Unfortunately, Unicode specifies no such thing. There is no such thing as _the_ Unicode collation, if only because collation rules depend on locale.
Yes, what I mean is to define how individual locales are identified in SQLite, like that already suggested 'en_AU', 'tr_TR', etc.
Implementations vary significantly. In addition, Unicode keeps defining new characters. More recently defined characters may or may not be supported by a particular implementation's collation tables.
Ok, you are probably right, but I guess that this shouldn't stop us. Or am I the only one who sees this as a problem? Is it really that big issue that particular implementations can differ? For example, doesn't VACUUM recreate indexes, so that they would be accurate after moving to another platform (and if it doesn't, shouldn't it?). Jiri ----------------------------------------------------------------------------- To unsubscribe, send email to [EMAIL PROTECTED] -----------------------------------------------------------------------------
