This is one of those ongoing topics that is yielding less and less further enlightenment. I am reminded of Aquinas's discussion of how ob Judg[e]ment Day to restore their bodies to exclusive cannibals properly and equitably. The treatment is lively and intelligent, but the issues are hard to engage with.
Shmuel is clearly right to emphasize that a 'roundtrip' facility must address not one but many requirements, some of them semantic and conflicting. The best that can be hoped for in these circumstances is 'engineering judg[e]ment' that yields a viable-in-some-circumstances compromise definition; and it is clear that in this context graphemes (glyphs) are more important than what were anciently called control characters. Moreover, complaint that that the intractable problem of fitting the contents of a pint pot into a quart jar has not been solved is bootless. SBCS variants of UNICODE are anyway at best dubious undertakings. If DBCS (or, in special circumstances, MBCS) UNICODE implementations are used instead, the problems that have been discussed in this thread at tedious length can be, indeed are, dealt with elegantly. John Gilmore, Ashland, MA 01721 - USA ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
