Make database tables accessible to a different database

[davidmartos96]

I have an use case of 2 separate tables which I would like to handle versioning, so drift versioning system works really nice. One would be the main one and the other would be for support data, which is attached.

How could I declare the code generation for those 2 database schemas, and make support.db tables available when building queries for main.db ? I usually use either raw queries or Dart query builder. I'd like to avoid passing down to functions 2 separate database objects, because most of the time "support" would be use to just reference the TableInfos and there is no actual open step for it. ATTACH DATABASE is used instead.

It sounds like "support" in this case could even be a global object with the generated table and column references, but I might be wrong.

[simolus3]

If I understand you correctly, enabling modular code generation could be helpful here? By default, drift generates a single .g.dart file for an entire database. With modular generation enabled, we instead generate code next to each table. This allows re-using generated table code across multiple drift databases, if that's what you're looking for here?

[davidmartos96]

We are using modular code already. It was mote a question about how to create queries combining schemas defined for different databases.

Wouldn't defining queries make the drift analyzer fail because it doesn't know a particular table? The table would only be available at runtime.

[simolus3]

What are you using for this today, raw queries? Drift doesn't really support attached schemas currently.

You could define those queries on a drift file importing definitions for support.db if they're known statically (even if that drift file is not referenced from the main database).

[davidmartos96]

I currently mostly use raw queries (not from drift files) and Dart query builder.

Right now I define both schemas in the same autogenerated database file but ideally I would like to separate them in order to not have name clashing and to be able to use drift schema versioning on each database.

What I would like to know is if there would be a way to statically access the support db drift schema to build queries that use both DBs using the Dart query builder API

[simolus3]

What I would like to know is if there would be a way to statically access the support db drift schema to build queries that use both DBs using the Dart query builder API

If it's acceptable for the queries not to use schema.tableName references (they would just to tableName), I think that should be possible. Unless I'm missing something, you could write another regular @DriftDatabase including the support schema and pass it a QueryExecutor that doesn't call QueryExecutorUser.beforeOpen in it's ensureOpen method (if you pass the query executor from another drift database that has already been opened, I believe that should just work).