I had the honor to be a guest on Java OffHeap (@offheap) while attending Goto Chicago 2017. I am also happy to announce that I am in talks with the OffHeap guys to have an occasional guest segment on their program, tentatively called "Mark and Sweep" where we will feature audio content from the existing interviews from my book Secrets of the Rockstar Programmers as well as new interviews with additional thought leaders.

The link to the episode is <http://www.javaoffheap.com/2017/05/episode-24-nothing-like-sweet-home.html>, but this post is my summary and notes, including the segment with Jim Weaver and Dean Wampler.

Part I, with Georges Saab and I, is here

Part II: Dean Wampler and Jim Weaver

51:18 Introductions

53:38 Bob polls the audience about alternative JVM languages.

54:12 What happens if JDK 9 flops? Dean asserts the microservices movement is a threat to the JVM because it exposes how bloated JVM apps typically are. Go and Rust: are a factor of 10/100 smaller. 55:04 But why does it matter if it's not a cost savings? 55:42 Dean points out that it matters at google scale. 56:00 Bob: jlink can help with this. 56:50 Freddy: if we use jlink with Jigsaw, will Scala be smaller? 57:24 Dean answers, but not quite, and points out that Lightbend does track JDK releases closely, citing lambda integration in a recent Scala release.

57:15 Freddy: other threats? 57:46 Bob: Azul's Paul Nauman: complexity of the JVM is the JVM's worst enemy. 58:28 Freddy, but that's true for every language.

59:20 Dean: another risk: Apparent with SPARK: the appeal of DSLs. 1:00:00 Bob: but what about debugging? Also, SPARK and massive heap sizes. 1:01:20 SPARK realized the JMM is great for general purpose, but if you have a billion of the same thing, there is no need for that complexity. 1:02:30 Bob: Code rot is another risk to the ecosystem. People stop maintaining stuff, but the stuff is still very widely used. 1:03:26 Freddy tells his JavaScript story: there's three of everything: npm, bower do the same thing. In spite of this JavaScript is still #1. 1:05 Dean: npm is using Rust for its scalability behind the scenes. 1:06:15 It's the old open vs. closed argument.

1:08:07 Bob ASF: Not a problem to have multiple projects that do the same thing. But you still need the JCP side to guarantee interoperability. 1:08:49 Jim Weaver adds that competition is good. 1:09:50 when you run JDK on AWS, it is OpenJDK. 1:10:26 Bob: there are things in place in the JDK to prevent fragmentation.

1:11:30 Freddy asks: What is your opinion of 9? Dean: They try to track the latest version of Java for their Scala users. 1:12:50 Lambdas did slow the adoption of Scala because it took away the worst pain point. 1:13:20 Jim: the Spring mantra make jar not war is well supported by jigsaw. 1:14:31 Bob points out that Spring uses a lot of reflection, but Jigsaw allows blocking code execution across package boundaries, won't that cause problems? 1:15:33 Freddy points out that 9 lets you get close to that 20M mark on size, but notes that with Spring boot, you may add a lot on top of that. Jim did not address this. 1:17:00 Freddy lists a nice process for optimization that starts with coarse grained and ends up with using JMH. This is a very valuable nugget.

1:19:24 Freddy asks to what extent Pivotal is going into machine learning. Jim answered that Pivotal is into data flows, such as Spring cloud data flow, which works with DeepLearning4J or Tensor Flow.

1:20:50 Freddy asks how Lightbend is doing machine learning. They are looking at how to make it accessible to their developers. 1:22:04 Bob ask how many are using Hadoop and is surprised with the small number of hands.

1:23:25 Freddy asks Dean to share the Lightbend strategy for world dominance. Answer: the tools they built, Akka, Play!, Lagom. An opinionated microservices framework. What they want to do is promote best practices and give them the tools to do it. Bob: we need to figure out a better way to structure the ideas so that regular developers can use it. 1:25:44 Freddy observes how hard it is to do the reactive manifesto correctly.

1:27:00 Discussion about the need to get knowledge out about best practices. 1:29:35 Bob observes that organizations need to change to adopt them.

1:32:10 Freddy asks, why is it so hard to get microservices right? James: 1:32:31 data is the hard part. 1:33:55 Dean: each microservices has its own persistence, but this poses a problem when you evolve a system and data needs to change.

1:35:55 How have alternative JVM languages played out? At Expedia, 85% is just regular Java. They do use a little Scala. The bit that they do use, some developers avoid touching because they are not familiar with Scala. 1:37:40 Dean: about half of their devs use Scala. 90% of the JVM usage is Java. 1:38:12 Container management and orchestration: Freddy: it's wild west right now. 1:39:45 Jim has a counter argument: perhaps use build packs instead of containers. 1:40:10 Freddy, I wish we could settle down with the Docker stuff.