imply-cheddar commented on PR #13580: URL: https://github.com/apache/druid/pull/13580#issuecomment-1358863442
I made a new PR (#13600) for this as this current one was hard to rebase to fix the conflicts. This PR doesn't impact the planning or structure by which Operators can be defined or chained together, so it's not attempting to do anything in that regard at all. Perhaps it's painting us into a corner, but I don't expect this to be so set in stone that it's unchangeable at the point that we actually start trying to plan logical steps directly into Operators. I'm not actually looking for the dataflow, I'm looking for the highest fidelity stacktrace that can relay the maximum amount of information of what lead up to an error/exception. The push-based model that you are talking about (with buffers and async) actually also eliminates the stacktrace as a way to understand where an exception came from. I'm focusing entirely on the pipeline of processing inside of a single thread where I think you are taking this and trying to extrapolate the whole thing to be used across process boundaries as well. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
