Supported functionality

Foundry Branching currently provides coverage for transforms code repositories, Pipeline Builder, Ontology Manager, and Workshop.

The current feature set includes the ability to:

• Create a Foundry branch from any transforms code repository, Pipeline Builder, the Ontology, Workshop, or from the Foundry Branching application. Branches are associated with a single Ontology.
• Access an existing Foundry branch from any transforms code repository, Pipeline Builder or Workshop resource, or from Ontology Manager.
• View modified resources on your branch at any point in your workflow using the branch taskbar or the Foundry Branching application.
• Preview data in Workshop. Workshop will load data from the branch for all resources that have been modified on that branch. For resources in the Workshop module that have not been modified, data will be loaded from Main.
• Run Actions on a branch. You may test Actions in a Workshop module to validate that they have been configured correctly. When all relevant object types are indexed on a branch, you can run the action type and see the edits on the branch.
  • Running Actions on a branch is intended as a testing mechanism, therefore no edits will be merged back into Main.
  • To view action edits on a branch, all object types edited by an action type must be indexed on that branch. You can do this through the individual object type page, or through the action type page in Ontology Manager.
  • When a TypeScript or Python function executes, it is branch aware. This means that when the function interacts with Object Storage, it will receive data and schema information specific to the branch in use. This ensures that any API calls to branch-aware services, such as Workshop, are handled correctly and reflect the current state of the branch.
    • In rare cases, if the function code interacts directly with the ontology, the code will retain the schema defined at the time of code compilation. This may result in the function not being aware of schema changes made on a branch, even when called from a branch-aware service.
    • Example scenario: Consider a function that edits a property of an object type. If you delete this property on a branch, executing the function will result in an error because the function tries to access a non-existent property. However, if your function lists properties without actually fetching objects, it will show the properties as they were at the time of the function's compilation, regardless of any branch-specific schema changes.
  • Some known limitations:
    • Webhooks and email notifications are not executed when an Action is run on a branch.
    • Functions that make calls to external systems are not supported on a branch.

• Track what is ready to be previewed on a branch.
• Manage the review process centrally. The review process will respect local Approvals settings for each application. Specifically, Code Repositories and Pipeline Builder will follow local policies set on individual repository or pipelines, and the Ontology will require an approval from at least one editor for each resource.
• Deploy changes centrally for Code Repositories, Pipeline Builder, the Ontology, and Workshop changes.
• Set retention policies for Foundry Branching. Branches that see no activity after a certain number of days are automatically marked as Inactive and later automatically closed. Resources of closed branches are deleted or de-indexed if not deployed to Main. Currently, you cannot reopen closed branches. Refer to the retention policy for additional details.
• Restricted views are in the experimental phase of development and may not be available on your enrollment. Restricted views must be added to a branch to build, modify the policy of, or index objects backed by the restricted view. To add a restricted view to a branch, select the branch in the dropdown menu from the restricted view. Then, use the Add resource option to add the restricted view to the branch.
  • Before an object type can be successfully indexed on a branch, its backing restricted view must be added to the branch. If an object type is indexed before its backing restricted view is added, a reindexing of the object type will be necessary.
  • To build a restricted view on a branch, the dataset from which the restricted view is created must also be built on that branch. This can be accomplished by building the dataset using either Pipeline Builder or Code Repositories on the appropriate branch.
  • Datasets that are not built using Pipeline Builder or Code Repositories cannot currently be added to a branch. Consequently, restricted views relying on such datasets cannot be built on a branch. Examples of these datasets include those backing edit-only workflows, datasets created via importing, or datasets built from a Fusion sheet.
  • There is no rebase workflow for branched restricted views. When a change to a restricted view policy is merged, it will overwrite the current policy on the main branch.

Unsupported functionality

• Object Views cannot currently be modified on branches other than the main branch. In Ontology Manager, the Object View will reflect data from the main branch, so changes to an Object View can only be made on the main branch.
• Typescript functions and AIP Logic functions: Currently, it is not possible to modify functions on a branch. Alternatively, you may reference a specific version of a function on a branch and test that version before merging it back to the Main branch.
• Outside of Workshop, any application that leverages the Ontology is not currently able to be branched. For instance, if your Workshop module contains non-Workshop elements such as Quiver dashboards, these will not be modifiable on a branch.
• Object Storage V1 backed object types.
• Materializations: Currently, Action edits submitted on a branch will not be reflected in any of the materialization datasets.
• Object types created in Pipeline Builder are not modifiable in Ontology Manager on a branch. Development for this feature is ongoing.
• Currently, you cannot reopen closed branches. This feature is coming later in 2025.