Recognises greenfield DDL aimed at a populated database
08-ddl/ handed to a release for an environment that already has data. This output builds a schema from nothing. The changes that look smallest in the diff are the ones that need a backfill plan — a nullable column tightened to NOT NULL, a widened or narrowed type, a split column — and on Databricks ALTER TABLE ... ADD CONSTRAINT validates the existing rows and fails on the first violation.
Can AI do this?
AI does this partly Some of the tested AI models master this; others do not show it. 3 of 10 models master it.
| Model | Tier | Probability of mastery | Status |
|---|---|---|---|
| gemini-4-argon | Flagship | 100% | Mastered |
| gemini-3.1-pro-preview | Flagship | 25% | Not shown |
| gpt-6-astra | Flagship | 100% | Mastered |
| claude-fable-5-1 | Flagship | 27% | Not shown |
| gemini-3.8-flash | Mid tier | 25% | Not shown |
| gpt-6.1-sol | Mid tier | 100% | Mastered |
| claude-sonnet-5-5 | Mid tier | 27% | Not shown |
| gemini-3.5-flash-lite | Low cost | 27% | Not shown |
| gpt-6-luna | Low cost | 25% | Not shown |
| claude-haiku-5-5 | Low cost | 27% | Not shown |
Mastered means a probability of at least 90% that the model has this knobit, estimated from its graded answers. “Not shown” means the answers do not settle it either way. How the models are tested.
What it looks like in practice
Expected result: The changes needing a backfill and a migration
How it is assessed: Flags the deficient case and passes the acceptable one. Deficient: account_number tightened to NOT NULL in the model and the regenerated DDL given to the release; 3 % of migrated accounts have no IBAN, the deploy fails halfway, and the schema is left with half its constraints applied Acceptable: The regenerated DDL diffed against the previous version as the statement of intent, and the migration hand-written for that diff where deployment lives (Flyway, Liquibase, dbt, Alembic) with backfill and rollback — then verified by rebuilding an empty database from the generated DDL and comparing it with the migrated one
Also known as: Greenfield DDL run against a populated database (generated DDL)
Related knobits
Builds on
All 33 knobits of Generate and review DDL.
Source: Review checklist 08-ddl, row 22; Ambler & Sadalage 2006; this skill (08-ddl.md)