Results returned from the task reporter can be Python objects other than primitive values (e.g., a numpy array). In the batch case, it doesn't make sense to put N number of these as proxies into the DB. Rather, create a list of them and proxy that list into the DB. Polaris-hpc code has an example of this.
Integrate the Polaris-hpc prototype example of this into eq-sql such that the result of eq.query_result is a future type object that can be iterated over to get the results. The resolution of the proxy should occur inside the API and all the user sees is the iterable result. This should be part of what distinguishes a batch result from the single "blown-up" results.
Results returned from the task reporter can be Python objects other than primitive values (e.g., a numpy array). In the batch case, it doesn't make sense to put N number of these as proxies into the DB. Rather, create a list of them and proxy that list into the DB. Polaris-hpc code has an example of this.
Integrate the Polaris-hpc prototype example of this into eq-sql such that the result of eq.query_result is a future type object that can be iterated over to get the results. The resolution of the proxy should occur inside the API and all the user sees is the iterable result. This should be part of what distinguishes a batch result from the single "blown-up" results.