Kathleen Labrie (Gemini Observatory), Craig Allen (Gemini Observatory), Paul Hirst (Gemini Observatory)
Gemini is developing a near real-time data quality assurance pipeline. The final product will support all facility instruments and all of their observing modes. Rather than developing separate pipelines for each instruments and modes, the choice was made early on to design an extensible pipeline environment where configuration plugins are added to a core infrastructure.
The iterative deployment of this extensible pipeline environment at Gemini has recently begun. The first deployment includes the core infrastructure support for the processing and image quality assessment of optical imaging GMOS data.
Support for an instrument and mode is added to the pipeline through a collection of configurations that must include data type definitions, data descriptors, and implementations for applicable processing steps, and that must respect interface control rules. The configurations provide the data abstraction necessary to the pipeline infrastructure, and the interface definitions allows for one simple, reusable lexicon describing the data and the processing of those data. The key benefits from a development point of view is maximum code reuse, with top-level recipes and many processing algorithms being completely instrument agnostic.
Paper ID: P075