Patch: FYI: PR 16134

Tom Tromey tromey@redhat.com
Tue Jun 22 18:59:00 GMT 2004


>>>>> "Bryce" == Bryce McKinlay <mckinlay@redhat.com> writes:

Bryce> Hannes Wallnoefer discovered that the encoder cache lookup in
Bryce> UnicodeToBytes and BytesToUnicode would always fail, because the cache
Bryce> code did not canonicalize the encoding name prior to lookup.

Whoops.

Bryce> +    encoding = canonicalize(encoding);

The original intent of this code was that the original (non-canonical)
name be passed to the iconv converter.  The idea is that our
canonicalization might result in a name not known by the underlying
iconv implementation.

This patch changes that.  And actually I don't know that this has ever
mattered.  But it would be simple to restore this behavior by keeping
the non-canonical name and passing it to the iconv converter
constructor.

Tom



More information about the Java-patches mailing list