Lucene: Status=Open

[Save this search]

Author relation

Updated ago

Comment count

Reaction count

Review Requested

User

Last comment user

Type

Filters: Status (Open), Issue type, Author relation, Created, Updated, Updated ago, Comment count, Reaction count, Review Requested, Mentioned, Reviewed, Commented, User, Last comment user, Draft, Component, Type, Labels, Commits?, Reporter, Assignee

#15620 PR: Add Temporal Merge Policy for time-series data

16.1 minutes ago 34 comments 0 votes 0 watches benwtrent, churromorales, github-actions[bot], msfroh, shubhamsrkdev

Description This PR introduces TemporalMergePolicy, a new merge policy designed for time-series ... The policy groups segments into time windows and merges segments within the same window, but never ... This preserves temporal locality and improves query performance for time-range queries. relates to ...

shubhamsrkdev 2.3 hours ago:	I guess it's fine - since its an advanced knob, I don't mind. If it wasn't, a preset (an enum of say 3 levels?) would have been better, not a concern here though..
shubhamsrkdev 2.5 hours ago:	Could this be done more efficiently? I understand we create multiple lists and then clear the elements from original list. Maybe this could be better done with pointers or Queue?

#15700 PR: Fix Tessellation error for holes that share the same leftmost vertex with an outer ring vertex

1.5 hours ago 0 comments 0 votes 0 watches github-actions[bot], roblburris

Description Apologies in advance if this is garbage, but I had Claude make an attempt at fixing ... If multiple holes share the same leftmost vertex, the ring accumulates multiple copies of that ... The angle-based getSharedInsideVertex tiebreaker could select the wrong copy when merging ...

#15699 PR: Update CHANGES.txt to reflect my optimization PR 15637

2 hours ago 2 comments 0 votes 0 watches countmdm, dweiss, github-actions[bot], shubhamsrkdev

Description Update CHANGES.txt as requested by reviewers of my PR 15637.

countmdm 2 hours ago:	> I think there is already a CHANGES.txt entry (targeting 10.5) for 15637 . I think @dweiss added it for your PR? I am somewhat confused by how to keep track of all Lucene branches.
shubhamsrkdev 2.1 hours ago:	I think there is already a CHANGES.txt entry (targeting 10.5) for #15637 . I think @dweiss added it for your PR?

#15697: For 4-bit quantized vectors, should we change the scheme of unpacking nibbles?

3.2 hours ago 2 comments 0 votes 0 watches benwtrent, kaivalnp

Description 4-bit quantized vectors are stored as nibbles, two packed in a single byte -- where ... Example: [1 5][2 6][3 7][4 8] For performing individual vector computations, we "unpack" the ... For higher dimension vectors, can this lead to CPU cache line misses?

kaivalnp 3.2 hours ago:	Also pasting a sample JMH benchmark (from 15698) run using: ` java --module-path lucene/ ... SinglePacked." -p size=1024 ` ` Benchmark ... Individual JMH iterations started strong, but performance dropped after some iterations.
benwtrent 3.3 hours ago:	@kaivalnp I agree, there is likely performance being left on the ground here. My intrinsics knowledge isn't strong enough to know where to look however :)

#15696 PR: Use SkipBlockRangeIterator in TermOrdValComparator

6 hours ago 1 comments 0 votes 0 watches benwtrent, github-actions[bot], romseygeek

DocValuesRangeIterator will duplicate lots of work here, in the worst case not doing any pruning at ... Switch to using SkipBlockRangeIterator instead, which only checks skip block boundaries and leaves ... Closes 15674

benwtrent 6 hours ago:

This significantly reduces the complexity of this individual class. But also allow better use of the bulk APIs. It makes me think that DocValuesRangeIterator is rather trappy when it comes to good performance :( ...

#15636 PR: Allow TermsEnum to express a preference for seekExact to FrozenBufferedUpdates

7.2 hours ago 3 comments 0 votes 0 watches ankesh-airon-mongodb, benwtrent, dabao521, github-actions[bot], kkmr, mccullocht

For Terms implementations that have an approximate membership filter, the current scheme of ... Allow TermsEnum to express a preference for seekExact() so that approximate membership filters ... Implement preferSeekExact() in BloomFilteringPostingsFormat where there is obvious benefit.

dabao521 2.3 days ago:

LGTM! Thanks for adding the extension and this will be super helpful to tune the bloom filter behavior ...

#15508 PR: Use native dot product in Lucene

9.6 hours ago 40 comments 0 votes 0 watches dweiss, github-actions[bot], goankur, rmuir, shubhamvishu

Description This is continuation of the work @goankur started in https://github.com/apache/ ... Quoting the comment below from the initial PR.

dweiss 9.6 hours ago:	If that's the case then sure - leave the branching - but only in those classes that are dedicated ... The bottom line for me is that if it isn't available (can't be loaded) there should be no change ... It should also make it easier to untagle those implementations, should someone retry my efforts ...
shubhamvishu 10.8 hours ago:	I agree we can have some scorers(dedicated to native; something like NativeDotProductScorer etc) ... Not having this branching would mean user needs to implement all the functions from VectorUtil ... Do you think we should mandate that or maybe what you are mentioning solves that problem also or ...

«
1
2
3
4
5
»

2,519 issues

[55.5 msec search, 56.4 msec total]