Some of the contrib material is very important.
I am still using ikvmc to compile lucene because I use the FSTLookup which
is only available in version 3. It provides guesses for misspelled words
for drop downs and such and is more efficient than a general fuzzy search.
On Tue, Nov 22, 2011 at 9:38 PM, Michael Herndon
mhern...@wickedsoftware.net wrote:
Some of the stuff currently in core were moved to contrib projects in newer
versions of Lucene
An example of this is the standard analyzer for 4x, is no longer in the
core project. So that is a scenario to be wary of.
As previously mentioned by chris, we should probably continue to maintain
what we have and slowly increase as contributors invest the time in porting
a contrib project, but otherwise concentrate on the essential.
Though I would imagine we should define what is considered to be essential
for a release and what is not. Its also possible to put contrib projects
not essential on a different release cycle altogether.
On Tue, Nov 22, 2011 at 1:38 PM, Prescott Nasser geobmx...@hotmail.com
wrote:
I'd like to port over as much as possible. There is a lot of great stuff
there and it only opens our user base up. That said I think they are nice
to haves not requirements
From: Christopher Currens
Sent: 11/22/2011 10:13 AM
To: lucene-net-dev@lucene.apache.org
Subject: [Lucene.Net] Contrib and releases
I wanted to take some time to discuss our position on the Contrib
projects.
Digy and I were a little off topic in the roadmap thread and I brought
it
up. Digy mentioned he always felt that it was a nice to have but
not necessarily required for a release. I can't say I disagree.
I had some time to look over a lot of the contrib code in Java, and there
are quite a few projects that cannot be ported directly, as they rely on
3rd party Java libraries that have no direct .Net equivalent. Porting
them
could delay releases a long time, which is why I think it hasn't really
been kept up to date as it is.
I think a requirement for releases should be to have whatever projects
that
are currently in contrib to be up to date and building, with valid and
passing tests (our 2.9.4 Contrib.Analyzers project is missing tests for 9
namespaces). I think it should be as simple as that. I don't think we
should worry about adding new projects, unless you feel compelled to do
so.
My opinion is that if someone wants a Contrib added to the project,
someone can port and donate the code to our project, or if they request
it,
someone can volunteer to port it themselves.
People do use our Contrib assemblies, I personally think this is a fair
trade-off to only have to maintain what we already have. I would like to
know how everyone else feels about it.
Thanks,
Christopher
