lidavidm commented on PR #34817: URL: https://github.com/apache/arrow/pull/34817#issuecomment-1907105616
> Ohhh I just saw exactly what you were talking about, I'd thought for whatever reason that the result stream was passed to DoAction by the transport but there it is, the stream is constructed by the handler... so ok, I can skip over the more involved stream-wrapper-triggered header ingestion and just move ahead with adding a defaulted flag to CheckAuth (and so on) to defer the ingestion to "elsewhere" and as you said above make that "elsewhere" L528 or so (for consistency do I not want to actually do this before handling an empty result stream, which it would currently happen before?) I am lost. You would stop calling SendingHeaders where it was previously called, and start calling it explicitly in each RPC handler at an appropriate time. I don't see why empty streams need special treatment. -- 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]
