Lurking on this, I have to underscore what Mark said. Andy, you are pretty correct on nearly every point that you made. But the things stated are more of refining your current process rather than taking RTC for current committers. You already had RTC with PRs from outsiders. If that slipped in, it just means that a trusted committer didn’t do their job. It happens. Breaking a trunk build for a day (or even a week) is ok. Thats why its trunk. I cannot tell you how many times I have downloaded a project’s trunk and things weren’t quite right.
Relative to what prompted this RTC discussion again, I think things got emotional and people slipped up afterwards. The beauty of all this is all parties shook hands and made up. Problem was more-or-less solved and the project was back on track. IMHO, taking on RTC is punitive. It means that you need to reset the way you do things because you cannot do it yourselves. Do you think you are at that point? It didn’t look that way to me… but its certainly possible based on what is being done behind the scenes. Just some food for thought. Jeff > On Jul 5, 2017, at 7:49 PM, Andy Gumbrecht <agumbre...@tomitribe.com> wrote: > > The main issue here is that both new and existing developers on the project > need breathing space in order to thrive and grow. > > The period between releases is for everyone and not just the few. It is > only 99.99% OK for one or two individuals. Everyone else seems to be > suffering behind closed doors or in silence, or fighting constant mobbing > to the point where 'our' fun project has become too tedious for many > people's free time. > > I'm not going to focus on the reasons behind the "Suffocating development > environment" thread, only that it was the web 'staging' environment used > for a review, but treated like it was North Korea production nuclear bomb > code. It should have been handled better. We found a resolution the long > way round (github web hosting). > > However, the situation has evolved where existing committers don't discuss, > create or assign tickets because they are literally mobbed or hijacked by > another committer within minutes. > That is currently so predictable that it has become a kind of un-funny joke > even outside of our community. Tickets are often created 'after' a commit > with a closed status. > > What needs to change is: > > Committers need to be able to take and work on a ticket in peace over > several days or even weeks, without being trumped due to impatience or the > notion of 'I know better'. > Many can only dedicate a finite amount of time, but still need to push > in-progress work regularly - Git makes that easier now. The review process > should be a fun and helpful thing. > > Committers and new contributors should be encouraged to take tickets. > > At most, impatience should be directed towards discussion, motivation and > encouragement - It's about team play on a global scale, not 'My way or the > highway'. > > It is often not viable to test the whole project for a small change - It > takes well over two hours. The buildbot is like our buzzer that says "fix > me" - Not revert me, or trash me, or trump me. > > Note to self: Why does the word 'trump' feel like it's been hijacked by > someone?... > > The 'buzzer' should be allowed to ring for a day or two, again not everyone > stays up the whole night ready to trash a breaking commit. They go to > sleep, get up, go to work, get home, eat... and then check the build if > they have time the next day. > > It is OK to break the build. Everyone gets to have a go at that and learn > from it. Over and over. We don't release broken builds, only the good ones > in-between. > > Any disagreement at any level goes to a vote. The majority wins. > > I think a trial RTC policy can help achieve these goals as it forces > community involvement - A good thing. > > Andy. > > > > On 5 July 2017 at 23:02, Jonathan Gallimore <jonathan.gallim...@gmail.com> > wrote: > >> Are you referring to the changes in the "Suffocating development >> environment" thread, or something else? In my view, the patch Andy applied >> last week had very limited review (1 person), and the revert had no review. >> We've seen contributions come in through GitHub PRs (which is great), but >> also applied directly to the repository by committers without further >> discussion (less great), effectively meaning just 1 reviewer - I'm not sure >> that's really the spirit of RTC. >> >> Jon >> >> >> On Wed, Jul 5, 2017 at 6:06 PM, Mark Struberg <strub...@yahoo.de.invalid> >> wrote: >> >>> As far as I recall the original issue was initially caused by applying a >>> PR. >>> That means we had this very issue with a commit which had RTC in place. >>> >>> Draw your own conclusions... >>> >>> LieGrue, >>> strub >>> >>> >>>> Am 05.07.2017 um 14:26 schrieb Romain Manni-Bucau < >> rmannibu...@gmail.com >>>> : >>>> >>>> Hmm, let put it in a raw way: can we skip the asf list on these >>>> discussions? Literally means can be use the way everybody uses for RTC, >>> ie >>>> github PRs *only*. If not I don't see the point to use it since >>>> contributors we got are mainly github/jira and I think it is natural >> as a >>>> contributor to use these media instead of the list. >>>> >>>> Can we somehow merge the github flow with the mailing in a smoother way >>>> than the jira integration - and even make jira optional? If not I'm >>> pretty >>>> sure it doesn't need any more evaluation, if we can then it can be >> great >>> to >>>> benefit from github well known flow. >>>> >>>> To rephrase it to maybe make it even more explicit: it is not about >>> making >>>> our - as committers - work easier but making contributions easier. >>>> >>>> >>>> Romain Manni-Bucau >>>> @rmannibucau <https://twitter.com/rmannibucau> | Blog >>>> <https://blog-rmannibucau.rhcloud.com> | Old Blog >>>> <http://rmannibucau.wordpress.com> | Github < >>> https://github.com/rmannibucau> | >>>> LinkedIn <https://www.linkedin.com/in/rmannibucau> | JavaEE Factory >>>> <https://javaeefactory-rmannibucau.rhcloud.com> >>>> >>>> 2017-07-05 13:23 GMT+02:00 Jonathan Gallimore < >>> jonathan.gallim...@gmail.com> >>>> : >>>> >>>>> As I see it, while the recent issue with the documentation was >> probably >>> the >>>>> trigger for discussing RTC on dev@, I think the general idea is >>> actually >>>>> to >>>>> get more discussion going on around features and fixes, and to >> encourage >>>>> more interaction in the review process. We are struggling as a >>> community in >>>>> that regard. The documentation issue might now be "fixed" by using >>> personal >>>>> html area usage, but I still think RTC is worthy of consideration. We >>> are >>>>> seeing some contributions come in from new contributors and we have an >>>>> opportunity to nurture them through this discussion and review >> process. >>>>> >>>>> Incidentally, if we trialled RTC and saw improvements, I'd vote to >> keep >>> it >>>>> after 3 months and not "safely return back". If we don't see >>> improvements, >>>>> I'd be trying to think of some other ideas to try. I think we'd all be >>> open >>>>> to other suggestions as well, but I'm of the view that if we don't try >>>>> something, then potentially nothing will change. >>>>> >>>>> Jon >>>>> >>>>> On Wed, Jul 5, 2017 at 9:25 AM, Romain Manni-Bucau < >>> rmannibu...@gmail.com> >>>>> wrote: >>>>> >>>>>> 2017-07-05 10:00 GMT+02:00 Gurkan Erdogdu <gurkanerdo...@yahoo.com. >>>>>> invalid>: >>>>>> >>>>>>> Romain>>>>@Gurkan: concretely nothing changed factually since we >>>>>> discussed >>>>>>> it and concluded we can't pay the overhead at the moment so why >>> pushing >>>>>>> it?I looked at the commit history from github ( >>>>>> https://github.com/apache/ >>>>>>> tomee/graphs/contributors). Only some couple of members provide >>>>>>> contributions in last couple of years. We need a more stable/healthy >>>>>>> community to increase the chance of long living the project. You are >>>>>> wrong, >>>>>>> the reason behind the such discussion is not related with prod, >>> website >>>>>> or >>>>>>> project source code. We are looking for some alternative solution >> (at >>>>>> least >>>>>>> temporarily) because of the mentioned problems. I suspect that this >>>>> type >>>>>> of >>>>>>> conflicts may occur in the future again. I am pushing this for the >>>>>> success >>>>>>> and future of Apache TomEE. I am not a PMC member or committer of >>> TomEE >>>>>>> project, but I just wanted to give my comments as ASF member. >>>>>>> >>>>>> >>>>>> Don't think so Gurkan, the problem was really bound to end user >> direct >>>>>> impact and we tackled it with the personal html area usage we need to >>>>>> document now. >>>>>> >>>>>> >>>>>>> Regards.Gurkan- >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> On Wednesday, July 5, 2017, 10:13:22 AM GMT+3, Romain Manni-Bucau < >>>>>>> rmannibu...@gmail.com> wrote: >>>>>>> >>>>>>> @Gurkan: concretely nothing changed factually since we discussed it >>> and >>>>>>> concluded we can't pay the overhead at the moment so why pushing it? >>>>>>> Concretely the issue was very particular in term of process cause >>>>>> affecting >>>>>>> almost directly our "prod" versus our project source doesn't and we >>> can >>>>>>> therefore tolerate more latency. And side note (probably some >> wording >>>>>> issue >>>>>>> but just to make it obvious if not): if it is to go back to the >> normal >>>>>>> process anyway after then we can gain these 3 months and already >> work >>>>> as >>>>>> we >>>>>>> and we'll do ;). >>>>>>> >>>>>>> >>>>>>> Romain Manni-Bucau >>>>>>> @rmannibucau <https://twitter.com/rmannibucau> | Blog >>>>>>> <https://blog-rmannibucau.rhcloud.com> | Old Blog >>>>>>> <http://rmannibucau.wordpress.com> | Github <https://github.com/ >>>>>>> rmannibucau> | >>>>>>> LinkedIn <https://www.linkedin.com/in/rmannibucau> | JavaEE Factory >>>>>>> <https://javaeefactory-rmannibucau.rhcloud.com> >>>>>>> >>>>>>> 2017-07-05 7:28 GMT+02:00 Gurkan Erdogdu <gurkanerdo...@yahoo.com. >>>>>>> invalid>: >>>>>>> >>>>>>>> Hi MarkThis is only for fixing the appeared (very important) >> problem >>>>> in >>>>>>>> the community. So, I don't see what will happen to the project in 3 >>>>>>> months >>>>>>>> period with RTC process? So, at least 3 months, every commit will >> be >>>>>>>> approved by the community via consensus. After that, we can safely >>>>>> return >>>>>>>> back to the normal process. >>>>>>>> >>>>>>>> Thanks.Gurkan >>>>>>>> On Wednesday, July 5, 2017, 1:15:31 AM GMT+3, Mark Struberg >>>>>>>> <strub...@yahoo.de.INVALID> wrote: >>>>>>>> >>>>>>>> RTC in my experience _only_ works on release branches, but is a >> total >>>>>>>> community killer on the mainstream branch (master, dev, whatever >> you >>>>>> call >>>>>>>> it). >>>>>>>> >>>>>>>> We usually don't have so many concurrent commits on the same topic. >>>>>> There >>>>>>>> was recently an exceptional case and it got resolved. >>>>>>>> Thus -1 >>>>>>>> >>>>>>>> Of course discussions might be done first. But not via PR but via >>>>> mail. >>>>>>>> Usually the devs have a good feeling about what is sensible and >> what >>>>>> not. >>>>>>>> For some deep change one usually sends a patch first for review. >> That >>>>>> is >>>>>>>> nothing we need to enforce - every good programmer will do just >> that! >>>>>>>> Otoh there are 99.99% of stuff which you just get done and commit >> it. >>>>>> And >>>>>>>> if there is something fishy, then it get's caught via the commit >> log >>>>>>> mails >>>>>>>> anyway. >>>>>>>> >>>>>>>> LieGrue, >>>>>>>> strub >>>>>>>> >>>>>>>> >>>>>>>>> Am 03.07.2017 um 10:05 schrieb Jonathan Gallimore < >>>>>>>> jonathan.gallim...@gmail.com>: >>>>>>>>> >>>>>>>>> On Mon, Jul 3, 2017 at 2:04 AM, David Blevins < >>>>>> david.blev...@gmail.com >>>>>>>> >>>>>>>>> wrote: >>>>>>>>> >>>>>>>>>> There’s a discussion on the private list on this topic, but given >>>>>> the >>>>>>>>>> recent thread I think it makes sense to move that here. >>>>>>>>>> >>>>>>>>>> The vote would be only on this question: >>>>>>>>>> >>>>>>>>>> - Is RTC worth trying for 3 months? (+1,+/-0,-1) >>>>>>>>>> >>>>>>>>>> I’ve seen some voices in favor, but do not want to propose a vote >>>>>>>>>> without a heads-up. Specifically, even if many people like the >>>>> idea >>>>>>>>>> we should talk about how we want to do it. >>>>>>>>>> >>>>>>>>>> # Review-than-commit >>>>>>>>>> >>>>>>>>>> For those that do not know, Review-than-commit is essentially >> what >>>>>>>>>> Github Pull Requests are. Prior to github, Apache describes them >>>>>> as: >>>>>>>>>> >>>>>>>>>> - Commit policy which requires that all changes receive consensus >>>>>>>>>> approval in order to be committed. >>>>>>>>>> >>>>>>>>>> I think we’ve seen evidence that: >>>>>>>>>> >>>>>>>>>> - Slowing ourselves down can be a good thing. >>>>>>>>>> >>>>>>>>>> - Moving ahead after discussion is a good thing. Discussion >>>>> should >>>>>>>>>> precede even the first commit. >>>>>>>>>> >>>>>>>>>> - More eyes and talk around commits can help documentation >>>>> efforts. >>>>>>>>>> >>>>>>>>>> - As 3 +1s are required, a one-to-one conversation with no one >>>>> else >>>>>>>>>> included is naturally discouraged. >>>>>>>>>> >>>>>>>>>> # Trial basis >>>>>>>>>> >>>>>>>>>> My thought is to go RTC for 3 months as a trial. After 3 months, >>>>> no >>>>>>>>>> action means we revert back to our present CTR. A new vote would >>>>> be >>>>>>>>>> required to continue RTC in any form, as-was or modified. >>>>>>>>>> >>>>>>>>> >>>>>>>>> Unless its obviously unanimous that everyone dislikes RTC at the >>>>> end >>>>>>> of 3 >>>>>>>>> months, I'd suggest we call a vote to decide how to proceed. Not >>>>>> quite >>>>>>>> sure >>>>>>>>> how that fits into +1/0/-1 however, so may be it should be a 3 >>>>> month >>>>>>>> trial, >>>>>>>>> followed by 2 weeks for review and discussion (during which we'd >>>>>> still >>>>>>> be >>>>>>>>> RTC) and then a vote? >>>>>>>>> >>>>>>>>> >>>>>>>>>> >>>>>>>>>> The trial-basis is to acknowledge that we are voting on a guess >> of >>>>>>>>>> potential benefits. This allows us to "try before we buy" and >> the >>>>>>>>>> vote really comes down to if we want to try. We need not make a >>>>>>>>>> decision based on other people's experience and have a means to >>>>> gain >>>>>>>>>> our own experience with a built-in escape clause that triggers >>>>>> lazily. >>>>>>>>>> >>>>>>>>>> RTC may sound like a good idea, but our implemention of it may be >>>>>> bad >>>>>>>>>> in practise. It may sound like a bad idea, but we may discover >>>>>>>>>> positives we hadn't anticipated. We don't currently know. >>>>>>>>>> >>>>>>>>>> # How would we do it? >>>>>>>>>> >>>>>>>>>> Some things that would be good to discuss: >>>>>>>>>> >>>>>>>>>> - How could we use github pull requests? Other communities do >>>>> use >>>>>>>>>> them and I suspect there are options we have not explored. >>>>>>>>>> >>>>>>>>> >>>>>>>>> I'd be in favour of that, as that process seems to be very well >>>>>> known. >>>>>>>>> >>>>>>>>> >>>>>>>>>> >>>>>>>>>> - Should all reviews be on the dev list? With Github PRs comments >>>>>>>>>> and JIRA comments, there needs to be a source of truth. >>>>>>>>>> >>>>>>>>> >>>>>>>>> I think both the discussion and review should happen on the dev >>>>> list. >>>>>>>>> GH/JIRA comments are fine in themselves, but there may be (should >>>>> be) >>>>>>>>> discussion on dev@ before a PR is opened, so having all that >>>>>>> discussion >>>>>>>> in >>>>>>>>> one place is important for me. Even if GH comments prove popular, >>>>> its >>>>>>> not >>>>>>>>> hard to copy/paste it to dev@ with a link. >>>>>>>>> >>>>>>>>> >>>>>>>>>> >>>>>>>>>> - Should we fully document the process before we try so we can >>>>> get >>>>>>>>>> the most value from a 3 month trial? >>>>>>>>>> >>>>>>>>> >>>>>>>>> I'd be in favour of discussing and documenting. >>>>>>>>> >>>>>>>>> >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> -- >>>>>>>>>> David Blevins >>>>>>>>>> http://twitter.com/dblevins >>>>>>>>>> http://www.tomitribe.com >>>>>>>> >>>>>>> >>>>>> >>>>> >>> >>> >> > > > > -- > Andy Gumbrecht > https://twitter.com/AndyGeeDe > http://www.tomitribe.com
