On Aug 11, 4:01pm, Peter Keller wrote: > Subject: Re: _item.mandatory_code can be undefined! (plus other developmen > > In the light of your comments, I guess that given that _item.name has to > be there, _item.mandatory_code is necessary to avoid the possibility of > partial row updates. Since I wasn't around for the discussion I don't know > the arguments for re-specifying the ITEM category at all, and I find it > hard to see why it was thought necessary. If there is simply a requirement > for the data item's name to be repeated somewhere within the save frame, > it might be better to do this with _item_description.name, rather than > _item.name. This would then avoid the need to re-specify > _item.mandatory_code. Both the ITEM_DESCRIPTION category, and the > _item_description.description item, are mandatory anyway, so this would be > a good way to do it. (Making this change might seem to involve some hard > work at first sight, but I could kludge my own code to do it fairly > quickly, if necessary.) > This is certainly a good alternative but the requirement was that there be a single standard data item like _item.name for each definition. > Personally I'd be very happy with that. I take it that you are suggesting > that _item.mandatory_code should be "inherit" for _item_type.code? From my > own work, I am starting to realise that these are the two items which are > really crucial in the handling of real data - if there are problems with > them, there is no way out. We will experiment with this a bit and then incorporate it in the next version of the DDL if there are no objections raised. > > Am I right in thinking that the _category.mandatory_code item for the > ITEM_TYPE category should then also be changed to yes, since all of its > member items would have a mandatory code of implicit or inherit? I guess > that the likely impact on the dictionaries should be the main > consideration here. > I do not think that we have to go this far. It may be that some folks would develop local dictionaries without this requirement. Regards... John -- **************************************************************************** * John Westbrook Ph: (908) 445-5156 * * Department of Chemistry Fax: (908) 445-5958 * * Rutgers University * * PO Box 939 e-mail: jwest@rutchem.rutgers.edu * * Piscataway, NJ 08855-0939 * ****************************************************************************