Another 646 patch and performance comparison with TOWER

[Tom Tromey]

>>>>> "Martin" == Martin Kahlert <martin.kahlert@infineon.com> writes:

Martin> Inside init (java/lang/natString.cc) we get into
Martin> BytesToUnicode::getDefaultDecode And after the system did not
Martin> succeed with any useful encoding, an
Martin> UnsupportedEncodingException is thrown which is handled by
Martin> substituting 646 by 8859_1 (see the code above these lines).

In gnu.gcj.convert.IOConverter I see this:

    // On Solaris the default encoding, as returned by nl_langinfo(),
    // is `646' (aka ASCII), but the Solaris iconv_open() doesn't
    // understand that.  We work around the problem by adding an
    // explicit alias for Solaris users.
    hash.put ("646", "ASCII");

Why doesn't the `646' get turned into `ASCII' in getDecoder?

>> Could you try this patch?  Be sure, of course, to back out your
>> current patch before applying this one.

Martin> I tried it but the patch seems to be buggy:

Yeah.  Let me work on it some more.

Tom