The Asia n Open VMS effort has been addressing vari
ous technical and engineering issues. 1'uc.. J- This sec
tion discusses the major ones.
Technical Issues
Loca l ization of the OpenVMS components to sup
port the Asian languages requ ires reengineering t he p rogram codes and text translation. The need to reengineer source code arises for two main reasons.
l . OpenVMS components make fu ndamental progra m mi ng assu mpt ions and practices based on t he ASCI I and DEC MCS character sets. For example,
- OpenVMS com ponents assume the character set to be ASCI I (plus D EC MCS in some cases), and b l i n d l y uppercase and lowercase charac
ters, val idate characters against the DEC MC:S, and define pri ntable characters accord ing ro the ASC I I and DEC MC:S encodi ngs.
- OpenVMS components assume characters to be 1 byte and use string manipulation algo
rithms based on 1-byte units.
- OpenVMS components assume the d isplay width of a character to be of fixed length (1 byte) and use screen d isplay management algo rithms based on the assu mption that
1 byte equals one d isplay col u m n .
- OpenVMS components assu me that t h e char
acter count, the byte count, a n d the d isplay width are the same. and use string manipu la
tion algorithms and character cel l term inal screen d isplay ma nagement based on this assu mption.
2. Some functio nal ity that is requ i red to support.
Asian l a nguages is missing in the standard Open VMS environment, For example,
- Keyboard input of Asian characters requ ires more compl i cated input method processing than is ava il able in the stand ard OpenVMS environment.
- Co l l ation ru les of Asian languages are radi
c a l ly d ifferen t from Engl ish col lation ru les, on which the stand ard OpenVMS environ
ment is based.
Digital Technical .Journal H1/. 5 No. 3 Sl//)/ll/!'r I'J'J.i
- The standard Open V MS environment does not support the appli cation-transparent process
i ng of l i iX:s.
- The writing d irection of Asian la nguages can be vert ical, i . e . , from top to bottom. The stan
dard OpenVMS environment assumes hori
zonta l , left -to-right languages.
Engineering Issues
H istorical ly, t he Asian loca l i za t ion of the OpenVJVIS system has been organized as an engineering effort that is separate from ma instream development. As a resu lt, a number of engineering constraints and overhead costs exist.
• Single la nguage support. The design goa l t()r the Asian OpenV\1S variants, as driven by the local market requi rements, has been targeted a t sup
port ing a single la nguage on one system, i.e., one l a nguage variant per system. As a result, no spe
cial design considerations are given to su pport
ing m u ltiple l a nguages on one system .
• Fu l l u pward compatibil ity. The top design requ irement is to keep fu ll downward compati
bil ity with original ASCII/DEC MCS OpenV\1S systems. A l l ASC II/DEC MCS appl ications with ex isting data must be able to ru n uncha nged on the Asian OpenVMS varian ts. In fact, an Asian OpenVMS system can, at any time, be reset to operate in the original DEC MCS mode, if desired . Therefore, most local ized components must be able to switch between the standard and Asian code paths. System mechanisms fo r determ in ing the current language varia nt and operating m ode are requ ired.
• Opti m al performan ce. Another design goa l is to minimize any pe rform ance i m pact on standard Engl ish components. As a resu l t , Asian codes are designed a round standard code paths. For exam
ple, branches fo r Asian code are p l aced at the end of a condi t ional statement, and Asian code branches out from the main l i ne code using spe
cial hooks.
• Limited or no kernel cha nges. Since Asian code cha nges are not merged int o the ma instream, kernel cha nges in Asian code wou l d be very diffi
cult to maintain with new OpenVMS releases. In addition, any kernel cha nges in the standard OpenVMS release wi l l l i kely hn:ak t h e Asian code. This puts a constraint on support ing Asian languages in OpenVJ\IIS kernel components.
77
Product Internationalization
• Com monality. Because the Asian languages share a lot of commonality, techniques such as common source are used for most Asia n local ized components to maximize engineering return by sharing common Asian local ization code.
Conclusions
Local language processing has become a mandatory functionality for computer systems sold in Asian markets. From the Open VMS operating system per
spective, the basic local Asian l anguage processing requirements are being addressed by its Asian language variants in a single-language-for-a-single
loca le manner. With globa l trade and the technol
ogy trend of distributed compu ting systems, the challenge for the future is to be able to provide OpenVMS services simultaneously to mu ltiple cl ients operating in different languages and code sets. Such a requ irement leads to the concept of a multil ingual operating system, which a llows soft
ware appl ications to run irrespective of the lan
guage and/or code set they support. With the avai lability of the ISO 10646 Universal Character Set (UCS) standard, the set of tools for building such a multilingual operating system has been enhanced. tH From an engineering perspective, the current Asian localization approach of Open VMS has been adopted historical ly because of a number of factors and constraints, such as the organization of engi
neering resources and the initial need to bring the capability rapidly to the market. The reengineering techniques are geared toward the character set encoding schemes cu rrently supported . The arrangement of performing localization remotely and independently from the original mainstream development has meant costly reengineering and maintenance overheads in the long term. With t he industrial trend of sh ipping global software sim u l
taneously satisfying mu ltiple different local market requirements, an international product engineer
ing approach must be taken to minimize the cost of worldwide system engineering to del iver a global product. In particu lar, the original product must be internationalized from the ground up, so that no separate reengineering is needed du ring local ization to support a loca l market. In addition, to achieve simultaneous worldwide del ivery, con
current engineering of loca lization needs to be performed closely in parallel with the product development.
78
References
I . T. Greenwood, " International Cu ltural Differ
ences i n Software," Digital Teclmical.Jounwl, vol . 5, no. 3 (Su mmer lS>93, this issue): 8 - 20.
2 . Code of the .Japanese Graphic Character Set for lnfonnation lnterchange, ]IS C 6226 -1978 (Tokyo: Japanese..: Standards Association, 1978)
3. Code of the .Japanese Graphic Character Set for b�(ormation Interchange, .JIS X 0208-1983 (Tokyo: Japanese Standards Association,
1983).
4. Code of the .Japanese Grapbic Cbaracter Set for lnj(mnation Interchange, .TIS X 0208-1990 (Tokyo: Japanese Standards Association, 1990).
5. Code of the Supplementary .fapa!lese Graphic Character Setfor lnfonnation In fer
change, )IS X 0212-1990 (Tokyo: Japanese Stan
dards Association, 1990).
6. Code for Information Interchange, ]IS X 020 1-1976 (Tokyo: Japanese Standards Association, 1976).
7. Code of Chinese Graphic Character Set for Information In terchange, GB 2312-1980 ( Bei
jing: Technical Standards Publ ishing, 198 1 ) . 8. Standard Interchange Code for General�y
used Chinese Characters, CNS 1 1643-1986 (lhipei: National Bureau of Standards, 1986) 9. Chi11ese Standard Interchange Code, CNS
1 1 6 43 -1992 (Taipei: National Bureau of Stan
dards, 1992)
10. Code /(Jr Inforrnation Interchange ( 1-Imz_r.:ul and Ha nja), KS C 5601-1987 (Seoul: Korean Industrial Standards Association, 1989).
1 1 . Information Processing-/SO 7-bit and 8-bit
Coded Character Sets-Code Extension Tech
niques, 3d eel., ISO 2022 (Geneva: Interna
tional Organization for Standardization/
International Electrotechnieal Commission, 1986).
12. T. Honma, H. Baba, and K. Ta kizawa,
"Japanese Input Method Independent of Applications," Digital Technical Journal, vol. 5, no. 3 (Summer 1993, this issue): 97- 107.
Vol. 5 No. 3 Summer 19'.!3 Digital TecfJIIical journal
Supporting the Chinese, japanese, and Korean Languages in the Open VMS Operating System
13. R. Haentjens, "The Ordering of Universal Character Strings," Digital Technical journal, vol 5, no. 3 (Summer 1993, this issue): 43-52 . 14. R. Scheifler and ). Gettys, X Window System, X 11, Release 5, 3d ed. (Burlington, MA: D igital Press, Order No. EY-.J802E-DP-EEB, 1992).
15. Introduction to Asian Language Software Localization (Maynard, MA: D igital Equip
ment Corporation, Order No. AD-PGOAA-TE, December 1990).
16. Technical Guide to Asian Language Software Localization (Maynard, MA: D igital
Equip-Digital Tech7lical Jnur11a.l 11>1. 5 No. 3 Summer 19?3
ment Corporation, Order No. AD-PGOBA-TE, December 1990).
17 Addendum to Technical G uide to Asian Lan
guage Software Local ization (Maynard, MA : Digital Equipment Corporation, Order No.
AD-PGOCA-TE, December 1990).
18. Information Technology- Universal Multi
ple-Octet Coded Character Set (UCS)-Part 1:
Architecture and Basic Multilingual Plane, ISO/lEC 10646-1 (Geneva: International Orga
nization for Standardization/International Electrotechnical Commission, 1993).
79