could you please send me example-files that fail, if possible ?
just sent
The reason is quite simple, I decided to also search for latin-2 encoding needed for langauges like Bosnian, Croatian, Czech, Hungarian, Polish, Romanian, Serbian (when in the Latin script), Slovak, Slovenian, Upper Sorbian, and Lower Sorbian, but it breaks some latin-1 encoded texts due to their similarity, not all and not always (depending of the used characters).
FireFox 3.5 still does not automatically detect these languages correctly (at least not in all cases).
I think I will add a configuration-option that the user can decide whether he/she wants to try to detect latin-2 also (with a hint, that it might break latin-1 detection).
I don't know a better solution, the alternative would be either not to detect latin-2 or some latin-1 encodings.
By the way: I did not stumble over the problem, because if you use german umlauts (äöü) or the german sharp s (ß), even single-byte texts are detected correctly, but I use utf-8 in almost any cases anyways (except for at work where I have to use windows tools, that do not all work with unicode correctly).