From keld@dkuug.dk Sun Dec 16 22:51:48 1990 Received: by dkuug.dk (5.64+/8+bit/IDA-1.2.8) id AA25533; Sun, 16 Dec 90 22:51:48 +0100 Date: Sun, 16 Dec 90 22:51:48 +0100 From: Keld J|rn Simonsen Message-Id: <9012162151.AA25533@dkuug.dk> To: i18n@dkuug.dk Subject: Danish enhanced model on i18n X-Charset: ASCII X-Char-Esc: 29 A model for internationalization within the IT area This paper discuss a possible model of approaching internationalization in the Information Technology area of ISO/IEC JTC1. It is aimed as a discussion paper for the New Work Item on internationalization of this committee. The internationalization issues are considering different cultural aspects within the IT usage, including character sets and its ordering, date and time, currency, number behaviour etc. The model consists of three dimensions, namely: 1. The user dimension. Different users need different ways of looking on data, a librarian needs another collating sequence of characters than do a systems programmer or a clerk looking up customers in his firm's database. There is a need for investigating these requirements by a broad survey. 2. The function type dimension. The processing dimension determines how data is used. Usage of data has been classified in the following categories: 1. Dialogue: data generation and presentation 2. Processing: data handling in a program 3. Storage: data storage and retrieval 4. Communication: data exchange Examples of differences are communication being stream oriented while storage is array oriented. Generation and presentation may be limited by hardware while storage and retrieval is not. 3. The function layer dimension. The conventions can be different due to where in the application hierachy the data is dealt with. Interface layers identified are: 1. interface to the hardware 2. interface to the operating system 3. Interface to system software and utilities 4. interface to applications An example is character collating sequence requirering different ordering on the hardware level being the hardware character set, the OS level requires a national collating sequence different from the binary values, but with all characters distinguishable, and the database library interface needs to treat several characters or character sequences as identical. Sven Thygesen and Keld Simonsen, DS, 1990-04-04