From jbettels@wadd.enet.dec.com Fri Sep 4 16:45:23 1992 Received: from inet-gw-2.pa.dec.com by dkuug.dk via EUnet with SMTP (5.64+/8+bit/IDA-1.2.8) id AA12027; Fri, 4 Sep 92 16:45:23 +0200 Received: by inet-gw-2.pa.dec.com; id AA19977; Fri, 4 Sep 92 07:45:20 -0700 Received: by vbormc.vbo.dec.com; id AA08180; Fri, 4 Sep 92 16:40:55 +0200 Message-Id: <9209041440.AA08180@vbormc.vbo.dec.com> Received: from wadd.enet; by vbormc.enet; Fri, 4 Sep 92 16:40:56 MET DST Date: Fri, 4 Sep 92 16:40:56 MET DST From: "Jurgen Bettels, TNSG/ISE 04-Sep-1992 1641" To: iso10646@jhuvm.bitnet, i18n@dkuug.dk Cc: jbettels@wadd.enet.dec.com Apparently-To: i18n@dkuug.dk, iso10646@jhuvm.bitnet Subject: request for feedback on character set identification proposal X-Charset: ASCII X-Char-Esc: 29 JTC1/SC2/WG3/N 156R Identification of Character Collections and Encodings ===================================================== 92-09-03 J|rgen Bettels (Digital Equipm. Co.) 1. Requirements There is a growing need to be able to indentify character sets using some specification mechanism such as ASN.1. The demand for this has been raised by SC18, SC21 and SC22 and also from other groups such as X/Open. A liason statement from SC22 to SC2 has expressed a need for character set names. This paper is proposing a possible approach to these requirements by revising ISO 7350 in order to allow an identification registry which can accommodate the requirements. 1.1 Type of Repertoires for which OID's Are Required It is recognized that ISO 10646 represents a registry of all individual characters (It is expected that any character not included in the first edition of ISO 10646 will be so in a later revision). Even though ISO 10646 does enumerate a number of character collections as sub-sets, it is clearly not a registry of all collections currently in use. This proposal would allow to generate identification for any collection of characters and their encodings. It must be possible to generate (and register) identifications for character repertoires and encodings from: 1. ISO standards (including 10646 and its subsets) 2. national standards 3. proprietary standards 1.2 Format It would be most useful to have the collection of all such identifiers in one document so that they are easily accessible by users. An alternative would be to have OID's added to each character set standard as it is revised. This would only address 1. above and even then it is not very likely that all standards will undergo a revision at which the OID could be added. Most flexible would therefore be a registry of identifiers under a revised ISO 7350 to which new entries can be added easily. 1.3 Content of the registry As a minimum, for each character set registered the following must be specified and part of the registry entry: - a reference and description of the standard - ASN.1 OID for the abstract syntax (i.e. the repertoire) - an optional ASN.1 OID for the transfer syntax (i.e. the encoding) - object descriptors of the OIDs for human reference and for programming languages Note that it is proposed here to separate the repertoire from the encoding. This allows more flexibility as multiple encodings exist for the same collection. There is particular interest from programming languages, databases and Operating systems (Posix) for such OID's. E.g. SC22 (Posix) as well as X/Open intend to register names for locales. These names will contain different standard fields specifying language, culture, character set, etc. It is expected that the object descriptors for abstract and transfer syntax proposed here be used in these fields. As other applications might have additional requirements it is expected that wide circulation and review of this paper will uncover these requirements. 2.0 Creation of the registry It is proposed that JTC1/SC2, after revision of ISO 7350, initially populates the register with a set of most frequently used character set repertoires and encodings Examples for the most urgently needed entries: 1. all parts of ISO 8859 2. all collections of ISO 10646 3. JIS X0208 and JIS X0212 4. national standard of Korea 5. " of PRC 6. " of Thailand 7. regional " of Taiwan 8. proprietary standards: - selected PC code pages - Microsoft Kanji code ("shift JIS") - EUC Japanese, Taiwanese... 9. other standards Note: the registry must be easily updatable and publicly available. 3.0 Examples for OIDs and descriptors: a. ISO standard 8859-1: OID descriptor { iso standard 8859 part(1) abstract-syntax(1) } "ISO 8859 part-1 rep" { iso standard 8859 part(1) transfer-syntax(0)} "ISO 8859 part-1 ENCoding" b. Repertoire of 8859-1 and UCS-2 encoding: the abstract syntax is the same as in example a. The transfer syntax would change to: { iso standard 10646 part(1) transfer-syntaxes(0) two-octet-form(2)} and "ISO 10646 part-1 form 2" b. JIS X0208: One possibility would be to use exactly the same OIDs as proposed for 2022: { iso standard 2022 abstract syntax(1) reg(...)} "ISO 2022 registrations ..." or as synonym something like: "JIS X-0208 edition..." A better alternative would be to use arcs which point to JIS. Such arcs have been defined in ISO 8824 (Annexe B) and use the country codes of ISO 3166. For this to work it is, however, necessary for the ISO member body (in this case JISC) to assign arcs to their standards: { iso member-body Japan(392) JIS X-0208-1990(n) } c. PC code pages Here we have the same problem as in example 2. Let us assume that the repertoire of the total set is that of 10646. Then a modified 7350 has to address the following issues: - extend the repertoire to 10646 (this would then also be the registry for 10646 subsets) - register also encoding methods (in case of code tables this means that the code table has to be in the registry together with the 10646 names or code points - definition of the format of the registry entry - synonyms must be unique. I.e. for a PC code page: { iso standard 7350 abstract-syntax(1) reg(...) } "ISO 7350 registrations ..." or as synonym: " some name for the code table" 4.0 Register entry: For each collection there shall be an entry for the abstract syntax (the character collection) and an (optional) entry for the transfer syntax (the encoding) as follows: registr. # | description | specification | OID | descriptor | | | | or synonym __________________________________________________________________________________________ | | | | reg. n | ............. | one or several of: | abstract syntax | string | | - reference to SC2 | | | | standard | | | | - 10646 collection # | | | | - list of 10646 code po- | | | | sitions or character | | | | names | | | | | | reg. m | ............ | - reference to specifica-| transfer syntax | string | | tion of register entry | | | | of the relevant collec-| | | | tion | | | | - code table | | It is possible that particular applications may require additional columns of information or provisions for private denominations. 5.0 Conclusions As this proposal is considered to cover urgent requirements JTC1/SC2/WG3 is planning to start work on this as soon as possible. Comments and possible additional requirements are invited and should be sent no later than 31 January 1993 to: J|rgen Bettels Digital Equipment Co. 12, av. des Morgines, CP 176 CH-1213 Petit-Lancy 1 Switzerland Fax: +41 22 709 41 40 or to the convenor of SC2/WG3: Jan van den Beld ECMA 114, rue du Rhone CH-1204 Geneva Switzerland Fax: +41 22 786 52 31 Email recipients of this paper are invited to reply by email instead.