On 2/11/2010 10:04 PM, Josh Handel wrote:
The vibe I am getting from this discussion is seems to be the following.. (this
is just a summary of the directions this conversation seems to me to be going).
1) A conceptual port of Lucene is probably not a good idea because search is
such a complex and difficult task.
2) Line by Line port works (as it has in the past) but is VERY time
consuming... If we can find automation tools and a way to parallelize the
process that would go a long way to helping this project survive.
3) We have LOTS of consumers of the tool, but limited (1) real committer,
that's because the line by line port isn't fun, exciting, and grueling.
....
7) There is an overwhelming desire to have .NETaffied API, perhaps as a
wrapper, or as a core deliverable.
8) There are a handful of automated tools to help with java to c# porting, but
we need to test them (and a fair bit of trepidation about if they will provide
value to the process)
I've been following this discussion, and couldn't resist the urge to
suggest the following, which would tackle all the points listed above
with one stone.
There exists a cross-platform C++ port of Lucene, called CLucene
(http://clucene.sourceforge.net). If Lucene.Net would like to change
course and wrap CLucene with a .NET-ish API instead of maintaining a
line-by-line port, that could benefit both projects. CLucene needs some
attention to get more people on deck (although there quite a few
already), and the .NET community would benefit from easier maintenance
and faster code-base.
What I'm offering is a model already in use by several OS projects
(SQLite.NET for example), and is quite successful. Let me know how this
sounds.
Itamar.
