What you do is indeed the correct way when you have metadata-variations for a single cross-domain variable. Within a domain/dataset you would use ValueListRef-ValueListDef constructs. I cannot see how your ItemDefs look like, do they all have the same “Name” attribute value for these variables?
If you could send me your define.xml (you can easily find my mail address), then I can have a look whether everything is fine and this is a “false positive”.
But that was not what I meant. What I meant was that the ItemDefs for the different variations of the same variable all have the same value for the “Name” attribute. For example:
I could reproduce your issue. I must categorize it as a false positive.
I first believed it had to do with the two different values for the “Length” attribute (but even than that would not be a non-compliance), but even if I set both “Length” attributes to the same value, the warning persists.
Please contact me personally for more details and how to avoid this false positive warning.
I was able to reproduce the issue by just changing two lines (adding an additional ItemDef and referencing it from an ItemRef) in the sample file that comes with the Define-XML 2.0 standard distribution. Anyone wanting to get a copy, just e-mail me.
I have a question regarding DD0039 (Variable is in wrong order within Dataset “XX”).
Our define file has a unique itemdef that is defined for each distinct combination of attributes for a variable.
For example, if Length or Comment variess across domains for a variable, in the XML, it is suffixed with a value (e.g. .V1, .V2) to uniquely identify the different combinations.
Is this a false positive error ? Or, Is there any trigger for DD0039?
Note: The variables are presented in the correct order in the SDTM datasets and SDTM define.xml.