Regarding normalization and https://github.com/airbytehq/airbyte/issues/1927, I know that a general solution might be far away, but I'm wondering if we could get to structured rows by hand-editing a DBT model to do the appropriate transformation for our specific tables. Is that something you think would be worth trying?

Yes, you can try editing and running your own queries to build the struct types with dbt!

If there exist any tips on how to get started with that, that would be awesome! Thanks 🙂

You can go over the serie of tutorials on transformations in airbyte: https://docs.airbyte.io/tutorials/transformation-and-normalization/transformations-with-dbt Here’s an example blog post on how to flatten struct/array (the contrary of what you actually want), but you’d have examples of how to build some too: https://shotlefttodatascience.com/2019/12/27/tutorial-bigquery-arrays-and-structs/

Wonderful! Thank you 🙂

@Ping did you get time to look into it. We have an immediate requirement here will be happy to know any details regarding this. If you have not started I can pick this.