Adrien Grand

Can UTF8StreamJsonParser#finishString be made faster with VarHandles?

2

I was looking at `UTF8StreamJsonParser#finishString` which seems to mostly consist of scanning for a trailing quote, with an optimized code path for ASCII strings. This optimized code path for ASCII...

TestIDVersionPostingsFormat failure

4

### Description This failure is not reproducible, which is maybe not too surprising given that the test involves concurrency. A few things are interesting: - This is the same failure...

### Description A long time ago, we introduced `Weight#scorerSupplier` to enable [query planning for range queries](https://www.elastic.co/blog/better-query-planning-for-range-queries-in-elasticsearch), using either points or doc values to run a range query depending on which...

Improve Lucene's I/O concurrency

11

### Description Currently, Lucene's I/O concurrency is bound by the search concurrency. If `IndexSearcher` runs on N threads, then Lucene will never perform more than N I/Os concurrently. Unless you...

Make `OneMerge#reorder` preserve blocks.

1

This updates `IndexWriter` to only call `OneMerge#reorder` when it has a chance to preserve the block structure, ie. either there are no blocks or blocks are identified by a parent...

Can we decrease the overhead of skipping?

1

### Description On top-k queries, Lucene is now competitive with Tantivy/PISA on https://tantivy-search.github.io/bench/, but it's still quite slower on counting queries. This made me want to run a similar experiment...

### Description @uschindler asked this question in https://lists.apache.org/thread/6o3hn3x8syfm8lj93kk5rrxb0kx701gp. In this discussion, we were looking for introducing the ability to iterate deleted docs, in order to compute (cheaply!) some facets across...

Cut over HNSW's neighbor lists to group-varint?

1

### Description HNSW's neighbor lists are like a short postings list, and we moved short postings lists from vints to group-varint recently, which looks like it yielded a small performance...

Since Anserini is often used for search performance benchmarks, enabling recursive graph bisection would help. For reference, most if not all PISA performance benchmarks seem to enable recursive graph bisection....

Use same BM25 k1/b parameters across engines.

1

The k1 and b parameters of BM25 can influence what hits may be dynamically pruned and thus performance numbers, so it would be good to use the same values across...