[DDI-SRG] [MRT] new model version / generated files / documentation work

[Wackerow, Joachim]

I created a new model version (2020-02-01) and the derived generated files.

The documentation work - which is based on the new set of rst files - could be started.

Please have a look at the model and the XSD and review.

All files are available in the subfolder "2020-02-01<https://drive.google.com/drive/folders/11kv0uLKq808i4ETOCq54MQKUGJNk6Rk3>" in the Google folder<https://drive.google.com/drive/folders/1XLElHvZTMRuTmwyev7kUHoWQVWwCN4XE>.

Model: DDI Core - Master 2020-02-01.eap
Related change log: ChangeLog_2020-02-01.txt
List of associations: DDI Core - Master 2020-02-01 CX association.html
Canonical XMI: DDI Core - Master 2020-02-01 CX.xmi (with given association names)
Canonical XMI: DDI Core - Master 2020-02-01 CX UniqueAssociationNames.xmi (with generated unique association names, used for XSD generation)
EA doc: DDI Core - Master 2020-02-01 CX EADoc.zip
XSD: DDI4Core_2020-02-01.xsd (errors happened, see next file, I set the error parts on comment in order to get a valid XSD)
Errors in XSD: DDI4Core_2020-02-01 error.txt
Oxygen doc of XSD: DDI4Core_2020-02-01_XSD_Documentation.zip
rst templates: DDI Core - Master 2020-02-01 CX rst.zip

I changed in the model following:

*        EA direction of aggregation/composition

*        Change of names which didn't comply to rules

*        Typos in names

*        New primitive XSD date, used instead of IsoDateType

*        Removal of packages RegularExpressions

*        Other model cleaning items
All the details are listed in the change log.

The rst files have errors in the generated listing of class attributes. Used UML primitives are not mentioned. Therefore the multiplicity moved into the previous column.
The error doesn't affect the parts which might be edited. The generating program will be fixed in future.

There are two versions of the Canonical XMI, one with the association names like in the model and one with generated unique association names for further processing.

XSD generation uses the Canonical XMI with the unique names. OWL generation will use the version with the original names. Then the associations can have reusable counterparts for common names like "has".
The XSD has three errors on ambiguous content of the complex types: ScopedMeasureType, CodeListType, StatisticalClassificationType. See related file DDI4Core_2020-02-01 error.txt.
Arofan and Oliver: please have a look at this and come back.

Some multiplicities might be not correct. Please check especially the cases of aggregations ("shared") where 0..1 is set. See file DDI Core - Master 2020-02-01 CX association.html.
Check also LongKey has LongMainKeyMember.

I changed some multiplicities where nothing was defined. Please check if this makes sense.
InstanceKey has Descriptor, Descriptor now set to 0..* (was 1)
LongKey has Descriptor, 0..*, 0..* (was undefined)
StatisticalClassification has ClassificationItemPosition, 0..1, 0..* (was undefined)
TimeComponent has TimeInstant, 0..*, 0..* (was undefined)
TimeComponent has TimePeriod, 0..*, 0..* (was undefined)

The work took longer than expected because I noted some errors regarding associations in the transform program to Canonical XMI and in the generation program of the association list.
This is fixed now but it took some time.

I'll send some suggestions for model cleanup on Monday. I'll then continue to work on the model cleanup until the documentation in the rst files is finished and Hilde takes over the then current EA file for updating the documentation (don't use the 2020-02-01 version for this!).

Achim


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.icpsr.umich.edu/pipermail/ddi-srg/attachments/20200201/66d3a0f5/attachment.html