Currently a query of datasets returns a dict, associating complete column names with requested values, whereas resolve_alias returns valuesid, ref_type (where ref_type is either "dataset" or "alias") and resolve_alias_fully returns just an id, since the id must come from the dataset table. Having different interfaces is not ideal in any case, but especially unfortunate now that dataset queries may return entries from both the production and working schemas, which is not supported by interfaces for the alias routines. It would be better if both alias routines returned a dict. Then the keys of the dict can fully describe where the returned id came from and it would be possible for queries to search both production and development schemas.
Such a change will entail modifications to gcr-catalogs as well since it makes use of the alias routines.
Currently a query of datasets returns a dict, associating complete column names with requested values, whereas
resolve_aliasreturns valuesid,ref_type(whereref_typeis either "dataset" or "alias") andresolve_alias_fullyreturns just an id, since the id must come from thedatasettable. Having different interfaces is not ideal in any case, but especially unfortunate now that dataset queries may return entries from both the production and working schemas, which is not supported by interfaces for the alias routines. It would be better if both alias routines returned a dict. Then the keys of the dict can fully describe where the returned id came from and it would be possible for queries to search both production and development schemas.Such a change will entail modifications to
gcr-catalogsas well since it makes use of the alias routines.