|Roy T. Fielding||Aug 3, 2004 5:29 pm|
|Gianugo Rabellino||Aug 4, 2004 1:14 am|
|Daniel Florey||Aug 4, 2004 1:51 am|
|Daniel Florey||Aug 4, 2004 1:58 am|
|Mats Norén||Aug 4, 2004 4:22 am|
|Stefan Guggisberg||Aug 4, 2004 6:13 am|
|David Nuescheler||Aug 5, 2004 3:01 am|
|Paul Russell||Aug 5, 2004 4:06 am|
|Andrew Savory||Aug 6, 2004 3:42 am|
|Subject:||proposal to move JCR RI to incubator|
|From:||Roy T. Fielding (fiel...@gbiv.com)|
|Date:||Aug 3, 2004 5:29:43 pm|
As you know, the JSR-170 JCR reference implementation is currently located in the Slide proposals subdirectory. The original purpose of that code was to provide a potential API for a Slide v++ back-end and to make it easier for all of the expert group (including Apache) to test out the RI and give more meaningful input back to the JSR. It was also hoped to be the basis of future collaboration on the RI and TCK. Unfortunately, due to the vagaries of the Java Community Process and being just a proposal within a larger Apache project, actual collaboration has been limited. I would like to change that.
The Apache board set up the incubator in order to handle incoming code bases that were developed outside of the ASF. Although the RI has been under Slide, there was a gap in which all of the code was pulled out and then later put back in. Even though I know there were good reasons for doing it, this effectively put its status back in the category of "not built within an Apache project", and thus subject to the requirement that all such code go through incubator first.
In addition, I think the lack of collaboration is troubling, particularly given the high level of interest in the JSR and the number of Apache-friendly companies that are in the expert group. In short, it seems that some people are interested in working on JCR and either not able to keep up with Slide development or unable to do so due to conflicts of interest at work. Likewise, it makes far more sense for Slide to focus on building the parts that might (or might not) make use of any JCR implementation, rather than on writing the RI implementation itself.
Therefore, I am proposing that the jcrri code be moved out of Slide and into the incubator as its own subproject, with explicit encouragement of anyone who wants to work on JCR itself to join that project as well. In that fashion, we can move all of the RI development out of Day internal lists and onto the public Apache list, move all of the public JSR-170 discussion onto the Apache development list, and encourage all of the expert group members who can and are allowed to work on open source to get involved. In other words, now that this JSR has passed public review and out of the old JCP constraints, let's finish it like a real open source project. Naturally, if we tried to do all that on slide-dev, then this project would get screwed by the unrelated traffic.
The first step is to write a proposal, which I have mostly done:
and the next step is to ask your opinions. Please let me know if anything about the proposal seems wrong or a bad idea.
Likewise, if you are already a Slide or other Apache project committer, please let me know if you would like to be added to the list of initial committers. If you are new to Apache and you want to be a committer, please send me information on why you should be on the initial list (e.g., you are working on a JCR implementation, already participating on the expert group, are well-known and trusted by the Apache community already, etc.). We are hoping for a diverse group, with development guidelines and committers working like the original Apache Group.
I'll send an invite to the JSR-170 expert group and other parts of Apache after people here have a chance to comment/volunteer.