Proposal: Add Flink SQL support

[campossalex]

Hi all,

I'd like to propose adding Flink SQL as a supported target for Ossie and get feedback before putting together a PR.

Context: every converter merged or in review today (dbt, GoodData, Salesforce, Polaris, Spark) targets a query-at-rest engine, and the project's roadmap names Flink as a future target with nothing concrete started yet. Flink SQL is a natural fit for Ossie specifically because it's a Calcite-based, largely ANSI-SQL-compliant surface - unlike Flink's DataStream or Table APIs, it can consume Ossie's SQL expressions almost directly. I'd propose scoping this to Flink SQL only (vendor_name FLINK_SQL, not just FLINK) to keep that distinction explicit, since the other Flink APIs aren't SQL-expression-based and would need a different kind of converter entirely.

I have a working prototype that converts a semantic model to Flink SQL DDL for Kafka, Pulsar, Postgres CDC, MySQL CDC, and filesystem sources, using a FLINK_SQL custom_extensions block to carry the connector/watermark info core-spec doesn't (and shouldn't) model. Happy to share it if useful for this discussion.

Main question for the group: is anyone already working on this, and does targeting Flink SQL specifically (as opposed to Flink more broadly) seem like the right scope for a first version?

Thanks, Alex Campos