0.2.1: - "ctype" should be known class. In the XMMS patch just describe how the playlist patch can be altered from "ctype" to "id3" in the case mpg123 patches are not applied. + Problem. Some times there are broken 'ogg' tags with CP1251 instead of unicode inside. Dissabling autodetection for them will cause incorrect encoding to be displayed. - If autodetected charset is disabled treat the autodetection failed. - If Locale Language is English, - resolve default charset to "Off". - Don't try to translate if postponed translate failed (for some time) - Solve mpg123 slowdown. 0.3.x: - Shall we always return result even if recoding is not required (not a NULL as now)? Currently we have no way to know if something is failed or if we just have a correct string already... - Language and Encoding autodetection improvements. + LibRCD should use DB4 with statistic for different languages + The statistic should be gathered using: * Aspell dictionaries. * Special program getting text on the standard input. * From LibRCC when language is preciesely detected. + The LibRCD engine should be used to fast language detection as well. * Just analyze output UTF8 string + Add ispell support - Translation improvemtns + Look if there are any offline translation libraries available. + Use stardict (or other dictionary) to translate on per-word basis. + Try to translate to first parrent encoding if translation to the current one is failed. + Transliterate support for more languages. 0.4.x: - Buffer managment: + SetBufferSize ( 0 - autogrow ) - Move all recoding functionality on the rccConfig Level - Revise Locking Subsystem - Load class configurations from the XML files. 0.5.x: - Special encoding. + Instead of IConv call considered function. * For example: Transliterate * For example: Translate to English + The options for encoding should be passed as a part of encoding name. * Develope naming conventions + Pluggable special encodings. 1.0.x: - Switch to Get/Ref/UnRef calls. - Drop down 'Class' keywords in all 'ClassCharset' function. Make it default behaviour. on request: - Multibyte(not-UTF8) support for FS classes