github-actions[bot] commented on PR #13822:
URL: https://github.com/apache/lucene/pull/13822#issuecomment-2403660295
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
mikemccand commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2403632181
I cherry-pick'd to 10.x as well. Let's see if CI builds are happy ... if
so, I think this is done!
--
This is an automated message from the Apache Git Service.
To respond to the me
mikemccand commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2403616445
OK I pushed 10.0.x fix ... working on 10.x now. Gradle sees the same weird
deadlock when I try to `clean` and `check` in one go. So weird.
--
This is an automated message from the
mikemccand commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2403538869
Although, strangely, when I run `./gradlew clean check` I get this very odd
(never seen this before) failure:
```
Unable to make progress running work. The following items ar
mikemccand commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2403535561
The `git cherry-pick` of the 9.12.x commit merged cleanly onto 10.0.x branch
... and `gradle check` is happy, so I plan to push that cherry-pick to 10.0.x
branch (and 10.x after that)
original-brownbear commented on PR #13864:
URL: https://github.com/apache/lucene/pull/13864#issuecomment-2403528442
Never mind the above, I think I found a neat solution :)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and u
mikemccand closed issue #13867: Lucene 9.12 fails reading older versions of
Lucene99HnswScalarQuantizedVectorsFormat
URL: https://github.com/apache/lucene/issues/13867
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
U
mikemccand closed issue #13880: HNSW BWC tests should validate that the
`int8_hnsw` zip filed index actually uses `int7` quantization
URL: https://github.com/apache/lucene/issues/13880
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to Git
mikemccand commented on issue #13880:
URL: https://github.com/apache/lucene/issues/13880#issuecomment-2403515707
Fixed by #13874
--
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 comme
mikemccand commented on issue #13867:
URL: https://github.com/apache/lucene/issues/13867#issuecomment-2403515200
Fixed by #13874
--
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 comme
mikemccand merged PR #13874:
URL: https://github.com/apache/lucene/pull/13874
--
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.
mikemccand commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2403494337
> I wish we had better automation here (check out release tag, build the bwc
index, copy zip files to newer release, etc.). I will open a spinoff issue...
Actually, this nice `d
mikemccand commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2403490011
OK I'll merge shortly once GitHub action is happy! Thanks everyone, phew!
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to Gi
mikemccand commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2403477991
> Oh! Is there an int8_hnsw.9.12.0.zip that can be deleted?
D'oh! I'll delete.
--
This is an automated message from the Apache Git Service.
To respond to the message, please
ChrisHegarty commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2403417121
Oh! Is there an int8_hnsw.9.12.0.zip that can be deleted?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use th
mikemccand commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2403377531
> I was not expecting to see changes in `index.9.12.0-cfs.zip` or
`index.9.12.0-no.cfs.zip`. Do you know why these are changed?
Net/net it's because I regen'd all 9.12.0 bwc zip
original-brownbear commented on PR #13864:
URL: https://github.com/apache/lucene/pull/13864#issuecomment-2403376299
> I suspect that the main offender are doc values terms dictionaries
Jup that's it :)
> and the storage overhead would likely be negligible without the heap
over
ChrisHegarty commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2403360433
I was not expecting to see changes in `index.9.12.0-cfs.zip` or
`index.9.12.0-no.cfs.zip`. Do you know why these are changed?
--
This is an automated message from the Apache Git S
dweiss commented on issue #13761:
URL: https://github.com/apache/lucene/issues/13761#issuecomment-2403307475
Thank you for reporting the problem, @bugmakerr
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL
dweiss closed issue #13761: IntObjectHashMap.values().toArray() method throws
ClassCastException
URL: https://github.com/apache/lucene/issues/13761
--
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 th
dweiss merged PR #13876:
URL: https://github.com/apache/lucene/pull/13876
--
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
mikemccand commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2403280987
OK phew I think I resolved the conflicts due to stale branch, and tests are
passing for me. I think this is ready? Let's let GitHub actions chew on it to
confirm...
--
This is an
BrianWoolfolk commented on issue #13840:
URL: https://github.com/apache/lucene/issues/13840#issuecomment-2403266928
Great! I'll look into this then and make a PR afterwards
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and u
mikemccand commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2403250756
Woops, I see my local `branch_9_12` (and where I branched my
`fix_back_compat` branch from for this PR) was out of date -- `branch_9_12` had
already bumped to 9.12.1 and added bwc ind
mikemccand commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2403082537
> > Ack, I'll re-revert the revert. Thanks @ChrisHegarty.
>
> Thanks. And sorry for the bad suggestion.
No worries!
> So, I think once the 9.12.0 int7 hnsw bwc file
mikemccand commented on issue #13880:
URL: https://github.com/apache/lucene/issues/13880#issuecomment-2403023334
OK I got this working, and confirmed that the new test case fails on the old
(`float32`) BWC indices and passes on the ones in this PR, phew. I pushed that
new test case onto th
dweiss commented on code in PR #13876:
URL: https://github.com/apache/lucene/pull/13876#discussion_r1793964617
##
lucene/core/src/test/org/apache/lucene/internal/hppc/TestIntObjectHashMap.java:
##
@@ -66,10 +69,8 @@ private static void assertSortedListEquals(int[] array,
int...
ChrisHegarty commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2402848199
> Ack, I'll re-revert the revert. Thanks @ChrisHegarty.
Thanks. And sorry for the bad suggestion.
So, I think once the 9.12.0 int7 hnsw bwc file is created, then this
itygh commented on PR #2682:
URL: https://github.com/apache/lucene-solr/pull/2682#issuecomment-2402841726
这是来自QQ邮箱的假期自动回复邮件。您好，我最近正在休假中，无法亲自回复您的邮件。我将在假期结束后，尽快给您回复。
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
UR
risdenk closed pull request #2682: Backport SOLR-14765 to branch_8_11
URL: https://github.com/apache/lucene-solr/pull/2682
--
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 un
mikemccand commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2402825006
> > Hmm the problem is, we need to regen the 9.12.0 `int8_hnsw` back compat
index using the bug fix from this PR (so that it actually has `int7` quantized
vectors, not unquantized `fl
mikemccand commented on issue #13880:
URL: https://github.com/apache/lucene/issues/13880#issuecomment-2402818214
OK I see -- I am checking the wrong class -- I have to get to the underly
`FlatVectorsReader` ...
--
This is an automated message from the Apache Git Service.
To respond to the
jpountz commented on PR #13871:
URL: https://github.com/apache/lucene/pull/13871#issuecomment-2402747421
Ah, nevermind, this PR was not included in the last nightly run, so we
should see its effect tomorrow.
--
This is an automated message from the Apache Git Service.
To respond to the me
jpountz commented on PR #13871:
URL: https://github.com/apache/lucene/pull/13871#issuecomment-2402745882
The speedup on nightlies is very small:
https://benchmarks.mikemccandless.com/IntervalsOrdered.html. I wonder if it's
due to a difference in hardware, or queries (nightlies have differen
jpountz commented on PR #13864:
URL: https://github.com/apache/lucene/pull/13864#issuecomment-2402721041
Looking at places where we used `DirectMonotonicReader`, I suspect that the
main offender are doc values terms dictionaries, which maintain two
`DirectMonotonicReader`s each. However, th
ChrisHegarty commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2402700651
> Hmm the problem is, we need to regen the 9.12.0 `int8_hnsw` back compat
index using the bug fix from this PR (so that it actually has `int7` quantized
vectors, not unquantized `fl
jpountz merged PR #13881:
URL: https://github.com/apache/lucene/pull/13881
--
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 closed issue #13818: TestTieredMergePolicy.testSimulateAppendOnly fails
with AssertionError
URL: https://github.com/apache/lucene/issues/13818
--
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
jpountz closed issue #13818: TestTieredMergePolicy.testSimulateAppendOnly fails
with AssertionError
URL: https://github.com/apache/lucene/issues/13818
--
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
gsmiller commented on issue #13840:
URL: https://github.com/apache/lucene/issues/13840#issuecomment-2402588570
@iverase I wonder if moving the logic is the right thing to do vs. having
the optimization in both places? I suspect the most likely way
`IndexSortSortedNumericDocValuesRangeQuery`
mikemccand commented on issue #13880:
URL: https://github.com/apache/lucene/issues/13880#issuecomment-2402543191
Hmm, well, I coded that up, on top of my PR from #13874, with this diff:
```
raptorlake:912x[fix_back_compat]$ git diff
diff --git
a/lucene/backward-codecs/src/test/o
iverase commented on issue #13840:
URL: https://github.com/apache/lucene/issues/13840#issuecomment-2402529934
I am not working on this at the moment.
This issue is an spinoff of this comment from Adrien so we might want to
pull the optimization to this class completly:
https://githu
jpountz closed issue #13854: TestDefaultCodecParallelizesIO.testStoredFields
fails
URL: https://github.com/apache/lucene/issues/13854
--
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 com
jpountz closed issue #13854: TestDefaultCodecParallelizesIO.testStoredFields
fails
URL: https://github.com/apache/lucene/issues/13854
--
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 com
jpountz merged PR #13875:
URL: https://github.com/apache/lucene/pull/13875
--
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 pull request, #13881:
URL: https://github.com/apache/lucene/pull/13881
The two seeds at #13818 had different root causes:
- The test allows the number of segments to go above the limit, only if
none of the merges are legal. But there are multiple reasons why a merge
gsmiller commented on issue #13840:
URL: https://github.com/apache/lucene/issues/13840#issuecomment-2402489157
@BrianWoolfolk this looks like a different optimization from what was done
in #13592. It doesn't look like it's worked on to me (at least I don't see this
optimization in the code;
mikemccand commented on issue #13880:
URL: https://github.com/apache/lucene/issues/13880#issuecomment-2402478358
> It's a bit tricky, or at least non obvious to me on a quick think, how to
introspect the Codec formats used in an index down to the `KnnVectorsFormat`...
Actually I think
javanna commented on PR #13825:
URL: https://github.com/apache/lucene/pull/13825#issuecomment-2402464116
Thanks @dweiss !!!
--
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.
mikemccand commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2402452969
> if u merge in `branch_9_12`, then
`dev-tools/scripts/addBackcompatIndexes.py 9.12.0` should be able to generate
the compat indices.
Hmm the problem is, we need to regen the 9
mikemccand commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2402425950
> I wonder if the simplest thing here is to just avoid the renaming (from
int8 to int7) in 9.x. Otherwise, I don't see how to get
`addBackcompatIndexes.py` to work - since 9.12.0 sour
bruno-roustant commented on code in PR #13876:
URL: https://github.com/apache/lucene/pull/13876#discussion_r1793571950
##
lucene/core/src/test/org/apache/lucene/internal/hppc/TestIntObjectHashMap.java:
##
@@ -66,10 +69,8 @@ private static void assertSortedListEquals(int[] array,
ChrisHegarty commented on code in PR #13879:
URL: https://github.com/apache/lucene/pull/13879#discussion_r1793484452
##
lucene/backward-codecs/src/test/org/apache/lucene/backward_index/TestGenerateBwcIndices.java:
##
@@ -82,6 +82,16 @@ public void testCreateSortedIndex() throws
ChrisHegarty commented on code in PR #13879:
URL: https://github.com/apache/lucene/pull/13879#discussion_r1793486021
##
lucene/backward-codecs/src/test/org/apache/lucene/backward_index/TestGenerateBwcIndices.java:
##
@@ -82,6 +82,16 @@ public void testCreateSortedIndex() throws
ChrisHegarty commented on code in PR #13879:
URL: https://github.com/apache/lucene/pull/13879#discussion_r1793484452
##
lucene/backward-codecs/src/test/org/apache/lucene/backward_index/TestGenerateBwcIndices.java:
##
@@ -82,6 +82,16 @@ public void testCreateSortedIndex() throws
ChrisHegarty commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2402235046
I wonder if the simplest thing here is to just avoid the renaming (from int8
to int7) in 9.x. Otherwise, I don't see how to get `addBackcompatIndexes.py` to
work - since 9.12.0 sour
ChrisHegarty commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2402227085
Argh!! this is very odd. We cannot just update the addBackcompatIndexes.py
script. Since it downloads the 9.12.0 version and tries to run gradle on it!!
and of course `testCreateIn
ChrisHegarty commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2402218226
if u merge in `branch_9_12`, then `dev-tools/scripts/addBackcompatIndexes.py
9.12.0` should be able to generate the compat indices.
--
This is an automated message from the Apache
ChrisHegarty commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2402214908
we need to update `dev-tools/scripts/addBackcompatIndexes.py`.
```diff
diff --git a/dev-tools/scripts/addBackcompatIndexes.py
b/dev-tools/scripts/addBackcompatIndexes.py
jpountz merged PR #13878:
URL: https://github.com/apache/lucene/pull/13878
--
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 commented on PR #13878:
URL: https://github.com/apache/lucene/pull/13878#issuecomment-2402145176
You are right, I pushed a change.
--
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 spec
mikemccand commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2402140838
> I can open the PR -- I still have the dev area where I got it working and
generated the bwc `.zip` in this PR.
OK I opened https://github.com/apache/lucene/pull/13879 -- this
mikemccand opened a new pull request, #13879:
URL: https://github.com/apache/lucene/pull/13879
Spinoff from #13867 to add the tooling back onto `branch_9_10` in case we
ever need to regenerate the `int7_hnsw.9.10.0.zip` bwc index again.
--
This is an automated message from the Apache Git
mayya-sharipova merged PR #13873:
URL: https://github.com/apache/lucene/pull/13873
--
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...@lu
mikemccand commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2402102296
> @mikemccand I can open PRs to the old versions to add the code. I did the
same "copy files and run it" dance that you did because I didn't want to commit
code to the old branches. B
mikemccand commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2402095420
> I'm try to determine why a bwc test fails in the CI.
Thanks @ChrisHegarty.
Can you point me to the CI build that is angry? I assume you set up a
9.12.x CI build somewh
benwtrent commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2402094834
@mikemccand I can open PRs to the old versions to add the code. I did the
same "copy files and run it" dance that you did because I didn't want to commit
code to the old branches. But
mikemccand commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2402071452
> Thank you for fixing my bwc mess! It was definitely a learning experience
and apparently I didn't learn enough!
No worries @benwtrent -- it is crazy complicated and trappy! I
benwtrent commented on PR #13872:
URL: https://github.com/apache/lucene/pull/13872#issuecomment-2402064518
> hm there is some functional problem with the change that yields terrible
recall for quantized vectors. I'll dig and fix and see if I can beef up the
unit test coverage as well.
benwtrent commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2402052859
Ah, I see 9.12.0 is already assumed by the bwc indices! I guess that means
we need to generate 9.12.0 indices for the hnsw bwc. You are correct
@ChrisHegarty .
--
This is an automa
benwtrent commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2402047972
I think you have to add it as a backcompat index in another file, but we
don't do that until 9.12.1 is cut and released! Since 9.12.1 hasn't gone
through the release process, according
benwtrent commented on issue #13867:
URL: https://github.com/apache/lucene/issues/13867#issuecomment-2402043612
Ack, I should have added more back compat indices MORE TESTING
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to Gi
iverase merged PR #13877:
URL: https://github.com/apache/lucene/pull/13877
--
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
iverase commented on PR #13877:
URL: https://github.com/apache/lucene/pull/13877#issuecomment-2401917665
I add an entry in 77b8098
--
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 com
iverase commented on PR #13877:
URL: https://github.com/apache/lucene/pull/13877#issuecomment-2401883251
I don't think we need to add an entry in changes as this is part of 9.11.1
and 9.12.0?
--
This is an automated message from the Apache Git Service.
To respond to the message, please lo
dweiss closed issue #13820: DataInput class can't be used with delegation
pattern
URL: https://github.com/apache/lucene/issues/13820
--
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 comm
dweiss commented on PR #13825:
URL: https://github.com/apache/lucene/pull/13825#issuecomment-2401859755
I've added migration and changes entry and cherry picked on branch_10_0,
branch_10x and main.
--
This is an automated message from the Apache Git Service.
To respond to the message, ple
javanna commented on PR #13877:
URL: https://github.com/apache/lucene/pull/13877#issuecomment-2401843256
Perhaps a changes entry would be good to add?
--
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 g
javanna commented on PR #13858:
URL: https://github.com/apache/lucene/pull/13858#issuecomment-2401838440
Heads up: I moved the CHANGES entry in main and branch_10x to the 10.0.0
section. See
https://github.com/apache/lucene/commit/111fc6f07857f909ad2a253cdcd6e948222ecf63
.
--
This is an
javanna commented on PR #13858:
URL: https://github.com/apache/lucene/pull/13858#issuecomment-2401818106
I merged this to branch_10_0. Does this also need to backported to 9_12 for
the next 9.12.1 patch release?
--
This is an automated message from the Apache Git Service.
To respond to th
javanna merged PR #13858:
URL: https://github.com/apache/lucene/pull/13858
--
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
javanna commented on PR #13825:
URL: https://github.com/apache/lucene/pull/13825#issuecomment-2401799763
I think we are also missing a CHANGES entry , @dweiss could you take care of
adding that please?
--
This is an automated message from the Apache Git Service.
To respond to the message,
ChrisHegarty commented on PR #13874:
URL: https://github.com/apache/lucene/pull/13874#issuecomment-2401798286
I think that this is a good change. I'm try to determine why a bwc test
fails in the CI.
--
This is an automated message from the Apache Git Service.
To respond to the message
javanna commented on PR #13825:
URL: https://github.com/apache/lucene/pull/13825#issuecomment-2401796608
I get confused that this PR was not merged yet it has a milestone set.
Should the milestone be set to the other PR that has been merged instead
(#13830) ?
--
This is an automated mess
dweiss commented on PR #13825:
URL: https://github.com/apache/lucene/pull/13825#issuecomment-2401573279
I've backported this to branch_10x and asked the RM to allow this to be
cherry picked for the subsequent respin of Lucene 10.0.
--
This is an automated message from the Apache Git Servi
dweiss opened a new pull request, #13876:
URL: https://github.com/apache/lucene/pull/13876
This method is only used in tests. I moved the toList collection utility to
the test class to minimize surface API.
Fixes #13761
--
This is an automated message from the Apache Git Service.
original-brownbear merged PR #13863:
URL: https://github.com/apache/lucene/pull/13863
--
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...
original-brownbear commented on PR #13863:
URL: https://github.com/apache/lucene/pull/13863#issuecomment-2401511333
Thanks Adrien!
--
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 com
88 matches
Mail list logo
