You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@lucene.apache.org by Michael Ryan <mr...@moreover.com> on 2014/07/15 15:24:25 UTC
NumericFacets assumptions
I'm looking at org.apache.solr.request.NumericFacets.getCounts(), trying to get an idea of how it works. Specifically looking at this bit of code:
final List<AtomicReaderContext> leaves = searcher.getIndexReader().leaves();
final Iterator<AtomicReaderContext> ctxIt = leaves.iterator();
for (DocIterator docsIt = docs.iterator(); docsIt.hasNext(); ) {
final int doc = docsIt.nextDoc();
if (ctx == null || doc >= ctx.docBase + ctx.reader().maxDoc()) {
do {
ctx = ctxIt.next();
} while (ctx == null || doc >= ctx.docBase + ctx.reader().maxDoc());
switch (numericType) {
case LONG:
longs = DocValues.getNumeric(ctx.reader(), fieldName);
Am I right that it is assuming that the docs are in order? This confused me because the javadoc for DocIterator says "The order of the documents returned by this iterator is non-deterministic". For most of the DocIterator implementations, the docs are returned in order, which I guess is how this works?
Also, am I right that it is assuming the index reader leaves are in order? That is, each leaf has higher doc ids than the previous leaf? I'm wondering too if that is always true, or if that is just how the implementation works now.
(I'm asking these questions because I want to do something very similar in a custom bit of code I'm writing, and would be nice if I could safely make these same assumptions.)
-Michael