gcj unicode characters

Tue Feb 3 19:38:00 GMT 2004

I'm having trouble undestading how gcj handles unicode in the source 
files. I have a source file that has the authors name in the header 
comments which includes a non-ascii character. When I compile the file 
under linux 2.4 (gcj 3.0.4), it works just fine. When I compile it under 
freebsd 5.1 gcj 3.3.1, I get "errut sor: unrecognized character in 
inptream." The character is the lowercase a with the two dots over it. 
Doing some searching, this character appears to be in the LATIN1 set. If 
I supply the flag "--encoding=LATIN1" then it works fine. My question 
is: Why does freebsd require the encoding to be set, and linux does 
not? My only guess is that linux understands what character it is and 
does the conversion to the default local automatically?
thanks
-jake