End users of the outputs of models, such as visualisation people, often need to use the inputs of a model run as well as the outputs to plot against each other. At the moment they will need to specify by hand the individual versions of the individual data products that they need. However, it would be easier to allow them to specify the data product output, and to pull the specific inputs used from config.yaml file uploaded to the registry. This could be achieved by pulling in the config file from the registry when needed.
So I propose extending the config.yaml syntax for the input files to allow a dependencies=true flag in a read: -> use: section, that would pull in all of the dependencies for that component (the inputs to the code run that produced that data product - i.e. the read: section of the registry's code_run.model_config entry).
End users of the outputs of models, such as visualisation people, often need to use the inputs of a model run as well as the outputs to plot against each other. At the moment they will need to specify by hand the individual versions of the individual data products that they need. However, it would be easier to allow them to specify the data product output, and to pull the specific inputs used from
config.yaml
file uploaded to the registry. This could be achieved by pulling in the config file from the registry when needed.So I propose extending the
config.yaml
syntax for the input files to allow adependencies=true
flag in aread:
->use:
section, that would pull in all of the dependencies for that component (the inputs to the code run that produced that data product - i.e. theread:
section of the registry'scode_run.model_config
entry).