zacharymorn commented on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748563011
Oh wow thanks for the updates there, and sorry to hear the current changes
are not yet ready! Yes the original implementation hasn't been used for a while
now (as
uschindler commented on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748539067
I am giving up for today. The whole implementation is plain broken and the
test never catched it. It was writen in earlier times of Lucene when
IndexOutputs were able to
uschindler commented on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748535427
Hi,
I modified the test to allow some enforced mode, so the direcory is fully
tests (basically, I made it to always return direct IndexInput/Outputs. After
this it
jpountz commented on pull request #993:
URL: https://github.com/apache/lucene-solr/pull/993#issuecomment-748532390
Sorry for the time to respond on this PR @kkewwei. It looks like the problem
you pointed out no longer exists in master?
uschindler commented on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748530259
I removed the file length handling and cleaned up the whole stuff. I am not
sure if the truncate is still needed, but I left it for now.
In addition, the
uschindler commented on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748529191
There's more broken, the file size on writing is incorrect sometimes. More
rewrite is required.
This is
uschindler commented on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748528098
Hi, I rewrote most of the ByteBuffer handling. The bugs came from the fact
that we used an additional outside bufferSize and bufferPos. All calcuations
were done based
murblanc commented on a change in pull request #2148:
URL: https://github.com/apache/lucene-solr/pull/2148#discussion_r546277620
##
File path:
solr/core/src/java/org/apache/solr/cloud/overseer/CollectionMutator.java
##
@@ -136,8 +154,13 @@ public ZkWriteCommand
uschindler commented on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748519083
I will rewrite that stuff a bit, so you can pass a mergeBufferSize when
creating the directory, but the value is adjusted according to the file system.
I don't want
murblanc commented on a change in pull request #2148:
URL: https://github.com/apache/lucene-solr/pull/2148#discussion_r546277475
##
File path:
solr/solrj/src/java/org/apache/solr/common/cloud/PerReplicaStates.java
##
@@ -0,0 +1,587 @@
+/*
+ * Licensed to the Apache Software
murblanc commented on a change in pull request #2148:
URL: https://github.com/apache/lucene-solr/pull/2148#discussion_r546277475
##
File path:
solr/solrj/src/java/org/apache/solr/common/cloud/PerReplicaStates.java
##
@@ -0,0 +1,587 @@
+/*
+ * Licensed to the Apache Software
murblanc commented on a change in pull request #2148:
URL: https://github.com/apache/lucene-solr/pull/2148#discussion_r546277024
##
File path:
solr/solrj/src/java/org/apache/solr/common/cloud/PerReplicaStates.java
##
@@ -0,0 +1,587 @@
+/*
+ * Licensed to the Apache Software
murblanc commented on a change in pull request #2148:
URL: https://github.com/apache/lucene-solr/pull/2148#discussion_r546277024
##
File path:
solr/solrj/src/java/org/apache/solr/common/cloud/PerReplicaStates.java
##
@@ -0,0 +1,587 @@
+/*
+ * Licensed to the Apache Software
[
https://issues.apache.org/jira/browse/SOLR-15052?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17252260#comment-17252260
]
Ishan Chattopadhyaya commented on SOLR-15052:
-
Thanks [~ilan] and [~mdrob] for the reviews.
uschindler edited a comment on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748517193
Yes that's the issue. On windows it looks like allocateDirect does not
allign to beginning of page at some memory location not aligned to 512 bytes.
The
murblanc commented on pull request #2148:
URL: https://github.com/apache/lucene-solr/pull/2148#issuecomment-748518170
Added my high level comments as a
[
https://issues.apache.org/jira/browse/SOLR-15052?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17252259#comment-17252259
]
Ilan Ginzburg commented on SOLR-15052:
--
I think this goes in the right direction and thank you for
uschindler commented on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748517370
Nevertheless what I don't like: why do we need this ALIGN masks and so on.
As we have the official block size we can use that for our calculations instead
of relying on
uschindler commented on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748517193
Yes that's the issue. On windows it looks like allocateDirect does not
allign to beginning of page at some memory location not aligned to 512 bytes.
The alignedSlice is
uschindler commented on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748516524
I think I know the problem: when you call alignedSlice, the buffer's
remaining() may be no longer the bufferSize. All calculations afterwards are
then using the
zacharymorn commented on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748509146
> > Hmm I don't have a windows box to test this. The blockSize is
dynamically computed though. If this works on linux/darwin but fails on
windows, does it signal
uschindler commented on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748508063
> Hmm I don't have a windows box to test this. The blockSize is dynamically
computed though. If this works on linux/darwin but fails on windows, does it
signal another
zacharymorn commented on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748506650
> Hi, I implemented the approach and also cleaned up the lookup code to not
require unchecked casts. I also made the code safer for exceptions, you only
catched
jpountz commented on a change in pull request #2141:
URL: https://github.com/apache/lucene-solr/pull/2141#discussion_r546265505
##
File path: lucene/core/src/java/org/apache/lucene/search/WANDScorer.java
##
@@ -271,6 +280,19 @@ public boolean matches() throws IOException {
[
https://issues.apache.org/jira/browse/LUCENE-9645?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17252248#comment-17252248
]
Adrien Grand commented on LUCENE-9645:
--
Would you like to open a pull request to fix this?
>
jpountz commented on a change in pull request #2141:
URL: https://github.com/apache/lucene-solr/pull/2141#discussion_r546262630
##
File path: lucene/core/src/java/org/apache/lucene/search/WANDScorer.java
##
@@ -130,10 +130,19 @@ private static long scaleMinScore(float
jpountz commented on a change in pull request #2141:
URL: https://github.com/apache/lucene-solr/pull/2141#discussion_r546262588
##
File path: lucene/core/src/java/org/apache/lucene/search/WANDScorer.java
##
@@ -130,10 +130,19 @@ private static long scaleMinScore(float
[
https://issues.apache.org/jira/browse/LUCENE-9617?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Michael McCandless resolved LUCENE-9617.
Fix Version/s: 8.8
Resolution: Fixed
Thank you [~msfroh]!
>
[
https://issues.apache.org/jira/browse/LUCENE-9617?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17252202#comment-17252202
]
ASF subversion and git services commented on LUCENE-9617:
-
Commit
[
https://issues.apache.org/jira/browse/LUCENE-9617?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17252201#comment-17252201
]
ASF subversion and git services commented on LUCENE-9617:
-
Commit
[
https://issues.apache.org/jira/browse/LUCENE-9617?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17252199#comment-17252199
]
ASF subversion and git services commented on LUCENE-9617:
-
Commit
[
https://issues.apache.org/jira/browse/LUCENE-9617?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17252200#comment-17252200
]
ASF subversion and git services commented on LUCENE-9617:
-
Commit
[
https://issues.apache.org/jira/browse/SOLR-11191?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
David Smiley updated SOLR-11191:
Labels: NestedDocuments (was: )
> Shard Split doesn't work on indexes that have nested documents
[
https://issues.apache.org/jira/browse/SOLR-11191?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
David Smiley reassigned SOLR-11191:
---
Assignee: David Smiley
> Shard Split doesn't work on indexes that have nested documents
>
uschindler commented on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748465821
Nevertheless, on Windows I see the test failing all the time (JDK 11 and JDK
15 were tested):
```
org.apache.lucene.misc.store.TestDirectIODirectory >
uschindler commented on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748465633
Hi, I implemented the approach and also cleaned up the lookup code to not
require unchecked casts. I also made the code safer for exceptions, you only
catched
dweiss edited a comment on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748454464
The master branch of the JDK always emits this exception, it seems:
dweiss commented on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748454464
The master branch of the JDK always emits this exception, it seems:
iverase commented on pull request #2154:
URL: https://github.com/apache/lucene-solr/pull/2154#issuecomment-748449367
good point @madrob, unit test added.
This is an automated message from the Apache Git Service.
To respond
zacharymorn commented on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748443350
> You can do it statically, look at how the unmapper in Mmapdir works:
> Look up the reflective in static intitalizer and save the result at very
end in static final.
uschindler commented on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748442053
You can do it statically, look at how the unmapper in Mmapdir works:
Look up the reflective in static intitalizer and save the result at very end
in static final.
zacharymorn commented on pull request #2052:
URL: https://github.com/apache/lucene-solr/pull/2052#issuecomment-748441602
> I'll try to help out later today, Zach.
Sounds good. Thanks Dawid!
This is an automated
42 matches
Mail list logo