On 1/21/07, Mike Klaas <[EMAIL PROTECTED]> wrote:
On 1/20/07, Yonik Seeley (JIRA) <[EMAIL PROTECTED]> wrote:
> > Looking at the negative filters stuff, I realized that andNot() had no
optimized implementation for HashDocSet, so I implemented that and union().
Out of curiosity, what is your current plan for this? Something along
the lines of storing a negated flag, which would be used to do
andNot() rather than intersection() in SolrIndexSearcher.getDocSet()?
I think it would be a great feature and can help out with devel or review.
There are two related things, and I'm only tackling one. I'm *not
*looking at a generated DocSet and then choosing to try and cache it
as an inverse if it would be smaller.
I am looking at queries, and determining if they are negative (no
positive elements, currently matches nothing in Lucene). If they are
negative, I generate and cache the positive version, and do andNot()
for operations.
Code is done but untested (no test code yet even). I'll add a draft
for your review now.
-Yonik
