I don't think we need to assume the dysfunctions of other communities are operative here. I trust this community to summarize a hangout discussion sufficient for me to follow along if I can't make it, no problem.
My point of view is pretty close to Sean's. I can see that Gradle is preferred by many, and looks like a plausible way forward. It's taken me some effort to make progress with the almost inscrutable makefiles before. That said I think we should have a release (0.8 looks like) where both build systems are available and one can do the same set of build and test actions with either. Then we can drop one in 0.9. Totally fine with a consensus decision on which one. On Mon, Jul 14, 2014 at 8:59 AM, Sean Mackrory <[email protected]> wrote: > I personally like that idea, and it would probably be beneficial even if it > wasn't the mechanism by which we made the decision, but the whole reason > we're having this discussion is because of concern raised about the > decision not being made on the mailing list. I know of one project where a > design document was posted for discussion, a call was scheduled on the > mailing list with international toll-free numbers and a recording, and > plenty of time for discussion after the call before a decision was made, > and this was still considered too closed of a decision. I suspect a hangout > might also violate the same criteria :) > > Perhaps if someone very familiar with the gradle code were to post a > screencast of them walking through the code and demonstrating the build, > that would allow all decision making to be on the list and help to resolve > the concerns already discussed? > > > On Mon, Jul 14, 2014 at 9:06 AM, jay vyas <[email protected]> > wrote: > > > I agree wiith mark that we need to make it easy and individuals in the > > community need to be comfortable with the transition. I propose a > solution > > at the end o this email. > > > > Heres where we are at: > > > > - Realistically, the debate on which is "better" is not going to be very > > fruitful.... We all know the high-level pros and cons of both Makefile > and > > build.gradle . > > > > - Neither is perfect,but > 50% of community parties agree gradle is a > step > > forward. > > > > jay +.1 (im netural with a slight lean to gradle) > > cos, roman, sean, +3 (all are ready to move forward) > > mark effectively might say is a -1 > > > > ************* - But we don't want to leave anyone behind! > > ********************** > > > > But we also need to move fast ! We dont want a schizophrenic or forked > > community. > > > > SO HERE IS MY SUGGESTION: > > > > 1) we schedule a meetup, or a screencast - specifically to go through the > > gradle code - from A to Z - > > 2) We validate and build all of bigtop, using gradle, during the > > screencast. NO EXCEPTIONS. That way we are all 100% sure that it > works, > > and we see it in action. > > 3) After that screen cast, we ensure that we have a unified community > which > > can self-sufficiently administer the gradle based build - as soon as > > possible. > > 4) If all parties to (1) agree that they are now ready , we delete the > > Makefile forever, in a patch which updates the README file with excellent > > explanation of how the build.gradle works. > > > > This is the most rapid way to move the entire community forward in unison > > and prevent code rot of maintaining duplicate features . > > > > Mark, and others, how do you feel about this idea? > > > > > > > > On Sun, Jul 13, 2014 at 2:16 PM, Mark Grover < > [email protected]> > > wrote: > > > > > Thanks for starting this thread, Jay and Cos. > > > > > > Here are my thoughts: > > > Gradle may actually be a better choice for Bigtop. However, in my > > opinion, > > > Make should be kept around for a little while to make the transition > easy > > > for the community. > > > > > > I would go as far as saying that we shouldn't be deprecating Make right > > > now. In my opinion, it's best to have a transition period where we all > > > support both Gradle and Make. During this time, contributors and > > committers > > > work (albeit, with a little extra pain) with both and develop their > own > > > opinion on which tool is the best for the project. Some of us may > already > > > have developed such experience by using various tools in the past while > > > others may have not. The idea is to give all members of the community > an > > > opportunity to make such a decision for themselves and share them with > > the > > > community. > > > And, consequently, the official decision to deprecate a tool from the > > > project shouldn't happen before this transition period, but after. > > > > > > Hope that makes sense. > > > Mark > > > > > > > > > > > > On Sat, Jul 12, 2014 at 6:34 AM, Sean Mackrory <[email protected]> > > > wrote: > > > > > > > I'm not super familiar with Gradle yet, but here are my thoughts: > > > > > > > > - It can't be any worse than make. I mean, $$($(1)-deb). I'd much > > prefer > > > > debugging a more modern language, and as Cos points out, it's > intended > > > for > > > > a JVM ecosystem but can do other things when needed. > > > > - I'd prefer we leave Make around for perhaps one more release just > to > > > > really solidify the Gradle system a bit more. However if we keep it > > > > deprecated, I see no point in "maintaining" both, meaning that if > > people > > > > want to add new features to the build (several JIRAs going on for > that > > > > right now) - there's no need to keep adding that to the Makefile, > let's > > > > just keep the versions and metadata for new projects up to date. I > > don't > > > > believe we routinely run into bugs, so I doubt much work will have to > > be > > > > put in outside of bigtop.mk. > > > > > > > > > > > > On Fri, Jul 11, 2014 at 11:54 PM, Roman Shaposhnik < > > [email protected] > > > > > > > > wrote: > > > > > > > > > On Fri, Jul 11, 2014 at 6:01 PM, Konstantin Boudnik < > [email protected]> > > > > > wrote: > > > > > > I am temporarily putting the [VOTE] thread to halt and instead > > > starting > > > > > > [DISCUSS]. Per Jay's: > > > > > > > > > > > > BIGTOP-1314 highlights the fact that we now have 2 build systems: > > > > > Makefile > > > > > > and build.gradle. And the fact that the Makefile approach is now > > > > > > deprecated. > > > > > > > > > > > > As per Mark's suggestion in that JIRA, the future of Makefile > > should > > > be > > > > > > decided on the mailing list before the next JIRA comes out to > > further > > > > > > distance from Make and embrace gradle. > > > > > > > > > > > > Should we continue to support the Makefile builder? > > > > > > > > > > I'd be more than happy to get rid of it right after we release > Bigtop > > > > 0.8.0 > > > > > > > > > > Thanks, > > > > > Roman. > > > > > > > > > > > > > > > > > > > > -- > > jay vyas > > > -- Best regards, - Andy Problems worthy of attack prove their worth by hitting back. - Piet Hein (via Tom White)
