On 03/10/2014 4:27 AM, Scott Gray wrote:
Currently, the connection between individual modules and the people who care is a bit fuzzy and has resulted in decisions being taken by people who do not have a real interest in the particular modules that are being dropped. They have no way to connect with the interested parties or even to be sure about who they are. One of the values of sub-projects is that you capture groups that have a narrow interest in particular areas but are not able to commit to the entire project.We have mailing lists and jira to gauge interest in a specific component. Code contributions in particular are very useful in determining interest, unless you're suggesting that the component is perfect and no further work is required.
It does not mean that end-users are not using the module. It is also a real-world application that people are using as is right now.
Everything in OFBiz has room for improvement and a lack of contributions is very much an indicator of a lack of interest in my opinion and experience with this project.
Of course, everything can be made better but that does not mean that you can throw out a perfectly good piece of code that is essential for some people just because no one is working on it. This would result in 90% of Linux commands being dropped from the OS. Just because no one is working on grep does not mean that I can live without it.
Sub-projects would also provide a way for users to interact with a team whose only function is to maintain and enhance that particular module.
It doesn't make any sense to create sub-projects in the hope that someone might show up and start community building. The ASF doesn't work that way for top-level projects and by the DB Project link you sent earlier implies they don't work that way for sub-projects either. A community must exist around a given component before it can have any hope of standing on its own.
You are absolutely correct.What is needed is a strategy and policy from the PMC to describe the conditions for establishing a sub-project. Once that is done, it will be up to the community to step up with proposals for establishing a sub-project that meets the criteria laid out. I am responding to the current need for a way for the PMC and contributors to attract people who have an interest in particular areas so that the group maintaining the core and the areas in which they are interested have an official team to work with to make sure that modules outside their interests are also maintained and are able to stay up to date with the Releases of the core functionality. The user community needs to have a way to understand what modules are in or out of the system and a mechanism to support their favorites.
Another alternative is to fork the project into a separate project that only maintains the modules that are not released as part of the core OFBiz. This is less attractive for the users but may be the only alternative to maintaining a full range of end-user functionality. It shifts more of the responsibility for integration to the end users but that is preferable to being unable to upgrade to the latest OFBiz due to features being lost between versions
I hope that the PMC members will do some investigation in this area and contact their Apache mentors and some of the projects that have extensive sub-project portfolios and see why they did this and what positive and negative results have been found over the years.
Regards Scott On 3/10/2014, at 7:50 pm, Ron Wheeler <rwhee...@artifact-software.com> wrote:Are there a lot of outstanding JIRA issues that users want fixed? It is not inconceivable that a module works as required. I am not sure that measuring the usefulness of a module by the number of bugs or deficiencies found recently is accurate. It seems to have been tested with the trunk as the core OFBiz has evolved. It appears that it may need some testing with the ccore 13.07.01 Release before PROJECTMGR can be either said to work as is with 13.07.01 or released as a new version of PROJECTMGR that does work with 13.07.01. If there was a sub-project with a following, there would be a group of people who want it to work and would be prepared to do what was required to keep the module functioning. It would be quite clear to the people interested in PROJECTMGR that it was their responsibility to make sure that it was functional with 13.07.01. Currently, the connection between individual modules and the people who care is a bit fuzzy and has resulted in decisions being taken by people who do not have a real interest in the particular modules that are being dropped. They have no way to connect with the interested parties or even to be sure about who they are. One of the values of sub-projects is that you capture groups that have a narrow interest in particular areas but are not able to commit to the entire project. The people working on the release of the core also have a clear project management group in each sub-project to consult when core functionality will affect individual modules or when planning a release and want to let the sub-project teams know that they must take some action in order to keep their module functional. It is not inconceivable that some sub-projects will die due to lack of interest. PROJECTMGR seems to have some life in it but without a formal sub-project structure it is hard to judge except from ML discussions and recent activity. Ron On 02/10/2014 3:02 PM, Scott Gray wrote:Surely the first step in considering a specialized component for sub-project creation would be the level of activity surrounding the component? Looking at the history of the projectmgr component I see 12 commits in the last TWO years 8 of which were global changes that coincidentally happened to touch on that component (translation work, global refactorings etc.). This leaves only 4 commits specific to the component and even those are minor UI adjustments. To be considered as a potential sub-project I would expect to see a hive of activity around that component with contributors specializing in solid contributions to further enhance it. "Build it and they will come" is not a valid approach to sub-project creation. If this component is so important to some of you, why are you not contributing to its enhancement? Regards Scott On 3/10/2014, at 2:56 am, Ron Wheeler <rwhee...@artifact-software.com> wrote:Of course, I see a lot of benefit in the Apache approach of sub-projects but perhaps the current group of committers should take some time to consider this and talk to the Apache Mentors assigned to the project as well as some of the project chairpersons from projects where sub-projects are in use. One of the advantages of being an Apache project is that there are many things for which there is an "Apache Way" and there are people in the broader Apache community that can provide information and guidance. To Jacopo's point about trust. I may trust someone to do one thing but not another. I may trust someone with a critical task that I would not entrust to another person who might be technically capable of doing it. As a project manager, I may trust someone to work on a particular part of an application but not on the data access. For the project to grow, the people working on the framework are going to have to get used to the idea that total strangers will be committing code to the project. The sub-project structure allows this to happen in a controlled way. It also allows sub-projects to attract the "right" mix of people which would be a totally different set of skills than the Framework project would want. Each sub-project will develop a team personality based on the sub-project's mission and the type of people required to implement the mission. I would expect the framework sub-project to be "hard core" technical people who know a lot about databases, security, entity modeling whereas the e-Commerce team will have people who are very knowledgeable about taxation, payment system integration, shopping cart design, user experience, and end-user documentation. The Project Management sub-project will attract people who know a lot about billing for consulting companies, accounting firms and legal offices as well project management, workflow, issue tracking, user interfaces, web services, etc. I would expect some overlap since many of the people here are very senior and have skills in multiple areas but I suspect that most new people will start in one sub-project and "cut their teeth" there before joining another. If it is done right it also makes everyone's job a lot easier and should reduce the amount of ML traffic for each person. Ron On 02/10/2014 9:22 AM, Jacopo Cappellato wrote:In my opinion we should avoid reconsidering the idea of creating committers with limited access; instead I would prefer to invite committers when we trust them as individuals, when they have demonstrated the right attitude and skills to work in our community etc... and demonstrate enough technical skills for the work they have to do; even if it is limited to a subset of the OFBiz codebase they will get full access to the repos but of course they will limit their field of action to the area they know, without requiring us to enforce commit rights limitations. As I said this can only work if we trust them 100% as persons at first. Jacopo On Oct 2, 2014, at 2:30 PM, Jacques Le Roux <jacques.le.r...@les7arts.com> wrote:That's an interesting idea. Now it also means more administration and we are already a bit sparse on the volunteering front. A simpler solution the OFBiz project used was to allow write access to only parts of the repo. This was before the Apache era. We gave up this way of doing because it was not the Apache way. I have not read it all yet but for instance I read in https://community.apache.org/newcommitter.html <<There may be extraordinary cases where we want limited work-related commit access. This will be resolved during the vote discussion. >> I don't know how technically this is possible in OFBiz trunk and branches, apart maybe asking the infra team? Which would most probably faces a veto... Jacques Le 01/10/2014 16:46, Ron Wheeler a écrit :The sub-project is a very useful Apache tool for helping projects grow. http://db.apache.org/newproject.html is interesting reading. http://ant.apache.org/antlibs/ very minimal description about Ant sub-projects but we all use their work. http://lucene.472066.n3.nabble.com/Close-of-Apache-Lucene-s-Open-Relevance-sub-project-td4141160.html a note about the official closure of a sub-project - very clear about why and what closure means. http://en.wikipedia.org/wiki/Apache_Ivy another popular sub-project. Description implies that it started in incubation and graduated to a top-level package and then became a sub-project of Ant. http://icodebythesea.blogspot.ca/2009/04/apache-servicemix-kernel-subproject.html is an example of a sub-project moving between two top-level projects. The sub-project structure allows for more specialization within the project resources so that people who are wizards with databases, kernels, etc get to worry about data access, performance, scalability, reliability, security while others who have more domain interest get to worry about features, usability, graphic design, workflow, reporting without getting in each other's hair. It also ensures a clearer demarcation between framework, core ERP and modules. I suspect that it would clean up project communication since people could subscribe to the sub-project lists that pertained to their interests. It might be easier for the existing community to accept new committers if the new people were part of a sub-project and were not committing to the particular codebase (framework, core, etc.) that the current committers are working on. It probably would help clarify the documentation since there would be a much clearer separation of framework from core from modules since each sub-project would have its own section in the project documentation. Each sub-project would have a much better defined target audience so writing docs would be a bit simpler and the language and terminology could be more relevant to the target audience. Ron On 01/10/2014 10:17 AM, Pierre Smits wrote:Ron, In the past there was a WIKI page decribing who was interested and who was willing to work on what. I don't know whether that page still exists. In the past we also had a system of having committers dedicated and committed to a subset of the trunk. This should still be feasible. But for that you need more committers. And to get more committers, this project needs to solicit and accept more. Regards, Pierre Smits *ORRTIZ.COM <http://www.orrtiz.com>* Services & Solutions for Cloud- Based Manufacturing, Professional Services and Retail & Trade http://www.orrtiz.com <http://www.orrtiz.com/> On Wed, Oct 1, 2014 at 4:10 PM, Ron Wheeler <rwhee...@artifact-software.com <mailto:rwhee...@artifact-software.com>> wrote: A defined method of deciding what moves from the trunk to a release would solve this. Back to my previous comment about 1 person to test and 1 person to fix bugs (could be the same person I suppose) would be a good starting minimum. Ron On 01/10/2014 2:56 AM, Pierre Smits wrote: The excuse of using PROJECTMgr in an older branch (12.x, the latest stable release) and testing it against trunk and therefor not including it in a release of a newer branch, is a lame one. We are diligent about this, meaning that we do follow up against any potential new release branch in order to be able to migrate to the newer branch when there is something released. Pierre Smits *ORRTIZ.COM <http://ORRTIZ.COM> <http://www.orrtiz.com>* Services & Solutions for Cloud- Based Manufacturing, Professional Services and Retail & Trade http://www.orrtiz.com On Wed, Oct 1, 2014 at 7:45 AM, Jacopo Cappellato < jacopo.cappell...@hotwaxmedia.com <mailto:jacopo.cappell...@hotwaxmedia.com>> wrote: The fact that someone is using it in an older branch and testing it in trunk is not enough to guarantee it works well with 13.07; the trunk and 13.07 are very different codebases. Additionally, the "projectmgr" component has 0 unit tests; I am not sure about about its stability, but for example comments in code like the following don't make me feel super confident: <!-- temporary disabled because it caused a db lock with the checkProjectMembership in projectpermission services --> One more point to note: since the component has not been in the 13.07 branch, it didn't undergo the 1-year long stabilization phase where only bug-fixes are backported: for example, one month ago, with revision 1618313, it was modified by a big commit to replace a series of Freemarker built-ins operation that we decided to not backport to 13.07 but only keep in the trunk. Jacopo On Sep 30, 2014, at 11:19 PM, Ron Wheeler <rwhee...@artifact-software.com <mailto:rwhee...@artifact-software.com>> wrote: So, as far as is known from Pierre's testing, there is no work required to "stabilize and bug fix" the module prior to including it in 13.07.01? Anyone else have any comments on the work required to include it in 13.07.01? Ron On 30/09/2014 5:13 PM, Pierre Smits wrote: Ron, All, We use the latest released branch, meaning 12.x. We don't expose our customers to an unstable unreleased branch, that is still undergoing significant changes. But, we test our solutions against trunk. This enables us to identify issues and register them in JIRA. And supply patches when workload allows it. So yes, PROJECTMGR, SCRUM, etc work also in r13.x Regards, Pierre Smits *ORRTIZ.COM <http://ORRTIZ.COM> <http://www.orrtiz.com>* Services & Solutions for Cloud- Based Manufacturing, Professional Services and Retail & Trade http://www.orrtiz.com On Tue, Sep 30, 2014 at 10:22 PM, Ron Wheeler < rwhee...@artifact-software.com <mailto:rwhee...@artifact-software.com>> wrote: Are you using it with a 12.04 or 13.xx? What work is required to get it into 13.07? Ron On 30/09/2014 3:06 PM, Pierre Smits wrote: Yes, I also have a vested interest in keeping this (PROJECTMGR) in the releases. It is part of our ORRTIZ:COM solution portfolio for our customers and we use it internally. And I have contributed to the improvement of the component. We, at ORRTIZ:COM, even use an extension to the code base to ensure that it also works for fixed price and internal projects. This extension includes generating the gl transactions regarding the cost price of each hour registered regarding a project. We also use the LDAP component to connect to our directory server (Apache Directory Server). Regards, Pierre Smits *ORRTIZ.COM <http://ORRTIZ.COM> <http://www.orrtiz.com>* Services & Solutions for Cloud- Based Manufacturing, Professional Services and Retail & Trade http://www.orrtiz.com On Tue, Sep 30, 2014 at 4:39 PM, Ron Wheeler <rwheeler@artifact-software. com wrote: It would be for me since it is one of the components that I want to use. Perhaps the more knowledgeable people might want to share a bit more of the background of the feature. Is it in 12.xx.xx? Is it currently in the 13.07 branch and therefor currently part of the 13.07 versions that people have put in production or is it just in the trunk that people are putting into production? What are the issues that need to be addressed before it is "stabilized and bug fixed"? Do any of these issues pose a significant risk to the stability of the rest of the functionality? Is anyone using it in production? What are their opinions of the state of the code and the degree of risk? Is anyone prepared to take on the task of getting it "stabilized and bug fixed" to a point where it can be safely included? What is the estimate of the minimum effort required? Ron On 30/09/2014 9:58 AM, Mike wrote: Why not deploy it as another hot-deploy component? Is it considered a "core" ERP component? On Tue, Sep 30, 2014 at 2:59 AM, Pierre Smits < pierre.sm...@gmail.com <mailto:pierre.sm...@gmail.com>> wrote: Jacopo, Back then there were already strong objections to excluding components from the release. I recall that Hans also wanted to keep the SCRUM component in the release, as well as there were proponents for BIRT and other components. These are good additions to the feature set of OFBiz and may be in use already by community members. It would be best that you solicit the advice of the entire community before a decision on excluding components from any release is taken. This affects more participants in this project than just you and the committers. Regards, Pierre Smits *ORRTIZ.COM <http://ORRTIZ.COM> <http://www.orrtiz.com>* Services & Solutions for Cloud- Based Manufacturing, Professional Services and Retail & Trade http://www.orrtiz.com On Tue, Sep 30, 2014 at 11:49 AM, Jacopo Cappellato < jacopo.cappell...@hotwaxmedia.com <mailto:jacopo.cappell...@hotwaxmedia.com>> wrote: Ok, got it. The release process that the OFBiz community is following is based on a feature freeze phase, that for the 13.07 branch started more than one year ago, during which only bug fixes are backported. This is done in order to stabilize the branch before an official release is done. Since the "projectmgr" component has never been part of the 13.07 branch then it may be unsafe to include it now just before the release is issued. It would be better to discuss its inclusion in the upcoming new release branch where it could be stabilized and bug fixed. Regards, Jacopo
-- Ron Wheeler President Artifact Software Inc email: rwhee...@artifact-software.com skype: ronaldmwheeler phone: 866-970-2435, ext 102
