[MIT Research Project] Identifying "features" in Cocoon 2.1

["Karim R. Lakhani" <la...@MIT.EDU>]

Dear Cocoon Community Members,

I am a PhD candidate at MIT researching new models of software development and in particular how open source communities innovate and create products in a distributed, global setting.  I have done extensive research on open source - some papers are here (http://web.mit.edu/lakhani/www/karimlakhaniresearch.html) and also run the MIT-based open source research community (http://opensource.mit.edu). 

For my doctoral dissertation I am studying software development practices in Apache Cocoon and PostgreSQL communities.  My method is to trace the social and technical history of each new feature developed by the community over an approximately 1 year period of time.  I do this by tracking software features to CVS/SVN commits and then to the related email discussion on the development list.  This allows me to create a history of each feature and change in the software. An overview of my research project and its method is here: http://web.mit.edu/lakhani/Public/LakhaniInforms.pdf

I have been in extensive conversations with Sylvain and Stefano and they recommended that I approach the community with my request to determine "important features" in Cocoon 2.1.  My plan is to create histories of each of the changes that exist up to the 2.1 (August 2003) release (http://cocoon.apache.org/2.1/changes.html).  However I would like to get the community to tell me which are the more important features in 2.1 - that were NOT present in the 2.0 release.  For instance in my conversation with Stefano and Sylvain - we determined that some of the important features in 2.1 included:

- Flow Script
- Cocoon Forms

It will probably be the case that some of the detailed changes will be absorbed into the "features" and some will be independent.

However all three of us felt that it would be good for the community to tell me what they considered to be important features in 2.1.  So I wanted to request the wider community to suggest features that should go in the analysis.  Think of features that you use in 2.1 or you tell your clients about - that were exclusively developed for 2.1

Please nominate features and I will summarize what was reported back to the community.  I will then do the "dirty work" of compiling and matching it to the CVS commits and the email discussion:)  Of course all my findings and writings will be public as well.

Thanks so much in advance.  
-- 
Karim R. Lakhani
MIT Sloan | The Boston Consulting Group
Mobile: +1 (617) 851-1224
http://spoudaiospaizen.net
http://web.mit.edu/lakhani/www | http://opensource.mit.edu 

Re: [MIT Research Project] Identifying "features" in Cocoon 2.1

["Karim R. Lakhani" <la...@MIT.EDU>]

Stefano Mazzocchi wrote:
> 
> Karim,
> 
> I was thinking that one possible way to do this, could be that you 
> compile a list, based on the cross-correlation between the CVS logs (too 
> granular) and the changes.xml file (eventually still too granular) and 
> then we filter it out here.

That is my plan any way.  To match changes.xml to cvs/svn logs.  This is the "dirty work!"

> 
> The reason for this is that, honestly, I don't remember *when* something 
> was added :-) so indicating a period of time is rather hard for us to 
> correlate with specific features (several blocks were added, for 
> example, but don't remember which ones).

Actually I don't so much need timing (hope I did not imply that in my email) - rather just a general sense of "features" above just the very granular changes file.  So for instance - if people were to ask you why should I use Cocoon 2.1? You might say......it can do blah blah blah - which is cool.

I wanted a general sense of features in Cocoon 2.1 as an anchoring point. If the community feels that this is not do able - I certainly understand and would be happy to show my matching of cvs to changes file.


> Seriously, I think asking for a list of features correlated with time 
> yields a "bah, can't remember" answer, while a list of things to look at 
> might allow us to say "oh yeah, that one".

will do - last call for nominations for features :)

> 
> WDYT?
> 

I am on board and happy to work with the community on it.

-- 
Karim R. Lakhani
MIT Sloan | The Boston Consulting Group
Mobile: +1 (617) 851-1224
http://spoudaiospaizen.net
http://web.mit.edu/lakhani/www | http://opensource.mit.edu 

Re: [MIT Research Project] Identifying "features" in Cocoon 2.1

[Stefano Mazzocchi <st...@apache.org>]

Karim R. Lakhani wrote:

> Please nominate features and I will summarize what was reported back to 
> the community.  I will then do the "dirty work" of compiling and 
> matching it to the CVS commits and the email discussion:)  Of course all 
> my findings and writings will be public as well.

Karim,

I was thinking that one possible way to do this, could be that you 
compile a list, based on the cross-correlation between the CVS logs (too 
granular) and the changes.xml file (eventually still too granular) and 
then we filter it out here.

The reason for this is that, honestly, I don't remember *when* something 
was added :-) so indicating a period of time is rather hard for us to 
correlate with specific features (several blocks were added, for 
example, but don't remember which ones).

CVS/SVN logs contain all that information, it's just a matter of 
filtering it out. A way to do this would be to remove all the logs that 
resulted in a few lines of code of change.

Ah, note that if you use "svn log -v --xml" you get an xml format of the 
logs and you can use cocoon to filter that out ;-)

Seriously, I think asking for a list of features correlated with time 
yields a "bah, can't remember" answer, while a list of things to look at 
might allow us to say "oh yeah, that one".

WDYT?

-- 
Stefano.