Sounds like they’re open sourcing it to get some help on it. I have to wonder if they’ve found it not worth the time to make it fully production ready.
It sounds like it's not compatible with all T-SQL commands and data types. It would be tough to get full feature completeness: Microsoft is constantly adding new stuff in each version, like the ability to run Java stored procedures, R & Python in the database, etc. Open sourcing it would let companies expand it to get specific features coded that they need - that otherwise might not get attention from the general public.
At launch Babelfish will be able to handle - with 100% correctness - the main semantics you'd want. HOWEVER, as you said (and as mentioned in the post), there's a large surface area, and a "long tail" of functionality that needs expertise from us and others to cover. So to do this right, it needs a community. It will be great for many use cases right from the start, but to ensure it's great/perfect for your use case...well, that just might take your help. Please work with us on this.
The only mentions so far are for Aurora for PostgreSQL, or self-hosted PostgreSQL.
So will all the other 'software interface' owners of the world.
https://docs.cloud.oracle.com/en-us/iaas/Content/Object/Task...
Announcing this first for SQL server might be a way to launch softly and learn.
>Those are the mechanics, but developers need to be certain that Babelfish truly speaks SQL Server’s language in a dependable, predictable way. As such, the guiding principle for Babelfish is correctness, with no compromises. What do I mean by correctness? Namely, that applications designed to use SQL Server semantics will behave the same on PostgreSQL as they would on SQL Server.
I am assuming they just mean in regards to types and the like - to get the same performance characteristics out of the code would be bonkers.
My take is that they are open sourcing it because they don't just want to use it to compete with Microsoft to host SQL Server workloads on a better basis than Microsoft's licensing policy has let them in the past, but also to just undercut Microsoft's SQL Server licensing revenue generally.
As a business strategy, sure, but given the way everyone I've heard from AWS talks about Microsoft and SQL Server licensing, I wouldn't be surprised if there was a good deal of spite involved, too.