Edgardo, You are correct that proper handling of pull requests is an important mechanism for the health of the community, bringing along new committers, and for adding benefit to the application. We've been flagging this as an important item to track and focus on in our quarterly board reports for the past six months. Frankly, I am extremely proud of the community and how much effort it puts into working with new and frequent contributors alike. This tweet and the response to it are things I think are wonderful examples of how this community is doing.
https://twitter.com/ktseytlin/status/786656997978419200 https://twitter.com/DanRosanova/status/786759566809702401 It is important though to also take a closer look at what is behind those PRs. As you mentioned main nifi developers and community engagement keep in mind that on a very conservative basis over more than 50 of those presently outstanding 77 PRs are from regular community contributors. The NiFi community follows a 'review then commit' model which means that no matter whether you're a new contributor or whether you're a PMC member and have been working on the project for more than 10 years your contributions go through the same process of peer review. In the past twelve months we have had about 92 different contributors and the month over month curve for unique contributors is growing. Therefore, it is going to be increasingly important we have a good community driven process in place. You brought up a couple of good points about the nature of the pull requests and I'll add some others. All of which we need to discuss and track over time. 1) They could be stalled by the originator 2) They could be stalled by license issues (very common) 3) They could be slowed by lacking alignment to a JIRA 4) We should consider some age-based kick out mechanism 5) Some PRs require far more testing than others and that could even include testing which requires money ...and there are other considerations. Some other important data points to keep in mind. In my opinion the number of open pull requests is an indicator worth observing but there are other indicators that are stronger. I just checked a few popular open source projects in terms of unique contributors to see their outstanding PR counts: Apache Spark = 472 Apache Kafka = 275 Apache Storm = 125 Apache Flink = 143 There are also some examples of really awesome communities that have been around for quite a while and have gotten the PR submission/review game down solid, HBase for example. We should and do look to them for examples of what we can be doing to get better. This community, like all apache way governed communities, is a reflection of what each of us puts into it. I am confident this community puts a great deal of effort and emphasis into honoring and encouraging contributors new and old and is working to improve that even more. Thanks Joe On Fri, Oct 14, 2016 at 8:30 AM, Edgardo Vega <edgardo.v...@gmail.com> wrote: > Joe - You are correct I was mentioning the PRs that are currently open. > > Regardless of how it happens reducing the count of open PRs I believe to be > extremely important. Maybe I was hoping that the release could be a forcing > function to make that happen. I believe that developers are more willing to > contribute when they see that their PRs will actually be able accepted and > merged into the code base. Having a low number of open PRs in progress is a > great indication that the main nifi developers are fully engaged with the > community. > > There are a few PRs that don't have any comments from committers at all. I > found one from August in that state. If that was my PR I don't think I > would be so willing to put another one in anytime soon. I do get that > sometime PRs get stalled by the originator, if so maybe a rule about > closing them after a certain amount of time or being taken over by a core > contributor if they think it worthwhile. > > I would like to shoutout to James Wing on my last PR he was quick to > review, provided great comments, testing, and even some additional code. It > was a great PR experience. > > Cheers, > > Edgardo > > > > On Thu, Oct 13, 2016 at 4:14 PM, Joe Percivall <joeperciv...@yahoo.com. > invalid> wrote: > >> Joe, I think you misread. Edgardo is referring to the Pull Requests that >> are currently open, not the tickets assigned to the 1.1.0 version. >> >> I think these goals (releasing 1.1.0 and cutting down the PR count) should >> be two different efforts. Doing a thorough job reviewing takes a >> significant amount of time from both the reviewer and contributor. In order >> to cut it down significantly would take much longer than a couple days. >> >> Also there has already been a lot of great new features and bug fixes >> contributed to the 1.X line and I don't think it's worth holding up a 1.1.0 >> release for tickets not assigned to this fix version. As an added bonus >> though, I think many of the tickets tagged as 1.1.0 have PRs already open >> so closing those will make a large dent in the PR count. >> >> >> Joe >> >> - - - - - - >> Joseph Percivall >> linkedin.com/in/Percivall >> e: joeperciv...@yahoo.com >> >> >> >> On Thursday, October 13, 2016 3:58 PM, Joe Witt <joe.w...@gmail.com> >> wrote: >> >> >> >> There are less than 30 right now. Many of the roughly 90+ JIRAs >> opened on 1.1.0 were easily dispositioned to 1.2.0 or closed or just >> had fix versions removed. >> >> We will need to have a push over the next bunch of days to deal with >> reviewing/merging/moving the remaining items. >> >> Thanks >> Joe >> >> >> On Thu, Oct 13, 2016 at 3:49 PM, Edgardo Vega <edgardo.v...@gmail.com> >> wrote: >> > Joe, >> > >> > There are 75 PRs currently open. Why not make a push over the next bunch >> of >> > days to get them closed and then cut the release after that. >> > >> > Cheers, >> > >> > Edgardo >> > >> > On Thu, Oct 13, 2016 at 12:44 PM, Joe Witt <joe.w...@gmail.com> wrote: >> > >> >> Team, >> >> >> >> There have been a ton of bugs fixed a few nice features. I would like >> >> to move to get Apache NiFi 1.1.0 release going pretty much based on >> >> where we are now and plan to move most tickets to a new Apache NiFi >> >> 1.2.0 version. We can try to get back on our roughly 6-8 week release >> >> schedule and shoot for a mid to late Nov release for NiFi 1.2.0 this >> >> way as well. Please advise if anyone has any other views on this. In >> >> the mean time I'll get the wheels in motion so you'll be seeing a lot >> >> of JIRA/issue updates to move version around. >> >> >> >> Thanks >> >> Joe >> >> >> >> On Thu, Oct 13, 2016 at 12:02 PM, Tony Kurc <trk...@gmail.com> wrote: >> >> > Sounds good Joe. I have no issue to you doing the rm'ing for it. >> >> > >> >> > On Oct 13, 2016 8:19 AM, "Joe Witt" <joe.w...@gmail.com> wrote: >> >> > >> >> >> Team, >> >> >> >> >> >> There are a lot of great fixes and improvements on the master line >> now >> >> >> and we're at a good time window to start pushing for a release. >> There >> >> >> are, however, about 90+ JIRAs assigned to 1.1.0 which are open. I'm >> >> >> going to go through them and remove fix versions where appropriate. >> >> >> >> >> >> I'm happy to take on RM task for this release though if someone else >> >> >> would like to take that on please advise. >> >> >> >> >> >> Thanks >> >> >> Joe >> >> >> >> >> >> > >> > >> > >> > -- >> > Cheers, >> > >> > Edgardo >> > > > > -- > Cheers, > > Edgardo
