bennr01

@thedrow I've added a regression test for the issue. The new test passes with the fix but fails without it. I wasn't exactly sure where to implement it, but as...

Hi, are there any problems with the changes in this pull request? While one of the test build fails, this is only caused by `bandit` complaining over the user of...

Hi @drallensmith, sorry for causing you some headaches. I think most of the issues can be fixed by moving the `_EvaluatorSyncManager` in the module namespace and also define another class,...

@d0pa The main advantage of using `multiprocessing` instead of `mpi4py` is that `multiprocessing` does not need any other dependencies. Most systems which have python installed also have the `multiprocessing` module...

@drallensmith the `digest sent was rejected` error is raised when `connection.recv_bytes(1024) != b"#WELCOME#"`. While `connection.recv_bytes()` should normally return the response, default socket behavior is to return an empty string when...

@drallensmith >What should the primary node do if there's an error that likely indicates the secondaries are out of contact? In particular, if they are out of contact (pipe for...

@drallensmith i have pushed an updated version of 'distributed.py' (based on your changes in 'config_work') to the 'distributed_rewrite' branch in my fork. It is still not finished, but here are...

> It seems that there are frequent failures of 1+ threads in test_distributed.py, which unfortunately don't get reported by travis, etc due to the lack of error status feedback. Interesting......

@CodeReclaimers `distributed.py` should be finished soon (thanks to @drallensmith for identifying some issues and trying to fix some of them). Currently, the only problems are the tests, which seem to...

@drallensmith The documentation looks good (as always). The documentation for `neat.distributed` does not contain any mistakes, but i added a few comments containing improvement suggestions to the commit. Thanks for...