You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@lucene.apache.org by jp...@apache.org on 2023/10/17 12:13:27 UTC

[lucene] branch main updated: Speed up TestIndexOrDocValuesQuery. (#12672)

This is an automated email from the ASF dual-hosted git repository.

jpountz pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/lucene.git


The following commit(s) were added to refs/heads/main by this push:
     new f055ac6e168 Speed up TestIndexOrDocValuesQuery. (#12672)
f055ac6e168 is described below

commit f055ac6e16818bb40311a8ec162f933af00a4a6b
Author: Adrien Grand <jp...@gmail.com>
AuthorDate: Tue Oct 17 14:13:21 2023 +0200

    Speed up TestIndexOrDocValuesQuery. (#12672)
    
    This changes the following:
     - fewer docs indexed in non-nightly runs,
     - `QueryUtils#checkFirstSkipTo` uses the `ScorerSupplier` API to convey it
       will only check one doc,
     - `QueryUtils#checkFirstSkipTo` no longer relies on timing to run in a
       reasonably amount of time.
---
 .../apache/lucene/search/TestIndexOrDocValuesQuery.java |  7 ++++---
 .../java/org/apache/lucene/tests/search/QueryUtils.java | 17 +++++++----------
 2 files changed, 11 insertions(+), 13 deletions(-)

diff --git a/lucene/core/src/test/org/apache/lucene/search/TestIndexOrDocValuesQuery.java b/lucene/core/src/test/org/apache/lucene/search/TestIndexOrDocValuesQuery.java
index e24c0cdb7c8..57ad6e24c82 100644
--- a/lucene/core/src/test/org/apache/lucene/search/TestIndexOrDocValuesQuery.java
+++ b/lucene/core/src/test/org/apache/lucene/search/TestIndexOrDocValuesQuery.java
@@ -112,14 +112,15 @@ public class TestIndexOrDocValuesQuery extends LuceneTestCase {
             newIndexWriterConfig()
                 // relies on costs and PointValues.estimateCost so we need the default codec
                 .setCodec(TestUtil.getDefaultCodec()));
-    for (int i = 0; i < 2000; ++i) {
+    final int numDocs = atLeast(1000);
+    for (int i = 0; i < numDocs; ++i) {
       Document doc = new Document();
-      if (i < 1000) {
+      if (i < numDocs / 2) {
         doc.add(new StringField("f1", "bar", Store.NO));
         for (int j = 0; j < 500; j++) {
           doc.add(new LongField("f2", 42L, Store.NO));
         }
-      } else if (i == 1001) {
+      } else if (i == numDocs / 2) {
         doc.add(new StringField("f1", "foo", Store.NO));
         doc.add(new LongField("f2", 2L, Store.NO));
       } else {
diff --git a/lucene/test-framework/src/java/org/apache/lucene/tests/search/QueryUtils.java b/lucene/test-framework/src/java/org/apache/lucene/tests/search/QueryUtils.java
index 0cd75f6b06c..5a454681f54 100644
--- a/lucene/test-framework/src/java/org/apache/lucene/tests/search/QueryUtils.java
+++ b/lucene/test-framework/src/java/org/apache/lucene/tests/search/QueryUtils.java
@@ -540,8 +540,10 @@ public class QueryUtils {
     s.search(
         q,
         new SimpleCollector() {
+          private final Weight w = s.createWeight(rewritten, ScoreMode.COMPLETE, 1);
           private Scorable scorer;
           private int leafPtr;
+          private long intervalTimes32 = 1 * 32;
 
           @Override
           public void setScorer(Scorable scorer) {
@@ -552,10 +554,11 @@ public class QueryUtils {
           public void collect(int doc) throws IOException {
             float score = scorer.score();
             try {
-              long startNS = System.nanoTime();
-              for (int i = lastDoc[0] + 1; i <= doc; i++) {
-                Weight w = s.createWeight(rewritten, ScoreMode.COMPLETE, 1);
-                Scorer scorer = w.scorer(context.get(leafPtr));
+              // The intervalTimes32 trick helps contain the runtime of this check: first we check
+              // every single doc in the interval, then after 32 docs we check every 2 docs, etc.
+              for (int i = lastDoc[0] + 1; i <= doc; i += intervalTimes32++ / 1024) {
+                ScorerSupplier supplier = w.scorerSupplier(context.get(leafPtr));
+                Scorer scorer = supplier.get(1L); // only checking one doc, so leadCost = 1
                 assertTrue(
                     "query collected " + doc + " but advance(" + i + ") says no more docs!",
                     scorer.iterator().advance(i) != DocIdSetIterator.NO_MORE_DOCS);
@@ -579,12 +582,6 @@ public class QueryUtils {
                     score,
                     advanceScore,
                     maxDiff);
-
-                // Hurry things along if they are going slow (eg
-                // if you got SimpleText codec this will kick in):
-                if (i < doc && System.nanoTime() - startNS > 5_000_000) {
-                  i = doc - 1;
-                }
               }
               lastDoc[0] = doc;
             } catch (IOException e) {