On Tuesday, January 11st, 2005 20:59Z Henry Spencer va escriure: > > Indeed, as the Unicode spec points out, you really want to do > case-insensitive comparisons by first mapping characters to > equivalence classes -- not by mapping to a particular case and > assuming that all equivalent letters will map to the same character.
But in a Turkish locale, 'I' and 'i' are NOT in the same equivalence class: the first is [.dotlessi.] (not sure it is the correct POSIX way to write this... hope you get the idea) while the second is in [.i.] Using the UCA primary keys will lead to a similar problem. Or did I miss something? Antoine -- Linux-UTF8: i18n of Linux on all levels Archive: http://mail.nl.linux.org/linux-utf8/
