[DISCUSS] Commit log Pattern Unification

classic Classic list List threaded Threaded
13 messages Options
Reply | Threaded
Open this post in threaded view
|

[DISCUSS] Commit log Pattern Unification

Brahma Reddy Battula
Hi All,


When we look at current git log,commit message patters are like following..

I feel, we should have unified one. Might not be very important,but it will be better have unified one.

a).    (Contributed by XXX via YYY)
b).   (XXX via YYY)
c).   Contributed by XXX
d).  Contributed by XXX.

Pattern 2 should be fine.

and one more thing how about author name as contributor name while committing..?

Once after conclusion we can update the follwoing WIKI also

https://wiki.apache.org/hadoop/HowToCommit#Committing a patch




Thanks And Regards
Brahma Reddy Battula
Reply | Threaded
Open this post in threaded view
|

Re: [DISCUSS] Commit log Pattern Unification

Akira AJISAKA
I'm +1 for using Git "Signed-off-by" feature.

Regards,
Akira

On 11/5/16 17:36, Brahma Reddy Battula wrote:

> Hi All,
>
>
> When we look at current git log,commit message patters are like following..
>
> I feel, we should have unified one. Might not be very important,but it will be better have unified one.
>
> a).    (Contributed by XXX via YYY)
> b).   (XXX via YYY)
> c).   Contributed by XXX
> d).  Contributed by XXX.
>
> Pattern 2 should be fine.
>
> and one more thing how about author name as contributor name while committing..?
>
> Once after conclusion we can update the follwoing WIKI also
>
> https://wiki.apache.org/hadoop/HowToCommit#Committing a patch
>
>
>
>
> Thanks And Regards
> Brahma Reddy Battula
>


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [DISCUSS] Commit log Pattern Unification

Daniel Templeton
In reply to this post by Brahma Reddy Battula
On 11/5/16 1:36 AM, Brahma Reddy Battula wrote:

> When we look at current git log,commit message patters are like following..
>
> I feel, we should have unified one. Might not be very important,but it will be better have unified one.
>
> a).    (Contributed by XXX via YYY)
> b).   (XXX via YYY)
> c).   Contributed by XXX
> d).  Contributed by XXX.
>
> Pattern 2 should be fine.

I agree.  Pattern 2 is fine, but the "via" is redundant with the info
already given by git.  Just "(XXX)" would be enough.

> and one more thing how about author name as contributor name while committing..?

Full name is most readable, but we have some collisions in the
community.  Email address is unique, but perhaps too unique as many
contributors have changed email addresses over the years.  Username is
also unique and may be the best answer, even though it's not as readable
as a name.  It's also how the committer is included in the log by git.

Daniel

---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [DISCUSS] Commit log Pattern Unification

Daniel Templeton
On 11/6/16 8:01 PM, Daniel Templeton wrote:
> It's also how the committer is included in the log by git.

OK, git actually shows name and email in the log.  It shows the username
in the annotations in NetBeans, which is what I was thinking of. :)

Daniel

---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [DISCUSS] Commit log Pattern Unification

Ravi Prakash-3
I have a preference for d) Contributed by XXX.

Wouldn't signed-off require the commit to come from the contributor? What
about people who submit patch files? I thought that was still the modus
operandi, no?

On Sun, Nov 6, 2016 at 8:18 PM, Daniel Templeton <[hidden email]>
wrote:

> On 11/6/16 8:01 PM, Daniel Templeton wrote:
>
>> It's also how the committer is included in the log by git.
>>
>
> OK, git actually shows name and email in the log.  It shows the username
> in the annotations in NetBeans, which is what I was thinking of. :)
>
>
> Daniel
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [hidden email]
> For additional commands, e-mail: [hidden email]
>
>
Reply | Threaded
Open this post in threaded view
|

Re: [DISCUSS] Commit log Pattern Unification

Andrew Wang
I've always done d), but isn't this information captured in JIRA anyway?

On Mon, Nov 7, 2016 at 11:29 AM, Ravi Prakash <[hidden email]> wrote:

> I have a preference for d) Contributed by XXX.
>
> Wouldn't signed-off require the commit to come from the contributor? What
> about people who submit patch files? I thought that was still the modus
> operandi, no?
>
> On Sun, Nov 6, 2016 at 8:18 PM, Daniel Templeton <[hidden email]>
> wrote:
>
> > On 11/6/16 8:01 PM, Daniel Templeton wrote:
> >
> >> It's also how the committer is included in the log by git.
> >>
> >
> > OK, git actually shows name and email in the log.  It shows the username
> > in the annotations in NetBeans, which is what I was thinking of. :)
> >
> >
> > Daniel
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: [hidden email]
> > For additional commands, e-mail: [hidden email]
> >
> >
>
Reply | Threaded
Open this post in threaded view
|

Re: [DISCUSS] Commit log Pattern Unification

Allen Wittenauer-6
In reply to this post by Ravi Prakash-3

> On Nov 7, 2016, at 11:29 AM, Ravi Prakash <[hidden email]> wrote:
>
> I have a preference for d) Contributed by XXX.
>
> Wouldn't signed-off require the commit to come from the contributor? What
> about people who submit patch files?

        If the patches are built with 'git format-patch', no.

        In Apache Yetus, we moved to requiring that patch format + committing with 'smart-apply-patch --committer'.  Makes life much easier, esp when it comes time to see who is actually contributing patches.
---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [DISCUSS] Commit log Pattern Unification

Ravi Prakash-3
In reply to this post by Andrew Wang
And sometimes there are multiple contributors, so it becomes Contributed by
XX1, XX2 and XX3.

I guess having the information in git logs makes for easy grepping, awking
and counting ;-)

On Mon, Nov 7, 2016 at 11:35 AM, Andrew Wang <[hidden email]>
wrote:

> I've always done d), but isn't this information captured in JIRA anyway?
>
> On Mon, Nov 7, 2016 at 11:29 AM, Ravi Prakash <[hidden email]>
> wrote:
>
>> I have a preference for d) Contributed by XXX.
>>
>> Wouldn't signed-off require the commit to come from the contributor? What
>> about people who submit patch files? I thought that was still the modus
>> operandi, no?
>>
>> On Sun, Nov 6, 2016 at 8:18 PM, Daniel Templeton <[hidden email]>
>> wrote:
>>
>> > On 11/6/16 8:01 PM, Daniel Templeton wrote:
>> >
>> >> It's also how the committer is included in the log by git.
>> >>
>> >
>> > OK, git actually shows name and email in the log.  It shows the username
>> > in the annotations in NetBeans, which is what I was thinking of. :)
>> >
>> >
>> > Daniel
>> >
>> > ---------------------------------------------------------------------
>> > To unsubscribe, e-mail: [hidden email]
>> > For additional commands, e-mail: [hidden email]
>> >
>> >
>>
>
>
Reply | Threaded
Open this post in threaded view
|

Re: [DISCUSS] Commit log Pattern Unification

Andrew Wang
Part of why I asked this question is because it's hard to get consistent
log messages until they can be enforced via precommit. We already have the
issue of typo'd or forgotten JIRA numbers in commit messages, and that'd
happen for attribution messages as well.

There's also always significant dedupe necessary to grep correctly. Until
someone is a committer, they don't have an Apache ID. In the meanwhile you
can use their JIRA username, but that can be different from their eventual
Apache ID, and people also change their JIRA usernames and emails.

IMO, if you have a usecase that requires counting contributors, go to the
source of truth: JIRA. The releasedocmaker.py script from Yetus builds the
changelog and release notes from JIRA info, and is pretty straightforward.
You can modify that, or grep the generated markdown output. Or, if it's
gone out in a release, you can grep the changelog from the release tarball.

On Mon, Nov 7, 2016 at 11:45 AM, Ravi Prakash <[hidden email]> wrote:

> And sometimes there are multiple contributors, so it becomes Contributed
> by XX1, XX2 and XX3.
>
> I guess having the information in git logs makes for easy grepping, awking
> and counting ;-)
>
> On Mon, Nov 7, 2016 at 11:35 AM, Andrew Wang <[hidden email]>
> wrote:
>
>> I've always done d), but isn't this information captured in JIRA anyway?
>>
>> On Mon, Nov 7, 2016 at 11:29 AM, Ravi Prakash <[hidden email]>
>> wrote:
>>
>>> I have a preference for d) Contributed by XXX.
>>>
>>> Wouldn't signed-off require the commit to come from the contributor? What
>>> about people who submit patch files? I thought that was still the modus
>>> operandi, no?
>>>
>>> On Sun, Nov 6, 2016 at 8:18 PM, Daniel Templeton <[hidden email]>
>>> wrote:
>>>
>>> > On 11/6/16 8:01 PM, Daniel Templeton wrote:
>>> >
>>> >> It's also how the committer is included in the log by git.
>>> >>
>>> >
>>> > OK, git actually shows name and email in the log.  It shows the
>>> username
>>> > in the annotations in NetBeans, which is what I was thinking of. :)
>>> >
>>> >
>>> > Daniel
>>> >
>>> > ---------------------------------------------------------------------
>>> > To unsubscribe, e-mail: [hidden email]
>>> > For additional commands, e-mail: [hidden email]
>>> >
>>> >
>>>
>>
>>
>
Reply | Threaded
Open this post in threaded view
|

Re: [DISCUSS] Commit log Pattern Unification

Akira AJISAKA
In reply to this post by Allen Wittenauer-6
In more detail, I'm +1 for requiring 'git format-patch' and commit with
'dev-support/bin/smart-apply-patch --committer', as well as Apache Yetus.

That way committers can use signed-off feature and don't need to add
'contributed by ...' to commit message by hand.

Thanks,
Akira

On 11/8/16 04:38, Allen Wittenauer wrote:

>
>> On Nov 7, 2016, at 11:29 AM, Ravi Prakash <[hidden email]> wrote:
>>
>> I have a preference for d) Contributed by XXX.
>>
>> Wouldn't signed-off require the commit to come from the contributor? What
>> about people who submit patch files?
>
> If the patches are built with 'git format-patch', no.
>
> In Apache Yetus, we moved to requiring that patch format + committing with 'smart-apply-patch --committer'.  Makes life much easier, esp when it comes time to see who is actually contributing patches.
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [hidden email]
> For additional commands, e-mail: [hidden email]
>


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

RE: [DISCUSS] Commit log Pattern Unification

Brahma Reddy
Thanks Akira Ajisaka, Ravi Prakash,Andrew wang,Allen Wittenauer and Daniel..

Some are favored for option d ( i.e" contributed by XXX.") and Akira and Allen are suggesting to use "signed-off-feature" with git-format-patch.

Many other communities, like Yetus,HBase,Spark all are enforcing for git-format-patch, can we also do this..?

So how to conclude on this..?

Seems to be we can't commit  author as contributor , as Andrew pointed.(contributors will not having the apache mail-id).

can we recommend contributors to keep consistent email-Id..?

Would like to add one more point ,  "smart-apply-patch" can improve like auto commit to specified branches , which makes things easier for committer..?



Regards
Brahma Reddy Battula



-----Original Message-----
From: Akira Ajisaka [mailto:[hidden email]]
Sent: 08 November 2016 10:05
To: Allen Wittenauer; Ravi Prakash
Cc: Daniel Templeton; Brahma Reddy Battula; [hidden email]
Subject: Re: [DISCUSS] Commit log Pattern Unification

In more detail, I'm +1 for requiring 'git format-patch' and commit with 'dev-support/bin/smart-apply-patch --committer', as well as Apache Yetus.

That way committers can use signed-off feature and don't need to add 'contributed by ...' to commit message by hand.

Thanks,
Akira

On 11/8/16 04:38, Allen Wittenauer wrote:

>
>> On Nov 7, 2016, at 11:29 AM, Ravi Prakash <[hidden email]> wrote:
>>
>> I have a preference for d) Contributed by XXX.
>>
>> Wouldn't signed-off require the commit to come from the contributor?
>> What about people who submit patch files?
>
> If the patches are built with 'git format-patch', no.
>
> In Apache Yetus, we moved to requiring that patch format + committing with 'smart-apply-patch --committer'.  Makes life much easier, esp when it comes time to see who is actually contributing patches.
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [hidden email]
> For additional commands, e-mail: [hidden email]
>


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]
Reply | Threaded
Open this post in threaded view
|

Re: [DISCUSS] Commit log Pattern Unification

Andrew Wang
This is a bit of an aside, but if anyone's interested in grepping git log,
I've written a python script that helps with this:

https://github.com/umbrant/versions

This includes a fixup file that patches typo'd commit messages as well as
commits that include multiple JIRAs. The script also detects reverts and
merges.

I wrote this to validate the 3.0.0-alpha2 fixversion matches the state in
the git repo, and it found a lot of mistakes. It should be easy to add
support for other releases, though someone would need to go through the
errors to make the fixup file.

Best,
Andrew

On Fri, Nov 18, 2016 at 1:08 AM, Brahma Reddy Battula <
[hidden email]> wrote:

> Thanks Akira Ajisaka, Ravi Prakash,Andrew wang,Allen Wittenauer and
> Daniel..
>
> Some are favored for option d ( i.e" contributed by XXX.") and Akira and
> Allen are suggesting to use "signed-off-feature" with git-format-patch.
>
> Many other communities, like Yetus,HBase,Spark all are enforcing for
> git-format-patch, can we also do this..?
>
> So how to conclude on this..?
>
> Seems to be we can't commit  author as contributor , as Andrew
> pointed.(contributors will not having the apache mail-id).
>
> can we recommend contributors to keep consistent email-Id..?
>
> Would like to add one more point ,  "smart-apply-patch" can improve like
> auto commit to specified branches , which makes things easier for
> committer..?
>
>
>
> Regards
> Brahma Reddy Battula
>
>
>
> -----Original Message-----
> From: Akira Ajisaka [mailto:[hidden email]]
> Sent: 08 November 2016 10:05
> To: Allen Wittenauer; Ravi Prakash
> Cc: Daniel Templeton; Brahma Reddy Battula; [hidden email]
> Subject: Re: [DISCUSS] Commit log Pattern Unification
>
> In more detail, I'm +1 for requiring 'git format-patch' and commit with
> 'dev-support/bin/smart-apply-patch --committer', as well as Apache Yetus.
>
> That way committers can use signed-off feature and don't need to add
> 'contributed by ...' to commit message by hand.
>
> Thanks,
> Akira
>
> On 11/8/16 04:38, Allen Wittenauer wrote:
> >
> >> On Nov 7, 2016, at 11:29 AM, Ravi Prakash <[hidden email]> wrote:
> >>
> >> I have a preference for d) Contributed by XXX.
> >>
> >> Wouldn't signed-off require the commit to come from the contributor?
> >> What about people who submit patch files?
> >
> >       If the patches are built with 'git format-patch', no.
> >
> >       In Apache Yetus, we moved to requiring that patch format +
> committing with 'smart-apply-patch --committer'.  Makes life much easier,
> esp when it comes time to see who is actually contributing patches.
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: [hidden email]
> > For additional commands, e-mail: [hidden email]
> >
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [hidden email]
> For additional commands, e-mail: [hidden email]
>
>
Reply | Threaded
Open this post in threaded view
|

Re: [DISCUSS] Commit log Pattern Unification

Brahma Reddy Battula
Dear All,


 Upon discussion till now , we can choose one of the following option.


a) Contributed XXX.

b) Signoff feature ( which might allow commits with non-apache mail,it might some more discussion)


Whether any vote thread required to conclusion for this..?




Thanks And Regards
Brahma Reddy Battula


________________________________
From: Andrew Wang <[hidden email]>
Sent: Tuesday, November 22, 2016 10:05 AM
To: Brahma Reddy Battula
Cc: Akira Ajisaka; Allen Wittenauer; Ravi Prakash; [hidden email]; Daniel Templeton; Brahma Reddy Battula
Subject: Re: [DISCUSS] Commit log Pattern Unification

This is a bit of an aside, but if anyone's interested in grepping git log, I've written a python script that helps with this:

https://github.com/umbrant/versions
[https://avatars1.githubusercontent.com/u/33948?v=3&s=400]<https://github.com/umbrant/versions>

umbrant/versions<https://github.com/umbrant/versions>
github.com
versions - Script for bulk updating JIRAs




This includes a fixup file that patches typo'd commit messages as well as commits that include multiple JIRAs. The script also detects reverts and merges.

I wrote this to validate the 3.0.0-alpha2 fixversion matches the state in the git repo, and it found a lot of mistakes. It should be easy to add support for other releases, though someone would need to go through the errors to make the fixup file.

Best,
Andrew

On Fri, Nov 18, 2016 at 1:08 AM, Brahma Reddy Battula <[hidden email]<mailto:[hidden email]>> wrote:
Thanks Akira Ajisaka, Ravi Prakash,Andrew wang,Allen Wittenauer and Daniel..

Some are favored for option d ( i.e" contributed by XXX.") and Akira and Allen are suggesting to use "signed-off-feature" with git-format-patch.

Many other communities, like Yetus,HBase,Spark all are enforcing for git-format-patch, can we also do this..?

So how to conclude on this..?

Seems to be we can't commit  author as contributor , as Andrew pointed.(contributors will not having the apache mail-id).

can we recommend contributors to keep consistent email-Id..?

Would like to add one more point ,  "smart-apply-patch" can improve like auto commit to specified branches , which makes things easier for committer..?



Regards
Brahma Reddy Battula



-----Original Message-----
From: Akira Ajisaka [mailto:[hidden email]<mailto:[hidden email]>]
Sent: 08 November 2016 10:05
To: Allen Wittenauer; Ravi Prakash
Cc: Daniel Templeton; Brahma Reddy Battula; [hidden email]<mailto:[hidden email]>
Subject: Re: [DISCUSS] Commit log Pattern Unification

In more detail, I'm +1 for requiring 'git format-patch' and commit with 'dev-support/bin/smart-apply-patch --committer', as well as Apache Yetus.

That way committers can use signed-off feature and don't need to add 'contributed by ...' to commit message by hand.

Thanks,
Akira

On 11/8/16 04:38, Allen Wittenauer wrote:

>
>> On Nov 7, 2016, at 11:29 AM, Ravi Prakash <[hidden email]<mailto:[hidden email]>> wrote:
>>
>> I have a preference for d) Contributed by XXX.
>>
>> Wouldn't signed-off require the commit to come from the contributor?
>> What about people who submit patch files?
>
>       If the patches are built with 'git format-patch', no.
>
>       In Apache Yetus, we moved to requiring that patch format + committing with 'smart-apply-patch --committer'.  Makes life much easier, esp when it comes time to see who is actually contributing patches.
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [hidden email]<mailto:[hidden email]>
> For additional commands, e-mail: [hidden email]<mailto:[hidden email]>
>


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]<mailto:[hidden email]>
For additional commands, e-mail: [hidden email]<mailto:[hidden email]>