At 10:35 17/02/04 +0000, Simon Marlow wrote:

We didn't choose Latin-1 over UTF-8: the current situation just reflects the fact that we haven't implemented UTF-8 yet. It's not trivial to expand GHC's current Alex specification to handle the whole Unicode character set. See this message for some musings:

http://www.haskell.org/pipermail/glasgow-haskell-users/2003-December/006 094.html

Ah, yes. It seems that there are two issues here: (1) a way to use alternative character set encodings for Haskell source programs. (2) using an extended repertoire of characters outside comment bodies in a Haskell source module. My comments were directed toward (1). I can see that (2) might be more challenging, but maybe less important to address? You mentioned the obvious place in which one might wish to include extended characters in Haskell non-comment code, viz. in strings. I'm not sure if that would be a good thing to do, for reasons of source code portability/exchangeability, at least until such time as text editors routinely support a fuller character repertoire. (I think that allowing escape sequences and a native Char representation that supports Unicode is basically a solved problem, is it not?) #g ------------ Graham Klyne For email: http://www.ninebynine.org/#Contact