ambari-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dmytro Grinenko <dgrine...@hortonworks.com>
Subject RE: New JIRA - Commit messages
Date Fri, 12 Jan 2018 13:09:26 GMT
More -1 than +1, below is why:

There is only one issue with Author/Committer fields. Obviously it is normal situation when
people working from several devices or environments
which were configured not at the same time. So one environment could have, for example Dmitry
Grinenko as an author; another dgrinenko .

With mail it is same situation, while previously were used same mail address for commits,
now it could be in @apache domain, @users.noreply.github.com or one from 
primary mail configured in github account settings.

All such thinks making search by author and email more complex, and probably hell if you doing
this in some tools. And in this situation maintaining same style naming 
convention for summary field could help

-----Original Message-----
From: Myroslav Papyrkovskyy [mailto:mpapyrkovskyy@hortonworks.com] 
Sent: Friday, January 12, 2018 2:48 PM
To: dev@ambari.apache.org
Subject: Re: New JIRA - Commit messages

+1 for removing redundant user names
author is always clearly visible via all cli tools / IDE.
Only reason I see we used it previously is that committers had to manually apply patch and
create commit.
This required more git knowledge and additional info from contributor to set author value
correctly. 
So name in commit message was nice and easy solution we don’t require anymore.

Myroslav


> 12 січ. 2018 р. о 01:50 Vivek Ratnavel <vivekratnavel@apache.org> написав(ла):
> 
> Hi Nate,
> 
> I find the addition of (username) to commit messages as redundant, 
> since the user who contributed can be identified from the "Author" tag 
> in git command line. I use "git log --grep=AMBARI-XXXXX" and get an 
> output with the author tag. Regardless of the committer who merged the 
> pull request, the "Author" tag will contain the original contributors 
> name. In your example, it will be "Author: Joe Smith". The PR might 
> have been merged by any committer with write access, but it becomes irrelevant.
> 
> commit 1eead54f467c7ff6abafa908abf5b66a7278a45a
> Author: Olivér Szabó <oleewere@gmail.com>
> Date:   Wed Jan 10 21:13:12 2018 +0100
> 
>    [AMBARI-22749] ADDENDUM: Create Pull Request Template - fix rat 
> check
> (#82)
> 
> commit 493612494740ab9b6a7b1e470951baf5cfbee78e
> Merge: 30f3a04 1bb2355
> Author: Vivek Ratnavel Subramanian <vivekratnavel90@gmail.com>
> Date:   Tue Jan 9 13:00:09 2018 -0800
> 
>    Merge pull request #70 from vivekratnavel/AMBARI-22749-trunk
> 
>    [AMBARI-22749] Create Pull Request Template
> 
> -Vivek Ratnavel
> 
> On Thu, Jan 11, 2018 at 3:37 PM, Nate Cole <ncole@apache.org> wrote:
> 
>> All,
>> 
>> 
>> 
>> Previous to the fork/pull request framework, we used to use commit 
>> message variations like so:
>> 
>> 
>> 
>> “AMBARI-XXXXXX. Great change for stack (ncole)”           (for committers)
>> 
>> or
>> 
>> “AMBARI-XXXXXX. Great change for stack (Joe Smith via ncole)”       (for
>> contributors)
>> 
>> 
>> 
>> Commits have started abandoning this syntax.
>> 
>> 
>> 
>> While I agree that the new workflow makes it clear the committer vs 
>> the author, it was useful to search just commit messages and pair 
>> them with the author.
>> 
>> 
>> 
>> I would propose we keep the “(username)” practice with commit message
>> 
>> 
>> 
>> since we don’t always use github for viewing history.  Contributors 
>> can use something like:
>> 
>> 
>> 
>> “(Joe Smith via pull request)” or just “(Joe Smith)”
>> 
>> 
>> 
>> Thoughts?
>> 
>> -Nate
>> 
>> 
>> 
>> 

Mime
View raw message