Thanks,   Jason..    I'll look into that.

+1 on Dave's comments -- the process shouldn't be that heavy-weight.
Yes -- there *should* be Jira tickets for anything not trivial,  but
should be at the discretion of the PR reviewer whether to request it
or not..

-dan

On Wed, Jul 19, 2017 at 7:55 AM, Jason Tucker <[email protected]> wrote:
> Re: the github PR trigger thing
>
> It looks like those hooks are available:
> https://blogs.apache.org/infra/entry/github_pull_request_builds_now
>
> Here's an example of it being used by the kafka project:
>
> https://github.com/apache/kafka/pull/3546
>
> Once those status checks are being returned, then the status gates can be
> turned on under branch protection.
>
> __Jason
>
> On Tue, Jul 18, 2017 at 10:33 PM, Dave Neuman <[email protected]> wrote:
>
>> I personally don't think that we should require a Jira for every single
>> PR.  I think blanket statements like "you must create a Jira ticket for
>> each PR", while made with good intentions, are too hard to enforce, don't
>> take every situation into account, and are sometimes a deterrent to new
>> contributors.  If Joe somebody from the internet wants to submit a PR, we
>> should welcome it and not say "we are not accepting this without first
>> creating a Jira account and then creating a Jira ticket".
>>
>> The changelog thing is a real problem.  We had a plan to have an
>> auto-generated changelog, but that was reliant on us moving to full github
>> which we haven't done yet.  I think maybe the changelog conversation might
>> need to be re-visited.
>>
>> Have you taken a look at our contributing guide[1] on github?  It already
>> outlines a lot of the things you are proposing here and I think its the
>> best place to have our PR requirements.
>>
>> I like Jason's idea, though not sure if it is possible.
>>
>> [1]
>> https://github.com/apache/incubator-trafficcontrol/blob/
>> master/CONTRIBUTING.md
>>
>> Thanks,
>> Dave
>>
>> On Tue, Jul 18, 2017 at 8:08 PM, Jason Tucker <[email protected]>
>> wrote:
>>
>> > Do you know if the the apache jenkins instance supports PR-triggered
>> > builds? If so, then repo branches can be protected so that PRs can only
>> be
>> > merged if the build status is successful. Might be a good safeguard to
>> have
>> > in place, if we have the ability to enable it on the jenkins side.
>> >
>> > __Jason
>> >
>> > On Wed, Jul 19, 2017 at 12:19 AM, Gelinas, Derek <
>> > [email protected]>
>> > wrote:
>> >
>> > > As the project grows in complexity and the speed of updates, we're
>> > finding
>> > > a real need for changelogs and a reduction in the number of commits.
>> At
>> > > this time, creating a changelog over even a short period of time is a
>> > > tedious and time consuming activity, and it is making even incremental
>> > > updates difficult for us here at Comcast.
>> > >
>> > > I would like to propose the following:
>> > >
>> > >
>> > >   *   Each PR must have a linked jira issue.  This is as simple as
>> > > creating the ticket and then adding [TC-XXX] to the beginning of the PR
>> > > title.  The jira ticket title should briefly describe the fix or new
>> > > feature in a manner which lends itself to inclusion in a changelog.
>> > >   *   Large changes should have as much information as possible about
>> the
>> > > nature of the change and how this change affects the system.
>> > >   *   Testing should be done on all changes before the PR is submitted!
>> > > Please do not submit anything that has not been fully tested,
>> regardless
>> > of
>> > > how simple or obvious it may seem.  You are the first line of defense
>> > > against bugs!  If the work is not yet complete, be sure to add [WIP] to
>> > the
>> > > title of the PR.
>> > >   *   Commits should be squashed as much as possible before the PR is
>> > > submitted.  27 commits for minor changes make review and later analysis
>> > > very difficult.  Commit messages should be descriptive so it is clear
>> > what
>> > > was changed.
>> > >   *   New features or changes to functionality should include
>> > > documentation changes whenever possible.
>> > >   *   Integration testing should be included whenever possible - we now
>> > > have automated integration testing, so this is another valuable method
>> to
>> > > isolate bugs before they hit the field.
>> > >   *   When in doubt, be clear about what you are doing!
>> > >
>> > > I would also like to propose that if these requirements are not met
>> > > without specific reasoning, PRs should be rejected until they are
>> > corrected.
>> > >
>> > > Derek
>> > >
>> >
>>

Reply via email to