Hi Alex, Thanks for your input. Please see my responses inline On 2020/09/04 00:39:54, "Goodman, Alexander (US 398K)" <alexander.good...@jpl.nasa.gov.INVALID> wrote: > I am sorry for not responding to this sooner. I am tentatively voting -1 > against retirement (even though I think my vote makes little difference at > this point if it still counts).
It absolutely does make a difference as it enables you express your opinion. That always counts :) > Frankly, when I initially saw this lengthy series of emails I wasn’t sure how > I should respond since I am pretty torn between both sides. I am not really > sure if my vote holds weight anymore since there are now already 3 votes in > favor of retirement, but I want to clarify on some points that Lewis and > Jimmy have previously stated as reasons for voting in favor retirement, since > I think they do not cover the full story. This will be lengthy so you may not > feel the need to read through all of this, but in summary I agree that > activity is an issue, but I disagree with the characterization of some that > we have not been transparent enough about our future plans. > First, I want to elaborate more on the refactoring that Kyo has mentioned. > This was an effort I had initially started in early 2018. This was originally > started by internal funding opportunities at JPL to create a replacement for > OCW which would use more modern libraries (xarray and dask), since it would > not only make it easy for us to handle big data use cases as we had long > desired but it would also make it easier for us to maintain the codebase in > the long run since it would make it much smaller. At JPL we have internally > called this prototype the “Big Climate Data Pipeline” (bcdp), and per JPL > policy it was not initially open source. We had to formally go through a > process (applying for a New Technology Release (NTR) and receiving approval. > We did not receive the formal approval to release BCDP to open source until > late last year. In the end, since I am the only person who is actively > developing it, I decided it would be more convenient to release it in its own > github repos itory separate from the main one (apache/climate) as this would allow me to make new releases very quickly which was necessary for the work I was doing in testing it on a Kubernetes cluster. Maybe it’s my misunderstanding, but some of the previous conversation seems to imply that we were not being transparent and keeping everything a secret which is not the case. We discussed BCDP many times on the mailing list and the plan to eventually transition the BCDP codebase was also made into a formal JIRA issue right around the time BCDP was released to open source for the first time. > https://github.com/bcdp/bcdp > If taking this approach goes against ASF principles, then I apologize, but I > do not recall being told that this was the case. There si no issue with any of the above Alex. The issue is the Apache OCW community and codebase is extremely quiet and we are therefore proposing it for retirement. This entire 'thing' is about community. Without community there is no real project. It's as simple as that. Nothing you see to have done is 'wrong' or intentionally going 'against' anything. > So with all that out of the way, let’s talk about the other issue which is > our low activity. This is an undeniable truth and reality that I don’t see > changing anytime soon whether BCDP gets integrated into OCW or not. My stance > from our very discussion on this last year hasn’t changed, and I encourage > anyone who has read this far to also see what I have said back then: > http://mail-archives.apache.org/mod_mbox/climate-dev/201911.mbox/browser > The amount of discussion on this mailing list is being used as one metric of > activity which I think is understandable given that this is an open source > project, but in practice we have only two core developers (me and Kyo) who > both work at the same place. We have tried our best to make our plans > transparent by making the appropriate JIRA issues, but at the end of the day > we have also had many offline discussions with each other about the project > out of convenience. We’ve had one non-JPL person (Michael Anderson) > contribute every now and then with a few bug fixes, but otherwise this has > been the reality of this project for a very long time. I don’t know if the > expectation for activity is for us to be as vibrant as we were in 2012-2015, > but that’s pretty unrealistic given that we just don’t have sufficient time > or funding, and that is also one of the reasons BCDP integration has been so > slow, and I don’t see this changing anytime soon. This happens Alex. People move on, some projects survive and other don't. Software sustainability is a huge issue. It's no ones fault here. Proposing OCW for retirement has nothing to do with anyone doing anything wrong. It is simply a realization that the community has seen a sustained period of inactivity which indicates that no-one is using the software or actively developing it. > So with all that said, although I currently am voting against retirement, I > would like to stress that I am just a PMC member and am ignorant on many of > the policies for ASF projects, and am unsure if OCW is the right fit given > that we are ultimately a small project with a niche userbase, and I can’t > really argue the fact that our activity is low. I would appreciate if Lewis > could provide his take on this question one more time. There are multiple metrics one can use to gauge community health. Low mailing list/comms activity is just one. At the ASF, typically the addition of Committers and/or PMC members is another. In the case of OCW the facts are as follows - No new PMC members. Last addition was Ibrahim Jarif on 2016-04-25. - No new committers. Last addition was Christopher Douglas on 2016-04-26. To many of us, this indicates that the project has not been attracting new committers or PMC members for atleast 4 years. Let's also look at project activity in terms of software releases. 1.3.0 was released on 2018-04-23. This has over 2 years ago and we've seen very little activity since then. At this stage, even with the work you are doing on BCDP, I still don't think that this justifies OCW continuing the way it has. Does this make sense? Do you agree/disagree with any of this? Thank you lewismc
