#!html
<h1 style="text-align: center; color: blue"> ACCESS Source Code and Input Data Management </h1>

The ACCESS NWP system is highly complex, and no single person knows all of it.

Not only are the multiple suites (Global, Regional, City, SREP etc.), but there are also

• Several large and complex components that are continually evolving (UM, VAR, OPS, ODB and SURF)
  • The bulk of the development on these is done at the Met Office
• A large number of small text data files used to configure the system
  • The structure of some of these depends on the version of the UM, VAR or OPS being used
  • These are suitable for inclusion in an svn repository
• Some large data sets, such as the ancillary files
  • These are not suited for inclusion in a software repository
  • Knowing how these data sets were created is essential
• A large number of scripts associated with cylc-scheduler tasks
  • available under the ROSE svn repository for a given suite
  • These are a mixture of Met Office and CAWCR developments
• A large number of scripts that may run in stand alone mode
  • Some of these are under ACCESS_UTILS (http://ngamai04.bom.gov.au:8011/utils)
  • These are generally developed in CAWCR.

An NWP suite consists of particular revisions of each of these repositories etc.

We need to know exactly the configuration we ran for either operations or an experiment for some time after the event, in order to meet our obligations to:

1. Reproduce our research
2. Recover operations from a catastrophic system failure
3. Meet demands of commissions of inquiry etc. But most importantly - to understand our own results!

The aim is to provide NWP suites that

1. Have a well defined environemnt (independent of the user)
2. Developments or changes to any component can be tested both individually and in combination
3. Is easy for every user to contribute updates to the main APS suites
4. Allows for easy excahnge of updates between users
5. Provides full provenance of code, scripts, data, etc.

• Source and Data Management Issues Some of the issues associated with source and input data management for ACCESS NWP systems

  • Complexities associated with SVN and the way we receive code
  • Dealing with entire NWP suites - not just the component source code

• Software Guidelines Software and Style Guidelines

  • Updating Suites
  • Updating Scripts
  • General good practices (commit early & commit often!)

• Data Guidelines Guidelines for managing NWP input data files