When JAF reads its resources from its jar file (or other jar files),
it doesn't specify a character set to use when constructing the
InputStreamReader. This cuases it to use the character set specified
by the property file.encoding, which is appropriate for files stored
in the local system's filesystem, but not for resources in jar files
that are supposed to be system independent. This is usually not a problem
because the resources contain only ASCII, which is a subset of almost all
character sets. Except for EBCDIC. Thus JAF doesn't read the characters
in the resource files properly on a system with file.encoding set to
EBCDIC (Cp1047), e.g., OS 390.