pitrou commented on PR #37202: URL: https://github.com/apache/arrow/pull/37202#issuecomment-1682362317
> The discussion is needed if we decide to put the definition of empty value (a better terminology than "default value" I used above) on the spec. No, this is entirely orthogonal. > My current position is that validation should be relaxed and not assume other implementations are producing non-null fields inside null structs. If we follow your position, other implementations will be exposed to such data. So they will be affected anyway. > For instance, I believe that a compute kernel, in the official implementation or other implementations, should be allowed to skip producing empty values inside of null structs. What do you mean with "skip producing empty values"? We _need_ to initialize memory anyway. Whether it's by a single `memset` call or per-value appends is an implementation detail. -- 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. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
