The operational problem
A process may need prepared data before generating its final file. Running each stage separately leaves the team to remember the order, confirm the previous task has finished and reconstruct the context when something fails.
Copying designs for every trial also multiplies changes that are hard to track. Reuse requires a distinction between transformation rules, execution conditions and the sequence coordinating the whole process.
How Dova helps
Dova separates Mapping, Session and Workflow. The Mapping defines the transformation; the Session binds that design to paths, parameters and runtime; the Workflow brings tasks together and defines their dependencies. The engine evaluates which tasks can start from their predecessors’ state.
The canvas lets you review the sequence and publish an executable version. Reuse Sessions in authorized processes and run a Workflow again with its intended context. Monitor retains information about the attempt so you can investigate the path without confusing it with the current draft.
How it works
Break the process into tasks
Define what each stage produces and what the next one needs. Start with two simple tasks whose outputs you can verify independently.
Prepare the Sessions
Publish the Mappings and configure their Sessions with paths, worker and parameters. Make sure later tasks can access the earlier outputs.
Connect dependencies
Create the Workflow and add tasks that invoke the Sessions. Connect the intended order and review each task’s options before validation.
Publish a consistent version
Validate and save the draft. Check the referenced Session and Mapping versions; publish the Workflow when the full set is ready.
Test the complete sequence
Run manually, record the attempt and check order, states and output. Test a controlled failure too, to understand which tasks continue or remain blocked.
What you need
- Published Mappings and Sessions, with permission to inspect their versions and execute the Workflow.
- An eligible worker and access to the files or systems used by each task.
- Clear dependencies, test destinations and an acceptance criterion for the combined result.
Limits to consider
- Saving a draft, publishing and executing are different actions. A Mapping change does not silently update all its consumers.
- A dependency coordinates tasks; it does not copy files between workers or prepare network access and credentials for a later task.
- A Workflow does not turn every external write into one transaction. Before repeating a failed task, check partial effects and possible duplicates.
Demonstration example
Illustrative example using synthetic data. This is not a customer result or a product screenshot.
Demonstration: prepare and deliver a file
A first Session normalizes a synthetic CSV and writes an intermediate file. A second reads that result and creates a delivery CSV. Configure a dependency so the second task waits for the first and verify both files. This is an illustrative design to execute in a laboratory.
w_DeliveryDemo
s_PrepareDemo
↓
s_DeliverDemo