OpenJDK Governing Board Minutes: 2018/8/23

The OpenJDK Governing Board met via conference call on Thursday, 23 August 2018 at 15:00 UTC with the following agenda:

  1. OCW comments/feedback
  2. OCTLA
  3. 100% complete OpenJDK
  4. Any other business

Four Board members were present: Georges Saab, Andrew Haley, Doug Lea, and Mark Reinhold.

One Observer was present: Steve Wallin.

The intent of these minutes is to capture the conversational flow of the Board's discussion and also to record decisions. If you are interested only in the latter then search for the word "AGREED" throughout the text.

1. OCW comments/feedback

Georges noted that this was the required quarterly meeting for calendar Q3 2018.

Moving to the OpenJDK Committers' Workshop, Georges outlined the Governing Board member's attendance (Andrew, Mark - all; Georges, Steve - some) and requested a brief summary. Andrew said that over the two days, it was clear that there were many important topics to discuss and misunderstandings to clarify. He observed many informal conversations in the corridors. Andrew believed that the OCW was worth doing and appreciated the "unconference" format. Georges agreed that there were a wealth of good topics. Mark said that the next OCW would be one or two days attached to FOSDEM '19 (Brussels, 2 & 3 Feb 2019). In response to Doug's request for attendance numbers, Mark replied that about 80 people registered; however, more than that showed up and there were a few no-shows. He also said that he thought it turned out really well and while there were some disagreements, everybody was constructive and positive. Andrew noted that there were quite a few people there who had never committed a patch.

Steve commented that recordings of several OCW talks were available and asked about the status of the others. Georges replied that the four introductory sessions were recorded and highly recommended. Mark said that the lack of recordings was intentional. Recording less formal talks was expected to be of dubious value and to suppress conversation. Andrew indicated that some people were hesitant to lead topics. In lieu of recordings, Mark encouraged those who lead topics to submit summaries to the workshop-discuss mailing list. Georges added that the summaries were not expected to be detailed. He observed that while live discussions were useful, the OCW was not a forum to make decisions and would not replace discussions which occur on the mailing lists.

2. OCTLA

Steve opened his topic by asking whether the Board viewed the OpenJDKCommunity TCK License Agreement (OCTLA) as an OpenJDK Project activity or an Oracle activity. He reported that there was an OpenJDK page which described the process for accessing the JCK and he wanted to know whether the Board had insight into the application and distribution process.

Georges replied that the OCTLA is an Oracle agreement and information was available on OpenJDK for historical reasons. When Sun Microsystems created the program in 2007, they wished to provide the TCK to those who were working on OpenJDK-derived implementations so that they could verify whether the implementation was compliant to the Specification. Doug viewed the fact that the TCK was not part of OpenJDK an oversight. Steve declared that IBM would like to contribute more to OpenJDK; however, without access to the OCTLA, their contributions were restricted.

3. 100% complete OpenJDK

Steve asked for the Board's perspective regarding the expectation that OpenJDK be 100% complete and independent of external components. In particular, he noted that the TCK was a dependency requiring a third-party commercial license. Doug agreed that this was a problem and he would support bring the TCK closer to OpenJDK if there was any single thing that would accomplish that. Steve believed that the TCK issue needed to be addressed to encourage a collaborative environment. Andrew expected that Steve's difficulty with TCK acquisition was due to being a corporation. Steve countered claiming that IBM was working on purely OpenJDK technology.

Andrew said that it was always his understanding that TCK access was at the good will of Oracle; however, he had not heard of it being a problem for anybody. Steve asserted that IBM's use was not any different than other OCTLA signatories. Mark disagreed, saying that the situation was materially different. Further, this had been explained privately and he didn't believe that a Board meeting was the forum for further discussion. Doug suggested public contribution of tests which would become part of the TCK as he did for concurrency. Mark emphasized that following Doug's advice required contributions. Steve said that they were working toward and open end-to-end environment and needed the tests to verify that they had compliant builds. Georges stated that the OCTLA made it possible for people doing an OpenJDK-derived implementation to acquire the TCK. Steve asserted that was what IBM was doing. Georges and Mark both disagreed. Continuing his comment about the OCTLA, Georges enumerated the list of 11 OCTLA signatories for Java SE 9, or later and began reading the list of 21 OCTLA signatories for Java SE 8. He stopped reading the list at Andrew's request.

Doug believed that the Compatibility Testing Scholarship Program was still operational and suggested that IBM consider applying. Steve said that he would need to investigate and did not understand how IBM's situation was any different than Red Hat or Azul Systems which did have access to the TCK. He declared that IBM's implementation was more than 90% derived from OpenJDK. Georges asked Steve whether he was saying that IBM's implementation was an OpenJDK derivative. Steve could not conclusively answer whether or not IBM's implementation is a derivative of OpenJDK. Steve requested clarification regarding whether this was at the VM level and received no response.

4. Any other business

Andrew reported that the Vulnerability Group was going well, though there was a small unanticipated problem. The Group did not see non-security-related patches until quite late in the release. He assumed that the critical patch release was the responsibility of the Vulnerability team; however, that did not appear to be the case. Mark noted that in principle, non-security patches needed to be handled differently and advised Andrew to speak to the JDK Updates Project Lead to request earlier communication of non-security patches. Andrew replied that the problem had already been discussed in the Group and earlier notice was expected.

Georges mentioned that he did not have an update on Project Skara but he expected it to move forward and wanted to raise awareness. Doug asked whether it was a side project or a serious contender. Mark replied that it was somewhere in between, noting that several people were investigating a move to Git and possibly a hosted provider. He added that there were a number of technical advantages in moving source to an essentially free content delivery network and while Git does have flaws, it had come a long way and was popular. Andrew observed that the potential issues with something like GitHub were obvious and he was not particularly happy. Mark said the source would not be bound to a single provider and another would be selected if there were problems. He concluded saying that in the worst case, self-hosting is always possible.

Andrew was perplexed that some people considered tool popularity to be a draw for developers when working on the JDK requires a great deal of understanding. Mark agreed saying that it was never a goal of his, or others in OpenJDK, to maximize the number of OpenJDK Contributors. The goal is to foster and mentor long-term, steady Committers.

Andrew recommended Karen Kinnear's Just a Small Class File Change -- Nestmates: A Case Study saying it was a fantastic description of what is actually involved in integrating features into the JDK. Mark agreed and noted that it was part of the JVM Language Summit.

At this point, the Board adjourned.