calcite-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jacques Nadeau <jacq...@apache.org>
Subject Re: The art of the commit comment
Date Fri, 31 Jul 2015 17:53:35 GMT
Fair point.  Although 2 is better than 3 (and actually I'd count only one
of those since rebasing does not alter the comment.

:D

Realistically, the "close pr" would preferably be in the pull request
(which I know is hard given the order of operations).

For this particular point (author paren), having a human repeat information
that the system already manages seems error prone and excessive.  To me, it
is like adding (7 files and 400 lines changed) to the end of the commit.
We could do it, but why add that manual task.

On Thu, Jul 30, 2015 at 7:49 PM, Julian Hyde <jhyde@apache.org> wrote:

> The committer has to edit the commit already, to add the “Close
> apache/incubator-calcite#…” line. And quite often to rebase.
>
> > On Jul 30, 2015, at 7:40 PM, Jacques Nadeau <jacques@apache.org> wrote:
> >
> > My goal is to avoid the expectation that a committer should be manually
> > editing every commit.  Ideally we should get to a point where that is
> > unnecessary.  Proposed patches/pull requests should be mergeable as is.
> >
> > Looking at both git log and github [1], an author is given clear credit
> > when different than committer.  I think the best type of credit.  It is
> the
> > way people's github profiles get associated with their work.  We should
> > require commits to have author tags to provide maximum recognition to
> > contributors.
> >
> > [1]
> >
> https://github.com/apache/drill/commit/a219f8784c55ce3bc15b9bb3a19d7b33e4021c00
> >
> > (note how Hsuan is given credit even though Parth is committer)
> >
> > On Thu, Jul 30, 2015 at 7:31 PM, Julian Hyde <jhyde@apache.org> wrote:
> >
> >> Not necessary, but where’s the harm in giving contributors obvious
> credit?
> >>
> >> (The change history in
> >> http://calcite.incubator.apache.org/docs/history.html <
> >> http://calcite.incubator.apache.org/docs/history.html> is culled from
> the
> >> git log, with just a little massaging to convert jira case numbers into
> >> links, so contributors get called out there too.)
> >>
> >>> On Jul 30, 2015, at 6:13 PM, Jacques Nadeau <jacques@apache.org>
> wrote:
> >>>
> >>> I don't follow.  If the author tag in the commit is correct, why is it
> >>> necessary to also add it in parentheses?
> >>>
> >>> On Thu, Jul 30, 2015 at 11:35 AM, Julian Hyde <jhyde@apache.org>
> wrote:
> >>>
> >>>> I agree with what Vladimir and Jinfeng have said.
> >>>>
> >>>> We should make it clear that we prefer contributions via pull request
> >> over
> >>>> patches. Then the authorship is already in the contribution.
> >>>>
> >>>> But if the contribution comes via a patch, I like Vladimir’s
> suggestion
> >> to
> >>>> use use “git —amend —author”.
> >>>>
> >>>> We had a similar discussion a few months back. James Taylor
> suggested[1]
> >>>> adding the author’s name in parentheses, which is what they do in
> >> Phoenix
> >>>> and HBase. I still think that is the right policy. Even though it is
> >>>> redundant, it gives people credit, and that helps to build community.
> >>>>
> >>>> Julian
> >>>>
> >>>> [1]
> >>>>
> >>
> http://mail-archives.apache.org/mod_mbox/calcite-dev/201412.mbox/%3CCAAF1JdjW2psSPLEGyM3YA_bCTVcNWTvNiwLHpL-CJ4-CFy-mbA@mail.gmail.com%3E
> >>>> <
> >>>>
> >>
> http://mail-archives.apache.org/mod_mbox/calcite-dev/201412.mbox/%3CCAAF1JdjW2psSPLEGyM3YA_bCTVcNWTvNiwLHpL-CJ4-CFy-mbA@mail.gmail.com%3E
> >>>>>
> >>>>
> >>>>
> >>
> >>
>
>

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