Following up today's discussion: * There is no universally accepted definition of diacritics we could use, in Unicode or elsewhere. (The original, linguistic definition is language-specific and thus useless for DNS.) Therefore we are forced to define it for ourselves. * The anonymous Unicode expert's argument that "LATIN WITH..." might include letters that aren't diacritic is, in this context, circular: if we include them in our defintion then they are diacritics by (our) definition! We should choose a definition that works for us. Well-defined, mechanically testable, predictable, easy to implement &c. What doesn't matter is whether or not all characters there are diacritics by some other definition, as long as they don't cause any problems for our purposes. -- Tapani Tarvainen