What's Happening in the JCP?

Print-friendly Version

Java Live Transcripts Index

Guests: Onno Kluyt and Aaron Williams
Moderator: Edward Ort (MDR-EdO)

This is a moderated forum.

MDR-EdO: Welcome to today's Java Live chat on "What's Happening in the JCP?" As many of you know, the Java Community Process is an open community process for developing and revising Java technology specifications, as well as reference implementations (RI) and technology compatibility kits (TCK). A lot has been happening in the JCP, and here to tell us about that, and to answer your questions, are Onno Kluyt, Chair of the JCP program, and Aaron Williams, Manager of the JCP Program Management Office. So let's start with the obvious question: what is happening in the JCP?

Aaron Williams: Thanks Ed. We finished the last update to the JCP process a few months ago - that's JCP version 2.6. With that update, we really focused on creating a more transparent process, with more visibility for all members of the community and the public. We also made some changes to encourage more participation and more efficient JSRs. If there are specific questions about the changes we made, I'd be happy to discuss them further.

Onno Kluyt: The other thing that's coming up are the elections for the executive committee. In the fall, the community elects new members to these two bodies.

Steeve_Coco: Could you elaborate on what you mean by "transparent"?

Aaron Williams: Sorry, transparent relates to providing community members and the public more access to drafts of the specs as well as the decisions the Expert Groups are making.

Capet: I don't know what the average time span is between the inception of a JSR until the final version of a spec, but it's way too long. Is JCP 2.6 actually going to make the process faster? If so, how?

Onno Kluyt: It differs a bit whether you're talking about a single API or one of the platforms. A single JSR (say like Java3D for ME) takes, on average, 18 months. The platforms take a little longer. We hope a couple of items in 2.6 will improve speed. The early draft review is a public review without a ballot. We hope this will encourage spec leads to go into this phase sooner than with community review. And in combination with transparency, what we're aiming for is that draft specs receive more feedback, and receive that feedback earlier. It seems though that we're getting better at it. In the first 2 years of the JCP it took something like 300 days to get into the first review. That's now down to about 200 days on average.

mister__m: Given the fact that a lot of them are still being developed, it looks like (1) not enough people know about it or (2) lots of spec leads don't want to make their EG work more transparent.

Aaron Williams: Expert Group transparency is important to the community. It does create some extra burden on the part of the Spec Lead, but for the benefit of the community, it is worth it. We're working on providing Spec Leads with more tools to make the communication process easier - and that is one of the reasons we're so excited about working with java.net. It provides Spec Leads and Expert Groups with one more mechanism for communication and collaboration.

mister__m: Hi Aaron and Onno, I'm the JSR Community Manager at java.net and I'd like to know what kind of other efforts are you planning to make in order to support a more open process (besides JCP 2.6 and the JSR community, obviously ;-))

Onno Kluyt: We're looking at a couple of things. For example, the default confidentiality clause in the JSPA, encouraging more intermediate drafts to be published by spec leads, in addition to the mandatory steps. But also we're taking a moment to see how 2.6 works out.

Capet: Is it true that there have been cases where an RI or a TCK has been made available through an open source license? If so, can you point me to where that's the case?

Aaron Williams: Yes, this is true. Open source is just one licensing model that Spec Leads can choose when it comes to licensing the output of the JSR - the source code for the RI and TCK. The JCP is designed to be an open standards organization, and work with open source as a licensing model. JSR 166 (Concurrency Utilities) and JSR 241 (Groovy) are two examples.

Onno Kluyt: Hey mister_m, a question for you. How is the JSR communities effort at java.net doing?

mister__m: Well, four jsrs have joined the community.

Onno Kluyt: Which ones are they?

mister__m: There are four JSRs that have joined the jsr community: jsf, jsp, jsr-107 (caching) and jstl. One of them was wrongly marked as private :-). There are some others that are intended to go public soon. Everyone can get an updated list of these projects and join them by going to the JSR Home on java.net.

Aaron Williams: Maybe I can ask a question to the participants: What aspects of the JCP process are most attractive, and what causes you the most frustration?

Capet: As a corollary to your question to the audience, what sort of things are you looking at as potential improvements to the JCP process, in other words, things that we might see in a JCP 2.7?

Aaron Williams: Cost of participation is something the PMO and the ECs are constantly trying to address. Along with the efficiency of the process - we want to make sure compatible implementations can make it to market as quickly as possible. We're also looking at ways to simplify the JSPA and make it more manageable going forward.

Aaron Williams: The Executive Committees of the JCP form the Expert Group, and the PMO serves as the Spec Lead, for any changes that happen to the JCP process or the JSPA. We are always looking for ways to make the JCP more efficient and more participatory. Any suggestions or ideas would be greatly appreciated.

Steeve_Coco: For what it's worth from a non-member (hopefully soon though): It's been a bit time consuming -- from where I seem to sit, it has seemed like a long way in -- or like most members had relationships prior to and outside the JCP. It's just been "pushed back" a few times. Listening to other comments too.

Onno Kluyt: Do you mean "long way in" as the time it is taking to become a member? Or are you talking about participating and commenting within an expert group?

Steeve_Coco: It was kind of metaphorical. Mostly it's been learning what the process was, how it works, and how I could be valuable.

Onno Kluyt: Aha. Were there things that could have been done to have made that easier? For example, the material on the web site, navigation, ...?

mister__m: There are a few things that I think that push people away from the JCP: one is the all the legal stuff (there should be a "simplified", non-technical version of the contract so people could understand what they should or should not do) and the lack of awareness of the fact you can join for free. What do you think about it?

Aaron Williams: Individuals are able to join now for free, and we want more real, committed individual participation. I agree, the JSPA's complexity, and the complexity of the process to get it signed and get going are both roadblocks that we would like to see reduced or removed in the near future.

mister__m: EJB 3.0 is indeed a great example of how the process can get more transparent.

Onno Kluyt: Yes. That one is certainly getting a lot of air time on theserverside, javalobby and other places :-)

mister__m: However, if feedback was open, that is, there was a ejb3-feedback@something everybody could sign up and if the discussion was made entirely public, it would be much more valuable to the community as a whole. What do you think about that?

Onno Kluyt: There's a lot of discussion on that. It could eventually go that way but at the same time there's some worry about "performing in front of the peanut gallery." However, JCP 2.6 does provide the spec lead with the ability set up such a mailing list. So I guess we're allowing it but not (yet) mandating it.

mister__m: Sometimes the Expert Group might think something is not important, but users see an idea rejected because EG members think it's not very useful and a lot of email starts reaching their mailboxes, it might make a difference. That's the kind of case I'm trying to illustrate here.

Onno Kluyt: Indeed.

mister__m: That's something we've envisioned while launching this community @ java.net, as you know, but i'm wondering if there's any kind of effort inside the JCP to promote this kind of thing in the EGs.

Onno Kluyt: Let's follow-up on that and see what we can do to help out.

Capet: So what is it about JCP 2.6 that will cut down on the 18 month (or longer) time span between JSR initiation and final spec?

Aaron Williams: Several things. First, by moving the ballot from after the first review, to after the second, we are providing a more "safe" review, so Spec Leads can feel more comfortable going to review sooner. Second, by being more transparent, we are expecting JSRs to get the good ideas and feedback earlier in the process, enabling them to finish faster. Finally, as part of 2.6 we're also providing Spec Leads with more information about how to create TCKs and RIs, hopefully making it easier for first time Spec Leads to be successful.

mister__m: EGs can hold their internal discussions in a private list, but use a public list as a way to communicating with users. What do you think about that?

Onno Kluyt: Yes, they can do that.

Steeve_Coco: Can you say something briefly about how the JCP and java.net work together specifically?

Aaron Williams: Sure. There are now communities on java.net for JSRs, so Spec Leads can choose to use those communities to collaborate with their Expert Group on the RI and TCK, or just use it as a conduit for communicating with the public. We're working to make the JCP work more closely with whatever tools our Spec Leads use, and java.net is certainly a good start for us.

red: The sheer volume of JSRs makes it challenging for an individual or organization to determine the relevancy. Is there a way in which this complexity can be communicated through diagrams (UML system/package models)?

Onno Kluyt: Yeah, you raise a good point. In practice these things are hard to do. Get a community to agree on the exact relationship between JDO and EJB for example. I always had this on my list of things to do eventually but time has escaped so far. Keeping up to date is another challenge.

mister__m: Maybe mandating could be good.

Onno Kluyt: Perhaps, let's first see what 2.6 leads to.

red: The Java names space is protected and the use of the namespace is managed through JSRs so even a composite listing of namespace capabilities for a given platform could have high utility.

Onno Kluyt: Can you elaborate on that? Composite listing, as in "javax.jsf does this. javax.swing" does this? Perhaps so that people can see which ones are taken?

red: Where does the JCP and PMO stand on JSR squatting (i.e., JSR with poor progress)?

Onno Kluyt: We frown upon it.

red: Frowned upon but actively managed?

Onno Kluyt: It's something that the ECs are discussing actually. In the earlier days of the JCP, the PMO had the authority to act on such JSRs. With the introduction of JCP 2 and the ECs, the PMO no longer has that. We now in the end depend upon the spec lead doing the right thing: withdraw or move forward.

mister__m: Making the process more transparent is definitely one of the goals of the JCP right now. What do you think about having a weekly/monthly status report of all JSRs?

Onno Kluyt: That reads like busy work. I want spec leads and expert groups to spend as much time working on the drafts, rather than working on reports.

Steeve_Coco: (Back on the public mailing list idea:) I was saying that it seems like a public list might possibly dilute the process -- Persons who have a desire to participate and commit to a project should prefer to join it so they can contribute more.

Aaron Williams: We agree! We want people who are committed and can provide valuable input to join the JSR and get involved. The mailing lists and comments aliases are designed to provide people with feedback mechanisms at review periods, not for day-to-day discussion of the JSRs.

Guest: I'm thinking that the JSRs are categorized into platforms -- thus a list of JSR class packages for a given platform will help developers and architects understand what is already defined through specifications and what is not.

Onno Kluyt: That breakdown is already on the web site. You can have the JSRs listed by platform, and by technology like for example XML or JAIN.

Guest: Thank you --- cheers!

Steeve_Coco: I wouldn't mind reports if they contained some aspect of what "red" said about organizing them -- so the update would have some kind of threading to it.

Aaron Williams: Reports are good for the community, but they put extra burden on the Spec Lead to produce. We try very hard to balance the needs of the community with the needs of the Spec Lead - we can't afford to make any changes that discourage people from becoming Spec Leads. We've tried so far to give the Spec Lead the flexibility to choose how they want to communicate, and that seems to be working out well so far.

Steeve_Coco: I definitely hear "not discouraging the spec leads".

Capet: You mentioned Groovy earlier. What's the status of the Groovy Programming Language? When can we expect to see an initial spec?

Aaron Williams: As with all specific JSRs, you would have to ask the Spec Lead for specific information like this. Spec Leads can be contacted through the jcp.org web site for more information.

mister__m: The real problem that motivated my question is: today it is really difficult to know whether a EG is active or not, maybe because of some legal issues. How can this problem be solved?

Onno Kluyt: We hope that the transparency aspects of 2.6 will provide at least a partial answer. There is more visibility into the whereabouts of the JSR and expert group and thus more social/peer pressure to work well.

Steeve_Coco: "Dead" JSR's is an interesting point. The process should not overburden the spec lead.

Onno Kluyt: Do you have a suggestion as how to deal with them?

Steeve_Coco: Provide a mechanism where the JSR initiator can take on tougher self-testing, to "prove" the concept more.

Onno Kluyt: You mean more elaboration on the technology that the person is proposing in the JSR submission?

Steeve_Coco: Maybe I'm just panicked :) If I have questions, I know who to ask!

Onno Kluyt: OK

Steeve_Coco: I am interested in the issue about how long the process takes -- From what's been said, I could see that faster turnaround time would result in a better product -- within the bounds of not detracting quality. What was the original goal for timeframes? What is the current idea of an optimum goal?

Aaron Williams: There really isn't any "optimal timeframe" for JSRs. Some JSRs are really big and take a long time for good reason, some are driven to get to market and finish very quickly. We want to provide a process that enables Spec Leads to go at the pace that is most appropriate for them, and still ensures we will have compatible implementations. The process really only requires about 3 months worth of time - you could complete a JSR in 3 months if you wanted to. The reason JSRs take longer is because people take time to reach consensus, and I don't think we can fix that (or want to) in the process.

Steeve_Coco: I see.

Steeve_Coco: That could be it -- a dying JSR could be envisioned to have a better fate: It seemed like there would need to be something more in the setup that gave the JSR a robustness that would inevitably lead to it working out in it's proper venue -- or with its proper lead or something. Hope this makes sense. Just if there were some tool to help the lead or other initiator to shape the JSR better initially.

Aaron Williams: Hmm. I'd love to hear more about what you would expect in that tool. Have you been a Spec Lead?

mister__m: Although I know this, I think it'd be important if you could tell people how they can contact the PMO for more info/suggestions.

Onno Kluyt: pmo@jcp.org or aaron@jcp.org or onno@jcp.org.

Steeve_Coco: Is there a URL pointer help page or something -- (seemingly presaged by mister m) -- Like: This aspect is handled using this or these URL's.

Aaron Williams: The web site for the JCP (jcp.org) is full of help for how to become a member, and how to be a Spec Lead. You can also get information about all of the JSRs and review them during their review periods. If there is something specific, I'll be happy to point you directly to the page.

red: Any comment on the rumor to a Solaris Community Process?

Onno Kluyt: Don't think there'll be such a thing. Not as in a context similar to the JCP. Sun does have plans to open source Solaris and will put a community effort around it but you would have to think more in the terms of something like NetBeans, mysql, eclipse, or apache.

mister__m: Could you clarify how JCP 2.6 can help to make Java open source?

Onno Kluyt: My favorite counter question: What do you mean when you say "make Java open source"?

mister__m: Could you briefly tell people what they have to do in order to join the JCP and/or to join an EG?

Aaron Williams: Sure. To join the JCP, you need to sign the JSPA and fax it into the PMO. Instructions for doing that can be found on the membership page. To join an EG, go to the JSR page and click on the link that says "Join this Expert Group." Not all JSRs will have that link, just the ones that are still accepting EG members.

mister__m: Before the most recent versions of JCP, as far as I know, it wasn't "legally" possible to produce an open source implementation of Java and claim it was compatible. Am I wrong?

Onno Kluyt: That was fixed in JCP 2.5, Oct '02.

mister__m: That's why I rephrased it to say "most recent versions" ;-)

mister__m: If Microsoft joined the JCP, would it be bad for Java?

Onno Kluyt: Well, they are welcome. Just like all the current 800 members of the JCP, they'll have to sign and agree to the JSPA first.

Aaron Williams: Thank you for participating. If you have any other questions or comments about the JCP, please feel free to send them directly to me or to the PMO alias and we will get back to you offline.

Onno Kluyt: Thank you all for the questions and suggestions. Take care!

MDR-EdO: Last moderator (me) signing off. The forum is now unmoderated.