chrish42 commented on pull request #7110: URL: https://github.com/apache/arrow/pull/7110#issuecomment-625943409
I'd also prefer not to add anything to the JSON, but I was raising that option because there were concerns about maintaining compatibility in the past when I initially raised this, so that would be a way to mitigate that. One way to add extra fields while still leaving all the parsing to the Flatbuffers library would be to create a new Flatbuffers schema with said additional fields. But the first choice would be to make sure that Flatbuffers won't be changing that format under Arrow's feet, and so not need those extra fields. Also, agreed about this not belonging in the ipc namespace. It should probably live closer to the Schema C++ Arrow class. But we can discuss all that more once there's agreement on the approach and I rework the pull request to be mergeable. I'm not sure what you mean by non-official function, though. Could you expand on that? For me, the goal is to have a (user-readable, non-code) way of expressing an Arrow schema that's supported across versions. I get that we'll get there progressively, however. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected]
