Github user vasia commented on the issue: https://github.com/apache/flink/pull/2053 Hi @joseprupi, thanks a lot for this PR. I have looked through the code and I'm quite worried about its performance and scalability. Have you tried it out with any big inputs? We will definitely have to get rid of the calls to 'collect()' and try to re-think whether we can do without so much state in the auxiliary vertices. Even though I am quite familiar with the original AP algorithm, I'm not sure I understand all the details of the binary AP. Could you maybe create and explain a small example, for a small input graph and say 2-3 iterations, and show the messages and computations that are required. You can add this to the design document. Thanks!
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. ---