github-actions[bot] commented on PR #14078:
URL: https://github.com/apache/lucene/pull/14078#issuecomment-2626006980
This PR has not had activity in the past 2 weeks, labeling it as stale. If
the PR is waiting for review, notify the d...@lucene.apache.org list. Thank you
for your contributi
sgup432 commented on issue #14183:
URL: https://github.com/apache/lucene/issues/14183#issuecomment-2625954720
@jpountz If you think it is feasible via above approach, I can quickly raise
a PR on this with some UTs.
--
This is an automated message from the Apache Git Service.
To respond t
sgup432 commented on issue #14183:
URL: https://github.com/apache/lucene/issues/14183#issuecomment-2625661420
@jpountz Yeah I meant we could use something like AtomicReference to set it
dynamically in a thread safe way.
I meant harmless in a sense that it should be easy to implement via
msokolov commented on PR #14187:
URL: https://github.com/apache/lucene/pull/14187#issuecomment-2625649749
I agree with the judgment, but maybe this just indicates we need to improve
the cache! Still, until we can figure out how to do so, +1 to disable by default
--
This is an automated me
jpountz commented on issue #14183:
URL: https://github.com/apache/lucene/issues/14183#issuecomment-2625648861
I am not entirely sure about the "harmless" part: this class is shared by
multiple threads so we would need to make sure that the value is updated in a
thread-safe way.
--
This i
jpountz opened a new pull request, #14187:
URL: https://github.com/apache/lucene/pull/14187
The query cache trades heap for faster queries. Given all the progress that
has been made on making uncached queries faster (`IndexOrDocValuesQuery`,
bitset encoding of blocks of postings, etc.), it'
jpountz commented on issue #14148:
URL: https://github.com/apache/lucene/issues/14148#issuecomment-2625595731
> That's the only thing that prevents MergePolicy from e.g. simply picking
that merge again.
I wonder if we need to prevent it from picking the same merge again. Could
we wai
mikemccand commented on issue #14148:
URL: https://github.com/apache/lucene/issues/14148#issuecomment-2625515650
Oh, that's a neat idea (`MergeScheduler` being able to "cancel" merge
choices from `MergePolicy`).
I think we would have to register the merge, immediately, on getting it f
jpountz commented on PR #14186:
URL: https://github.com/apache/lucene/pull/14186#issuecomment-2625505749
Interestingly, this makes queries run faster:
```
TaskQPS baseline StdDevQPS
my_modified_version StdDevPct diff p-value
jpountz commented on PR #14186:
URL: https://github.com/apache/lucene/pull/14186#issuecomment-2625445431
I am annoyed that I'm not able to create a proper test for this. It seems to
require very specific conditions, I found it while playing with score
quantization.
--
This is an automate
jpountz opened a new pull request, #14186:
URL: https://github.com/apache/lucene/pull/14186
This is a rare bug (for instance none of the queries in nightly benchmarks
return different top hits with the fix, and I haven't been able to create a
proper test) but still a bug.
--
This is
benwtrent commented on code in PR #14181:
URL: https://github.com/apache/lucene/pull/14181#discussion_r1936117214
##
lucene/core/src/java/org/apache/lucene/util/hnsw/HnswGraphBuilder.java:
##
@@ -296,6 +282,30 @@ to the newly introduced levels (repeating step 2,3 for new
levels
benwtrent commented on code in PR #14181:
URL: https://github.com/apache/lucene/pull/14181#discussion_r1936115985
##
lucene/core/src/java/org/apache/lucene/util/hnsw/HnswConcurrentMergeBuilder.java:
##
@@ -190,12 +191,25 @@ private int getStartPos(int maxOrd) {
}
}
benwtrent commented on code in PR #14181:
URL: https://github.com/apache/lucene/pull/14181#discussion_r1936115985
##
lucene/core/src/java/org/apache/lucene/util/hnsw/HnswConcurrentMergeBuilder.java:
##
@@ -190,12 +191,25 @@ private int getStartPos(int maxOrd) {
}
}
benwtrent commented on code in PR #14181:
URL: https://github.com/apache/lucene/pull/14181#discussion_r1936115373
##
lucene/core/src/java/org/apache/lucene/util/hnsw/HnswConcurrentMergeBuilder.java:
##
@@ -190,12 +191,25 @@ private int getStartPos(int maxOrd) {
}
}
benwtrent commented on code in PR #14181:
URL: https://github.com/apache/lucene/pull/14181#discussion_r1936114735
##
lucene/core/src/java/org/apache/lucene/codecs/hnsw/DefaultFlatVectorScorer.java:
##
@@ -90,23 +91,29 @@ public String toString() {
private static final class B
msokolov commented on code in PR #14181:
URL: https://github.com/apache/lucene/pull/14181#discussion_r1935932869
##
lucene/core/src/java/org/apache/lucene/codecs/hnsw/DefaultFlatVectorScorer.java:
##
@@ -90,23 +91,29 @@ public String toString() {
private static final class By
jpountz commented on PR #14185:
URL: https://github.com/apache/lucene/pull/14185#issuecomment-2625283827
Benchmarks suggest that the additional overhead in nextDoc() is fine:
```
TaskQPS baseline StdDevQPS
my_modified_version StdDev
jpountz opened a new pull request, #14185:
URL: https://github.com/apache/lucene/pull/14185
The recent optimization from #14164 interfered in a bad way with a prior
optimization.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHu
jpountz commented on PR #14164:
URL: https://github.com/apache/lucene/pull/14164#issuecomment-2625125069
This seems to have introduced a bug:
https://ci-builds.apache.org/job/Lucene/job/Lucene-Check-main/11964/. I think
it's related to how advanceShallow() sometimes refills the buffer.
--
benwtrent commented on PR #14160:
URL: https://github.com/apache/lucene/pull/14160#issuecomment-2624958162
I ran this over the "nightly" dataset (8M 768 dim vectors). No force
merging. I think this is the nightly behavior. I ran over various filter
criteria (I think nightly is 5%).
benwtrent commented on PR #14173:
URL: https://github.com/apache/lucene/pull/14173#issuecomment-2624759420
I like where this PR is going.
> Note: This change does not include dependent multi-valued vectors like
ColBERT, where the multiple vectors must used together to compute similari
mikemccand commented on PR #14101:
URL: https://github.com/apache/lucene/pull/14101#issuecomment-2624467638
Wow, this looks really awesome! I wish we could retroactively apply it to
PRs missing their module labels ... but let's start with this change (so new
PRs are labeled).
--
This is
kaivalnp commented on code in PR #14178:
URL: https://github.com/apache/lucene/pull/14178#discussion_r1935570579
##
lucene/sandbox/src/java22/org/apache/lucene/sandbox/codecs/faiss/LibFaissC.java:
##
@@ -0,0 +1,268 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) unde
pseudo-nymous commented on code in PR #14101:
URL: https://github.com/apache/lucene/pull/14101#discussion_r1935416351
##
.github/labeler.yml:
##
@@ -0,0 +1,134 @@
+# This file defines module label mappings for the Lucene project.
+# Each module is associated with a set of file g
kaivalnp commented on code in PR #14178:
URL: https://github.com/apache/lucene/pull/14178#discussion_r1935407529
##
lucene/sandbox/src/java22/org/apache/lucene/sandbox/codecs/faiss/LibFaissC.java:
##
@@ -0,0 +1,268 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) unde
kaivalnp commented on code in PR #14178:
URL: https://github.com/apache/lucene/pull/14178#discussion_r1935304089
##
lucene/sandbox/src/java22/org/apache/lucene/sandbox/codecs/faiss/LibFaissC.java:
##
@@ -0,0 +1,268 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) unde
kaivalnp commented on code in PR #14178:
URL: https://github.com/apache/lucene/pull/14178#discussion_r1935303469
##
lucene/sandbox/src/java22/org/apache/lucene/sandbox/codecs/faiss/FaissKnnVectorsWriter.java:
##
@@ -0,0 +1,204 @@
+/*
+ * Licensed to the Apache Software Foundatio
pseudo-nymous commented on code in PR #14101:
URL: https://github.com/apache/lucene/pull/14101#discussion_r1935253167
##
.github/labeler.yml:
##
@@ -0,0 +1,134 @@
+# This file defines module label mappings for the Lucene project.
+# Each module is associated with a set of file g
stefanvodita commented on code in PR #14101:
URL: https://github.com/apache/lucene/pull/14101#discussion_r1935220020
##
.github/labeler.yml:
##
@@ -0,0 +1,134 @@
+# This file defines module label mappings for the Lucene project.
+# Each module is associated with a set of file gl
pseudo-nymous commented on PR #14101:
URL: https://github.com/apache/lucene/pull/14101#issuecomment-2623844525
> This is great @pseudo-nymous! I left minor comments that I would like to
see addressed, but this is very good already. I retract my concern about the
action budget. This is runni
pseudo-nymous commented on PR #14101:
URL: https://github.com/apache/lucene/pull/14101#issuecomment-2623829618
Did further testing with new sync flag, change is working as expected.
Ref: https://github.com/pseudo-nymous/lucene/pull/3
--
This is an automated message from the Apache Git S
32 matches
Mail list logo
