What's the best way to monitor stats about upstream datasources? Useful when things go wrong to be able to point to the inputs, but AssetMaterialization feels like it's not designed for the problem because I haven't generated that data. GreatExpectations? Cheers!

Hey George - when you say "monitor stats", what would this ideally look like for you? I.e. would you like something like the assets page where you could go to see a history of stats you've logged about an upstream datasource?

Yeah tbh, I think AssetMaterialization does the trick perfectly, but I think you guys are strong in the 'Thought Leadership' space and I want to follow recommended design patterns as closely as possible. Congrats on the series A!

Thanks! Yeah, AssetMaterialization would be my recommendation for the time being. I think we'd ultimately like to add some sort of other event that represents observations (instead of materializations) about assets

@George Pearse - I'm busy working on an example that deals with this kind of "monitoring upstream datasources" problem. See https://github.com/mrdavidlaing/dagster/tree/test-driven-software-defined-assets/examples/test_driven_software_defined_assets I'd love to hear your thoughts (via PR#5781 ) on the problem space and how I'm tackling things.