mrkm4ntr commented on PR #13066:
URL: https://github.com/apache/lucene/pull/13066#issuecomment-1947844218
This? https://github.com/mikemccand/luceneutil
--
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
mrkm4ntr commented on PR #13066:
URL: https://github.com/apache/lucene/pull/13066#issuecomment-1947841380
@jpountz Thank you for your review.
> I see that we have a rewrite rule that inlines nested disjunctions, so
presumably this is more about disjunctions within conjunctions than
d
uschindler commented on PR #13108:
URL: https://github.com/apache/lucene/pull/13108#issuecomment-1947486317
> That's great! If we extract the JDK version number as well, maybe we can
commit this.
Extracting the JDK version number is already implemented, because there's a
check for Ja
stefanvodita commented on PR #13108:
URL: https://github.com/apache/lucene/pull/13108#issuecomment-1947475358
That's great! If we extract the JDK version number as well, maybe we can
commit this.
@uschindler - would you want the parameter for a specific JDK version
removed on both `main`
uschindler commented on PR #13108:
URL: https://github.com/apache/lucene/pull/13108#issuecomment-1947365781
Smoke tester succeeded (see mailing list).
--
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
jpountz commented on code in PR #13066:
URL: https://github.com/apache/lucene/pull/13066#discussion_r1491684699
##
lucene/core/src/java/org/apache/lucene/search/DisjunctionSumScorer.java:
##
@@ -43,10 +47,20 @@ protected float score(DisiWrapper topList) throws
IOException {
benwtrent opened a new pull request, #13111:
URL: https://github.com/apache/lucene/pull/13111
Getting the access of a FeatureField#value is useful for deduplicating. If
you have a sparse vector model and you want to handle multiple inputs from
them, you want flexibility in how you de-duplic
dweiss opened a new pull request, #13110:
URL: https://github.com/apache/lucene/pull/13110
This extracts a common build setup composite action and cleans up existing
workflows a bit. A functional change is that gradle/actions/setup-gradle is
used to set up gradle instead of a manually-craft
bjacobowitz opened a new pull request, #13109:
URL: https://github.com/apache/lucene/pull/13109
### Description
Create a `WrappedCandidateMatcher`, along with a factory for creating these
objects and a default implementation of the factory, to assist users in
creating their o
uschindler commented on issue #13107:
URL: https://github.com/apache/lucene/issues/13107#issuecomment-1946879030
@stefanvodita 's PR branch works on Policeman Jenkins (we had to disable the
version check):
https://jenkins.thetaphi.de/job/Lucene-Release-Tester-v2/3/console
Once this s
uschindler commented on PR #13108:
URL: https://github.com/apache/lucene/pull/13108#issuecomment-1946765849
Runs: https://jenkins.thetaphi.de/job/Lucene-Release-Tester-v2/2/console
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitH
uschindler commented on PR #13108:
URL: https://github.com/apache/lucene/pull/13108#issuecomment-1946763411
Thanks, will retrigger the job!
--
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 spe
uschindler commented on PR #13108:
URL: https://github.com/apache/lucene/pull/13108#issuecomment-1946757305
waaah it doesnt like the version :-(
Can you comment out the version check in your branch?
--
This is an automated message from the Apache Git Service.
To respond to the messa
uschindler commented on PR #13108:
URL: https://github.com/apache/lucene/pull/13108#issuecomment-1946751077
It's already in testing:
https://jenkins.thetaphi.de/job/Lucene-Release-Tester-v2/1/console
--
This is an automated message from the Apache Git Service.
To respond to the message, p
stefanvodita commented on issue #13107:
URL: https://github.com/apache/lucene/issues/13107#issuecomment-1946722027
> Can I pass multiple --test-alternative-java
You can. For example, I ran:
```
JAVA_HOME=/usr/lib/jvm/java-11-amazon-corretto python3 -u
dev-tools/scripts/smokeTest
stefanvodita commented on PR #13108:
URL: https://github.com/apache/lucene/pull/13108#issuecomment-1946714708
We can, that would be cleaner. But this should work for testing 9.10, right?
Marking as draft, since I don’t think we would merge as-is.
--
This is an automated message from the
uschindler commented on issue #13107:
URL: https://github.com/apache/lucene/issues/13107#issuecomment-1946709421
Thanks @stefanvodita. I will give it a try now on Policeman Jenkins! Maybe
remove the hard coded java 17. I think instead of printing Java version it
could just print the command
uschindler commented on PR #13108:
URL: https://github.com/apache/lucene/pull/13108#issuecomment-1946691964
Can we remove the java 17 one? Or at least map the java 17 command line
parameter as additional alternative.
--
This is an automated message from the Apache Git Service.
To respond
stefanvodita commented on issue #13107:
URL: https://github.com/apache/lucene/issues/13107#issuecomment-1946685598
I was looking at this in the background and wrote a quick hack to add
`--test-alternative-java` for branch_9x in #13108. I think it does what
@uschindler wanted except print th
stefanvodita opened a new pull request, #13108:
URL: https://github.com/apache/lucene/pull/13108
Introduce `--test-alternative-java` to`smokeTestRelease.py`, allowing the
tests to run on multiple JDKs.
--
This is an automated message from the Apache Git Service.
To respond to the mess
uschindler commented on PR #13068:
URL: https://github.com/apache/lucene/pull/13068#issuecomment-1946674143
Backport was easy, no conflicts. I also checked with `fgrep -R brToString *`
but no occurrences found.
Thanks @sabi0 !
--
This is an automated message from the Apache Git Ser
uschindler merged PR #13068:
URL: https://github.com/apache/lucene/pull/13068
--
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.
uschindler commented on issue #13107:
URL: https://github.com/apache/lucene/issues/13107#issuecomment-1946639414
Add a command line parameter to disable tests. The we could test it.
Anyways: If you have a branch I can try it on policeman.
--
This is an automated message from the Apa
rmuir commented on issue #13107:
URL: https://github.com/apache/lucene/issues/13107#issuecomment-1946634786
well, i think part of what makes changing the smoketester scary is that
there's no validation of it in any PR I make. So if the github action would
"verify smoketester works" but simp
uschindler commented on issue #13107:
URL: https://github.com/apache/lucene/issues/13107#issuecomment-1946620997
> I agree - I've just taken a look and tests consume a lot of time on gh
jobs. The rest should be doable.
It does not need to be Github (it would consume ASF's credits, whi
uschindler commented on issue #13107:
URL: https://github.com/apache/lucene/issues/13107#issuecomment-1946614889
> There is value in people running smoke tester in various environments and
configurations but I think we could also define a github workflow which would
run the smoke tester aga
dweiss commented on issue #13107:
URL: https://github.com/apache/lucene/issues/13107#issuecomment-1946607741
I agree - I've just taken a look and tests consume a lot of time on gh jobs.
The rest should be doable.
--
This is an automated message from the Apache Git Service.
To respond to t
rmuir commented on issue #13107:
URL: https://github.com/apache/lucene/issues/13107#issuecomment-1946585638
You could even disable the running of tests (which might be the slowest
part) for such a github workflow job. The real value is in the release checks
and the packaging and all that, t
dweiss commented on issue #13107:
URL: https://github.com/apache/lucene/issues/13107#issuecomment-1946558279
There is value in people running smoke tester in various environments and
configurations but I think we could also define a github workflow which would
run the smoke tester against a
rmuir commented on issue #13107:
URL: https://github.com/apache/lucene/issues/13107#issuecomment-1946349998
yeah, i think for the smoketester we should just test only java versions
supported by gradle? Since this is different than just running tests but trying
to check that much more stuff
uschindler commented on issue #13107:
URL: https://github.com/apache/lucene/issues/13107#issuecomment-1946346647
> gradle should fail with a clear error though if the version is not
supported.
Correct.
My problem is in addition: I want to also test java 22, but Gradle refuses
uschindler merged PR #13104:
URL: https://github.com/apache/lucene/pull/13104
--
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.
rmuir commented on issue #13107:
URL: https://github.com/apache/lucene/issues/13107#issuecomment-1946308313
gradle should fail with a clear error though if the version is not supported.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to
uschindler commented on issue #13107:
URL: https://github.com/apache/lucene/issues/13107#issuecomment-1946306542
I think it may still parse the Java version passed as alternative java
homes, but just ensure they are >= the base version.
--
This is an automated message from the Apache Git
uschindler commented on PR #12753:
URL: https://github.com/apache/lucene/pull/12753#issuecomment-1946295995
Thanks Robert for the cleanup!
I would like to propose to step over Java 17 completely and change main
branch to 21. We would have the first 10.x release (based on branch_10x) w
rmuir opened a new issue, #13107:
URL: https://github.com/apache/lucene/issues/13107
### Description
Currently it is difficult for smoketester to "keep up" with supported java
versions, which are moving faster these days.
@uschindler ran into it with the release vote for 9.10,
stefanvodita commented on PR #12547:
URL: https://github.com/apache/lucene/pull/12547#issuecomment-1946277632
Thank you for the approval! I want to leave this open for now while
iterating over #12966. I think I prefer doing #12966 first, since it's a more
complicated change and doing it fir
rmuir commented on PR #12753:
URL: https://github.com/apache/lucene/pull/12753#issuecomment-1946274403
@uschindler hit some related difficulties with 9.x release vote and these
JDK versions in the smoketester (I still have avoided touching it here but it
needs java 21 support). So maybe it
mikemccand commented on issue #13106:
URL: https://github.com/apache/lucene/issues/13106#issuecomment-1946265552
Whoa, very cool @jpountz! This reminds me of [this longstanding
issue/paper](https://github.com/apache/lucene/issues/4036) which also inlined
skip data directly in the postings,
uschindler commented on PR #13104:
URL: https://github.com/apache/lucene/pull/13104#issuecomment-1946149224
Let's open another issue about this. I think we did the final only fields
for query instances, but token streams are not behaving bad as they are not
cached and have state.
But
easyice commented on PR #13104:
URL: https://github.com/apache/lucene/pull/13104#issuecomment-1946004282
Thank you for the great suggestions!
@uschindler You are right, all the setters are used after the construction
only, It makes sense if they are only set on construction. The only
slow-J commented on code in PR #12930:
URL: https://github.com/apache/lucene/pull/12930#discussion_r1490898433
##
dev-docs/codec-version-bump-howto.md:
##
@@ -0,0 +1,74 @@
+
+
+# Lucene Codec Version Bump How-To Manual
+
+Changing the name of the codec in Lucene is required for
uschindler commented on issue #13101:
URL: https://github.com/apache/lucene/issues/13101#issuecomment-1945923223
Hi,
I think you should also open an issue in Apache Solr, because basically the
problem is more on Solr's side. As said before th
I also thought yesterday evening about
stefanvodita merged PR #13041:
URL: https://github.com/apache/lucene/pull/13041
--
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...@lucen
uschindler commented on PR #13068:
URL: https://github.com/apache/lucene/pull/13068#issuecomment-1945649280
To the other reviewers: I would like to merge this as a first step. To me
this looks fine as it removes the code duplication and we have a better method
name. Ok?
If we remove
jpountz opened a new issue, #13106:
URL: https://github.com/apache/lucene/issues/13106
### Description
On top-k queries, Lucene is now competitive with Tantivy/PISA on
https://tantivy-search.github.io/bench/, but it's still quite slower on
counting queries. This made me want to run a
46 matches
Mail list logo
