Why mot use comment-annotated .sql like liquibase or flyway database migration tools do? How does yaml help?
As someone who did migrations wrapped in XML back in the day, those parsing wrappers around sql (xml, yaml, json) add complexity for the user and create opaque error messages at times due to the wrapper format. (quoting, escaping, indenting, allowed characters, etc)
I get that as a certain fraction of your userbase may like yaml and I am not saying you shouldn't support it. And I get that the yaml parsing library can accelerate certain dev steps for you. But the impedence mismatch between yaml and sql and error handling will make your product usability worse.
You are already building a domain specific language of sorts. You will build a better product if you keep your "language" as close to the problem domain (sql-centric migrations) as possible. Yaml just adds complexity.
As someone who did migrations wrapped in XML back in the day, those parsing wrappers around sql (xml, yaml, json) add complexity for the user and create opaque error messages at times due to the wrapper format. (quoting, escaping, indenting, allowed characters, etc)
I get that as a certain fraction of your userbase may like yaml and I am not saying you shouldn't support it. And I get that the yaml parsing library can accelerate certain dev steps for you. But the impedence mismatch between yaml and sql and error handling will make your product usability worse.
You are already building a domain specific language of sorts. You will build a better product if you keep your "language" as close to the problem domain (sql-centric migrations) as possible. Yaml just adds complexity.
My two cents anyway.