damccorm commented on PR #17267: URL: https://github.com/apache/beam/pull/17267#issuecomment-1099392945
> I think the difficulty here is that we can't meaningfully test the effectiveness of the watermark estimation until we have ProcessContinuations in, but I see (other than the last two suggested changes) no reason not to get it in at the moment. I would argue that we can to some degree with splitting and a little hacking to set the mode to streaming (I've run this with dataflow and seen it correctly update the watermarks on splits as described in my description). That isn't a complete demonstration of functionality, but the implementation isn't complete either since it doesn't try to handle processContinuations yet (that will need to be a future pr). Either way, I'm on board with your overall assessment that we shouldn't block this change on that, if for no other reason than it blocks future watermark work. Thanks for the linting callouts, I committed both of them -- 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]
