[Haskell-cafe] Problem with text and text-icu packages

[David Carter]

Hi,

I'm struggling with text-0.1 and text-icu-0.1, as announced at

http://www.serpentine.com/blog/2009/02/27/finally-fast-unicode-support-for-haskell/

The code in

http://pastebin.com/m7d8d9f91

is intended to read in a UTF-8 file a1.txt, reverse it twice, and
write it out to another UTF-8 file a2.txt, which I would have thought
should make a2.txt identical to a1.txt. Mostly it is, but sometimes it
isn't, e.g.:

% echo a > a1.txt
% tryicu   # <- my code
% cat a2.txt
ࠋ퐤

Specifically, a1.txt contains bytes 97 and 10, while a2.txt contains
bytes 224 160 139 237 144 164.

Have I misconstrued things, or is this a bug?

I am running ghc 6.10.2 and ICU 4.0.1 on SuSE.

Thanks for any help

David