Subject: AW: [ubl-ssc] derivation methodology Spreadsheets are not the instrument to maintain data properly.
Hi Anne, imho the spreadsheet models have to be identical in respect of data content with the ef2spreadsheet output. Ad the end this means that - the EF stored data are the master data, - the cercle EF-->Spreadsheet; Spreadsheet--->EF; EF--->Spreadsheet allows automated comparision between the Spreadsheet and checks whether EF works correctly - the maintenance has to be done there - and that everybody doing UBL maintenance has to get a free EF licence to do so. Best Regards, Michael btw: The cercle I' talked about is what I do for EDIFACT for many years now; as a result we do not have technical errors in the directories. -----Ursprungliche Nachricht----- Von: Anne Hendry [mailto:email@example.com] Gesendet: Mittwoch, 16. Juni 2004 07:57 An: Michael Dill Cc: firstname.lastname@example.org; email@example.com Betreff: Re: [ubl-ssc] derivation methodology Hi Michael, Yes, thanks for your thoughts. What you propose sounds like a good plan. The primary goal, for both the final 1.0 CD and any future development is to make sure the schemas are an accurate reflection of the spreadsheet (business) models. While the single source model you curently employ avoids some problems, there is still the necessity to check whether the output matches the spreadsheet models (and to correct it if not). There would be several ways to check this, including the one you have suggested. Regarding your last statement about the purpose of doing a derivation, it's not so much the derivation that is the point, but the question came up as to the best way to accomplish 1.1 schema generation. Yes, we can continue to discuss this at the next meeting. Your input on this would be very helpful. It's good to hear that you and David will both be in Copenhagen! Thanks, Anne Michael Dill wrote: >Hi Anne, >regarding the last minutes: >We will need Michael's thoughts on how best to make 1.1 happen: > through derivation? Perhaps a derivation methodology should be > developed? It should start with the models. 1.1 will be different > than changes for a future major release. Since this is mostly > done in the model there should be no impact that would be an > issue with the rules. Stephen will check to see. > >Both David and I will be in Kopenhagen in order to support the further >development of UBL. > >I agree that it makes sense to derive schemas from models directly. >Currently we derive both spreadsheets and schemas from UML/CCTS data models. >This single source helps alot to avoid problems. > >Also I see that a cercle (EFmodel->spreadsheet->EFmodel->spreadsheet) could >be implemented in order to check the correctness of the EF processes as >well. Thus we can compare both spreadsheets automatically, which gives a >better and faster QA. > >Please let us discuss what the purpose of a derivation should be: 1) >generate 1:1 schemas from UBL models or/and 2) generate profile = customized >schemas. > >Best regards, >Michael > > > > >