[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Analysis and Response to Gunther's last minute RT, CCT & code lists
Hello folks, I was thinking earlier about creating a subdirectory under "xsd/cct/" in which Gunther will be fully responsible for presence, conformance with NDR and compatibility with other contents in "xsd/" for whatever he puts in there. The good points in all earnest and favor brought by Gunther has been to introduce the concept that UBL models should build upon RT which builds upon CCT. As of model draft-11, I believe Tim & Stephen have worked out the Representation Terms to take only from those listed in RT as provided by Gunther's schemas. The last minute information brought by Gunther is, however, not so welcome. It has caused great disturbance to LC and FP as a chain reaction. More responsible behavior would have been much appreciated. Irregardless of the late communication, the changeover from CCT to RT is deemed sufficiently important that we lose more nights of sleep and weekends to try to cope with the changes. But on second thought, I think I don't wish to let the schemas in "xsd/" get tainted due to his disregard for things like: - some of the schemas, most importantly RT & CCT and also the Gunther's proprietary code list schemas, are not parseable XML files. They lack proper namespace definitions, and in certain places, proper schema import elements. - introducing his own code list namespaces that is different from the conclusion from code list group, - introducing on his own accord new code lists (MIME?) that nobody from code list group seems to have known, - in introducing MIME, it induces questions about other IETF code lists that remain not included, - questionable code list implementations (MIME code list maxLength having 3 when all the enumerated values mostly exceed 3) - inconsistent implementation of code list against other UBL schema components (RT has GraphicType, PictureType, VideoType and SoundType, yet his implemented MIME code list has absolutely NO support of image/*, audio/*, video/*, and appears to be a haphazard and arbitrary listing of whatever happens to be there). - incomplete implementation of code list against usage in other UBL schema components. (MIME type definitions are structure-rich [RFC-2045 & related standards]. Minimally, used as content-type descriptions, there needs to be a corresponding minor type for each major type. Gunther's list does not permit clear use of the values, does not permit clear implementation of validation, and when used in the only place of RT:BinaryObjectType, does not describe clearly nor is useful for understanding exactly what content type the binary object contains). - improper and non-uniform treatment of code lists (RT depends selectively on CurrencyCode and LanguageCode lists, but then CountryCode list depends in turn on RT. This effectively means, absurd as it sounds, without CurrencyCode nor LanguageCode, we cannot use CountryCode.) - not following NDR's filenaming but uses his own proprietary naming that does not following "UBL-" + componentName + version, and has mixed use of underscores "_" and dashes "-". This, although seemingly a minor fix, has major impact on all higher-level schemas that import them. - not following NDR's global element definition rule. There's clear absence of elements defined in RT for each complexType. - not following NDR's ContentType naming requirement for simpleType restrictions. These are just *some* of the problems observed. It is beginning to take toll on various people when Gunther provides basic components that require massive editing each time he sends an update. The point of listing the above is not to help Gunther edit his schema design problems, but to illustrate that an independent submission of work that doesn't go along with discussions and conclusions reached so far causes more problems to production team of people than help. We will, however, take what good there is in those schemas, namely the need to take names from RTs only. It has therefore come to my reluctant conclusion that at this point, we can only adopt a simplified version of his RT and CCT contribution (because CCT is used quite heavily by RT). This would meet the primary aim of aligning LC's draft-8 with CCT requirements. And that is about all I think we can do at this point. The simplified version will discard code list implementations. Best Regards, Chin Chee-Kai SoftML Tel: +65-6820-2979 Fax: +65-6743-7875 Email: cheekai@SoftML.Net http://SoftML.Net/
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]