Does anyone do any testing in downstream BI tools when pushing changes to dbt? My use case would be for Looker specifically. Occasionally someone may think it’s safe to delete or rename a column, and a dimension or measure may stop working in Looker.
An idea would be to have something that tries to query important explores via the API, selecting all columns, and maybe limiting to 1 row for performance. In Looker’s case, I think you would also have to tell Looker to look at a staging schema where dbt has been run to reflect the changes, and also possibly tell it to run on a development branch in case you’ve made corresponding changes in Looker you want to test simultaneously?
I’d be interested in things that people might be doing in this area, and thoughts generally around how people think something could be implemented.