> My solution is "subset-of-UTF-8 corresponding to the native language of user".
I already demonstrated that this was not true, and so did your own remark about code page 936 . You are not rational about this, and you avoided a civil discussion of the points in my previous message, so I am not willing to take this discussion any further.
> This Windows-specific part would take a lot of LOC in Lua source.
A patch has been posted in this same thread, so we are talking about homework already done.