Comment by 0cf8612b2e1e

5 days ago

Heh, I was just starting to look at SDF and SQL Mesh to see if they actually addressed any of the dbt pain points.

dbt needs to play catch up against SQLMesh feature-wise, so they bought their other competitor SDF. SQLMesh seems to have more development velocity, and dbt will need to execute a smooth transition and integration to catch up.

For context, the team behind SQLMesh also develops SQLGlot, which powers the features dbt attempts to implement

  • To expand on this, dbt uses Jinja templating in your SQL to allow the developer to modify the query so it can be expanded at "compile-time" into the target database SQL dialect. (uni-directional, write once deploy anywhere). The key features are CICD, test automation, and transformation sequence automation.

    SQLGlot is a Abstract Symbol Tree SQL Dialect transpiler that could (in theory) convert from one dialect to another (bi-directional).

    SQLMesh appears to combine both of the above into one tool that sounds like it's even better.

What is your initial impression and pros and cons?

(Asking as I kinda wish my company’s opinionated dbt had made some different choices)