Practices involving source-code repositories

[Marcos César de Oliveira]

Hi, everybody.

I'm a Brazilian software engineering researcher, and currently I'm working
on a paper where we are conducting an investigation about evolution of
Enterprise Information Systems. One of the case studies in this study is
Apache OFBiz.

After analyzing the  data extracted from github, we realized that the
majority of issues were resolved with a single commit (84%), and that the
majority of issues have commits from a single user (96%). We conjecture
about the reasoning about this numbers, and guessed that would be that the
OFBiz development community has a convention to commit only when they are
done with the issue. This is true? Is there such convention or practice?

Any help will be very appreciated and I thanks everybody in advance.

---
Marcos César

[taher]

Hi Marcos,

you are looking in the wrong place, OFBiz repository exists in subversion, not github, and git is only cloning subversion. To the study the project history review the logs and details in subversion and the JIRA issues.

Cheers,

Taher Alkhateeb

----- Original Message -----

From: "Marcos César de Oliveira" <[hidden email]>
To: [hidden email]
Sent: Sunday, 1 November, 2015 7:42:15 PM
Subject: Practices involving source-code repositories

Hi, everybody.

I'm a Brazilian software engineering researcher, and currently I'm working
on a paper where we are conducting an investigation about evolution of
Enterprise Information Systems. One of the case studies in this study is
Apache OFBiz.

After analyzing the data extracted from github, we realized that the
majority of issues were resolved with a single commit (84%), and that the
majority of issues have commits from a single user (96%). We conjecture
about the reasoning about this numbers, and guessed that would be that the
OFBiz development community has a convention to commit only when they are
done with the issue. This is true? Is there such convention or practice?

Any help will be very appreciated and I thanks everybody in advance.

---
Marcos César

[Pierre Smits]

Hi Marcos,

Your questions cannot be answered as easy as you have posted your
conjecture. Would any other investigation into OFBiz issues, and by whom
and how the are resolved, yield the same results?
Any investigation into a product of a project under the umbrella of the
Apache Software Foundation should not be based on information/data in
GitHub.

Best regards,

Pierre Smits

*OFBiz Extensions Marketplace*
http://oem.ofbizci.net/oci-2/

On Sun, Nov 1, 2015 at 5:42 PM, Marcos César de Oliveira <[hidden email]
> wrote:

> Hi, everybody.
>
> I'm a Brazilian software engineering researcher, and currently I'm working
> on a paper where we are conducting an investigation about evolution of
> Enterprise Information Systems. One of the case studies in this study is
> Apache OFBiz.
>
> After analyzing the  data extracted from github, we realized that the
> majority of issues were resolved with a single commit (84%), and that the
> majority of issues have commits from a single user (96%). We conjecture
> about the reasoning about this numbers, and guessed that would be that the
> OFBiz development community has a convention to commit only when they are
> done with the issue. This is true? Is there such convention or practice?
>
> Any help will be very appreciated and I thanks everybody in advance.
>
> ---
> Marcos César
>

[Marcos César de Oliveira]

Hi, Taher and Pierre.

I collected the commit log from GitHub and the issues from Jira. I assumed
that GitHub mirrors the main subversion repository and that the mapping of
commits from subversion and GitHub is one to one. I confirmed this by
comparing the commit log from svn and git, and both repositories have the
same number of commits. This is correct? The association of commits and
issues was extracted from the comment field of the commits, specifically
searching for the pattern OFBIZ-\d+. Accordingly, I found that 23.3% of
commits are associated with issues. I also confirmed this number using the
svn log. From this subset, 83.74% of the issues have only one commit
associated. This finding make me think if the developers of OFBiz try to
follow some recommendation with regard to the size of commits. One
hypothesis is that the developers try to commit only when the issue is
resolved. Another hypothesis is that the issues tend to be small and
require only one commit to be resolved. For now, I would like to know if
exists any recommendation that asks for to commit only after the issue is
resolved, or if exists any recommendation that can influence the number of
commits associated with an issue.

Best regards,

Marcos César


Em dom, 1 de nov de 2015 às 17:46, Pierre Smits <[hidden email]>
escreveu:

> Hi Marcos,
>
> Your questions cannot be answered as easy as you have posted your
> conjecture. Would any other investigation into OFBiz issues, and by whom
> and how the are resolved, yield the same results?
> Any investigation into a product of a project under the umbrella of the
> Apache Software Foundation should not be based on information/data in
> GitHub.
>
> Best regards,
>
> Pierre Smits
>
> *OFBiz Extensions Marketplace*
> http://oem.ofbizci.net/oci-2/
>
> On Sun, Nov 1, 2015 at 5:42 PM, Marcos César de Oliveira <
> [hidden email]
> > wrote:
>
> > Hi, everybody.
> >
> > I'm a Brazilian software engineering researcher, and currently I'm
> working
> > on a paper where we are conducting an investigation about evolution of
> > Enterprise Information Systems. One of the case studies in this study is
> > Apache OFBiz.
> >
> > After analyzing the  data extracted from github, we realized that the
> > majority of issues were resolved with a single commit (84%), and that the
> > majority of issues have commits from a single user (96%). We conjecture
> > about the reasoning about this numbers, and guessed that would be that
> the
> > OFBiz development community has a convention to commit only when they are
> > done with the issue. This is true? Is there such convention or practice?
> >
> > Any help will be very appreciated and I thanks everybody in advance.
> >
> > ---
> > Marcos César
> >
>

[Adrian Crum-3]

We follow a best practice of limiting a commit to fixing only one thing.
That approach provides the ability to revert the commit should something
go wrong. Reverting a single fix is easier than reverting a small
portion of a larger commit.

Adrian Crum
Sandglass Software
www.sandglass-software.com

On 11/1/2015 12:30 PM, Marcos César de Oliveira wrote:

> Hi, Taher and Pierre.
>
> I collected the commit log from GitHub and the issues from Jira. I assumed
> that GitHub mirrors the main subversion repository and that the mapping of
> commits from subversion and GitHub is one to one. I confirmed this by
> comparing the commit log from svn and git, and both repositories have the
> same number of commits. This is correct? The association of commits and
> issues was extracted from the comment field of the commits, specifically
> searching for the pattern OFBIZ-\d+. Accordingly, I found that 23.3% of
> commits are associated with issues. I also confirmed this number using the
> svn log. From this subset, 83.74% of the issues have only one commit
> associated. This finding make me think if the developers of OFBiz try to
> follow some recommendation with regard to the size of commits. One
> hypothesis is that the developers try to commit only when the issue is
> resolved. Another hypothesis is that the issues tend to be small and
> require only one commit to be resolved. For now, I would like to know if
> exists any recommendation that asks for to commit only after the issue is
> resolved, or if exists any recommendation that can influence the number of
> commits associated with an issue.
>
> Best regards,
>
> Marcos César
>
>
> Em dom, 1 de nov de 2015 às 17:46, Pierre Smits <[hidden email]>
> escreveu:
>
>> Hi Marcos,
>>
>> Your questions cannot be answered as easy as you have posted your
>> conjecture. Would any other investigation into OFBiz issues, and by whom
>> and how the are resolved, yield the same results?
>> Any investigation into a product of a project under the umbrella of the
>> Apache Software Foundation should not be based on information/data in
>> GitHub.
>>
>> Best regards,
>>
>> Pierre Smits
>>
>> *OFBiz Extensions Marketplace*
>> http://oem.ofbizci.net/oci-2/
>>
>> On Sun, Nov 1, 2015 at 5:42 PM, Marcos César de Oliveira <
>> [hidden email]
>>> wrote:
>>
>>> Hi, everybody.
>>>
>>> I'm a Brazilian software engineering researcher, and currently I'm
>> working
>>> on a paper where we are conducting an investigation about evolution of
>>> Enterprise Information Systems. One of the case studies in this study is
>>> Apache OFBiz.
>>>
>>> After analyzing the  data extracted from github, we realized that the
>>> majority of issues were resolved with a single commit (84%), and that the
>>> majority of issues have commits from a single user (96%). We conjecture
>>> about the reasoning about this numbers, and guessed that would be that
>> the
>>> OFBiz development community has a convention to commit only when they are
>>> done with the issue. This is true? Is there such convention or practice?
>>>
>>> Any help will be very appreciated and I thanks everybody in advance.
>>>
>>> ---
>>> Marcos César
>>>
>>
>

[Marcos César de Oliveira]

Thanks Adrian.

Em dom, 1 de nov de 2015 às 18:42, Adrian Crum <
[hidden email]> escreveu:

> We follow a best practice of limiting a commit to fixing only one thing.
> That approach provides the ability to revert the commit should something
> go wrong. Reverting a single fix is easier than reverting a small
> portion of a larger commit.
>
> Adrian Crum
> Sandglass Software
> www.sandglass-software.com
>
> On 11/1/2015 12:30 PM, Marcos César de Oliveira wrote:
> > Hi, Taher and Pierre.
> >
> > I collected the commit log from GitHub and the issues from Jira. I
> assumed
> > that GitHub mirrors the main subversion repository and that the mapping
> of
> > commits from subversion and GitHub is one to one. I confirmed this by
> > comparing the commit log from svn and git, and both repositories have the
> > same number of commits. This is correct? The association of commits and
> > issues was extracted from the comment field of the commits, specifically
> > searching for the pattern OFBIZ-\d+. Accordingly, I found that 23.3% of
> > commits are associated with issues. I also confirmed this number using
> the
> > svn log. From this subset, 83.74% of the issues have only one commit
> > associated. This finding make me think if the developers of OFBiz try to
> > follow some recommendation with regard to the size of commits. One
> > hypothesis is that the developers try to commit only when the issue is
> > resolved. Another hypothesis is that the issues tend to be small and
> > require only one commit to be resolved. For now, I would like to know if
> > exists any recommendation that asks for to commit only after the issue is
> > resolved, or if exists any recommendation that can influence the number
> of
> > commits associated with an issue.
> >
> > Best regards,
> >
> > Marcos César
> >
> >
> > Em dom, 1 de nov de 2015 às 17:46, Pierre Smits <[hidden email]>
> > escreveu:
> >
> >> Hi Marcos,
> >>
> >> Your questions cannot be answered as easy as you have posted your
> >> conjecture. Would any other investigation into OFBiz issues, and by whom
> >> and how the are resolved, yield the same results?
> >> Any investigation into a product of a project under the umbrella of the
> >> Apache Software Foundation should not be based on information/data in
> >> GitHub.
> >>
> >> Best regards,
> >>
> >> Pierre Smits
> >>
> >> *OFBiz Extensions Marketplace*
> >> http://oem.ofbizci.net/oci-2/
> >>
> >> On Sun, Nov 1, 2015 at 5:42 PM, Marcos César de Oliveira <
> >> [hidden email]
> >>> wrote:
> >>
> >>> Hi, everybody.
> >>>
> >>> I'm a Brazilian software engineering researcher, and currently I'm
> >> working
> >>> on a paper where we are conducting an investigation about evolution of
> >>> Enterprise Information Systems. One of the case studies in this study
> is
> >>> Apache OFBiz.
> >>>
> >>> After analyzing the  data extracted from github, we realized that the
> >>> majority of issues were resolved with a single commit (84%), and that
> the
> >>> majority of issues have commits from a single user (96%). We conjecture
> >>> about the reasoning about this numbers, and guessed that would be that
> >> the
> >>> OFBiz development community has a convention to commit only when they
> are
> >>> done with the issue. This is true? Is there such convention or
> practice?
> >>>
> >>> Any help will be very appreciated and I thanks everybody in advance.
> >>>
> >>> ---
> >>> Marcos César
> >>>
> >>
> >
>