Pipeline to support un-interupted deployment to production - Deployment Settings to be based on previous steps dataset

When you define the Deployment Settings for a step (e.g. the Test step), the Deployment Settings are using the dataset in the "current" step. This way you can't configure the Deployment steps before you actually have deployed something to that step. This is true both for new deployments as well as when you have added a new parameter to the dataset.

To build a better solution, it shall be possible to define the Deployment settings already before you deploy, and then this needs to be done on the previous steps dataset.

For already existing deployments, I do understand that you might both want to change the Deployment settings for the already deployed dataset so that this takes effect at the next refresh. But this makes it complicated when you want to deploy new features in the dataset requiring a Deployment Setting to be configured, because if e can't configure the deployment settings on the previous step, the reports will show wrong data from the time we deployed, until we have configured the Deployment Settings and performed a refresh. And this is not CI/CD. In test this is not a big issue, but in production it is.

My suggestion is that it shall be possible to choose what dataset to be used for defining the Deployment Settings, either the already deployed dataset, or the dataset in the previous phase.

Another possibility is to always use the previous phase dataset for defining the Deployment Settings. This way, there will be a change in the current behavior and the Deployment Settings will not take effect until you do your next deployment. But this is kind of what the name defines (Deployment Settings).

There's probably other ways to implement this as well, but the main thing here is to make sure we can deploy new versions to production without the need for a service window!!