Tables not defined in 'tableOrdering' option should be used in resulting dataset

Hi everyone. I noticed strange behavior with using ‘tableOrdering’ option. For example I have 3 tables in my dataset.yml file: first_table, second_table and third_table. And in annotation ‘@DataSet’ I set ‘tableOrdering’, but add there 2 tables instead of 3: first_table and second_table. As a result, delete and insert operations isn’t executed for this table and test failed. I didn’t found any warning or error logs.

Database rider version: 1.16.0

dataset.yml

first_table:
  - name: first_record
  - name: second_record

second_table:
  - name: first_record
  - name: second_record

third_table:
  - name: first_record
  - name: second_record

@Test
@DataSet(value = "dataset.yml", tableOrdering = {"first_table", "second_table"})
@ExpectedDataSet(value = "dataset.yml")
void retrieveAll() throws Exception {

After debugging I found, that in ‘DataSetExecutorImpl.performTableOrdering(DataSetConfig dataSet, IDataSet target)’

private IDataSet performTableOrdering(DataSetConfig dataSet, IDataSet target) throws AmbiguousTableNameException {
        if (dataSet.getTableOrdering().length > 0) {
            target = new FilteredDataSet(new SequenceTableFilter(dataSet.getTableOrdering(), dbUnitConfig.isCaseSensitiveTableNames()), target);
        }
        return target;
    }

We set FilteredDataSet and his field _filter contains tables only from ‘tableOrdering’ property. As a result all Iterators of this dataset iterate only by those tables.

[RomashkoVictor]

yes, I agree with you. Current situation is bad and error with incorrect ordering is much better=)

[rmpestano]

Sure, in this case add all tables present in dataset in tableOrdering to avoid surprises. Also you’ll probably get a constraint violation error if the order is incorrect and then you can fix it. In current situation the table which is not present in tableOrdering is simple removed from the dataset so you don’t even get an error because the table is not populated.