From: "olivier Heintz" <[email protected]> > Thank you Jacques for your comment. > > I added some comment in-line to clarify what i meant > > Le 07/12/2012 09:13, Jacques Le Roux a écrit : >> From: "olivier Heintz" <[email protected]> >>> The thread title is confusing for this discussion. >>> >>> I reformulate my last mail : >>> Sort from the more important to the less >>> 1) give a process to promote contribution. Contribution should be sent >>> before quality process review >> I see roughly 3 types of contribution >> 1) Bug fixes >> 2) Improvements of existing features >> 3) New features >> >> In OFBiz standard contribution process >> https://cwiki.apache.org/confluence/display/OFBADMIN/OFBiz+Contributors+Best+Practices >> 1) are straightforward => create a Jira, attach a patch >> 2) Don't need to be discussed 1st on the dev ML, except if the improvement >> is really a big change >> 3) Should always be discussed 1st on dev ML to avoid disappointments >> >> Those are OFBiz and not Apache conventions, but could still be used as >> template for Apache OFBiz Extras > You are right, i did not detailled enough about contribution types > 1) Bug fixes, current ofbiz process is clear > 2) Improvements of existing features with a good quality level, current > ofbiz process is clear > 3) New feature (small or large) not already done, current ofbiz process > is clear > 4) New feature (small or large) already developped within contributor > project. > I wanted to insist on the necessity to have a way to contribute. > Obviously, it must be identified as such.
I don't see the difference between 3 and 4. From a committer POV it's the same. So I must be missing something. You mean in the only context of Apache OFBiz Extras? >>> 2) Improve OFBiz Quality, and so accept only contribution with quality >>> review >> In OFBiz standard contribution processn this is already the case, a >> committer should always review before committing. >> In OFBiz we use the Review Then Commit (RTC) procedure and not the Commit >> Then Review (CTR) http://www.apache.org/foundation/glossary.html > I wanted to point out the fact that all contributions have to respect > quality rules. For instance, every new service (other than auto-entity) > must have a Junit test provided. That would be wonderful, but so far we never reached this stage ("must have"). Also I don't think the OFBiz project wants to force people to provide Junit test for each feature. And finally there are already a lot of contributions waiting. The contributors should 1st understand that not only committers can review and test. When a contribution is reviewed and/or tested by another contributor than the author the committers work is much reduced and the quality is improved. We are still in the slimdown phase effort. And this means that we (committers) favour bug fixes. >>> 2.1) Quality for an ERP should be for technical and functional at the >>> same level >>> 2.2) Quality criteria must be clear and well defined >>> 3) be more modular than component level, to be able to measure quality >>> more easily and precisely >> At this stage I wonder if your discussion (Apache OFBiz Extras? Still not >> quite clear in the subject ;o) is not implicilty related to Neogia addons? > only talking about OFBiz I then wonder how (resources) you envision to reach such a challenge... We have already some difficulties to cope with the curren contributions. How would you decide on quality? To be frank, this is freightening to me. I foresee administrative work in your intention, but I must be wrong, right? So far we decided on quality by peer review and lazy consensus, what would you want to add? >>> 4) slim down ofbiz and put not mandatory function in an option area >> slimdown: https://issues.apache.org/jira/browse/OFBIZ/fixforversion/12320551 >> Apache OFBiz Extras: >> http://code.google.com/a/apache-extras.org/hosting/search?q=label%3aOFBiz >> >>> 5) give a clear process to validate a contribution. Multiple status with >>> a clear definition for each. Don't you fear a too much admistrative work? >>> 6) give a plan with timelines to classify, on quality criteria, each >>> existing apache-ofbiz functions >> Seems a bit complicated :) Our limited community cannot reasonably sustain >> too much "paperwork". This has already been expressed by experienced OFBiz >> committers about this subject. We just need to keep things realistic... > I tried to explain that OFBiz slim-down process have to keep going > beyond the components, and for this purpose we should start by > discussing function by function. > For each function, the quality-level should be estimated. I think that > this kind of contribution could also help the community. How and by who the "quality-level should be estimated" is the basic question. Not even sure the OFBiz team agree about that, sounds like a tremendous work for "existing apache-ofbiz functions" >> >>> 7) add more functions // enhance quality of existing functions // move >>> function from one area to an other (kernel, optional function at hight >>> quality level, optional function on quality review process, ...) >>> >>> so, first clarification : ofbiz-extra is a mean and not an end >>> second clarification : Apache-ofbiz must be for all hight quality ofbiz >>> piece, kernel or additionals functions. >> Totally agreed >> >>> To be very clear, In My Opinion, the main advantage for ofbiz-extra is ONLY >>> 1) to be able to give a commit authorization for new contributor, to >>> motivate them to share their current realization Actually this is not related to commit authorization at all. It's just a place to share things between contributor (which include committers). We can already do the same things by other means. It's just a bit more official and (hopefully) better organised. >>> 2) to have a unique place for contribution before being evaluate by the >>> community on quality review process. Yes that's it. I see you much insist on quality. Do you mean that we should use a tools like Selenium or (better IMO) webdriver? Should we dedicate specific human resources for OFBiz quality assurance? This would be a good thing to discuss at least. Nowadays all serious projects are guaranteed by a quality process with dedicated person/s. Just stumbled upon http://wiki.openoffice.org/wiki/Quality_Assurance. But it seems to me that infra does not provide tools and especially enough resources for that. So would this be done externally, something to discuss... Well, when I really think about it, as I said it's freightening :D >> Still this seems a bit complicated to me. The higher the barriers you put, >> the less contributions you will get >> >>> If we want a hight level of quality, we should have process to be able >>> to remove a function from OFBiz-Kernel or "optionals functions", >>> BECAUSE all code on trunk should be evaluate with the same criteria, >>> existing from a long time is not a quality criteria. It's not because >>> something was with a hight quality level that it is always with it. >> Sounds right indeed >> >>> Last point, maybe quality was not considered as a priority by very many >>> or we'd see more people (committers and non-committer contributors) >>> working on it. >>> But I'm sure it is only related to the development phase where was OFBiz >>> - increase number of function - >> Yes I agree, earlier, and even last, years were more in this mood. Now that >> OFBiz is "mature" less new features are proposed. But I think also that >> something else happened/is happening. I'm not yet sure what, but it's like >> OFBiz has a smell... Let me clarify "OFBiz has a smell..." I refer to Frank Zappa: "Jazz isn't dead. It just smells funny." I say that because I see less activity and I wonder why. Of course David's departure is a part of the cause, but it does not explain all. >>> Now I'm sure many of us to be confident that the quality will enable us >>> to increase our business. >> Yes agreed, we already focus on higher quality than more features. This must >> no say that no new features should appear... > +1 ;-) Now it seems that Jacopo proposed something new recently (to address Paul's and others concern). I can't find it, but IIRW it was about extras in repo but not in releases, an intermediate stage. This sounds like an adequate proposition to me (not related to quality) Jacques >> Jacques >> >>> Le 30/11/2012 09:13, Paul Piper a écrit : >>>> Unfortunately, I would have to second David's opinion. As mentioned in the >>>> other mailing-thread, I cannot see any benefit from migrating parts of the >>>> source into a google repository. Instead I think that the effects will >>>> result in lesser quality product, not higher ones, as discussed here: >>>> http://ofbiz.135035.n4.nabble.com/Slim-down-effort-current-situation-td4637617.html#a4637828 >>>> >>>> >>>> So I would argue that it is best to maintain everything in the same trunk >>>> as >>>> part of the ASF. I would rather like to discuss less enforced guidelines or >>>> subproject structures for the apache extras subproject so that those can >>>> reach maturity through other means. Don't get me wrong: I do think that a >>>> lot of the points & questions you raise are valid, Olivier, and I also >>>> agree >>>> that we need a structure that would be beneficial to the subproject... but >>>> within the same svn trunk and apache ofbiz brand. >>>> >>>> That being said: I like the condition-set you gave to identify product >>>> maturity. If we can extend the 5week rule to something more suitable for >>>> this community (5 weeks is rather short), I believe that those could easily >>>> be adapted for a full subproject. >>>> >>>> >>>> >>>> -- >>>> View this message in context: >>>> http://ofbiz.135035.n4.nabble.com/Summary-of-ApacheCon-Eu-conference-Why-ofbiz-extra-tp4637910p4637949.html >>>> Sent from the OFBiz - User mailing list archive at Nabble.com. >>>> >
