Source code is closed source and at an old employer, unfortunately, but I once used Ragel to write a non-blocking MySQL client library using the regular expression features for packet parsing and, separately, a state chart for query and session state--basically, which type(s) of packet to expect next and otherwise documenting how session states transitioned.

If I had to do it again[1] I might not use Ragel for either packet parsing or session state. The MySQL wire protocol is rather simple, and Ragel has a steep learning curve, which made hacking on the library unnecessarily difficult for others. At the time I was already knee-deep in Ragel for other stuff, so it was the path of least resistance for me. But for more complex state management I would definitely return to Ragel as the expressive syntax is worth all the documentation and commentary in the world. And because of how well Ragel supports embedded code blocks (including for non-blocking/asynchronous I/O designs), you can maintain proximity between critical code and the state chart directives, which improves both expressiveness and self-documentation. That's the real power of Ragel--it's source code-level interfaces. On the surface the interface seems a little clunky (no interoperability at code AST level), but in practice it's second to none.

[1] Of course, the second time around you already know the pain points and contours of the problem so it's much simpler to open-code an elegant, minimal solution. So it's not much of a comment on Ragel to say that I wouldn't do it again.