Adam, mine were just some general guidelines, but George didn't take time to confirm them so we still don't know whether they are valid or not. About literal conversion from Java, I am aware that Lucene.Net would benefit very much from pure .NET features, but afaik there are a couple of reasons why it's like that. First, Java Lucene documentation is pretty good while .NET's is lacking, so keeping the same API brings some advantages when it comes to understand what a class or method does. Second, a 1-1 port implies index compatibility, which is a great advantage for interoperability and since you can use tools written in Java, like Luke, to analyze your indexes, since such tools don't exist for .NET. Finally, it looks like George is practically alone as a committer, which is not good, but since this is the situation creating a new API would require too much work for him.
Simone Adam Hurwitz - BIA wrote: > To continue a thread from earlier: Simone Busoli recommends that in > order to get involved with Lucene.Net, we should compare the source code > with the java 2.1 version and find any differences and patch them. Does > this mean that there is a running list somewhere of classes that have > already been compared? Which ones need work? > > Also, in response to George's request for developers: it would be a lot, > lot easier to get people to work on Lucene.Net if it wasn't supposed to > be such a literal translation of the java code. Straight translation is > boring. I don't see why a .net style can't be introduced immediately. I > like and use Lucene.net and would like to see it stay healthy.
