dweiss commented on pull request #634:
URL: https://github.com/apache/lucene/pull/634#issuecomment-1029716363
Ping, @rmuir - WDYT? Should I merge this in or leave as a convenience patch
in jira?
--
This is an automated message from the Apache Git Service.
To respond to the message,
jtibshirani opened a new pull request #645:
URL: https://github.com/apache/lucene/pull/645
This PR proposes some renames to clarify the code structure. The top-level
`KnnGraphValues` is renamed to `HnswGraph`, since it now represents a
hierarchical graph. It's also moved from
mocobeta commented on pull request #643:
URL: https://github.com/apache/lucene/pull/643#issuecomment-1029592938
@uschindler @msokolov
Would you take a look at the revised constructors' API design when you have
some time? If this gains basic consensus, I will add tests for the newly
[
https://issues.apache.org/jira/browse/LUCENE-10404?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Julie Tibshirani updated LUCENE-10404:
--
Description:
While searching each layer, HNSW tracks the nodes it has already
Julie Tibshirani created LUCENE-10404:
-
Summary: Use hash set for visited nodes in HNSW search?
Key: LUCENE-10404
URL: https://issues.apache.org/jira/browse/LUCENE-10404
Project: Lucene - Core
jtibshirani commented on pull request #641:
URL: https://github.com/apache/lucene/pull/641#issuecomment-1029566793
I filed https://issues.apache.org/jira/browse/LUCENE-10404 so we don't
forget about the hash set idea.
--
This is an automated message from the Apache Git Service.
To
gautamworah96 commented on a change in pull request #632:
URL: https://github.com/apache/lucene/pull/632#discussion_r799102705
##
File path: lucene/facet/src/java/org/apache/lucene/facet/DrillSideways.java
##
@@ -285,53 +289,42 @@ public DrillSidewaysResult search(
}
gautamworah96 commented on a change in pull request #632:
URL: https://github.com/apache/lucene/pull/632#discussion_r799102042
##
File path: lucene/facet/src/java/org/apache/lucene/facet/DrillSideways.java
##
@@ -192,7 +192,11 @@ protected Facets buildFacetsResult(
*
gautamworah96 commented on a change in pull request #632:
URL: https://github.com/apache/lucene/pull/632#discussion_r799091771
##
File path: lucene/facet/src/java/org/apache/lucene/facet/DrillSideways.java
##
@@ -351,45 +344,36 @@ public DrillSidewaysResult search(ScoreDoc
[
https://issues.apache.org/jira/browse/LUCENE-10391?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17486759#comment-17486759
]
ASF subversion and git services commented on LUCENE-10391:
--
Commit
[
https://issues.apache.org/jira/browse/LUCENE-10403?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17486756#comment-17486756
]
Greg Miller commented on LUCENE-10403:
--
For what it's worth, I recently narrowly avoided an
[
https://issues.apache.org/jira/browse/LUCENE-10391?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17486752#comment-17486752
]
ASF subversion and git services commented on LUCENE-10391:
--
Commit
jtibshirani merged pull request #641:
URL: https://github.com/apache/lucene/pull/641
--
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:
gsmiller opened a new pull request #644:
URL: https://github.com/apache/lucene/pull/644
# Description
Add utility method to `ArrayUtil` allowing the user to omit a min size for
generically typed arrays, adding parity with capabilities for primitive type
arrays.
# Solution
jtibshirani commented on a change in pull request #641:
URL: https://github.com/apache/lucene/pull/641#discussion_r799057820
##
File path:
lucene/core/src/java/org/apache/lucene/util/hnsw/HnswGraphBuilder.java
##
@@ -93,6 +96,11 @@ public HnswGraphBuilder(
this.random =
Greg Miller created LUCENE-10403:
Summary: Add ArrayUtil#grow(T[])
Key: LUCENE-10403
URL: https://issues.apache.org/jira/browse/LUCENE-10403
Project: Lucene - Core
Issue Type: Improvement
mayya-sharipova commented on a change in pull request #641:
URL: https://github.com/apache/lucene/pull/641#discussion_r799027275
##
File path:
lucene/core/src/java/org/apache/lucene/util/hnsw/HnswGraphBuilder.java
##
@@ -93,6 +96,11 @@ public HnswGraphBuilder(
jtibshirani edited a comment on pull request #641:
URL: https://github.com/apache/lucene/pull/641#issuecomment-1029284791
Extracting `HnswGraphSearcher` is a lot nicer, I will push a refactor.
On the topic of hash sets, I tried switching to `IntIntHashMap` on top of
this PR and it
jtibshirani commented on pull request #641:
URL: https://github.com/apache/lucene/pull/641#issuecomment-1029284791
Extracting `HnswGraphSearcher` is a lot nicer, I pushed that refactor.
On the topic of hash sets, I tried switching to `IntIntHashMap` on top of
this PR and it gives a
[
https://issues.apache.org/jira/browse/LUCENE-8739?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17486619#comment-17486619
]
Adrien Grand commented on LUCENE-8739:
--
Robert disagreed with introducing a requirement on libzstd
[
https://issues.apache.org/jira/browse/LUCENE-8739?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17486614#comment-17486614
]
Praveen Nishchal commented on LUCENE-8739:
--
As we observed earlier, Zstd is at par with
dnhatn commented on pull request #640:
URL: https://github.com/apache/lucene/pull/640#issuecomment-1029191014
@jpountz Thanks for reviewing. I've pushed
https://github.com/apache/lucene/pull/640/commits/85d294ee611a22e3d0fb80347fcf2bc244e2fa5d
to simplify the fix.
--
This is an
[
https://issues.apache.org/jira/browse/LUCENE-10385?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Adrien Grand resolved LUCENE-10385.
---
Fix Version/s: 9.1
Resolution: Fixed
> Implement Weight#count on
[
https://issues.apache.org/jira/browse/LUCENE-10002?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17486570#comment-17486570
]
ASF subversion and git services commented on LUCENE-10002:
--
Commit
[
https://issues.apache.org/jira/browse/LUCENE-10385?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17486571#comment-17486571
]
ASF subversion and git services commented on LUCENE-10385:
--
Commit
[
https://issues.apache.org/jira/browse/LUCENE-10385?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17486567#comment-17486567
]
ASF subversion and git services commented on LUCENE-10385:
--
Commit
jpountz merged pull request #635:
URL: https://github.com/apache/lucene/pull/635
--
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:
jpountz merged pull request #639:
URL: https://github.com/apache/lucene/pull/639
--
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:
[
https://issues.apache.org/jira/browse/LUCENE-10002?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17486566#comment-17486566
]
ASF subversion and git services commented on LUCENE-10002:
--
Commit
mocobeta commented on pull request #638:
URL: https://github.com/apache/lucene/pull/638#issuecomment-1029151546
I opened a draft PR: https://github.com/apache/lucene/pull/643.
Would you take a look at it?
--
This is an automated message from the Apache Git Service.
To respond to the
mocobeta opened a new pull request #643:
URL: https://github.com/apache/lucene/pull/643
I drafted an idea to revise the dictionary constructors.
Main changes:
1. always use Class.getResourceAsStream() to load class resources. (don't
delegate it to the parent class)
2.
Alan Woodward created LUCENE-10402:
--
Summary: Intervals.prefix() handles multicharacter unicode
incorrectly
Key: LUCENE-10402
URL: https://issues.apache.org/jira/browse/LUCENE-10402
Project: Lucene
Adrien Grand created LUCENE-10401:
-
Summary: Seeking on empty doc-value terms dictionaries fails with
AIOOBE
Key: LUCENE-10401
URL: https://issues.apache.org/jira/browse/LUCENE-10401
Project: Lucene
[
https://issues.apache.org/jira/browse/LUCENE-10401?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Adrien Grand updated LUCENE-10401:
--
Issue Type: Bug (was: Task)
> Seeking on empty doc-value terms dictionaries fails with
jpountz commented on pull request #641:
URL: https://github.com/apache/lucene/pull/641#issuecomment-1029048199
I've become a bit cautious with codec-level caching. E.g. we have
threadlocals for stored fields which end up storing `num_search_threads *
num_indices * num_segments_per_index`
msokolov commented on pull request #641:
URL: https://github.com/apache/lucene/pull/641#issuecomment-1029006052
Thanks for working on this, and it shows a nice result. I had been
discussing with some folks at work and we were considering whether it would be
possible to maintain the state
msokolov commented on pull request #638:
URL: https://github.com/apache/lucene/pull/638#issuecomment-1028994632
I think the only reason why we have the file loading option is to support
unit testing the dictionary-loading components. Maybe there's a better way?
--
This is an automated
[
https://issues.apache.org/jira/browse/LUCENE-10400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Tomoko Uchida updated LUCENE-10400:
---
Description:
It was suggested in a few issues/pr comments.
* do not delegate to load
Tomoko Uchida created LUCENE-10400:
--
Summary: Clean up the constructors' API signature of dictionary
classes in kuromoji and nori
Key: LUCENE-10400
URL: https://issues.apache.org/jira/browse/LUCENE-10400
javanna commented on pull request #639:
URL: https://github.com/apache/lucene/pull/639#issuecomment-1028873856
This should be good to go now ;)
--
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
uschindler commented on pull request #638:
URL: https://github.com/apache/lucene/pull/638#issuecomment-1028862245
Thanks. This really needs motivation, one reason why I did not do it!
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to
mocobeta closed pull request #638:
URL: https://github.com/apache/lucene/pull/638
--
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:
mocobeta commented on pull request #638:
URL: https://github.com/apache/lucene/pull/638#issuecomment-1028858655
I will open an issue to clean up the constructors' signatures for the
dictionaries if I have a chance. I might need some time and good motivation for
the refactoring.
--
This
mocobeta edited a comment on pull request #638:
URL: https://github.com/apache/lucene/pull/638#issuecomment-1028811056
Personally, I don't care about the API or two-args constructor to load
external resources (the main interest/concern here) too, but I know there are
people who need and
mocobeta commented on pull request #638:
URL: https://github.com/apache/lucene/pull/638#issuecomment-1028811056
Personally, I don't care about the APIs (two-args constructors to load
external resources) too, but I know there are people who need and add the
feature to kuromoji and nori (I
uschindler commented on pull request #638:
URL: https://github.com/apache/lucene/pull/638#issuecomment-1028808508
I would really not move this dictionary code to commons module and add
opens. This is as bad as having the (now deprecated) methods in IOUtils to load
stopwords files. Loading
uschindler commented on pull request #638:
URL: https://github.com/apache/lucene/pull/638#issuecomment-1028805953
> I'll explore if we can remove the delegation around resource loading
without changing the API interface (for now).
I won't care about API, it is all not really public,
javanna commented on a change in pull request #635:
URL: https://github.com/apache/lucene/pull/635#discussion_r798385420
##
File path: lucene/CHANGES.txt
##
@@ -128,6 +128,9 @@ New Features
based on TotalHitCountCollector that allows users to parallelize counting the
javanna commented on a change in pull request #639:
URL: https://github.com/apache/lucene/pull/639#discussion_r798383042
##
File path: lucene/core/src/test/org/apache/lucene/search/TestBooleanQuery.java
##
@@ -492,33 +493,43 @@ private void
mocobeta edited a comment on pull request #638:
URL: https://github.com/apache/lucene/pull/638#issuecomment-1028790164
Thanks for your comment. Minor correction - we currently don't even have
DictionaryLoader but the resource loading methods are embedded in two abstract
BinaryDictionarys
javanna commented on a change in pull request #639:
URL: https://github.com/apache/lucene/pull/639#discussion_r798382358
##
File path: lucene/core/src/test/org/apache/lucene/search/TestBooleanQuery.java
##
@@ -492,33 +493,43 @@ private void
mocobeta commented on pull request #638:
URL: https://github.com/apache/lucene/pull/638#issuecomment-1028790164
Thanks for your comment. Minor correction - we currently don't even have
DictionaryLoader but the resource loading methods are embedded in two
BinaryDictionarys themselves. My
uschindler commented on pull request #638:
URL: https://github.com/apache/lucene/pull/638#issuecomment-1028771777
> Please don't do the current code or add opens. Just simple remove the
DictionaryLoader class completely and just use getResourceAsStream() directly.
Most important: Remove
mocobeta commented on pull request #638:
URL: https://github.com/apache/lucene/pull/638#issuecomment-1028736869
Hi @uschindler, I think I need your review or opinion.
This reduces code duplication and technically works, but adds "open"
directives in the module descriptors.
If this
54 matches
Mail list logo
