Hi all, since this thread has become a little dormant I'll try to sum up what I gathered from it so far. If no one has any objections I'll create a [VOTE] thread of the back of that. Not all of the below has been strictly speaking agreed upon, so please read critically :)
Fundamentally we will follow a Review-then-commit workflow with two notable exceptions: - Trivial changes (jira issue classified as trivial) - these still need to be posted for review but can be merged after 72 hours by lazy consensus - To fix a broken build - to be reviewed later, if possible Review requirements are separated, for code, tooling, website etc. the following applies: If the pull request was opened by a committer, the reviewer can be a non-committer (chosen by the committer). For pull requests opened by non-committers the reviewer must be a committer. For pull requests that change the content of training material the usual rules don't apply, as we probably won't have experts for all fields in the team from the get-go. Not sure how best to handle this, should we initially try going for simply 2 lgtm without any requirements around commiter status? For every commit there has to be either a jira or a pull request (both is fine too). I think that covers the basics, if everybody is happy with that fundamentally, I'd suggest voting on it at this level and I'll then draft a contributor guide based on that, which we can refine. Best regards, Sönke On Fri, Mar 15, 2019 at 6:33 PM Dmitriy Pavlov <[email protected]> wrote: > Hi Kenneth, > > Thank you for sharing, it is a very good topic. I've copy-pasted it at > least to Ignite > > https://cwiki.apache.org/confluence/display/IGNITE/How+to+Contribute#HowtoContribute-PeerReviewandLGTM > and > I agree with this approach for Training(incubating). > > Sincerely, > Dmitriy Pavlov > > пт, 15 мар. 2019 г. в 20:02, Kenneth Knowles <[email protected]>: > > > Here it is: > > > > > https://beam.apache.org/contribute/committer-guide/#always-get-to-lgtm-looks-good-to-me > > > > But it should also be in the beginner contributor guide, that they can > > review code as a way to get started. Of course, unsolicited feedback is > not > > always welcome, so a code author needs a way to connect with them and > > request their review. > > > > Kenn > > > > On Fri, Mar 15, 2019 at 12:22 AM Lars Francke <[email protected]> > > wrote: > > > > > Oh, that's a good idea. I remember the discussion, you're doing lots of > > > good things in the Beam project. That's a good way to get new > > contributors > > > on board and in the watchlist for new committers. > > > > > > As far as I can tell this is not written down in your Contributors > guide. > > > Do you have it somewhere else or is this "informal"? > > > > > > On Fri, Mar 15, 2019 at 5:02 AM Kenneth Knowles <[email protected]> > wrote: > > > > > > > Interesting idea about the 72 hour lazy consensus for committing. > > Vaguely > > > > related is that you can commit without review if the build is broken, > > > since > > > > it is emergency. But you are still required to open a pull request > and > > > > merge your own pull request, so everyone can see and comment on it > > later. > > > > > > > > Also, a thing that works really well for Beam is RTC where if the > > author > > > is > > > > a committer, the reviewer can be a non-committer. You could also have > > > lazy > > > > consensus after 72 hours for minor changes, but might not need it if > > you > > > > can trust a committer to choose a good enough non-committer reviewer. > > > > > > > > Kenn > > > > > > > > On Wed, Mar 13, 2019 at 10:36 AM Craig Russell <[email protected] > > > > > > wrote: > > > > > > > > > Hi Lars, > > > > > > > > > > I think we might codify this as RTC with an exception that if > someone > > > > > wants to commit a trivial or non-controversial patch, they still > need > > > to > > > > > post the patch for review and specifically request lazy consensus > > after > > > > 72 > > > > > hours or some other time we deem relevant to this project. > > > > > > > > > > Craig > > > > > > > > > > > On Mar 13, 2019, at 2:21 AM, Lars Francke < > [email protected]> > > > > > wrote: > > > > > > > > > > > > On Mon, Mar 11, 2019 at 4:34 PM Craig Russell < > > [email protected] > > > > > <mailto:[email protected]>> wrote: > > > > > > > > > > > >> > > > > > >> > > > > > >>> On Mar 11, 2019, at 6:35 AM, Lars Francke < > > [email protected]> > > > > > >> wrote: > > > > > >>> > > > > > >>> Okay, any other opinions? > > > > > >>> > > > > > >>> I don't necessarily need "Bylaws" but a good "Contributors > Guide" > > > is > > > > > one > > > > > >> of > > > > > >>> my pet peeves. So as soon as we have the website up and running > > I'd > > > > > like > > > > > >> to > > > > > >>> get some content up there. > > > > > >> > > > > > >> Yes, a contributors' guide is a great thing to add to the > > > user-facing > > > > > site. > > > > > >>> > > > > > >>> I'm usually in favor of review-then-commit but in our case I'm > > not > > > > sure > > > > > >> if > > > > > >>> that's always going to work. > > > > > >>> Ideally we'd have people contribute stuff that neither of us > is - > > > > > >>> technically I mean - able to review as we won't have experts > for > > > > every > > > > > >>> single field in the committership initially. > > > > > >>> But we can of course review for form and style etc. > > > > > >>> > > > > > >>> Other than that I'm in favor of requiring a +1 from another > > > committer > > > > > to > > > > > >>> commit anything but this can be painful for small things (e.g. > a > > > > single > > > > > >>> typo). For Hive we established a rule that for small changes > (at > > > > > >>> contributors discretion) after a delay/request for reviews of > 72h > > > or > > > > so > > > > > >> it > > > > > >>> can be committed without feedback. > > > > > >> > > > > > >> This sounds reasonable. It's a combination of RTC and CTR with > > lazy > > > > > >> consensus. Formally, I think it would be described as CTR with > > > > > discretion > > > > > >> to wait until another committer has reviewed a "large" change. > > > > > >> > > > > > > > > > > > > I understand what you mean but I (others might disagree) would > like > > > to > > > > > see > > > > > > it codified as RTC and have the commit without a review as the > > > > exception > > > > > > not the other way around. This has been a major pain for me as a > > > "lone" > > > > > > non-big-corporate contributor to various projects. You see things > > > > > happening > > > > > > without discussion on any public mailing list/Jira. > > > > > > > > > > > > > > > > > > > > > > > >> Regards, > > > > > >> > > > > > >> Craig > > > > > >>> > > > > > >>> Cheers, > > > > > >>> Lars > > > > > >>> > > > > > >>> On Tue, Feb 26, 2019 at 10:11 PM Sharan Foga < > [email protected]> > > > > > wrote: > > > > > >>> > > > > > >>>> Hi > > > > > >>>> > > > > > >>>> I'd be happy with having both models of Jira and Github in > place > > > > too. > > > > > >>>> > > > > > >>>> Thanks > > > > > >>>> Sharan > > > > > >>>> > > > > > >>>> On 2019/02/25 19:06:28, Lars Francke <[email protected]> > > > > wrote: > > > > > >>>>> On Mon, Feb 25, 2019 at 5:37 AM Kenneth Knowles < > > [email protected] > > > > > > > > > >> wrote: > > > > > >>>>> > > > > > >>>>>> On Fri, Feb 22, 2019 at 2:31 PM Lars Francke < > > > > > [email protected]> > > > > > >>>>>> wrote: > > > > > >>>>>> > > > > > >>>>>>>> > > > > > >>>>>>>> Bylaws have gone out of fashion and it’s generally > recommend > > > > that > > > > > >>>>>>> podlings > > > > > >>>>>>>> (and TLP) don’t have them and use the “default”. > > > > > >>>>>>>> > > > > > >>>>>>> > > > > > >>>>>>> Really? I didn't notice. > > > > > >>>>>>> > > > > > >>>>>>> > > > > > >>>>>>>> As the default is often unclear I've created a default > > simple > > > > set > > > > > >>>> that > > > > > >>>>>>>> graduating projects can use, [1] Legal and board have > > reviewed > > > > > >>>> this. > > > > > >>>>>>>> > > > > > >>>>>>> > > > > > >>>>>>> That's a good starting point. What I struggle with is that > > > every > > > > > >>>> project > > > > > >>>>>>> works slightly different in things like: Commit-then-review > > or > > > > > >>>>>>> review-then-commit or whether ReviewBoard needs to be used > or > > > > > patches > > > > > >>>>>> need > > > > > >>>>>>> to be attached to Jira etc. > > > > > >>>>>>> These rules are often not written down which makes it hard > to > > > > > >>>> contribute. > > > > > >>>>>>> This might be a particular pet-peeve of mine because I do - > > > > nature > > > > > >>>> of the > > > > > >>>>>>> job - lots of "drive by" contributions but I'd still love a > > > > clearly > > > > > >>>>>> defined > > > > > >>>>>>> set of rules on how we want to operate. > > > > > >>>>>>> > > > > > >>>>>>> This includes - and I don't actually know what works there > > > these > > > > > >>>> days and > > > > > >>>>>>> what doesn't - the Github use. > > > > > >>>>>>> If anyone knows what's allowed and possible it'd be great > if > > > you > > > > > >>>> could > > > > > >>>>>>> share. > > > > > >>>>>>> > > > > > >>>>>> > > > > > >>>>>> At this point, the GitHub pull request model is a de facto > > > > standard > > > > > in > > > > > >>>> my > > > > > >>>>>> mind. It works great with ASF infra. > > > > > >>>>>> > > > > > >>>>>> The most important thing being that you don't have to read a > > > > > project's > > > > > >>>>>> contribution guide to know how to *request* that they *pull* > > > from > > > > > your > > > > > >>>> fork > > > > > >>>>>> of their code. Great for drive-by contributions. If a > project > > > > > doesn't > > > > > >>>>>> basically follow this model, I don't trust them to accept > > > outside > > > > > >>>> input. We > > > > > >>>>>> should definitely use it. I would guess users will be much > > more > > > > > likely > > > > > >>>> to > > > > > >>>>>> also want to casually contribute bits here and there, > compared > > > to > > > > > >> other > > > > > >>>>>> projects. Let's make it easy and fun for them (and bring > them > > on > > > > > board > > > > > >>>> :-). > > > > > >>>>>> > > > > > >>>>> > > > > > >>>>> I agree that lots of projects use it but most projects I work > > > with > > > > > work > > > > > >>>>> differently (e.g. Hadoop, Kafka, HBase etc.). Some do have a > > > model > > > > > >> where > > > > > >>>>> both ways are accepted (Jira & Github). I myself am not the > > > biggest > > > > > fan > > > > > >>>> of > > > > > >>>>> Pull requests. I'm against using it as the only option. I > like > > > the > > > > > >>>>> "old-fashioned" way of attaching patches to Jira. It doesn't > > lock > > > > me > > > > > >>>> into a > > > > > >>>>> workflow provided by a 3rd party. I can download everything > > > offline > > > > > and > > > > > >>>>> prepare a review offline as well. That's not possible with > > Github > > > > (I > > > > > >> can > > > > > >>>>> download the code but I can't prepare a review for the > website > > > > > >> offline). > > > > > >>>>> > > > > > >>>>> But as I said: I definitely agree that it makes "drive-by" > > > > > >> contributions > > > > > >>>>> easier so I'm in favor of having both models. > > > > > >>>>> > > > > > >>>>> Lars > > > > > >>>>> > > > > > >>>>> > > > > > >>>>>> > > > > > >>>>>> Kenn > > > > > >>>>>> > > > > > >>>>> > > > > > >>>>> > > > > > >>>>>> > > > > > >>>>>> > > > > > >>>>>>> > > > > > >>>>>>> > > > > > >>>>>>> > > > > > >>>>>>>> Thanks, > > > > > >>>>>>>> Justin > > > > > >>>>>>>> > > > > > >>>>>>>> 1. > > https://wiki.apache.org/incubator/DefaultProjectGuidelines > > > > > >>>>>>> > > > > > >>>>>> > > > > > >>>>> > > > > > >>>> > > > > > >> > > > > > >> Craig L Russell > > > > > >> Secretary, Apache Software Foundation > > > > > >> [email protected] <mailto:[email protected]> <mailto:[email protected] > > > > <mailto: > > > > > [email protected]>> http://db.apache.org/jdo < > http://db.apache.org/jdo> > > < > > > > > >> http://db.apache.org/jdo <http://db.apache.org/jdo>> > > > > > > > > > > Craig L Russell > > > > > Secretary, Apache Software Foundation > > > > > [email protected] <mailto:[email protected]> http://db.apache.org/jdo < > > > > > http://db.apache.org/jdo> > > > > > > > > > > > > > > > -- Sönke Liebau Partner Tel. +49 179 7940878 OpenCore GmbH & Co. KG - Thomas-Mann-Straße 8 - 22880 Wedel - Germany
