lucene-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Grant Ingersoll <gsing...@apache.org>
Subject Re: [VOTE] merge lucene/solr development
Date Thu, 04 Mar 2010 16:24:22 GMT

On Mar 4, 2010, at 7:30 AM, Jukka Zitting wrote:

> Hi,
> 
> On Thu, Mar 4, 2010 at 3:33 PM, Mark Miller <markrmiller@gmail.com> wrote:
>> On 03/04/2010 09:21 AM, Simon Willnauer wrote:
>>> Are we still voting or is this already a discussion?
>> 
>> As far as I'm concerned, a vote is on - the vote was started and people are
>> voting.
> 
> I haven't followed this too closely, but the large amount of confusion
> seems to indicate that calling a vote on this may be premature.
> 
> As an outside observer (I'm not actively involved in Solr or Lucene
> Java development) this proposal seems like a pretty monumental change
> for the projects. Would it make sense to rather start with smaller
> steps that everyone can agree on?
> 

I don't see that smaller steps would work.  For instance, I'm -1 on just the analyzers move,
b/c I don't think it will work in practice and it is just something that will end up gutting
Solr over time as more things are moved out.  It will require Solr to always have to be exactly
in sync with the same versions of Lucene otherwise there will never be agreement on releasing
the Analyzer artifact between Lucene and Solr committers, which is what this bigger vote is
about.

My only concern with the current vote is the lockstep releases, but I don't know if that is
actually something that has to be voted on now anyway.  I don't see any benefit from deciding
that now.  I suspect that the committers can work that out when the time comes.

Taking a step back, my view as PMC Chair and given the Board's penchant for "no umbrella"
projects, I also think the merge makes sense at the ASF level.  Mahout will be spinning out
to a TLP after 0.3 (assuming the board approves it) and I think Tika likely makes sense as
a TLP too.  One could make a case for Nutch being spun out due to it being more focused on
crawling these days (but I'm not pushing it either, as it still fits well in Lucene IMO too),
but I think Solr, Lucene and the Ports are intimately connected and are solely focused on
one thing: search.  This, in my mind, is a nicely focused community with a few subprojects
with a very large set of overlapping committers, a well focused PMC and an aligned community.
  At the same time, the PMC needs to be respectful of the very large community of users who
have already built Lucene implementations.  I think this proposal does that.

So, in the end, +1 on the vote, w/ the exception of the lockstep releases, which I think we
should defer to the community once it's fully formed to work out the details.

-Grant
Mime
View raw message