Back to Volume
Paper: DRAGONS: One Pipeline to Rule them All
Volume: 522, Astronomical Data Analysis Software and Systems XXVII
Page: 583
Authors: Labrie, K.; Cardenes, R.; Anderson, K.; Simpson, C.; Turner, J. E. H.
Abstract: Several years ago, Gemini embarked on the development of a pipeline infrastructure, written in Python, that would allow reduction of data from its multiple facility instruments. This contrasts with the more common approach to have a specialized pipeline for each instrument and mode. Our approach is instead based on only one infrastructure with access to often reusable, sometimes customized reduction step units and recipes. This strategy helps keep the codebase to a minimum, speed up development, and in the long-term saves a lot of effort on the maintenance. Building on this idea, the plan was (and still is) to use the same core software for night-time quality assessment pipeline and for the user science quality reduction pipeline. After running the nighttime quality assessment pipeline for several years and adding support for new types of data, we eventually encountered serious limitations with the initial implementation. It was time to study our use cases and rewrite it. Here we present our new, much improved, iteration of the project. We will also highlight some missteps from the first iteration and discuss how we decided to correct them.
Back to Volume