http://bugs.winehq.org/show_bug.cgi?id=5469
------- Additional Comments From charles@diagnos.co.uk 2006-30-06 11:15 ------- Righto, I think I understand. Sorry if I talk a bit of nonsense here, I'm just trying to get things clear in my head...
As long as the individual characters stay within the first 256 points, should there be any danger of different sequences matching, as I don't think there are any decomposed characters within this range?
So, from looking at the code for wine_compare_string, there are three levels of comparison (compare_unicode_weights, compare_diacritic_weights, compare_case_weights) - am I right in assuming these correspond to the order of
1. alphabetic ordering 2. diacritic ordering 3. case ordering
(I'm copying from: http://www.unicode.org/reports/tr10/#Scope)
Should there be a 4th tie-breaking level here?