Here are some questions that might arise:
Japanese language support must be provided by the operative system. If you can see Japanese characters in your web browser, then they should be rendered correctly in this tool. Otherwise, it's a bug (please let us know about it!!).
Thanks to Zsolt for noticing.
Since version 0.4.1, romaji is converted to katakana where capital letters are inputted.
Thanks Zsolt for the idea.
The old browse mode is still there. To activate it, set in the file 'config.properties' this line:
In exercises where several kanji are presented and the user chooses the correct one, kanji similar to the correct answer could be choosen as wrong answers, in order to have a more challenging exercise. In this version, having 'similarity' enabled only means that similar kanji are more likely to be choosen as wrong answers in those exercises.
The algorithm is based on radicals and it's not perfect, of course. Next releases might have better methods. If you don't like the idea, you can switch it off in the file 'config.properties':
Thanks Lars Yencken for his paper on kanji similarity models and his help.