Your preferred solution has advantages only if the long term costs are real. But how often is it necessary to apply casing rules to existing documents? Quite rarely, I would think. Search engines might want to, I agree, but I would expect a basic search engine to fold dotted and dotless i on the basis that they cannot be distinguished reliably. On your solution the costs must be borne for all documents before moving to Unicode or its putative successor; with the solution chosen by Unicode, the costs need be borne only for the minority of documents which actually need casing rules applied to them....
Mixed Turkish and other European language documents that are without language markup have the same problem, no matter where the burden is placed. Some I's will receive inappropriate glyphs when a casing rule is applied. The problem is just as pronounced with either method, and the need to rewrite such documents to ensure proper casing is the same.
I will admit that my preferred solution has higher initial costs, but lower long term costs that cause me to favor it. In any case, changing to my preferred solution now would not be worth the confusion that would be caused. If there ever is a successor to Unicode, then it would be worth examining this idea, but such an event is at least twenty years away.
We might hope that within twenty years almost all new documents will be marked with their language.
-- Peter Kirk [EMAIL PROTECTED] (personal) [EMAIL PROTECTED] (work) http://www.qaya.org/
