Target Audience: Software Engineer

Level of Session: Intermediate

Unicode has already become the encoded character set of choice in the Internet and World Wide Web environment. Internet and Web applications can now handle characters of many people's native languages without changing the character encoding. Character encoding was one of the most important and difficult issues to internationalize the Internet and Web applications, and Unicode solved it.

Once character encoding issue has been solved, the next issue for internationalizing Internet and Web applications is to correctly handle date/time/number/currency formats, collation, etc. that depend on the language and the location of the user. Also, when translated user interface and messages are available, we should utilize them based on the language of the user.

In order to achieve this goal, we need to detect user's locale on the client machine, send this locale information to the server machines that process user's data. The server machines need to interpret user's locale which is sent from the client machine, and need to act as if user's data were processed on the client machine. The server machines then return results to the client machine with the translated user interface and messages, date/time/number/currency formats, collation, etc. of user's locale.

If all the above were done on a single operating system, we may not have any difficulty. However, the real world is heterogeneous. Different software systems represent an equivalent locale differently. Therefore, we have to convert locale among different software systems.

In this paper, we show how we can perform locale conversion among Java, Oracle, Solaris and Win32 in order to get what users want, regardless of where the server machines reside or what operating systems server machines use.