Draft Charter 6/21/93 - mdw We are in the process of building global directory systems and other information services on the global Internet. In many parts of the world it is seen as essential for the success of the global services that they should be able to recognize, store, and present textual information like personal and organizational names, represented in the character sets used by those concerned. This means that the Directory must be able to handle national characters not found in the US-ASCII repertoire. The same applies to the other global information services on the network (e.g., the databases used in many information servers). This is especially a problem as information services are provided for clients on various different hardware architectures. Currently, for the Western European languages at least 5 different encodings are in use on the network: ISO-7 National Variants, ISO 8859/1, ROMAN8, T.61, and RC850. (See RFC1345 for further information on these character sets.) If you consider the other scripts used in Europe and the other encodings the number of different character set codes rise to as many as 40. This is the real (and messy) world we live in. Changing the character sests in this world is not an option, as curent systems run applications which can support only the character sets used by that system. However, a universal encoding has begun to appear: UCS (ISO 10646). Initial experience with this solution has been positive. However, there are still many issues to be addressed in the context of ISO 10 646 and the other character set codes, which will exist on the Internet in the future: (1) Can we agree on some common network services/model for character set handling? (2) Should a general-purpose SW tool be designed that will support both UCS and regional character sets? (3) Is there a solution that will make character set convertors for different codes ``plug-and-play'' (i.e., an API) without specifying the actual underlying implementation? Can we use UCS as a common denominator for that? (4) Is it necessary to have a document identifying the language and the character sets which cater to a particular language? (5) If we need to solve these problems and UCS (ISO 10 646) is the only available general option today which is maybe close to be sufficient, can we start with UCS and make minimal changes or specifications which will be sufficient for our needs. Can we discuss the missing agreement/specifications required in the communication protocols such as: (a) The order of octets in the interchange of data is left to be specified by the sender and the recipient in UCS. What are the ``sender'' and ``recipient'' on the Internet? Can we define a mechanism to identify the serialized byte order of a data stream? (b) Additional encoding mechanisms for the UCS have been proposed. Do these schemes have any merit? (c) Some amount of profiling may be necessary for UCS use in some particular countries, do we need to specify that globally or we can leave it to a particular region to be solved as regional matter? (d) Do we need to differentiate or specify how tagged data (i.e., the field type in a database) and how ``serialized byte order'' data are treated in a communication protocol or will some common specification for the tag and the type be suffient? The goal of the BOF is to test the interest for the various issues. If possible a clear set of issues could be identified then working group(s) will be defined.