Why I left Warehouse out of our Fabric deployment scope

The author has chosen to exclude Warehouse from the Microsoft Fabric deployment pipeline for a specific reason. The DEFAULT_ITEM_TYPES list, which includes DataPipeline, Lakehouse, Notebook, and SemanticModel, intentionally omits Warehouse. This deliberate exclusion serves to prevent schema reset risk during the publish process. Publishing Warehouse through the current pipeline path could lead to unintended structure removal, rendering the deployment additive when it should be additive.

The author explains that there are two options once this issue is identified: include Warehouse in the deployment scope despite the risks or exclude it and handle Warehouse deployment separately with its own tooling. The author opts for the latter choice, taking the stance that an automation covering most cases but potentially corrupting others is preferable to one that covers most cases and silently corrupts the rest.

The exclusion is made visible by placing the comment inside the DEFAULT_ITEM_TYPES list, rather than in a separate documentation document that may not be read. This exclusion is also highlighted in the README under known limitations and tested to ensure that Warehouse is absent from the default scope. The test asserts that the string "Warehouse" is not present in the deploy.DEFAULT_ITEM_TYPES list.