With any luck, this will prompt the library writers to produce a fix.
Keep your fingers crossed. :)
Update: It seems that the library writers need to fix their documentation. Apparently, if you want to send a UTF-8 encoded string that contains non-ASCII characters, you must send through the Unicode BOM as the first string in the sequence. If that were documented anywhere, I would have done so long ago. But that appears to be secret knowledge, held only by a few.
Thud, thud, thud.