Open xuhdev opened 3 years ago
We'll check with the OpenAIHub team to see if this would pose a problem. My understanding is that we used this convention to mimic what was implicitly done in https://github.ibm.com/IBMCode/Code-Data, where the id is the slug name.
Related issue: https://github.com/machine-learning-exchange/mlx/issues/209
For MLX we need to ensure that given an id
that contains underscores (_
), the generated DLF/DataShim YAML has a metadata.name
that conforms to DNS-1123 subdomain naming restrictions (must consist of lower case alphanumeric characters, -
or .
)
Currently IDs use dashes instead of underscores:
https://github.com/CODAIT/exchange-metadata-converter/blob/7160fd4c5c2cc54225e0677689a0be7c52ed4773/dax-data-set-descriptors/jfk.yaml#L1
Using underscores potentially eases Python users' life by allowing them to type IDs as variable names (say `dax.datasets.noaa_weather_data_jfk_airport). If it's not possible, we would have to do a dash-underscore conversion on one end.