failed validation due to change in dataset composition

[travissolt]

We have a chain of DS-jobs that build datasets which are used for different jobs. When changing a dataset downstream (say we add a new field) and then we try to re-validate the job the job fails validation because the field does not exist in the dataset. Is there a reason that during validation the job will not create a dummy dataset based on the fields provided rather then reference back to the old dataset used in the last run? Even after deleting the datasets the job still failed validation saying that the datasets did not exist. Is their a way around this issue? If I compile and run the job no issues. It's only if I try to validate prior to running that it fails validation.

Thanks

[ray.wurlod]

Data Sets' structure is locked in once they're created. You must delete the Data Set if you wish to change its structure.