Closed elray1 closed 1 year ago
I don't remember discussion of the schema changing too.
My understanding was that it was the way the schema was already laid out (with type_id
and type
as sub-properties of output_type
) that was causing folks to want columns in model_output data to be output_type
and output_type_id
.
The way I have thought about this, this configuration setting in a hub's tasks.json file is specifying details about the values in the output_type_id
column of model output files, so it would be cleanest/clearest if the json field name was also output_type_id
, to establish a clear connection between the names of things in the config file and the name of the column in the model outputs that it is describing.
Sounds fair indeed. It will mean breaking changes in all current example/test hubs and hubUtils
functionality though.
I'm actually considering making hubUtils
functionality back compatible with the current version of the schema, at least for a while?
e.g. see this line
Elsewhere, we have decided to name this
output_type_id
rather thantype_id
, but it seems like the schemas here have not been updated yet. Is this intentional, or just an oversight?