All aboard: Rebuilding dbt in the open

Rewriting a 9 year old code base is hard. Rewriting it in the open over ~1 billion dbt installs later? Harder. Why did we do it? For faster runtimes, SQL comprehension, shared adapters, better docs, and a modernized developer experience.

We'll be rooting our session in a metaphor from a similar engineering problem, but for trains back in the 1800s: the break of gauge. I'll teach you how we aren't going to pay the cost of transfer as we modernize onto OSS v2.

This session will be demo packed:

• v1 → v2 conversion using autofix, and the v2 parser already running inside v1.12
• Side by side parse times in v1 and v2
• The power of one binary shipped with every adapter
• Docs v2 because we both know your `manifest.json` is too big
• Time allowing: how the distributions differ at catching an error (spoiler: it's 1.68s vs. 0.04s — wouldn't you like to know how to catch an error 42x faster?)

By the end of this session you should know:

1. Why you would want to upgrade to v2 and how to use the flagship v2 upgrades
2. How to upgrade to v2 — a project upgrade will be shown
3. How to choose the right distribution for you
4. More of the v2 mechanics and how it works under the hood

I hope to see you there!