lucy-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Marvin Humphrey <>
Subject Self-governance
Date Sat, 19 Jun 2010 02:55:03 GMT

A thread about Lucy recently wrapped up on the Lucene general@ list.

The board at Apache is breaking up umbrella projects.  The Lucene TLP (top
level project, as opposed to the Lucene Java subproject) has already spun off
Tika, Mahout, and Nutch; Lucy is next.  However, Lucy is not ready to become a
TLP yet, so that spawned a discussion of what should happen.

To summarize the outcome of the discussion:

  * Lucy must aim to "graduate" and become a TLP.
  * Lucy will stay where it is for now, rather than e.g. moving to the
  * Lucy must regularly assess its progress towards graduation using the
    incubation checklist at
    and report to the Lucene PMC.
  * No official deadline was set by the PMC, but Doug Cutting threw out "six
    months" and so we should probably start with that as our target.

Here's a template for us laid out by Grant Ingersoll, the Lucene PMC chair:

  1. Doing a release. 
  2. Showing some user list traction (i.e. real users) 
  3. Identifying and cultivating other contributors (via patches, discussions,
     JIRA issues, helping others, etc.) who can then become committers. 

The primary measure of success for a project at Apache is not technical: the
goal is to create a robust meritocratic development community.  In order to
graduate, we will need not only committers and users, but our own PMC which
takes responsibility for governing the project.

Over the next week, we should perform a self-critique and create our first
formal report, judging ourselves by the incubator criteria.  Our biggest
weakness from the ASF perspective, clearly, is that the project is too reliant
on me and would be unlikely to survive my departure at this time.  We should
set a goal of making me dispensible.

Here's where I think we stand with regards to Grant's plan, in a nutshell:

Lucy's high-level design is done, and a lot of the coding is done as well.
Developing and refining the Clownfish object model was the hardest part, and
that finished up last fall.  I think an initial release within 3 months is
doable, and we may be able to accelerate that.

I expect acquiring users after the release to be relatively straightforward.
Near-realtime search powered by mmap is a killer feature.

As for cultivating committers and PMC members, that will come naturally if we
acquire a significant user base.  I don't think we need to do anything special
for now beyond releasing good software, publicizing it effectively, and making
it easy and rewarding to contribute -- as we have tried to do over the last
year with our existing contributor base, with some success.

Marvin Humphrey

View raw message