I need an advice
Thu Aug 16 09:07:00 GMT 2001
>>>>> "Martin" == Martin Kahlert <firstname.lastname@example.org> writes:
Martin> The system property "file.encoding" is thus set to "646" in
Martin> natSystem.cc This encoding seems to be either incorrect or it
Martin> cannot be handled,
Yes. `646' is not an encoding we recognize internally. And, Solaris
seems to let nl_langinfo(CODESET) return a name which iconv_open()
doesn't recognize. (IMHO that is a serious bug.)
Martin> - The quick fix for me was
Martin> #define DEFAULT_FILE_ENCODING "8859_1" instead of
Martin> #define DEFAULT_FILE_ENCODING file_encoding ()
Martin> in natSystem.cc
Martin> That's not nice, but it works.
What if we add `646', along with a comment explaining why, to the
alias table in gnu/gcj/convert/IOConverter.java? I assume `646' is
`ASCII', based on the presence of `iso646-us' in the table. Is that
true? Or is it really 8859-1?
I think adding 646 to the alias table is reasonable since this is a
known problem for many Solaris boxes.
More information about the Java