mxnet-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Suneel Marthi <smar...@apache.org>
Subject Re: [VOTE] tracking code changes with JIRA by associating pull requests
Date Tue, 27 Feb 2018 22:13:17 GMT
On Tue, Feb 27, 2018 at 10:50 PM, Nan Zhu <zhunanmcgill@gmail.com> wrote:

> Thanks, Suneel!
>
> the vote still remains sense on its major points
>
> "
> 1. most of PRs should be titled as [MXNET-???][MODULE_NAME] PR short
> description, where MXNET-??? is the JIRA ID, MODULE_NAME can be python,
> scala, symbol, etc.
>

Any reason u need the [MODULE_NAME] in there - I would -1 that

[MXNET-XYZ] is unique enuf to identify the specific module - not to mention
that the different modules can be setup to label each jira - so
[MODULE-blah] is unnecessary.


> 2. only the very tiny fix, e.g. fix a typo, remove some unused variables,
> or the hotfix which brings the broken build back to track, can be filed
> without JIRA ID in title,
>

Agreed - and in this case the convention has been to use [NO-JIRA] in title.

> "
>
> though we do not need additional efforts to make it happen,
>
> the only thing we need to get a consensus on is that, we need to use JIRA
> to track work items and title PRs with JIRA ids
>

What ??? elaborate please??

>
> Hi, all, a friendly reminder, the vote will be ended at 12:00 p.m. on this
> Friday
>
>
> Best,
>
> Nan
>
>
>
> On Tue, Feb 27, 2018 at 1:44 PM, Suneel Marthi <smarthi@apache.org> wrote:
>
> > Suggest you see how other projects are doing it - Flink, Kafka or any
> other
> > project.
> >
> > Yes u r right.
> >
> > When u make a github PR with PR label in title like [Flink-3456] for eg:
> -
> > that way the corresponding JIRA - Flink-3456 here would be automatically
> > updated.
> >
> > On Tue, Feb 27, 2018 at 10:28 PM, Nan Zhu <zhunanmcgill@gmail.com>
> wrote:
> >
> > > Hi, Suneel,
> > >
> > > how can we enable it? when we titled JIRA id in pull request, it will
> > > synchronized automatically?
> > >
> > > Best,
> > >
> > > Nan
> > >
> > > On Tue, Feb 27, 2018 at 1:23 PM, Suneel Marthi <smarthi@apache.org>
> > wrote:
> > >
> > > > All projects on Apache have Jira <---> Github integration in place.
> > > >
> > > > So its a solved problem - look at Flink, Kafka, Mahout, OpenNLP,
> > > > PredictionIO and every other Apache project - all of them have this
> > > > working.
> > > >
> > > >
> > > >
> > > > On Tue, Feb 27, 2018 at 8:35 PM, Steffen Rochel <
> > steffenrochel@gmail.com
> > > >
> > > > wrote:
> > > >
> > > > > Nan - have you looked at plugin's to make the integration and
> > > > > synchronization between Jira and github easier? E.g.
> > > > > https://www.atlassian.com/blog/jira-software/connecting-
> > > jira-6-2-github
> > > > f
> > > > > Ideally one has one button in github to create a Jira and
> afterwards
> > > > > changes on either github or Jira get synchronized.
> > > > > What tools is ASF infra recommending?
> > > > > Have you used
> > > > > https://github.com/apache/spark/blob/master/dev/github_jira_
> sync.py
> > > and
> > > > > what is the recommended use case? How do get github issues updated
> > from
> > > > > Jira?
> > > > >
> > > > > Steffen
> > > > >
> > > > > On Tue, Feb 27, 2018 at 10:31 AM Indhu <indhubharathi@gmail.com>
> > > wrote:
> > > > >
> > > > > > +1 to the proposal
> > > > > >
> > > > > >
> > > > > > On Tue, Feb 27, 2018, 9:20 AM Nan Zhu <zhunanmcgill@gmail.com>
> > > wrote:
> > > > > >
> > > > > > > ideally,
> > > > > > >
> > > > > > > users report something fishy in github issue
> > > > > > >
> > > > > > > when confirmed that it is a bug or something to be improved,
we
> > > > should
> > > > > > > create JIRAs
> > > > > > >
> > > > > > > On Sun, Feb 25, 2018 at 9:31 AM, Chris Olivier <
> > > > cjolivier01@gmail.com>
> > > > > > > wrote:
> > > > > > >
> > > > > > > > i believe that JIRAs are “work items@, while github
issues
> are
> > > > more
> > > > > > like
> > > > > > > > reporting. at least this is what Infra sort of claimed.
> > > > > > > >
> > > > > > > > On Sun, Feb 25, 2018 at 9:30 AM Chris Olivier <
> > > > cjolivier01@gmail.com
> > > > > >
> > > > > > > > wrote:
> > > > > > > >
> > > > > > > > > +1
> > > > > > > > >
> > > > > > > > >
> > > > > > > > > On Fri, Feb 23, 2018 at 9:56 AM Marco de Abreu
<
> > > > > > > > > marco.g.abreu@googlemail.com> wrote:
> > > > > > > > >
> > > > > > > > >> Hello Nan,
> > > > > > > > >>
> > > > > > > > >> Good suggestion!
> > > > > > > > >>
> > > > > > > > >> "hotfix which brings the broken build back
to track"
> > > nitpicking,
> > > > > > but I
> > > > > > > > >> wouldn't consider this a tiny fix. There
should also be a
> > jira
> > > > > that
> > > > > > > > >> reported the build being broken, so that
shouldn't be a
> > > problem
> > > > to
> > > > > > > link.
> > > > > > > > >>
> > > > > > > > >> Very good idea with the automated script!
> > > > > > > > >>
> > > > > > > > >> How would we handle permissions? Which actions
are
> > > > non-committers
> > > > > > able
> > > > > > > > to
> > > > > > > > >> execute and in which cases would a committer
be required?
> > > > > > > > >>
> > > > > > > > >> How would we treat GitHub issues in future?
As a board for
> > > users
> > > > > to
> > > > > > > ask
> > > > > > > > >> usage questions?
> > > > > > > > >>
> > > > > > > > >> In order to improve user experience for new
developers,
> I'd
> > > like
> > > > > to
> > > > > > > > >> suggest
> > > > > > > > >> that more experienced people might create
jira tickets on
> > > behalf
> > > > > of
> > > > > > > > others
> > > > > > > > >> instead of telling them "please create a
ticket". I think
> we
> > > all
> > > > > > made
> > > > > > > > the
> > > > > > > > >> experience with people from it department
who blocked
> every
> > > > > request
> > > > > > > > until
> > > > > > > > >> a
> > > > > > > > >> ticket was created and assigned :P
> > > > > > > > >>
> > > > > > > > >> Best regards,
> > > > > > > > >> Marco
> > > > > > > > >>
> > > > > > > > >> Am 23.02.2018 6:07 nachm. schrieb "CodingCat"
<
> > > > > codingcat@apache.org
> > > > > > >:
> > > > > > > > >>
> > > > > > > > >> Hi, all
> > > > > > > > >>
> > > > > > > > >> To make the changes in code base more trackable,
> > > > > > > > >>
> > > > > > > > >> I would propose to link each PR with the
a JIRA from now
> on:
> > > > > > > > >>
> > > > > > > > >> 1. most of PRs should be titled as
> [MXNET-???][MODULE_NAME]
> > PR
> > > > > short
> > > > > > > > >> description, where MXNET-??? is the JIRA
ID, MODULE_NAME
> can
> > > be
> > > > > > > python,
> > > > > > > > >> scala, symbol, etc.
> > > > > > > > >>
> > > > > > > > >> 2. only the very tiny fix, e.g. fix a typo,
remove some
> > unused
> > > > > > > > variables,
> > > > > > > > >> or the hotfix which brings the broken build
back to track,
> > can
> > > > be
> > > > > > > filed
> > > > > > > > >> without JIRA ID in title,
> > > > > > > > >>
> > > > > > > > >> In JIRA side, to link the issue with an arrived
PR, we can
> > > run a
> > > > > > > script
> > > > > > > > >> like https://github.com/apache/
> > spark/blob/master/dev/github_
> > > > > > > > jira_sync.py
> > > > > > > > >> to
> > > > > > > > >> update JIRA status (can be run in Jenkins)
> > > > > > > > >>
> > > > > > > > >>
> > > > > > > > >> The benefits of applying such a flow include
(but not
> > limited
> > > > to)
> > > > > > > > >>
> > > > > > > > >> 1. facilitate the release process:
> > > > > > > > >>
> > > > > > > > >> e.g., as long as tagging each JIRA with the
correct
> affected
> > > > > version
> > > > > > > and
> > > > > > > > >> fixed version, the release manager can quickly
identify
> what
> > > are
> > > > > the
> > > > > > > > >> changes applied in this version; or we can
quickly
> identify
> > > > > whether
> > > > > > > > there
> > > > > > > > >> is any blocker issue in the JIRA
> > > > > > > > >>
> > > > > > > > >> 2. trackable changes
> > > > > > > > >>
> > > > > > > > >> any changes in MXNET can be quickly searched
through JIRA
> or
> > > > even
> > > > > > > > through
> > > > > > > > >> Google (JIRA looks like did something makes
it more
> > indexable
> > > by
> > > > > > > > Google),
> > > > > > > > >>
> > > > > > > > >>
> > > > > > > > >> If the vote gets pass,  the plan in the near
horizon
> > includes
> > > > > > > > >>
> > > > > > > > >> 1. update JIRA with the modules names
> > > > > > > > >>
> > > > > > > > >> 2. write runbook for release manager/committer
for
> releasing
> > > new
> > > > > > > > >> version/merging patches, as well as contributors
about the
> > > usage
> > > > > of
> > > > > > > JIRA
> > > > > > > > >>
> > > > > > > > >> 3. push the changes to the existing and coming
PRs (this
> > also
> > > > > needs
> > > > > > > the
> > > > > > > > >> collaboration across the whole committer
team)
> > > > > > > > >>
> > > > > > > > >> The vote will end at 12:00 p.m. of March
2nd, 2018
> > > > > > > > >>
> > > > > > > > >> Best,
> > > > > > > > >>
> > > > > > > > >> Nan
> > > > > > > > >>
> > > > > > > > >
> > > > > > > >
> > > > > > >
> > > > > >
> > > > >
> > > >
> > >
> >
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message