Open nreinicke opened 4 months ago
and assuming we could wire this in via the CompassApp.run method:
class CompassApp:
def run(..., chunksize: Optional[int]):
if chunksize is not None:
return run_in_batches(...)
thinking: i feel like stacking alternatives into the core run method would probably be more easily discoverable for users. same goes for the "to geopandas" behaviors, and, maybe all of that wiring-in is just one more downstream issue to collect these all as behaviors of CompassApp.run
, which gives us one of those big APIs like Pandas.read_csv
or similar that serves all users.
maybe all of that wiring-in is just one more downstream issue to collect these all as behaviors of CompassApp.run, which gives us one of those big APIs like Pandas.read_csv or similar that serves all users.
Yeah I like that idea. I changed to scope of this to just complete the function and then we can build out a new issue for wiring it in (along with any other run methods we discover)
zenon18 (Mark W.) will like to work on this issue
The current workflow when using the python API is to load a
CompassApp
object from a config and then call therun()
method, passing a set of queries as python dict objects. We could also add another way to interact with the application which would be to point to a config file and a query json file and then have the application read the queries in batches and also write to an output file in batches. This could take the form of a single function that looks like this: