See http://lists.w3.org/Archives/Public/public-html/2009Oct/0376.html The Belarusian localization has ISO-8859-5 as the character encoding fallback. This seems odd, since Russian and Ukranian default to Windows-1251. How was the ISO-8859-5 default reached? Wouldn't defaulting to Windows-1251 give a better user experience with unlabeled legacy Web content?
It seems reasonable.
Also I will turn on heuristic detector by default intl.charset.detector = ruprob
Corrected in 1.9.2 branch (Firefox 3.6) and in head trunk.