dweiss merged PR #13177:
URL: https://github.com/apache/lucene/pull/13177
--
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
jpountz commented on PR #13178:
URL: https://github.com/apache/lucene/pull/13178#issuecomment-1993797886
I'm curious what problem you are trying to address. It looks like you're
trying to avoid the overhead of checking the number of clauses, but intuitively
this wouldn't help much as we hav
jpountz merged PR #13125:
URL: https://github.com/apache/lucene/pull/13125
--
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.apa
jpountz opened a new issue, #13179:
URL: https://github.com/apache/lucene/issues/13179
### Description
Currently, Lucene's I/O concurrency is bound by the search concurrency. If
`IndexSearcher` runs on N threads, then Lucene will never perform more than N
I/Os concurrently. Unless yo
jpountz commented on PR #13153:
URL: https://github.com/apache/lucene/pull/13153#issuecomment-1994160216
Have you seen interesting performance numbers with this change?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use t
benwtrent commented on code in PR #13124:
URL: https://github.com/apache/lucene/pull/13124#discussion_r1523142360
##
lucene/core/src/java/org/apache/lucene/codecs/lucene99/Lucene99HnswVectorsFormat.java:
##
@@ -152,7 +154,25 @@ public Lucene99HnswVectorsFormat() {
* @param b
easyice commented on PR #13172:
URL: https://github.com/apache/lucene/pull/13172#issuecomment-1994345997
Thanks @dweiss , I found a seed `837FF885325AC743` that can reproduce this
failure on the main branch, but it's not stable, it may fail once in about 10
runs. with this patch, I ran it 1
benwtrent commented on PR #13124:
URL: https://github.com/apache/lucene/pull/13124#issuecomment-1994372130
@zhaih @jpountz I am going to create a separate issue around making HNSW
worker slicing automatic. It will require a bunch of its own benchmarking and
work and honestly seems orthogona
jpountz commented on PR #13124:
URL: https://github.com/apache/lucene/pull/13124#issuecomment-1994373820
Thanks for sharing performance numbers @benwtrent, very interesting. Also
double checking if you saw my above comment:
https://github.com/apache/lucene/pull/13124#pullrequestreview-19304
benwtrent commented on PR #13124:
URL: https://github.com/apache/lucene/pull/13124#issuecomment-1994399018
@jpountz would you prefer something like the original patch from @dweiss ? I
can submit the merging actions independently to the intra-merge executor.
Anything more (like figurin
jpountz commented on PR #13124:
URL: https://github.com/apache/lucene/pull/13124#issuecomment-1994458100
Yes exactly, something very simple, mostly to exercise intra-merge
concurrency with more than just vectors.
--
This is an automated message from the Apache Git Service.
To respond to t
benwtrent opened a new pull request, #13181:
URL: https://github.com/apache/lucene/pull/13181
With quantized vectors, and with current vectors, we separate out the
"scoring" vs. "iteration", requiring the user to always iterate the raw vectors
and provide their own similarity function.
jpountz commented on issue #13152:
URL: https://github.com/apache/lucene/issues/13152#issuecomment-1994488165
I agree that the penalty feels too high. We have challenges with queries
like this one because there is no good theoretical basis for the right way to
score such queries (at least t
jpountz commented on PR #12842:
URL: https://github.com/apache/lucene/pull/12842#issuecomment-1994545075
It looks like `writeGroupVInt` has room for improvement. Can we improve it
by making it look a bit more like the read logic?
--
This is an automated message from the Apache Git Service
jpountz commented on PR #13149:
URL: https://github.com/apache/lucene/pull/13149#issuecomment-1994666062
Would we be subject to the same issue if/when 3+ different implementations
of `DocIdSetIterator` get used in `IntersectVisitor#visit`?
--
This is an automated message from the Apache G
benwtrent opened a new issue, #13182:
URL: https://github.com/apache/lucene/issues/13182
### Description
Opening an issue to continue discussion originating here:
https://github.com/apache/lucene/pull/13076#issuecomment-1930363479
Making vector similarities pluggable via SP
benwtrent commented on PR #13124:
URL: https://github.com/apache/lucene/pull/13124#issuecomment-1994921965
> Yes exactly, something very simple, mostly to exercise intra-merge
concurrency with more than just vectors.
Latest commit adds `TaskExecutor` actions to merge to allow differen
zhaih commented on code in PR #13124:
URL: https://github.com/apache/lucene/pull/13124#discussion_r1523620861
##
lucene/core/src/java/org/apache/lucene/codecs/lucene99/Lucene99HnswVectorsFormat.java:
##
@@ -152,7 +154,25 @@ public Lucene99HnswVectorsFormat() {
* @param beamW
antonha commented on PR #13149:
URL: https://github.com/apache/lucene/pull/13149#issuecomment-1995011184
> Would we be subject to the same issue if/when 3+ different implementations
of `DocIdSetIterator` get used in `IntersectVisitor#visit`?
Yes.
Your question makes me think th
jpountz commented on PR #13149:
URL: https://github.com/apache/lucene/pull/13149#issuecomment-1995156682
Relatedly indeed, I was wondering if the API should expose an IntsRef or
something like that, so that there is a single virtual call per block of doc
IDs anyway (IntsRef cannot be extend
benwtrent commented on PR #13124:
URL: https://github.com/apache/lucene/pull/13124#issuecomment-1995183835
@jpountz ok, that naive attempt failed as norms & terms apparently need to
be merged in order (one or the other would fail due to missing files...).
I am not sure if this is tru
rishabhmaurya commented on code in PR #13123:
URL: https://github.com/apache/lucene/pull/13123#discussion_r1523704731
##
lucene/misc/src/java/org/apache/lucene/misc/index/BPIndexReorderer.java:
##
@@ -341,116 +344,94 @@ protected void compute() {
*/
private boolean sh
antonha commented on PR #13149:
URL: https://github.com/apache/lucene/pull/13149#issuecomment-1996041622
@jpountz I had a quick look at the code, and it seems to me like there are,
with this PR, only two implementations used for the DISI used for the
`IntersectVisitor#visit` method - which
rmuir commented on PR #13177:
URL: https://github.com/apache/lucene/pull/13177#issuecomment-1996150774
Thanks for doing this. I was disappointed that java doesn't allow this:
```
jshell> ProcessHandle.current().destroyForcibly()
| Exception java.lang.IllegalStateException: destroy o
github-actions[bot] commented on PR #13142:
URL: https://github.com/apache/lucene/pull/13142#issuecomment-1996171442
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
github-actions[bot] commented on PR #12985:
URL: https://github.com/apache/lucene/pull/12985#issuecomment-1996171619
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
26 matches
Mail list logo
