Re: [ubl-dev] Details on the Methodology

[Tim McGrath <tmcgrath@portcomm.com.au>]

I am happy to answer these for you Tom,

Tom Gaven wrote:
>   I realize that the purpose of Chapter 3 of the UBL committee draft [1] was
> not to define the detail use of the methodology used, but I would be
> extremely interested in learning more about the methodology,
> specifically....
>
>   1) Is the code to convert the spreadsheets to XSD publicly available?
Yes, the link is in Appendix D of the specification. The code generator was
provided by SoftML and their URL is http://SoftML.Net/jedi/ubl/sw/UBLish
(it also requires a runtime - which you can find there too). There was plans
for a Perl code generator as well.  We used this for UBL 0p70 but it now
needs updating.
>   2) Is there documentation on how to set up the spreadsheets?
This may be deliverable from the UBL Customization process, but we cannot
say yet.  Other groups have copied the current spreadsheets and replaced
the entries with their own BIEs.  For example, the University of California
Berkeley in  its Center for Document Engineering has done this and also produced
a guide (i attach a copy).
>   3) What are some of the limitations?  
>         (for example, is 'xsd:choice'/'xsd:all' supported?)
Not my area, but if you check the Naming and Design Rules document (file:///C:/xml/ubl/1.0/beta/distribution/UBLv1-beta/rn-ndrsc-v1-0-beta.htm)
it says...
"The xsd:all 
 element MUST NOT be used."(GXS7 ) and 
"The xsd:choice 
  element MUST NOT be used."(GXS8)
>   4) Is the transform from XSD to UML publicly available?  
>         Is it automated? How much manual effort was applied to the UML
> models?
Again this is covered in Appendix D of the specification (D2.1).  We used Ontogenics Corporation's hyperModel tool.  My
understanding was that once the tool was configured for the UBL profile,
it required no manual effort - but you can contact them directly for more
details.
I should add that there are two UML models for UBL 1.0.  The first is a conceptual
model that was hand built and manually transformed into the UBL spreadsheet
models.  It is a design form.  The second is an implementation UML model
generated from the final schemas.   The latter is the one I assume you meant.
>   5) Is round tripping supported?
This follows from the point I just made.  I am not sure if round tripping
from XSD to UML and back is yet built into the Ontogenics 
Corporation's hyperModel tool, but
I am sure it could be.
One of the goals for the UBL tools and techniques subcommittee is to have
round tripping between UML-Spreadsheets-XSD, that is conceptual model to
logical model to implementation model.
Perhaps someone from the tools and techniques subcommittee may have more
details of this.
> Thanks, Tom
>
> [1] http://www.oasis-open.org/committees/ubl/lcsc/UBLv1-beta/

-- 
regards
tim mcgrath
phone: +618 93352228  
postal: po box 1289   fremantle    western australia 6160

modeling_guidelines-1-cde.pdf