacter" there, you will find which ones were designated in which Unicode > version. (Only two were designated in Unicode 1.)
Thanks, I forgot to check this file, which was introduced later to allow applications certified to comply to a particular version to be used with later versions (for example to allow emulation of previous versions using default character properties and absence of decompositions or case mappings with non compliant texts that were encoded with unassigned code points). As I don't need support for non compliant texts, I simply dropped this reference. If this text file is really normative, then it means that texts using unassigned code points should be rejected in compliant applications and not accepted silently, not even by using default properties, which may break in a later version of Unicode... You also use the term "designated", isn't it also saying that it is "assigned" (for example codepoints are assigned to surrogates, even if they are not characters, and the FAQ already states that not all assigned code points are characters.