Description of problem: Some very simple Japanese words don't covert to their expected values, requiring selection of candidates and editing, making input tedious. Version-Release number of selected component (if applicable): Canna-3.6-24 How reproducible: Always Steps to Reproduce: 1. Inside gnome-terminal running in default Japanese environment (romaji input)with Canna running, press Shift-Space to turn on editing. 2. Type "omoimasu" ("(I/you/we/they) think"), then press space to convert, enter to commit 3. Type "wakarimashita" ("(I/you/we/they) understand"), then press space to convert, enter to commit Actual results: "å°¾æ¯ä»é " for "omoimasu" "ååãçä¸" for "wakarimashita". Expected results: "æãã¾ã" for "omoimasu" "ãããã¾ãã" or "åããã¾ãã" for "wakarimashita" Additional info: It is possible to produce the expected results, but it requires post- editing. Also, Canna will temporarily cache the corrected result, but the user will either need to modify the dictionary or redo the next time a canna connection is made. Additionally, one should not For simple common words, the most obvious candidate should appear first (like FreeWnn/Wnn, ATOK, and MS-IME) for efficient Japanese input
Well, actually it was already fixed in 3.7-1. but it depends on your .canna file. you have to enable fuzokugo dictionary in .canna.