Mark Davis <[EMAIL PROTECTED]> wrote: > Ken is right; collation is *quite* complex.
I don't doubt that, but I don't see the relevance to the reordering proposal, which would map each code point to another code point (in a reversible way) just before ACE encoding, and would apply the reverse mapping just after ACE decoding. Complexity is indeed a concern, but this is nowhere near as complex as collation, because it operates on individual code points, not strings. Do the NICs have some numbers for us? Soobok Lee has used the names in the VGRS testbed both to train his reordering tables and to evaluate their effectiveness. It would be very interesting to see how effective the latest tables are for names from other registries, like the Japan, China, and Taiwan NICs. AMC
