> On 2008.08.09 23:48:33 +0100, "Eric Y. Kow" <[EMAIL PROTECTED]> scribbled > 1.3K characters: >> On Sat, Aug 09, 2008 at 11:33:02 -0600, [EMAIL PROTECTED] wrote: >> > Adds support for utf8-string if the user has it installed. >> >> As Gwern suggests, it may just be worthwhile to make this mandatory, in >> the interest darcs code distillation. Any comments on encode and >> errors, Rob? >> >> -- >> Eric Kow <http://www.nltg.brighton.ac.uk/home/Eric.Kow> > > Well, as I've said on #darcs: I'm not a UTF expert, but it seems to me > that since we only use 'encode', and since apparently 'encode' only > operates on [Char]s and the IO functions which could give us Chars are > runtime are not supposed to ever give us bad Chars, error handling isn't > an issue. But I'm not sure of this which is why I was hoping for input for > other people before I did anything more than just clear cruft out of > UTF8.lhs. > > -- > gwern > JFKSWC X Montenegro press-release conspiracy Pine Bellsouth CIO Gulf site >
Well, I do think the "use utf8-string if specified" idea is good. But I think leaving the error handling code in is still a good idea, since I've since too many "This should never happen!"s happen. -Rob _______________________________________________ darcs-users mailing list [email protected] http://lists.osuosl.org/mailman/listinfo/darcs-users
