A referential integrity violation should be classified into two different subtypes:
- blank RI violation
- non-blank RI violation
These subtypes can be detected in the feature that retrieves the samples. There should be a separate specific query for the blank condition on the many-side of the relationship, and the existing SAMPLE should exclude the blank value from the extraction.
These two properties should be combined into one single visualization of the samples (for compatibility with the existing clients) that always shows the blank value as the first invalid value (if detected), followed by the others found in the sample extraction.
A referential integrity violation should be classified into two different subtypes:
These subtypes can be detected in the feature that retrieves the samples. There should be a separate specific query for the blank condition on the many-side of the relationship, and the existing SAMPLE should exclude the blank value from the extraction.
These two properties should be combined into one single visualization of the samples (for compatibility with the existing clients) that always shows the blank value as the first invalid value (if detected), followed by the others found in the sample extraction.