ARM’s First Value-Added Product ‘Code Sprint’: a successful development acceleration paradigm

 
Poster PDF

Authors

Karen Lee Johnson — Brookhaven National Laboratory
Pavlos Kollias — Stony Brook University
Eugene E. Clothiaux — Pennsylvania State University
Tami Fairless — Pacific Northwest National Laboratory
Meng Wang — Brookhaven National Laboratory
Katia Lamer — Brookhaven National Laboratory
Mariko Oue — Stony Brook University
Michael Jensen — Brookhaven National Laboratory
Scott Giangrande — Brookhaven National Laboratory
Edward Luke — Brookhaven National Laboratory
Yaosheng Chen — University of Colorado Boulder

Category

ARM infrastructure

Description

Scientific teams frequently develop algorithms that use ARM observations to retrieve higher-level products. These valuable algorithms are typically not immediately suitable for operational implementation, do not produce ARM standard output, and are not readily implemented as Value-Added Products (VAPs) within the ARM production environment. Typically, a considerable amount of developer effort and time is required to incorporate a science code into the ARM VAP framework. Developer resources are limited and there are many scientific algorithms worthy of becoming VAPs. As a result, there is considerable interest in expediting the VAP development process. We experimented with a new approach during a highly focused product development workshop, or “code sprint,” held at Stony Brook University in summer, 2016. For seven days, science leads and ARM developers met face to face for intensive collaborations and code development. Our goal for this period was to convert three existing scientific algorithms into three ARM VAP codes running within the ARM development environment. We describe the code sprint activities and the results of the collaborations, then make recommendations for future similar workshops.