lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Christine Poerschke (BLOOMBERG/ LONDON)" <>
Subject Re: [JENKINS-EA] Lucene-Solr-trunk-Linux (64bit/jdk1.9.0-ea-b60)
Date Thu, 06 Aug 2015 20:29:52 GMT
Hi Adrien,

No, it was just 2 documents in 1 segment yesterday. Just returned to this and yes, you're
right, placing the 2nd document into a 2nd segment (prior to the force merge) seems to work,
at least with the seed that previously caused the test to fail.

If that sounds about right, i'd be happy to properly patch for this tomorrow.


--- a/lucene/misc/src/test/org/apache/lucene/search/
+++ b/lucene/misc/src/test/org/apache/lucene/search/
@@ -115,6 +115,8 @@ public class TestEarlyTerminatingSortingCollector extends LuceneTestCase
       // the index, although want want a sorted segment so it needs to be merged
       iw.getReader().close(); // refresh
       iw.addDocument(new Document());
+      iw.commit();
+      iw.addDocument(new Document());
     else if (random().nextBoolean()) {

----- Original Message -----
To: Christine Poerschke (BLOOMBERG/ LONDON),
At: Aug  6 2015 10:27:38

Hi Christine,

On Wed, Aug 5, 2015 at 11:42 PM, Christine Poerschke (BLOOMBERG/
LONDON) <> wrote:
> Locally changed createRandomIndex to add not the 1 but 2 documents before the forceMerge.
Still ends up with an unsorted segment:

Did you ensure that there is a reopen between these two documents are
added, so that we not only have 2 docs but also 2 segments? This can
be done eg. by calling iw.getReader().close(); on the


To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message