Detailed description of query processing functionality

[Joel Mackenzie <jo...@gmail.com>]

Hi all,

I'm trying to understand how Lucene deals with particular query types. 
My main case is a query where there is a mixture of Should and Must 
clauses, potentially with some MinimumShouldMatch criteria.

Firstly, I understand that Lucene operates as a documented-ordered 
index, and hence processes queries in a Document-at-a-Time manner. My 
question is how does Lucene manage these types of query under the hood? 
For example, do we first intersect the Must lists and build a bit-vector 
which is then used to jump to the matching documents, which can then be 
scored and returned? Or do we simply do this on-the-fly as a single pass 
traversal? Perhaps a combination of both?

Any information would be greatly appreciated, and of course, any 
examples or relevant code/snippets are a bonus.

Joel