RE: [ubl-ndrsc] Defining the logical model

["CRAWFORD, Mark" <MCRAWFORD@lmi.org>]

Tim wrote - 
> i am not sure what some of the terms you use are, but  from the very 
> first version (april 2002) we have had a clear policy on this 
> in terms 
> of the models.  I dont see why the schemas should not match this - in 
> fact i suggest it is a good idea they dont change it (and 
> that we keep 
> consistent terminology).
> 
> 	1) For all ABIEs that have the potential to be used in 
> schemas that cross functional boundaries (i.e. 
> transportation, procurement) - place them in the CAC module.
> 
> (what is CAC? this is called the Reusable schema. )

Several points here.  First LC is very inconsistent in what they do and don't place in the CAC.  Second, it is NDR's responsibility to define the overall schema structure, modularity, and naming as part of its NDR effort.  Third, The modularity has been consistent since at least Montreal. 
> 
> 	2) For all ABIEs that are part of a unique subset (only 
> of value to a subset of business processes within a 
> functional area) - place in the internal schema 
> 
> (i have no idea what this is)

I would suggest that you look at NDR.  There is a strong business case for having the capabilities of an internal schema and we have defined this for at least two years.
> 
> 	3) For all ABIEs that are unique to a particular 
> message - place in the control schema. 
> 
> (we starting out this way but from the earlest library it 
> became clear we could not easily segregate unique ABIEs.  
> What has evolved is that only the document level ABIE are in 
> this schema.  furthermore, if we want to encourage re-use 
> then it makes sense to make every ABIE potentially re-usable. 
> NB even apparently obvious 'unique ABIEs' like OrderLine  was 
> found to be re-used!)

Tim, the problem is that you have no consistency in your approach.  Michael's tool is model driven, not spreadsheet driven and the current inconsistencies do not support schema generation.
> 
> I can understand Michael asking the question, but is this an 
> NDR issue 
> at all?  All ABIEs in the Reusable model go in the Reusable 
> schema, all 
> ABIEs in the individual document models go in their 
> respective document 
> schema.  We have been doing this for 2 years and it works fine.

It does not work fine.  The inconsistencies show that.  The inability of a model driven tool (which is our only hope at the moment to approach realizing the current deliverable date) to generate schemas without some structure clearly shows that.  


> 
> As Mark says, it is the LC model that determines where these things 
> should go, so why duplicate/confuse everyone by having potentially 
> different rules in the NDRs?  
> Am i missing something here?

It is the LC model that determines where within the XSD modularity from NDR the stuff goes.

M