I've got a "very liberal" comparator in production at the Linux Counter
project.
It mostly works, because the people who assign names tend to not assign
names that collide under the comparator.
But the strings that result from the comparator are quite distant from
"expected" strings in many cases - in particular, "asciifying" Japanese
with a heuristic that works OK for Chinese Mandarin gives output that's
simply meaningless.
Better Than Nothing Matching......
Harald
_______________________________________________
Ietf@xxxxxxxx
https://www1.ietf.org/mailman/listinfo/ietf