Jonathan Ellis
Jonathan Ellis
> There are three times that are useful to understand the execution behavior in ANN: (1) time to materialize the indexed predicates, (2) time to perform the ANN search, (3)...
Troubleshooting in the wild has shown that having sstable-level details is actually useful. So let's put this on the shelf for now.
So we have a 14% win on the hornet lat/lon simulated queries and a bunch of changes that could be responsible: 1. the nextPosting vs advance optimization in PLRI 2....
this is reasonable, but giving it a (possibly incorrect) default value seems fragile for a public method. short of deeper surgery, I think making it a Boolean and leaving it...
@tjake do you have time to review this?
> I guess, if we had a way to tell in advance if reading the row from the first few sstables will trigger an S3 download, then we could decide...
No preference. On Tue, Feb 17, 2015 at 11:54 AM, Lyuben Todorov [email protected] wrote: > Is there a preference to one large patch or aggregated commits per client? > >...
(low priority imo b/c the reranker only calls getVectorInto and not getVector)
If/when we add back concurrent removeDeletedNodes, this is the PR where it got removed: https://github.com/jbellis/jvector/pull/273
cc @mdogan since I think you're potentially interested in having a concurrent `removeDeletedNodes`