incubator-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Daniel Kulp <dk...@apache.org>
Subject Re: FW: (qpid) Diversity
Date Thu, 06 Mar 2008 19:04:55 GMT
On Thursday 06 March 2008, Martin Ritchie wrote:
> - Firstly not all work is done on trunk so purely looking at trunk is
> not a good metric

That's fair.  I forgot about the work on the branches.  That's a very 
valid point.    Updated list with branches added from Jan 1:

Alan Conway        RH       84
Aidan Skinner      ?        51
Arnaud Simon       RH       48
Andrew Stitcher    RH       
Carl Trieloff      RH       6
Gordon Sim         RH       40
Jim Meyering       ?         
John O'Hara        JPMC     
Marnie McCormack   JPMC     (maternity)
Martin Ritchie     JMPC     42
Kim van der Riet   RH       
Nuno Santos        RH       6
Rafael Schloming   RH       56
Rajith Attapattu   RH       24
Robert Godfrey     JPMC     14
Robert Greig.      JPMC     
Rupert Smith       ?        36

264 RH
143 Non-RH

Back to Nov 1:
Alan Conway        RH       167
Aidan Skinner               51
Arnaud Simon       RH       105
Andrew Stitcher    RH       5
Carl Trieloff      RH       22
Gordon Sim         RH       107
Jim Meyering                
John O'Hara        JPMC     
Marnie McCormack   JPMC     (maternity)
Martin Ritchie     JMPC     81
Kim van der Riet   RH       10
Nuno Santos        RH       7
Rafael Schloming   RH       60
Rajith Attapattu   RH       58
Robert Godfrey     JPMC     20
Robert Greig.      JPMC     
Rupert Smith                63

541 RH
215 Non-RH

> - Looking at the last two months especially as that includes the start
> of the year which is typically a quite period also will show poor
> commit numbers.

Yes, lower numbers, but they should still be relatively proportional as 
it should be just as poor for RH folks as for non RH folks, right?    
The issue isn't the number of commits, it's who is doing them.  Anyway, 
I added a 4 months set shown above.  

Yes, # of commits is not a "good" metric as different people work 
different ways.  Example: I tend to commit small things several times a 
day.   If I find a spelling mistake, it gets committed.   Other people 
save things up and do larger commits.   It's all personal.   The thing 
that the IPMC needs to know more about are "trends".  Is the community 
getting better or worse from  a diversity standpoint.   That IS 
subjective to some extent and each IPMC member may interpret the data 
differently, but the point is they need the data.    Thus, having the 6 
month, 4 month, and 2 month numbers is a start.


> - We are preparing for an M2.1 release so a lot of effort is being
> expended on that branch.

Fine.   I included the entire trunk + branches dirs.

> My take would be to look at the last 6 months, which admittedly
> includes a number of holiday periods so the count of commits may be a
> little low. Since 2007-09 for the commits on the qpid repository
> shows:
>
> aconway 272
> aidan 54
> arnaudsimon 230
> astitcher 16
> cctrieloff 45
> gsim 201
> kpvdr 11
> nsantos 8
> rajith 169
> rgodfrey 99
> rgreig 98
> rhs 151
> ritchiem 593
> rupertlssmith 468
>
> 1103	RedHat
> 1312    Non-Aligned
>
> So Redhat have less than half the commits to Qpid so I don't think
> this is something we should worry about. 

It IS something to worry about if all the "Non-Aligned" folks really are 
aligned together under one entity.  The questions that each IPMC member 
needs to answer for themselves based on all the metrics are:

1) If RedHat decides tomorrow to pull all it's engineers off of qpid, 
would the project survive?

2) If the "entity" behind the "non-aligns" gets bought out by another 
entity that has no interest in AMQP and pulls the engineers off, will 
the project survive?

Those are the things the IPMC must consider (and the board if graduating 
top level) as it affects the long term viability of the project.  

> Keeping an eye on for sure, 
> but over analysing the alignment of those people that don't want to or
> are not allowed to say who they work for is not something that Apache
> requires nor would it be beneficial. Saying that we are not diverse
> because a lot of work on trunk has been done by a small set of people
> over a two month period is not helpful to the discussion.

But it helps provide insite into the above questions, so it is helpful.   
Other things that aren't even mentioned in this are things like 
participation on the dev list, documentation updates, wiki updates, JIRA 
cleanups, etc....   Those are all ALSO important metrics and maybe some 
of the folks above are more valuable in areas other than the raw code.  
There are lots of roles in a community, the code is just a single metric 
amoungst many.  For example:
http://markmail.org/search/?q=qpid+type%3Adevelopment+date%3A200709-200803
is another metric.

> It is also worth noting that volume of commits does not in anyway
> correspond to the quantity of code changed and even looking at lines
> changes cannot say anything for the quality. I think the fact that we
> have an active project that has matured to the point that where we
> feel as though we can self regulate in the Apache Way speaks far more
> to the community of the project than identifying who is paying the
> bills.
>
> The whole project worked very hard to pull together two servers and
> five client libraries for the M2 release all talking AMQP 0-8.

And that is all to be commended.  It was a good effort.

> We are 
> again working as a community to provide a M2.1 release that will inter
> operate at AMQP 0-9 with other AMQP products outside the Apache world.
> I for one am looking forward to our future releases where we can again
> move the entire project on to the wholly different AMQP 0-10.

And nothing stops you from doing all of that while still in the incubator 
while working on trying to furthur diversify your community.  Working 
with other AMQP products could be a perfect opportunity to find other 
interested people and get them involved with Qpid.  Why rush?  


From experience with CXF, after we got 2.0 and 2.0.1 out the door, we 
thought the same way.  "Hey, we worked hard to get the TCK passing, 
integrated with Geronimo, and two releases out, we're ready to go!" but 
Jim (one of our mentors) still had concerns about diversity.   Thus, we 
stuck with it and worked hard on getting other people more involved.   
It has paid off as we have added several very good folks that have 
brought fresh ideas and perspectives to the project.  If we HAD 
graduated, I'm not sure if we would have spent the time/effort on the 
mentoring and such that was required to bring them on board.  We've 
learned a lot in the process.  In hind sight, Jim was completely 
correct.  Part of the "Apache way" you talk about is being concious of 
things like that.  


Another metric I personally like to look at is how well the community has 
interacted with the "Apache community as a whole."  Aka: have they 
worked well with other Apache projects and such.   While not a 
requirement, this helps to show how well the folks "get it".  Looking at 
the list of committers, only Rajith is a committer on stuff other than 
qpid.  When you found bugs in the projects you depend on (like Mina), 
did you file patches and stuff to help out those communities or did you 
just file a bug and say "fix it"?  Did you work with them to try and get 
a release incorporating your changes/fixes?  Again, not a requirement, 
just another metric that is weighed in with everything else.   In the 
other direction, when other projects wanted to use qpid, did you help 
them out?   This is an important part of trying to grow your community. 
If more people use it, the more likely they will contribute patches and 
ideas and potentially become committers.  In my experience, the answer 
to #2 was "barely".  I wanted to add qpid JMS testing to CXF and I WAS 
working with qpid folks to help get M2 artifacts publishable/releasable 
to maven repos so maven based projects could use it, but a last minute 
decision (off list, that's another issue) killed that and I'm still 
stuck.  Again, that was MY experience.  Other folks may have had a 
better experience.  I know Rajith has done some work with Synapse.   Not 
sure the extent or result.  Again, not "required", just another data 
point.

"Is the community ready to graduate?" is a very subjective opinion.   
There are a LOT of variables that go into it and each IPMC member will 
weigh them differently.  Each IPMC member has different past experiences 
and different priorities.   The community thinking they are ready is 
only one small part of it.

That all said, I'm NOT on the IPMC.   Thus, my thoughts don't really 
count other than to provide insight based on MY experiences.  I don't 
have a binding vote.   The IPMC folks are the ones that will need to 
figure out if the diversity of the qpid community is a concern and, if 
so, if the qpid community has taken steps to rectify that.

Anyway, that's my inflation adjusted $.02 worth.  

-- 
J. Daniel Kulp
Principal Engineer, IONA
dkulp@apache.org
http://www.dankulp.com/blog

---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@incubator.apache.org
For additional commands, e-mail: general-help@incubator.apache.org


Mime
View raw message