Introduction

The LSST Pipeline Harness is a package for constructing pipelines out of reusable scientific processing modules and executing pipelines within an parallel processing environment. The harness serves to separate middleware functionality—management of parallel processing, I/O functionality, access to persistence, aggregation and connection of processing stages, etc. The Pipeline class encapsulates the main executing pipeline; it spawns a number of parallel workers denoted Slices. Both Pipeline and Slice workers execute a loop of application stages that have been provided with pipeline policy. The harness provides a Stage API for hosting an algorithm written by an application developer and integrating the application code into the LSST Pipeline Harness framework. Classes are provided to facilitate the encapsulation of serial code to be executed by stages within the central Pipeline and parallel code to be executed by stages within the Slice workers. The following cases show examples of application stages extending these classes for integration into the harness:

Template for testing an application Stage in a Pipeline : Template for testing an application Stage in a Pipeline