Template Version: @(#)sac_nextcase 1.64 07/13/07 SMI
This information is Copyright 2008 Sun Microsystems
1. Introduction
1.1. Project/Component Working Name:
Korean Input Method Enhancement
1.2. Name of Document Author/Supplier:
Author: Aijin Kim
1.3 Date of This Document:
09 February, 2008
4. Technical Description
ntroduction
1.1. Project/Component Working Name:
Korean Input Method Enhancement
1.2. Name of Document Author/Supplier:
Aijin Kim (Aijin.Kim at Sun.COM)
2. Project Summary
2.1. Project Description:
This project enhances user experience and functionality of the Korean
input method
on Solaris by implementing a new Korean LE(Language Engine),HangulLE
and replacing the existing KOLE.
HangulLE imports libhangul which is an open sourced library for hangul
processing.
Please see 4.1.2 for more detail about libhangul.
In Nevada, KOLE will be replaced with HangulLE, while in S10URs, KOLE
will be kept
and HangulLE will be added to the existing package so that we can
guarantee
safety in user experience perspective during S10URs with a transition
stage
of the replacement.
Once the implementationis completed, HangulLE will also be open sourced
to openi18n.org.
http://www.openi18n.org
The OSR numbers related with the projects are as below:
- OSR 7595 : in-bound of Libhangul
- OSR 7747 : out-bound of HangulLE
2.2. Risks and Assumptions:
As this project is specific to Korean, engineer should have basic
knowledge of Korean language and understand the functions of Korean
input method.
Libhangul is an open sourced library and the latest version is 0.0.6.
Further maintaining work may be required according to the upgrade
of libhangul.
There are some changes in user's usage between KOLE and HangulLE and
all the
changes will be mentioned in Korean User Guide and OLH of HangulLE.
Please see 4.1.6 for more detail about usage compatibility.
As KOLE won't be delivered any more in Navada, EOFs of KOLE and Hanja
Tool which
is used for user-defined Hangul-Hanja dictionary in KOLE are needed.
More specifically, EOF announcement is planned at the next available
S10UR and EOF execution is planned at Solaris Nevada.
3. Business Summary
3.1. Problem Area:
Currently, KOLE's UI is still based on Motif while other LEs (e.g. CLE,
TWLE,HKLE) were upgraded to be based on GTK+ in IIIMF r12. It causes
UI inconsistency of input methods on Solaris.
In the aspect of implementation, KOLE is not upgraded to use Common-LE
API
and it may cause difficulty with code maintenance.
Current KOLE has limited functionality for the major functions
compared to other major input methods such as Microsoft Korean IME
and SCIM(Smart Common Input Method).
Major functions which should be improved are as below:
- Hangul(Korean) input
- Hangul/Hanja(Korean/Chinese) conversion
- keyboard layout selection
3.5. Opportunity Window/Exposure:
Solaris Nevada
Solaris 10U6
3.6. How will you know when you are done?:
HangulLE replaces KOLE's all features and is integrated to
Solaris Nevada and Solaris 10U6.
4. Technical Description:
4.1. Details:
4.1.1. Basic concept of Korean language
Hangul is the native alphabet of the Korean language.
One hangul syllable character is composed of two or three phonemic
elements by composition rule.
Syllable = Leading consonant + Vowel (+ Trailing consonant)
4.1.2. External technology - Libhangul
Libhangul is a generalized and portable library for hangul processing.
It is developed by Choe Hwanjin(choe.hwanjin at gmail.com) as
Open Hangul Project under GNU LGPL. Currently, libhangul is imported to
several Korean input methods. Libhangul supports various kinds of
keyboard layout automata and also provides Hanja dictionary.
http://kldp.net/projects/hangul/
4.1.3. Structure of HangulLE
HangulLE imports libhangul as an IME(Input Method Engine).
During IME initialization, LE checks if libhangul is installed
previously.
If libhangul.so is not found, IME can not be started. Hanja dictionary
is also
loaded during IME registration.
+-------------+
| HangulLE |
+-------------+
| +--------------+
+-------------+ | libhangul.so |
| Hangul IME |<--------- +--------------+
+-------------+ | hanja.txt |
+--------------+
4.1.4. Functionality
(1) Hangul input
Hangul input is the most basic function of Korean input method. HangulLE
provides enhanced user conveniences for Hangul input.
Generally, user should use 'Shift+single consonant' to input a double
consonant. HangulLE supports the double consonant input by pressing
single consonant key twice and it's more convenient than using Shift
key.
For example, if user want to input '???', (s)he can input '???' twice
as well
as 'Shift+???'.
- Shift + '???' -> '???'
- '???' + '???' -> '???'
Vowel can't be the first of Hangul syllable in the composition rule of
Korean. If user makes a mistake with input order, input characters are
not
composited to a syllable. Then user should delete them and input with
correct sequence again. It often annoys users.
- '???' + '???' -> '???' : correct order
- '???' + '???' -> '???''???' : incorrect order
HangulLE provides auto composition function which automatically corrects
input sequence and composites to a syllable. Even if user inputs a vowel
first and then inputs a consonant, (s)he doesn't have to delete and
re-input.
- '???' + '???' -> '???'
- '???' + '???' -> '???'
User is still able to input intended character sequence even if it's
an invalid order of syllable.
(2) Hangul/Hanja conversion
Hanja dictionary of libhangul has not only Hanja character candidates
but also
meaning and pronunciation information of each character. All the
information
is displayed in the candidate list and it provides much convenience
during Hanja input.
To input Hanja, user should input Hangul syllables first and convert
them to
proper Hanja characters by selecting a candidate from the candidate
list.
Hangul/Hanja conversion is executed by shortcut key or icon on the
auxiliary window. Hanja key is used as a shortcut key.
(3) Keyboard layout selection
Keyboard layout for Korean is switched in the properties window.
HangulLE
supports below keyboard layouts:
- dubeol
- sebeol final
- sebeol 390
- sebeol sunarea
Default value is dubeol layout which is most commonly used. And
User can switch keyboard layout in the properties window.
4.1.5. User Interface - Auxiliary window
Auxiliary window is an additional GUI For CJK LEs. User can switch
input methods and change properties of LE via the auxiliary window.
The layout of HangulLE's auxiliary window as below:
+-----------------------------------------------------------------+
| Mode Switch | Full/Half | Hanja | Virtual Keyboard | Properties
|
+-----------------------------------------------------------------+
- IM Switch: Input mode switching
- Full/Half: Full/Half width switching
- Hanja: hangul/hanja conversion
- Virtual Keyboard: Virtual Keyboard window
- Properties: properties setting
4.1.6. Usage Compatibility
All the features of existing KOLE will be replaced by HangulLE or
alternative
ways. However, there are some changes in user's usage.
(1) Input Mode
KOLE has four different input modes, which are exclusive each other:
- Hangul
- Hex(KS X 1001)
- Hex(UTF-8)
- Symbol
Hangul and Hex(KS X 1001) mode is kept in HangulLE.
Hex(UTF-8) mode will be covered by UNICODE engine of IIIMF.
Instead of exclusive symbol input mode, HangulLE provides a feature of
symbol input
with virtual keyboard, which is consistent with CLE,TWLE. It allows
user to input
hangul and symbols without switching input modes. And also
'consonant+Hanja key'
is supported for symbol input just like Windows IME and SCIM-hangul.
(2) Shorcut Key Definition
KOLE is using several shortcut keys with 'Ctrl+', which are already
defined
in some applications. It causes conflicts between KOLE and applications
and
this issue was already reported as CR6267800.
HangulLE doesn't use 'Ctrl+' key sequences to avoid conflicts with
applications.
Instead, HangulLE defines shortcut keys to be similar with Windows IME
and SCIM.
(3) User-defined Hangul-Hanja Dictionary
As all the hangul/hanja processings and hanja dictionary are managed by
libhangul,
the existing Hanja Tool won't be used for user-defined Hangul-Hanja
dictionary
in HangulLE.
libhangul doesn't support the feature of user-defined dictionary in the
current version. However, community discussions are currently in
progress
to support the feeature in libhangul. Once the feature is provided in
the new
version of libhangul, it will replace Hanja Tool.
4.1.7. Selection of A Language Engine in S10UR
In S10UR, HangulLE will be the default Korean language engine. When
users switch
to Korean input mode with 'Ctrl+Space' or 'Hangul/English' key,
HangulLE will
be luanched.
Users can select a language engine easily with below steps:
- Left-click the Input Method icon on the gnome-terminal.
- From the language list, select 'Korean' and preferred language
engine.
4.2. Bug/RFE Number(s):
4903905 Hex input mode algorithm needs to be revised
4919114 Sebul keyinput needs separated automata
6267800 shortkey for hanja conversion can not work in some gnome
application, such as gedit
4.3. In Scope:
N/A
4.4. Out of Scope:
N/A
4.5. Interfaces:
- Interface Imported:
Interface Stability Note
-----------------------------------------------------------------------------------
IIIM protocol and libraries Committed, PSARC/2005/525
Consolidation
Private, and
Project Private
- Interfaces exported including to be Obsoleted:
Interface Stability Note
-----------------------------------------------------------------------------------
HangulLE GUI and behavior Uncommitted GUI and
methods of
new Hangul
engine
/usr/lib/iiim/le/hangul/ Consolidation Hangul
directory where
Private various
Hangul engines and
their files
reside
/usr/lib/libhangul.so.0.0.0 Volatile libhangul
library
/usr/share/libhangul/hanja/hanja.txt Volatile Hanja
dictionary
KOLE GUI and behavior Obsolete GUI and
methods of the
Uncommitted current
Hangul engine
/usr/lib/iiim/le/kole/* Obsolete The current
KOLE and
Consolidation its directory
Private
/usr/dt/bin/sdthanja Obsolete Hanja Tool
Uncommitted
/usr/lib/mle/{ko,ko.UTF-8}/syshjd Obsolete System Hanja
dictionary
Consolidation
Private
$HOME/.usrhjd Obsolete User Hanja
dictionary
Consolidation
Private
/usr/dt/app-defaults/ Obsolete Hanja Tool
resource files
{ko,ko.UTF-8}/Sdthanja Consolidation
Private
/usr/openwin/lib/locale/{ko, Obsolete Hanja Tool
L10N files
ko.UTF-8}/LC_MESSAGES/sdthanja.cat Consolidation
Private
- Release binding:
Project team asks for Micro/Patch release binding for newly exported
interfaces. For the interfaces to be obsoleted, the project team ask
for
EOF announcement at Micro/Patch (i.e., S10UR) and EOF execution at
Minor
(i.e., Nevada).
4.6. Doc Impact:
OLH of Korean Input Method
Korean Solaris User's Guide
EOF announcement into Release Notes
4.7. Admin/Config Impact:
N/A
4.8. HA Impact:
N/A
4.9. I18N/L10N Impact:
This project is a part of internationalization effort.
4.10. Packaging & Delivery:
For S10UR, both KOLE and HangulLE will be delivered in the existing
package.
SUNWkleu Korean locale core package
For Nevada, HangulLE will be delivered as a new package and KOLE will
be removed.
The name of the new package is not decided yet.
4.11. Security Impact:
N/A
4.12. Dependencies:
N/A
5. Reference Documents:
PSARC 2005/525 IIIMF upgrade to revision 12
Internet/Intranet Input Method Architecture:
http://www.openi18n.org/subgroups/im/iiimf/whitepaper/whitepaper.html
Internet-Intranet Input Method Protocol (IIIMP) specification:
http://www.openi18n.org/subgroups/im/iiimf/protocol/spec.html
ASARC 1999/276 /usr/lib/im directory for platform independent Input
Method Systems
ASARC 1998/515 Multi Script Input Support in UTF-8 locales
ASARC 1998/422 IIIMP - Internet/Intranet Input Method Protocol
ASARC 1997/359 Multilingual Input Method Server for X/Java/Corba
6. Resources and Schedule:
6.1. Project Availability:
03/24/2008, Nevada build 86
S10U6 build 1
6.3. Cost of Capital Resources
G11nws.sfbay as repository server. Existing
Desktop development machine. Existing
6.5. ARC review type:
FastTrack
7. Prototype Availability:
7.1. Prototype Availability:
09/30/2007
7.2. Prototype Cost
15 person.days
6. Resources and Schedule
6.4. Steering Committee requested information
6.4.1. Consolidation C-team Name:
Globalization
6.5. ARC review type: FastTrack
6.6. ARC Exposure: open
