What’s the best way to build fault tolerance into a computationally expensive asset? This asset will process many rows, and should it fail (or our deployment crash etc) I would like the results to be saved and resumed later. So far I’m leaning toward partitioning the asset by blocks of rows, but was wondering if there was a more “dagstery” way of yielding results during materialisation

Hi @Guy McCombe - I think that partitioning the assets by blocks of rows would be the most "dagstery" way of doing this at this time