📓 Fusion Diaries: UDF support, Iceberg materializations, and an incremental compile cache in the CLI

Author: Anders Swanson, Senior DX Advocate at dbt Labs

Not much new shipped last week features-wise. But as always, we’ve squashed many bugs and vanquished some dragons.

However, there’s a huge amount of things landing over the next few weeks that are still underway. See “Work in Progress” for more information.

Also check out the new Agentic workflow for migrating to the new authoring layer in the 👓 Stuff you should read section below. And as always, there’s a meme at the end.

Velocity

• 50 issues closed across the dbt-fusion and internal repos
• 5 new preview releases

Community contributions

Slack

22 contributors over the past week!

GitHub

Four folks opened 13 issues on dbt-fusion last week: @joar, @jecolvin, @jsnb-devoted, and @CyanoFresh!

Special shoutout to @joar who reported 9 of them himself!

(Abridged) Release notes

There were four releases last week: preview.25, preview.26, preview.27, preview.28 and preview.29.

We won’t go over each. For version-specific information, check out dbt-fusion’s CHANGELOG

🚧 Work in progress

Here’s an ambitious list of all the things we’re targeting to land in time for Coalesce.

A good number of these should have initial later this week!

Fusion

• initial support for UDFs (Core and Fusion)
• Iceberg Catalogs and materializations for Snowflake and Databricks
• CLI will soon have the same incremental compile cache that the Language Server does!

VS Code extension

Partial LSP compile

Some larger dbt projects still take minutes to compile in the background initially before you can make use of the extensions features like Previewing a CTE or viewing column-level lineage.

Support for analyses

I’m very excited for this. You can think of analyses as a query editing experience that is disconnected from your DAG, but can still make use of jinja.

codelens for when static analysis is disabled

Not only do we allow users to manually disable static analysis, but Fusion will also intelligently do this when it comes across a good reason to do so. This is great, but the big UX drawback is that it’s hard to know when you have a file open if features like Intellisense will be supported! This is because static analysis is a requirement for this.

So we are working to ship a codelens that shows users at the top of the file if Intellisense is even an option.

🤔 Looking for feedback

Has Fusion caused you to do analytics engineering differently? We want to hear from you.

• Are you running less compile commands in the terminal now?
• Do you find yourself toggling less to your data platform’s web UI now that you can easily query inside the dbt VS Code extension?
• Can you accomplish your tasks faster with Fusion’s SQL understanding?

Let us know!

👓 Stuff you should read

dbt-autofix isn’t a panacea! There’s a long-tail of changes that cannot be automatically be resolved by moving YAML and config blocks around. Even if it were possible, this doesn’t mean it’s advisable. That’s why we’ve cooked up an AGENTS.md for migrating to complement the CLI tool.

Check it out!

So if you have yet to migrate, if you have access to Cursor or Claude or any agent, you can simply pass it that prompt, and the LLM should be able to make significant progress in making your project compatible with the new authoring layer.

Because the information is just text, it’s also a great source of information. Check out the prompt for the list of challenges some projects will experience when migrating to the new authoring layer. We also welcome contributions if you’ve come across other scenarios that others may benefit from!

🏁 Made it to the meme

We’re working hard to make compiles as fast as possible, but the reality is there’s a lot to statically analyze for most projects. The good news is that once the first compile is done, you’re off to the races!

I’m not sure about you, but it gets me dancing!

You’re going to want to turn on your volume for this (but not too loud!).

Original version on X