Prepare a release with a known destination
PipelineX prepares Fabric notebook and pipeline definitions with deployment scripts, environment settings and automation templates. Review the package files and destination settings before taking the next step.
Release review is tied to the prepared package and its destination. Source or destination changes require a new package and renewed review, so an earlier approval cannot silently carry over to different deployment inputs.
Use your repository review process
Configured connections support GitHub repositories and Actions, or Azure DevOps Services with Azure Repos and Azure Pipelines. An administrator selects and verifies an operator-authorized repository connection; several named connections can serve different teams or repositories.
An engineer saves the generated package to a new review branch and opens a draft pull request. Existing repository files are retained. Reviewers use the provider’s code review and branch policies, and merge through that provider.
Check the commit you intend to release
Configured automation can run package checks when the review branch is created. Engineers can also request checks explicitly and refresh the recorded commit’s status, with a link to the provider’s report and the last refresh time.
The supplied package checks cover package integrity, Fabric item structure, conversion guards and notebook syntax. A successful check is useful preparation for delivery; it does not execute the migrated workload or establish equivalence with DataStage output.
Keep runtime validation in the release flow
Use the customer runner and runtime evidence workflow to execute configured validation and return evidence for the release. Fabric access, capacity, connection bindings, representative inputs and a real DataStage baseline must be prepared in the customer environment.
Simulation and synthetic examples remain distinct from production evidence. Deployment review, runtime comparison and human migration acceptance each answer a different question and stay visible as separate decisions.
Choose the delivery path that fits your environment
Downloadable packages remain available for teams that manage delivery through their own automation. Connected repository delivery requires administrator setup and server-provisioned credentials; it does not use an engineer’s local repository login.
The connected release workflow described here is for Microsoft Fabric. PipelineX also generates migration code for Databricks and Snowflake, whose deployment configuration and operational process must be planned for the selected environment.