[GitHub] [lucene] benwtrent commented on a diff in pull request #11860: GITHUB-11830 Better optimize storage for vector connections

benwtrent commented on code in PR #11860: URL: https://github.com/apache/lucene/pull/11860#discussion_r1039687827 ## lucene/core/src/java/org/apache/lucene/codecs/lucene95/Lucene95HnswVectorsFormat.java: ## @@ -0,0 +1,192 @@ +/* + * Licensed to the Apache Software Foundation (AS

[GitHub] [lucene] luyuncheng commented on pull request #11987: Make Decompressor release memory buffer

luyuncheng commented on PR #11987: URL: https://github.com/apache/lucene/pull/11987#issuecomment-1337530535 > the idea is not to have one instance per segment. > but it means we can reuse buffers per-search and per-merge rather than creating per-document garbage. basically it would work

[GitHub] [lucene] rmuir commented on pull request #11987: Make Decompressor release memory buffer

rmuir commented on PR #11987: URL: https://github.com/apache/lucene/pull/11987#issuecomment-1337607709 sorry, we definitely should not be adding arrays to bytesref here. Like i said, we can just remove the threadlocal. The issue to me has nothing to do with buffers. it has to do with

[GitHub] [lucene] luyuncheng commented on pull request #11987: Make Decompressor release memory buffer

luyuncheng commented on PR #11987: URL: https://github.com/apache/lucene/pull/11987#issuecomment-1337627048 @rmuir Thanks for your replying! i am looking forward to your no-threadlocal-design. i have no question about it, i'll close this issue -- This is an automated message fr

[GitHub] [lucene] luyuncheng closed pull request #11987: Make Decompressor release memory buffer

luyuncheng closed pull request #11987: Make Decompressor release memory buffer URL: https://github.com/apache/lucene/pull/11987 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment.

[GitHub] [lucene] fcofdez opened a new pull request, #11997: Add IntField, LongField, FloatField and DoubleField

fcofdez opened a new pull request, #11997: URL: https://github.com/apache/lucene/pull/11997 This commit adds new IndexableFields that index both points and doc values at once. Closes #11199 -- This is an automated message from the Apache Git Service. To respond to the message, plea

[GitHub] [lucene] zhaih commented on issue #11913: lucene-replicator PrimaryNode unsafely publishes reference during construction

zhaih commented on issue #11913: URL: https://github.com/apache/lucene/issues/11913#issuecomment-1337918394 Can we have a `initialized` field inside `PrimaryNode` and let `PreCopyMergedSegmentWarmer` check against it before making any call to `PrimaryNode`? -- This is an automated messag

[GitHub] [lucene] mdmarshmallow commented on pull request #11901: Github#11869: Add RangeOnRangeFacetCounts

mdmarshmallow commented on PR #11901: URL: https://github.com/apache/lucene/pull/11901#issuecomment-1338159980 Hey @gsmiller, sorry for the delay but finally got around to your feedback. I decided to just add multidimensional support since it was fairly straightforward (if not a bit tedious

[GitHub] [lucene] stevenschlansker commented on issue #11913: lucene-replicator PrimaryNode unsafely publishes reference during construction

stevenschlansker commented on issue #11913: URL: https://github.com/apache/lucene/issues/11913#issuecomment-1338221987 I think that would fix this specific problem, although where would you set `initialized`? The end of the PrimaryNode constructor is too soon, since the subclass constructor

[GitHub] [lucene] benwtrent commented on a diff in pull request #11860: GITHUB-11830 Better optimize storage for vector connections

benwtrent commented on code in PR #11860: URL: https://github.com/apache/lucene/pull/11860#discussion_r1040218979 ## lucene/core/src/java/org/apache/lucene/util/hnsw/HnswGraph.java: ## @@ -144,6 +145,10 @@ public NodesIterator(int size) { this.size = size; } +p

[GitHub] [lucene] rmuir opened a new pull request, #11998: Migrate away from per-segment-per-threadlocals on SegmentReader

rmuir opened a new pull request, #11998: URL: https://github.com/apache/lucene/pull/11998 Currently the stored fields and term vectors apis on the index are "stateless". Unlike the other parts of the APIs, users can't call any iterators/enumerators, they just do stuff like: ``` ind