Hi a quick question. Is it the case that if a launch plan has custom setting of

output_location_prefix

, all the tasks in the launch plan will use that location to store and read task outputs, regardless of the type of the output (Python native, vs. complex structure), instead of using the default one that comes with Flyte installation? Or rather the type of the output matters?

output type matters

https://docs.flyte.org/en/latest/concepts/data_management.html#raw-data-prefix this seems to describe a bit more.

primitive values will continue to be stored in the same place

Do we have a full list of supported types?

All the schema, blob, FlyteFile, FlyteDirectory types?

yes, those will be offloaded to output_location_prefix, and pickle, structured dataset, and torch.Tensor.

Thank you!