Foundry Branching is a Beta feature, and is in active development. Some functionality may change before this feature becomes generally available.
The scope of Foundry Branching is currently limited to Pipeline Builder, the Ontology, Workshop, and running Actions on a branch. Anything outside of this scope is not supported. For instance, if your Workshop module contains non-Workshop elements such as Quiver dashboards, these will not be modifiable on a branch.
Contact your Palantir representative to enable the Foundry Branching application on your enrollment. Before using Foundry Branching, familiarize yourself with the full scope of functionalities, especially what is not currently supported.
Foundry Branching enables you to develop and test comprehensive end to end workflows in the Palantir platform that might otherwise be too disruptive or risky for a live production environment. While data pipelines and the Ontology support separate branching workflows, Foundry Branching provides a unified experience to make modifications across multiple applications on a single branch, test those changes end-to-end without disrupting the production environment, and merge those changes with a single click. For instance, take a workflow that consists of a simple pipeline in Pipeline Builder that outputs a dataset used to back an object type. With Foundry Branching, you can make changes to the logic and schema of your output dataset on a branch, see these changes in Ontology Manager on that same branch, and modify the object type definition as a result.
Foundry Branching also supports a review process; Developers can add reviewers for different resources depending on each resource's approval policy. Once changes are finalized and approved, they can be deployed into the Main branch. This feature ensures a safe and controlled approach to updating and improving data pipelines, the Ontology, and Workshop applications.
Some considerations before getting started:
Scope: You can make changes to Pipeline Builder pipelines, the Ontology, and Workshop modules on a single branch. You can also run Actions on branches without writing back edits to Main.
Review the full list of known limitations. A selection of unsupported features include:
Branch cost and retention: Each branch has associated compute and storage costs, and modifying large object types on a branch can incur significant additional costs. Branch cost insights in Resource Management are currently in progress. Branches are intended to be relatively short-lived, and retention policies are available to automatically close stagnant branches.
Foundry Branching enables development and testing of workflow changes on a separate branch. This is ideal for managing changes in a development environment, as it allows developers to work in isolation and only merge changes into the Main branch when the feature is complete. Release management is the process of managing multiple versions of resources across distinct environments that serve different purposes.
Release management and Foundry Branching can work together harmoniously. They should not be seen as alternative solutions to the same problem, but rather complementary solutions to different problems. For example, you can use release management and Foundry Branching together when developing a large feature that needs to be added to a workflow. Larger features could take a few weeks to develop and require foundational changes to dataset and object type schemas. You can develop this feature on a Foundry branch and merge the changes into the development environment when it is completed. Then, you can use release management to deploy these changes to your test and production environments.