Dependencies

Dependencies control execution order between activities. The library provides small helpers that read closer to workflow code than raw dependsOn arrays.

Chain activities

lookup.then(copy).then(notebook)

By default, .then() creates a dependency on Succeeded.

Use shorthand chaining

lookup >> copy >> notebook

>> is equivalent to .then() and works well for linear flows.

Branch on non-success conditions

copy.then(notify, on="Failed")
copy.then(cleanup, on="Completed")

Supported conditions mirror Fabric dependency conditions such as Succeeded, Failed, Completed, and Skipped.

Fan in multiple predecessors

join.after(copy_a, copy_b)

This is useful when a later activity should wait for several parallel branches.

Nested scopes matter

Control-flow activities such as IfCondition, ForEach, Switch, and Until contain nested activity scopes. Validation prevents cross-scope dependencies that Fabric would not accept.

Validation catches graph mistakes

Before serialization, the pipeline graph is validated for:

• duplicate activity names
• unknown dependency targets
• cross-scope dependencies
• dependency cycles

That lets you catch orchestration mistakes in code review or CI rather than in Fabric after deployment.

examples/etl_with_lock_pattern.py shows cleanup wiring with .then(..., on="Completed"). examples/parameterized_elt_controller.py shows nested ForEach / Switch scopes.