Dat VS Liquibase

Yes there are some really interesting projects, also in the ML replicability space. One really nice approach is the DAT project [1]. The protocol [2] looks pretty sensible and useful. Unfortunately, the tooling has been in such a state of permanent flux (i.e. Perpetual deprecation) that I've never bothered to invest much time. [1] https://datproject.org/ [1] https://datproject.org/. - Source: Hacker News / about 2 years ago

As far as keeping track of domain changes you can store DDL files in version control like you mention or use tools like Flyway (https://flywaydb.org) or Liquidbase (https://liquibase.org) which takes care of database migrations. Source: about 2 years ago

I just use SQL directly (or something like JOOQ). For database migrations I use Liquibase. Source: about 2 years ago

Regarding the migrations, there are tools such as https://liquibase.org/ or FlyAway that handle this. Heck, you can even use an ORM that has a migration baked-in but that defeats the purpose of having the migrations in a separate project. Source: about 2 years ago

I've trialled schemachange and liquibase which are change script based tools. I've ruled out a whole load of other tools that are either change script based tools or don't support Snowflake, including the following:. Source: about 2 years ago

Nowadays I prefer to automate database updates and deployment, using Liquibase and its relational database vendor agnostic syntax for that. Especially on production systems. But on local dev environments, I can still use the occasional SQL in a pinch. Source: over 2 years ago