Concepts¶

AntFlow separates the description of scientific processing from the service that exposes it and the infrastructure that executes it. Understanding these boundaries helps you choose a workflow, interpret a job, and prepare packages that other users can run.

From definition to execution¶

A workflow author describes processing in CWL and stores its application package in GitLab. AntFlow discovers the workflow and exposes a process with an input/output interface. Submitting inputs creates a job, which a launcher executes through a CWL engine. Published results provide access to the outputs and, when configured, their STAC metadata.

These are separate activities: validating a CWL file does not publish a process, and publishing a process does not execute a job.

Topics¶

• Architecture: platform components, connected services, and the workflow lifecycle.
• Workflows, processes, and jobs: how definitions, interfaces, executions, and results relate.
• Common Workflow Language: how tools and data dependencies describe processing independently of a particular platform.
• Application packages: how workflow descriptions, image references, and metadata make an application reusable.
• Launchers and execution engines: where orchestration and application commands run, and how infrastructure affects execution.
• OGC API - Processes: the standard process/job model and its relationship to AntFlow's API.

For short definitions, use the Glossary.

Put the concepts into practice¶

• Run an existing workflow.
• Create your first workflow.
• Publish a workflow.

Field-level input, resource, authentication, and publication details belong in Reference.