dweiss commented on PR #15012:
URL: https://github.com/apache/lucene/pull/15012#issuecomment-3138707532
It's fine, I just mentioned it because it's a relatively new thing and while
tests may pass locally, they'll fail on gh because of extra checks performed
only there.
--
This is an auto
dVenkatNaveen commented on issue #15000:
URL: https://github.com/apache/lucene/issues/15000#issuecomment-3138626263
Thanks for the update @msokolov. I have another follow-up question.
The following statement retrieves my top 20 (maxHits)search results based on
score:
ScoreDoc[
markrmiller commented on PR #15012:
URL: https://github.com/apache/lucene/pull/15012#issuecomment-3138540515
Yeah, this isn't something that is about to be committed any second now.
This is something for those looking at the Github issue to take a look it -
hopefully not for the formatting
markrmiller commented on issue #13797:
URL: https://github.com/apache/lucene/issues/13797#issuecomment-3138499219
@rmuir does this match what you were thinking?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL a
github-actions[bot] commented on PR #15015:
URL: https://github.com/apache/lucene/pull/15015#issuecomment-3138379692
This PR does not have an entry in lucene/CHANGES.txt. Consider adding one.
If the PR doesn't need a changelog entry, then add the skip-changelog label to
it and you will stop
corecursion opened a new pull request, #15015:
URL: https://github.com/apache/lucene/pull/15015
This PR provides `MultiIndexMergeScheduler` for multi-tenant merge
scheduling for Lucene. A version of `MultiIndexMergeScheduler` has been in
production use successfully at MongoDB for many years
rmuir merged PR #15002:
URL: https://github.com/apache/lucene/pull/15002
--
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.
To unsubscribe, e-mail: issues-unsubscr...@lucene.apach
rmuir closed issue #14999: Performance bottleneck in
IndexReader.registerParentReader with many parentReaders
URL: https://github.com/apache/lucene/issues/14999
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL abov
msokolov opened a new pull request, #15014:
URL: https://github.com/apache/lucene/pull/15014
This is a followon from https://github.com/apache/lucene/pull/15013 rebased
on branch 10x. I added grouping by block using updateDocuments so that
parent/child docs would stay together. I haven't se
msokolov commented on issue #15000:
URL: https://github.com/apache/lucene/issues/15000#issuecomment-3137942060
Searching for the term 'e' among documents that all have a single 'e' in
them is going to result in a more or less random order, as you've observed. I
would say the behavior is con
msokolov closed issue #15000: Lucene 5.x to 9.12.1 Migration – Change in Search
Result Ordering
URL: https://github.com/apache/lucene/issues/15000
--
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
dweiss merged PR #15006:
URL: https://github.com/apache/lucene/pull/15006
--
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.
To unsubscribe, e-mail: issues-unsubscr...@lucene.apac
dweiss merged PR #15008:
URL: https://github.com/apache/lucene/pull/15008
--
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.
To unsubscribe, e-mail: issues-unsubscr...@lucene.apac
dweiss commented on PR #15012:
URL: https://github.com/apache/lucene/pull/15012#issuecomment-3137693395
(in fact, if you have eclint and run ```./gradlew tidy
-Plucene.tool.eclint=eclint```, it'll also run the fixes.
--
This is an automated message from the Apache Git Service.
To respond
dweiss commented on PR #15012:
URL: https://github.com/apache/lucene/pull/15012#issuecomment-3137687513
Hi Mark.
https://github.com/apache/lucene/blob/main/help/formatting.txt#L32-L40 - if you
have the tool installed, eclint -fix should also apply the required fixes (if
possible).
--
Th
msokolov commented on issue #15005:
URL: https://github.com/apache/lucene/issues/15005#issuecomment-313769
So I think this query will be broken if any parent documents have the vector
field defined for them, which seems a bit weird, but ... OK
--
This is an automated message from the
msokolov commented on issue #15005:
URL: https://github.com/apache/lucene/issues/15005#issuecomment-3137627553
I'm starting to try and grok how this Query works, but it is tricky and I do
see that when there are multiple leaves it relies on a global heap
(BlockingFloatHeap)
--
This is an
dweiss merged PR #15009:
URL: https://github.com/apache/lucene/pull/15009
--
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.
To unsubscribe, e-mail: issues-unsubscr...@lucene.apac
ChrisHegarty commented on issue #15005:
URL: https://github.com/apache/lucene/issues/15005#issuecomment-3137597239
The test in Elasticsearch that reproduces this consistently has 2 segments:
1. Segment 1 has two docs. The first doc has 2 children with a single vector
each, the second doc
msokolov opened a new pull request, #15013:
URL: https://github.com/apache/lucene/pull/15013
First stab at a unit test to reproduce problem reported in
https://github.com/apache/lucene/issues/15005
The test fails with some assertion in
`DiversifyingNearestChildrenKnnCollector.collect
msokolov commented on issue #15005:
URL: https://github.com/apache/lucene/issues/15005#issuecomment-3137521742
It looks as if `ParentBlockJoinKnnVectorQueryTestCase` is the place to look,
but when I look there I don't see any test case with more than a handful of
vectors being indexed. I c
jpountz commented on code in PR #15011:
URL: https://github.com/apache/lucene/pull/15011#discussion_r2243394074
##
lucene/core/src/java/org/apache/lucene/search/AcceptDocs.java:
##
@@ -0,0 +1,194 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * c
ChrisHegarty commented on issue #15005:
URL: https://github.com/apache/lucene/issues/15005#issuecomment-3137008733
Yeah, I think so. The PR that added support has some more information, see
`Add ParentJoin KNN support` #12434.
My next steps would be modify one of the tests added by #1
jpountz commented on code in PR #14996:
URL: https://github.com/apache/lucene/pull/14996#discussion_r2243070631
##
lucene/core/src/java/org/apache/lucene/util/FixedBits.java:
##
@@ -31,8 +33,17 @@ public boolean get(int index) {
}
@Override
- public void applyMask(Fixed
jpountz commented on code in PR #14996:
URL: https://github.com/apache/lucene/pull/14996#discussion_r2243068238
##
lucene/core/src/java/org/apache/lucene/index/SoftDeletesDirectoryReaderWrapper.java:
##
@@ -188,11 +189,11 @@ private static boolean assertDocCounts(
static fi
kaivalnp commented on PR #15010:
URL: https://github.com/apache/lucene/pull/15010#issuecomment-3136781663
Thanks @msokolov!
--
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.
msokolov commented on issue #15005:
URL: https://github.com/apache/lucene/issues/15005#issuecomment-3136725287
hm, this is disturbing - I wonder if we are lacking some test coverage for
nested vector field search? TBH I'm not sure what Lucene class implements that.
Is it the `DiversifyingCh
msokolov commented on PR #15010:
URL: https://github.com/apache/lucene/pull/15010#issuecomment-3136700904
I got rid of most of the labels
--
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 speci
msokolov merged PR #15010:
URL: https://github.com/apache/lucene/pull/15010
--
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.
To unsubscribe, e-mail: issues-unsubscr...@lucene.ap
markrmiller commented on issue #13797:
URL: https://github.com/apache/lucene/issues/13797#issuecomment-3136505419
Hi all,
So the proposal is a simple change: **stop forcing a full re‑index on every
major release unless we actually break the on‑disk format**.
All we need is to maint
markrmiller opened a new pull request, #15012:
URL: https://github.com/apache/lucene/pull/15012
…ple Major Versions #13797
--
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.
T
shubhamvishu commented on PR #14963:
URL: https://github.com/apache/lucene/pull/14963#issuecomment-3136470025
Thanks for the review @vigyasharma!
> What happens if I create an index with bypassTinySegments=true, but later
read it in an application with the flag set to false? I think w
shubhamvishu commented on code in PR #14963:
URL: https://github.com/apache/lucene/pull/14963#discussion_r2242742833
##
lucene/core/src/java/org/apache/lucene/codecs/lucene99/Lucene99HnswVectorsWriter.java:
##
@@ -598,10 +629,21 @@ static FieldWriter create(
FieldInfo f
shubhamvishu commented on PR #14963:
URL: https://github.com/apache/lucene/pull/14963#issuecomment-3136408719
> Or some higher-level abstraction that can either be consumed in a
random-access fashion (Bits) or sequential (DocIdSetIterator)?
Thanks @jpountz. I opened https://github.com
github-actions[bot] commented on PR #15011:
URL: https://github.com/apache/lucene/pull/15011#issuecomment-3136408257
This PR does not have an entry in lucene/CHANGES.txt. Consider adding one.
If the PR doesn't need a changelog entry, then add the skip-changelog label to
it and you will stop
shubhamvishu opened a new pull request, #15011:
URL: https://github.com/apache/lucene/pull/15011
### Description
Addresses [this
comment](https://github.com/apache/lucene/pull/14963#issuecomment-3092568029)
on the PR #14963 to allow both sequential and random access consumption f
ChrisHegarty commented on PR #14978:
URL: https://github.com/apache/lucene/pull/14978#issuecomment-3136203382
I added a note to the 10.3 API section of the change log, since we'll likely
backport this.
--
This is an automated message from the Apache Git Service.
To respond to the message,
ChrisHegarty commented on PR #14978:
URL: https://github.com/apache/lucene/pull/14978#issuecomment-3136179000
This is looking very good. I'm just doing some additional testing and
benchmarking before final review.
Additionally, I added a unit test for the bulk scorer, that verifies bu
github-actions[bot] commented on PR #14978:
URL: https://github.com/apache/lucene/pull/14978#issuecomment-3135996074
This PR does not have an entry in lucene/CHANGES.txt. Consider adding one.
If the PR doesn't need a changelog entry, then add the skip-changelog label to
it and you will stop
msokolov commented on code in PR #14996:
URL: https://github.com/apache/lucene/pull/14996#discussion_r2242295341
##
lucene/core/src/java/org/apache/lucene/index/SoftDeletesDirectoryReaderWrapper.java:
##
@@ -188,11 +189,11 @@ private static boolean assertDocCounts(
static f
N624-debu commented on PR #14900:
URL: https://github.com/apache/lucene/pull/14900#issuecomment-3135258237
> This PR does not have an entry in lucene/CHANGES.txt. Consider adding one.
If the PR doesn't need a changelog entry, then add the skip-changelog label to
it and you will stop receivi
github-actions[bot] commented on PR #14900:
URL: https://github.com/apache/lucene/pull/14900#issuecomment-3135188800
This PR does not have an entry in lucene/CHANGES.txt. Consider adding one.
If the PR doesn't need a changelog entry, then add the skip-changelog label to
it and you will stop
42 matches
Mail list logo
