At 2002-09-03 01:27, Martin Norbäck wrote:

Hehe, I was about to commit my version yesterday, but decided to wait until today to check some things. Well, I'll commit anyway, UTF8norpan.hs and well see what happens :)

They are quite different, so you could compare them. Must work now.

Yeah, I've got one too in HBase, but I'd need to port it to the regular Prelude, and there's no point if there are two other versions... Anyone want to do a UTF-16? It's dealing with "surrogates", probably a bit simpler than sucking different numbers of bits out of UTF-8 octets. UTF-32 should be trivial. Once upon a time there was UCS4, a 31-bit space for characters. Then apparently they decided they didn't need more than 17*2^16 (=1,114,112) characters, and now we just have what UTF-16 can reach. Then there are charsets. Data for them might be on the web somewhere. -- Ashley Yakeley, Seattle WA