Build failed in Jenkins: SystemML-DailyTest #808

See -- [...truncated 12700 lines...] 17/02/13 03:48:40 INFO scheduler.DAGScheduler: Submitting ResultStage 243 (MapPartitionsRDD[669] at collectAsList at MLContextTest.java:1337), which has

Re: Removal of workaround flags

Thanks for bringing up the topic. Our deep learning scripts (i.e. algorithms with several intermediate transformations) have shown cache release times to be a major bottleneck, thus leading to the creation of SYSTEMML-1140. Specifically, what did you use to attempt to reproduce 1140? -Mike

Re: Pull Request Reviews

Thanks, Deron, for bringing up this topic! PRs, and the associated discussions, are a critical part of any modern, successful open source project. As Deron stated, anyone in the community should feel free to review PRs -- we want your thoughts and opinions and greatly appreciate your help! -

Re: Namespace handling w/ imports

Thanks, Matthias for bringing this up. As Glenn pointed out, the full file path as the namespace is needed so that we can effectively build libraries/packages for SystemML, rather than just single-file scripts. If you truncate the namespace down to just the name of the specific file, then you

Re: Pull Request Reviews

Apache usually have two ways for revieweing: REview then commit, Commit then review. If review is becoming a bottleneck, just use CTR, maybe waiting for a PR build to be green. On Mon, Feb 13, 2017 at 10:03 AM wrote: > Thanks, Deron, for bringing up this topic! PRs, and the associated > disc

Re: Namespace handling w/ imports

thanks for the clarifications - given that we don't support creating namespaces other than through imports, I think we can leave this for now (potentially along with a note in our future explain documentation). @Mike: Indeed, it should have been mnist_lenet instead of .defaultNS, referring to the

Re: Removal of workaround flags

Well, I used exactly the mnist_lenet scenario discussed in the JIRA, but what I've observed are eviction times <2.5% of total execution time, almost no sparse intermediates, and the script execution time being dominated by con2d_bias_add. Again, the discrepancy might very well stem from changes mad