hi Max, Note that Flatbuffers and the Arrow columnar format are certainly not competing with each other. While we use Flatbuffers to serialize schemas, data headers, and other metadata, the data itself is not stored inside a Flatbuffer or similar serialized structure. This is a common misconception so worth restating.
You can see the serialization details in the columnar format documentation http://arrow.apache.org/docs/format/Columnar.html#serialization-and-interprocess-communication-ipc Each library generally has functions to serialize schemas into language-independent protocol messages. Regarding "binding generation tools", as I recall there has been some discussion in the past of generating schema-specific accessor APIs similar to how Protobuf/Thrift/Flatbuffers generates accessor APIs, this seems like an interesting direction to explore for a volunteer in the community. - Wes On Fri, Oct 25, 2019 at 1:49 PM Max Burke <[email protected]> wrote: > > Hello! > > I had a couple questions. I have a system that is currently using Flatbuffers > because they provide a couple features: 1) flatbuffers support generating > flatbuffer-encoded representations of schemas (ie: reflection) and 2) > flatbuffers have a bindings generator for type-safe access in client > languages. > > That said, the ecosystem evolving around Arrow is really intriguing. > > Is there any equivalent now for serializing schemas/reflection? > > Are there any plans for supporting binding generation tools? > > -- > -Max
