On Thu, Mar 09, 2006 at 11:38:21PM +0900, Stephen J. Turnbull wrote: > >>>>> "Jamie" == Jamie Webb <[EMAIL PROTECTED]> writes: > > Jamie> e.g. we can always insert new lines rather than adding > Jamie> characters to existing ones, perhaps like this: > > Recognizing a line requires recognizing a newline,
Yes, but that issue is nothing new. Darcs already AFAIK will not work with character sets which are not ASCII-compatible, because it already has to recognise newlines. My point is that character-based diffs do not make the situation any worse. > which requires > recognizing a character. Not with UTF-8. > Eventually you have to put constraints on what Darcs considers to be > text. I would recommend "UTF-8-compatible and not declared binary" as That's a bad way of putting it. Very little is 'UTF-8-compatible'. The point is that UTF-8, ISO8859-x, etc. are backwards-compatible with ASCII. > a reasonable heuristic for most purposes, with "UTF-8-compatible and > declared text" as the bullet-proof alternative, and "declared unibyte > text with newline = LF" as a reasonable backward compatibility > compromise (note that this is fairly safe for Windows since you can > think of CR as a nuisance whitespace character, although > darcs-inserted newlines without CR would probably confuse most Windows > editors). I'd suggest that the current design is better. -- Jamie Webb _______________________________________________ darcs-users mailing list [email protected] http://www.abridgegame.org/mailman/listinfo/darcs-users
