[an error occurred while processing this directive] [an error occurred while processing this directive]

Minutes of the CCSM Software Engineering Working Group Meeting
4 and 5 February 2003
NCAR, Boulder, Colorado

Pat Worley opened the meeting at 8:30 a.m. on 4 February 2003.

Tony Craig presented an overview of the status of CCSM Software Engineering. Topics presented included the status of the CCSM Software Engineering Group (CSEG), a review of and plans for CCSM releases, an update on process improvement, testing, vectorization, an overview of three new projects, update of the CCSM GUI and Data Mover, getting CCSM documents under control, and a rewrite of the data models.

John Drake summarized the status of many activities in SciDAC including CAM performance, cpl6 status, rewrite of atmosphere/land coupling and associated performance upgrades, and chemistry.

Cecelia DeLuca gave an update on ESMF. The project will have a prototype release on April 30th and will be holding a community meeting at GFDL on May 15th to present the release. The release will consist of the ESMF software and a number of simple examples that demonstrate component coupling, halo operations, array redistribution, and other basic functions of the framework.

In the afternoon, Erik Kluzek presented an overview of incorporation of ESMF into CCSM. Erik gave a overview of how CCSM will use ESMF in CCSM and the schedule for implementing it. Using ESMF allows CCSM to move away from our multi-executable model, and use a single executable while still running concurrently. Erik also went over plans for subroutinization (creating an initialize, run, and finalize subroutine for each model), coupling using ESMF, and using ESMF utilities. CCSM plans to use most if not all of the ESMF infrastructure to some degree. CCSM plans to subroutinize the models in April 2003, and then work on prototyping on an ESMF development branch until the second quarter 2004 when, if ESMF performance, reliability, and robustness is satisfactory, CCSM may move the ESMF version to our production models.

Tom Henderson went over plans, process, and redesign of CAM. A new process has been put in place to serialize check-ins to the CAM development branch with the goal of reducing risk prior to the IPCC deadline. Several check-ins have already gone through the process, which has been fine-tuned along the way. There is now a CAM Check-In page that displays planned, current, and completed check-ins to aid developers with planning (http://www.cgd.ucar.edu/~cam/cam_checkins/).

Jeff Anderson presented a talk entitled "Incorporating CAM into a Data Assimilation Testbed". He provided input on how the data assimilation community uses CAM and what things might be useful in a redesign.

On 5 February 2003, Tony Craig summarized the status of CCSM testing and then Pat Worley led a discussion on that topic. Notes and recommendations are included at the end of Tony's talk. The recommendations include:

In the final hour, there was an open discussion. A number of points were made:

Participant List:
Tony Craig, NCAR
Cecelia DeLuca, NCAR
Matthew Cordery, Cray
Erik Kluzek, NCAR
Pat Worley, ORNL
John Drake, ORNL
Phil Jones, LANL
Tom Henderson, NCAR
Bill Preeg, Southwestern Bell
Forrest Hoffman, ORNL
Mariana Vertenstein, NCAR
Bill Putman, NASA DAO
Phil Merilees, NCAR
Brian Kauffman, NCAR
Keith Lindsay, NCAR
Byron Boville, NCAR
Helen He, LBL
Wei Yu, NCAR
Vince Wayland,NCAR
Woo-Sun Yang, LBNL
Jay Larson, ANL
Rob Jacob, ANL
Bob Malone, LANL
Matthew Hecht, LANL
Mat Maltrud, LANL
Art Mirin, LLNL
Jeff Anderson, NCAR/GFDL
Kevin Raeder, NCAR