Because OpenAI used specific training data for those programming languages?

It can also handle follow up questions to your initial prompt. But I somehow doubt that the follow up questions become training data outside of your current session.
You'll probably have to fork over millions of $$ to OpenAI for them to consider adding official DataFlex support, plus a lot of code.

I also doubt you want everything to look like the available open source code as that is not always following best practices.