Fair enough. I do agree that multiple sources of truth is not great (even if avoiding it means some copy/pasta between PRs and JIRAs).
For now, I guess we can just ask people to open JIRAs, or file them for the casual contributors. Thanks all! -rgs On 8 June 2015 at 09:27, Patrick Hunt <[email protected]> wrote: > Agree with the single record of truth. All changes currently go > through Jira. It's part of the process and documented in the "how to > contribute" page: > https://cwiki.apache.org/confluence/display/ZOOKEEPER/HowToContribute > > The git mirror at github is just that, a mirror of svn and meant to be > a convenience. We don't use the PR process there, etc... > > Patrick > > On Mon, Jun 8, 2015 at 9:06 AM, Chris Nauroth <[email protected]> > wrote: > > FWIW, the JIRA requirement is typical of Apache projects. Most projects > > have a strong preference that this single record of truth lies on Apache > > infrastructure, hence the use of Apache JIRA and Apache's hosted git > > rather than GitHub. The idea is that a full permanent record of all > > project decisions resides in Apache infrastructure, maintained by the > ASF, > > and not subject to external forces like a company folding and needing to > > shut down its site. (I don't have any reason to suspect this of GitHub, > > but the point is that it's something outside of ASF control.) > > > > I don't know for sure that this is a required policy, but I wanted to > > point out that it's consistent with the Apache projects I've seen. > > > > --Chris Nauroth > > > > > > > > > > On 6/8/15, 5:19 AM, "Camille Fournier" <[email protected]> wrote: > > > >>I personally don't think that having a single record of truth and asking > >>people to use that record is asking too much. I'm not in favor at all of > >>removing the requirement for tickets to track work. Perhaps if we were > >>entirely in git hub it would be one thing but we aren't. > >> > >>C > >>On Jun 8, 2015 12:42 AM, "Raúl Gutiérrez Segalés" <[email protected]> > >>wrote: > >> > >>> Hi Camille, > >>> > >>> On 7 June 2015 at 14:59, Camille Fournier <[email protected]> wrote: > >>> > >>> > I personally think all PRs should have an associated JIRA. This is > >>>also > >>> the > >>> > requirement I have on my engineering team at work, and it seems > >>>totally > >>> > reasonable to me. > >>> > > >>> > >>> Thanks for sharing! Is it with a git pull request work flow though? > Take > >>> this ‹ small albeit important ‹ > >>> pull request for instance: > >>> > >>> https://github.com/apache/zookeeper/pull/32 > >>> > >>> I think it would be nice to avoid asking casual contributors to file > >>>JIRAs > >>> for those small patches so > >>> the impedance for contributing is reduced. > >>> > >>> Also, having the committer do the paperwork sounds like too much red > >>>tape > >>> given that > >>> the pull request is already pretty well documented. I at least would be > >>> very happy > >>> if we could just push those patches by referencing the PR in the > >>>comment, > >>> instead of a JIRA. > >>> > >>> I also suspect that eventually we might end up moving to git, so I > think > >>> there is value in allowing > >>> pull requests as a submission mechanism (for some cases?), since it'll > >>>make > >>> the eventual transition > >>> smoother and with less unknowns. > >>> > >>> > >>> -rgs > >>> > >>> > >>> > >>> > C > >>> > > >>> > On Sun, Jun 7, 2015 at 2:24 PM, Raúl Gutiérrez Segalés < > >>> > [email protected]> > >>> > wrote: > >>> > > >>> > > Heya, > >>> > > > >>> > > there's an increasing number of pull requests (PRs) coming through > >>> github > >>> > > (great! more contributions!). How do we deal with them? Do we need > >>>to > >>> > file > >>> > > a corresponding JIRA before we merge them or can we just reference > >>>the > >>> > PR? > >>> > > > >>> > > I rather not tax the contributors with having to file the JIRA, but > >>> > taxing > >>> > > the committer is also not great.. Thoughts? > >>> > > > >>> > > > >>> > > -rgs > >>> > > > >>> > > >>> > > >
