Encoding determination algorithm in XMLReader class has a 'dotless-i' problem for Turkish locale systems. In getContentTypeEncoding method;
encoding = (encoding!=null) ? encoding.toUpperCase() : null;
which results in a problem with iso-encoded feeds. Because in Turkish locale, uppercase of "i" (lowercase i) is "İ" (uppercase I with dot). Since encoding types use western encoding, the right code should be;
encoding = (encoding!=null) ? encoding.toUpperCase(Locale.US) : null;
About dotless-i problem, you may review my blog about it.
Operating Systems with Turkish locale