kl0u commented on pull request #12549: URL: https://github.com/apache/flink/pull/12549#issuecomment-641499385
@sjwiesman thanks for the feedback. As for the answers: 1. The blocking or non-blocking is not affected by executing in application mode or not. It depends on the API call. `execute()` is blocking while `executeAsync()` is non-blocking. So your code and the interleaving of jobs depends on which call you are using in the case of a multi-`execute()` job. I can include a sentence, here. 2. In this case, I do not think there any functional differences from a job cluster. The only stuff I can find is stem from the fact that the `main()` is not executed on the cluster, so, for example, if you are registering cache files in your environment, the paths must be accessible. I can also include a sentence about that. 3. One of the problems here is what happens with a multi-`execute()` or `executeAsync()` application when there are also finite jobs in it. There is no way currently to tell if the job finished or it never launched. So currently it is not supported and there are no plans to add it in the foreseeable future. Given the above, I would feel more comfortable saying nothing instead of either "closing the door" or promising something that may not come (at least not soon). ---------------------------------------------------------------- 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]
