At 13:12 +0200 2001/05/27, Marcel Oliver wrote: >1. Store the full language context with every character token sequence > along the lines that Javier suggests. In other words, treat the > language context as part of the input encoding. It would seem that > if Frank's requirement for an ICR ("a single item must have a > unique and well-defined meaning") is to be met, it would > essentially mean that every character needs to be tagged for > language context. Perhaps I misread this as you are saying that every character should be stamped with language context: I do think that the language context should be everywhere present, though, and that one should pick it up as early and conveniently as possible. But I do not know how that should be done. Could one not classify all available (human) languages, as one in Unicode attempts to classify all characters, and make that available as a code? -- The number of languages should be much fewer than the characters. In such a model, there will be a generic language code corresponding to old TeX. Hans Aberg