Database features and API stability

Each new release breaks build, so using the library is like sitting on a time ticking bomb. It's fine when you need to parse literally any SQL query, e.g. you have wild mix of SQLite, MySQL, DuckDB and PostgreSQL commands, but it's rarely needed this way.

It would be better if each database-specific field will be feature-gated, as new syntax added doesn't affect previously added features.

Here struct builders mentioned in another issue would be greatly appreciated.

I would like to contribute into builders part.

The issue names no files or tests, so first map the existing database-specific fields, feature-gating setup, and struct-builder work referenced here and in the other issue. Narrow the proposal to a concrete builder or gating milestone, then define tests showing supported database features remain buildable without unrelated syntax changes.

Written by the indexing model from the issue text.