Mike, thanks for moving out of the JIRA issue. For completeness I just add the link to the issue where this thread started though. --> https://issues.apache.org/jira/browse/LUCENE-2279
I also think we need a solution for this problem but it does not seem to be that easy. Would moving the analysis be compatible with the lucene core having no dependencies? Not that I do not favor that solution I really think we should move all that out but I'm not sure about the place for this to live. My first impression would be a lucene contrib module but that would raise other issues like all solr committers then need access to that contrib. A new project would surely make sense but is also quite an overhead isn't it?! simon On Wed, Feb 24, 2010 at 5:40 PM, Michael Busch <[email protected]> wrote: > +1! I think that's the way to go. It's also confusing currently that some > analysers are in Lucene's core jar, and that there is an additional contrib > analysis jar. Your proposal would solve this problem too. > > Michael > > On Feb 24, 2010, at 8:32 AM, Michael McCandless <[email protected]> > wrote: > >> I think, in order to stop duplicating our analysis code across >> Nutch/Solr/Lucene, we should separate out the analyzers into a >> standalone package, and maybe as its own sub-project under the Lucene >> tlp? >> >> The goal would be eventually to have a single source for all our >> analysis needs, and for all Lucene projects to eventually cutover to >> this source (deprecating their current analysis code). >> >> We could also at this time fix some of the known problems in the >> analysis APIs, eg that the Analyzer base class confusingly exposes >> both non-reuse and reuse APIs, that not all Analyzers are final, etc. >> >> What do people think...? >> >> Mike >
