This feature covers the integration of libkkc, a new Japanese Kana Kanji input library, into Fedora, along with:
- IBus input method engine using libkkc as the backend (ibus-kkc)
- language model data (libkkc-data)
- Name: Daiki Ueno
- Email: ueno at gnu.org
- Targeted release: Fedora 19
- Last updated: 2013-01-15
- Percentage of completion: 60%
There are currently two options for typical users to input Japanese sentences: ibus-anthy or ibus-mozc. However, both have issues:
- Anthy, the backend library, has been dead upstream for years.
- The accuracy is not good because of bugs in the core algorithm of Anthy.
- Contributions to the input method are limited to Google employees.
- There are no library interface. That means it cannot easily be used by other input method frameworks than IBus, such as Fcitx and uim.
ibus-kkc will be an open and better replacement of those.
Benefit to Fedora
- Provide an intelligent and openly developed Japanese input method.
- Support for phrase dictionary
- Support for Japanese numeric conversion (done)
- Provide Anthy compatible library for smooth transition from ibus-anthy
- Import marisa-trie, libkkc, ibus-kkc packages to Fedora (packages are available in private repo)
How To Test
- Try to add "Japanese (Kana Kanji)" input source from gnome-control-center Region panel.
- Enable "Japanese (Kana Kanji)" input source from the gnome-shell.
- Input some Japanese text, say "watashinonamaehanakanodesu", to input a Kana-string "わたしのなまえはなかのです"
- Type Space to convert it to a Kana-Kanji-mixed-string "わたしの名前は中野です".
- With the libkkc back end, Japanese users can speed up the Kana Kanji input thanks to its high accuracy algorithm than ibus-anthy.
- libkkc (not packaged yet)
- marisa-trie (not packaged yet)
- In worst case, we could drop the package and hide it from the gnome-control-center.
- ibus-kkc with new libkkc backend provides a more intelligent Kana Kanji input method and speed up the user input of Japanese characters.