lucene-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Marvin Humphrey <>
Subject Re: [PMC] [DISCUSS] Lucy
Date Sun, 13 Jun 2010 18:42:08 GMT
On Sun, Jun 13, 2010 at 07:23:09AM -0400, Grant Ingersoll wrote:
> The thing it needs to do, in my mind, is prove their a community here more
> than just Marvin (and despite the discussion on IRC, the mail archives
> paint a different story:                          

I hope Peter Karman will forgive me for citing him as an example.

The majority of Peter's work on Lucy over the last year has been on the
compatibility layer.  His patches typically originate as KinoSearch patches;
Peter works on KS rather than Lucy because he needs something that gets the
job done today.  However, because the KS compatibility layer consists entirely
of code which has been contributed to Lucy, his patches also apply against
Lucy.  See for example

Peter has also been rolling KinoSearch releases for a while now, freeing me up
to do things like fix bugs for which he's supplied the test cases :)
<>. He's had a commit bit on
KS for years.  

Additionally, Peter has released multiple CPAN distros which extend
KinoSearch, including a more sophisticated query parser:

Those of you who know my hatred for query parser holy wars can appreciate how
happy it made me to see someone else write that module. :)

> ASF projects must be able to survive the withdrawal of a contributor and I
> don't think Lucy would.  

Lucy has already survived the effective withdrawl of a major contributor:
Dave Balmain.  

While Dave was active, though, we got some excellent work done.  

This year, Peter has ramped up his participation and made many contributions
-- but perhaps he will move on, too.  Sadly, Nate Kurz, mmap's early champion,
has been too busy with sorbet of late to sustain his participation at past
levels.  People come and go.

So while I agree that Lucy is too reliant on me and would probably not survive
my departure at this time, it is a collaborative effort.  Always has been, and
always will be -- collaboration is baked into the language of the original
Lucy proposal.  And as the informal project lead, I've thought very hard about
how to make it as easy as possible for others to contribute.  

For instance, I've tried to ensure that the four layers that make up Lucy (C
configuration probing, object model, core indexing and search code, bindings)
are as loosely coupled as possible -- so that it's feasible for someone to
achieve mastery over individual project subcomponents and work on them
effectively in isolation.  More concretely, last year I aggressively engaged
Peter and Nate in an effort to refactor the C configuration probing layer
solely for the sake of simplicity and grokkability. 

I expect that in time this sustained emphasis on collaboration will pay off,
that Lucy will survive its present developer community bottleneck, and that I
will become dispensible frighteningly quickly as the community grows.  :)

> The Incubator is where this stuff gets sorted out at the ASF.  

Moving to the Incubator is probably the most appropriate move for Lucy at this
time.  It's not ready to graduate to TLP status, but under the new policy it
apparently cannot stay under the Lucene umbrella.

> I was only suggesting the other options as something Lucy might
> find easier to deal with since it won't have ASF restrictions, 

Lucy is and ought to be a community project.  Moving to somewhere like Google
code would impede that.  It belongs at Apache.  

> but if the project wants to stay in the ASF than by all means to the
> Incubator. 

Within the near future, I will take up the matter on the Incubator list.  I'd
do it today, but there are some private conversations I would like to conclude
first (and some people appear to be on vacation).

Marvin Humphrey

View raw message