Don't do iso-to-utf conversion on text that's already UTF8, as that corrupts the text; the CHANGES file is a mixture of different encodings, which is why we need a patch to fix it rather than just running it through iconv in %prep.