When building a simple test program (at least when using libedit's readline compatibility code) it mishandles Unicode characters. % touch räksmörgås.txt % cc -I/usr/local/include -DUSE_LIBREADLINE=1 -o tr tr.c -L/usr/local/lib -lreadline % ./tr >räksmörgås.txt 18 bytes % cc -I/usr/local/include -DUSE_LIBEDIT -o tr tr.c -L/usr/local/lib -ledit % ./tr >r\U+00C3\U+00A4ksm\U+00C3\U+00B6rg\U+00C3\U+00A5s.txt 18 bytes (type "r" then TAB for tab-completion). Also libedit doesn't really allow me to enter non us-ascii characters at all. This is different behaviour from gnu readline (see the attached simple test program). Behavior is the same on FreeBSD 11.3-RELEASE and CURRENT.
SOrry for the delay responding to that PR, to you have your tr.c file here so I can test ?