Merge pull request #29548 from vespa-engine/balder/control-create-postinglist-when-non-strict

Balder/control create postinglist when non strict

9 files changed, 40 insertions, 9 deletions

diff --git a/searchlib/src/vespa/searchlib/attribute/postinglistsearchcontext.h b/searchlib/src/vespa/searchlib/attribute/postinglistsearchcontext.h
index f5683546eea..eefb2ea77d5 100644
--- a/searchlib/src/vespa/searchlib/attribute/postinglistsearchcontext.h
+++ b/searchlib/src/vespa/searchlib/attribute/postinglistsearchcontext.h
@@ -436,6 +436,8 @@ NumericPostingSearchContext<BaseSC, AttrT, DataT>::use_posting_lists_when_non_st
     // Based on this we see that LMC = 5 * PLMC.
     // The same relationship is found with the test case range_hits_ratio=[200].
 
+    if ( ! info.create_postinglist_when_non_strict()) return false;
+
     constexpr float lookup_match_constant = 5.0;
     constexpr float posting_list_merge_constant = 1.0;
 
diff --git a/searchlib/src/vespa/searchlib/fef/indexproperties.cpp b/searchlib/src/vespa/searchlib/fef/indexproperties.cpp
index 9c986d0bc63..d503a955afc 100644
--- a/searchlib/src/vespa/searchlib/fef/indexproperties.cpp
+++ b/searchlib/src/vespa/searchlib/fef/indexproperties.cpp
@@ -461,6 +461,12 @@ bool AlwaysMarkPhraseExpensive::check(const Properties &props, bool fallback) {
     return lookupBool(props, NAME, fallback);
 }
 
+const vespalib::string CreatePostingListWhenNonStrict::NAME("vespa.matching.create_postinglist_when_non_strict");
+const bool CreatePostingListWhenNonStrict::DEFAULT_VALUE(true);
+bool CreatePostingListWhenNonStrict::check(const Properties &props, bool fallback) {
+    return lookupBool(props, NAME, fallback);
+}
+
 } // namespace matching
 
 namespace softtimeout {
diff --git a/searchlib/src/vespa/searchlib/fef/indexproperties.h b/searchlib/src/vespa/searchlib/fef/indexproperties.h
index 1921f52276f..7262e599235 100644
--- a/searchlib/src/vespa/searchlib/fef/indexproperties.h
+++ b/searchlib/src/vespa/searchlib/fef/indexproperties.h
@@ -347,6 +347,16 @@ namespace matching {
         static bool check(const Properties &props) { return check(props, DEFAULT_VALUE); }
         static bool check(const Properties &props, bool fallback);
     };
+
+    /**
+     * When enabled posting lists can be created on the fly even if iterator is not strict.
+     **/
+    struct CreatePostingListWhenNonStrict {
+        static const vespalib::string NAME;
+        static const bool DEFAULT_VALUE;
+        static bool check(const Properties &props) { return check(props, DEFAULT_VALUE); }
+        static bool check(const Properties &props, bool fallback);
+    };
 }
 
 namespace softtimeout {
diff --git a/searchlib/src/vespa/searchlib/fef/ranksetup.cpp b/searchlib/src/vespa/searchlib/fef/ranksetup.cpp
index d6b0b900516..ed74386b2fa 100644
--- a/searchlib/src/vespa/searchlib/fef/ranksetup.cpp
+++ b/searchlib/src/vespa/searchlib/fef/ranksetup.cpp
@@ -61,6 +61,7 @@ RankSetup::RankSetup(const BlueprintFactory &factory, const IIndexEnvironment &i
       _compileError(false),
       _degradationAscendingOrder(false),
       _always_mark_phrase_expensive(false),
+      _create_postinglist_when_non_strict(true),
       _diversityAttribute(),
       _diversityMinGroups(1),
       _diversityCutoffFactor(10.0),
@@ -135,6 +136,7 @@ RankSetup::configure()
     _mutateOnSummary._operation = mutate::on_summary::Operation::lookup(_indexEnv.getProperties());
     _mutateAllowQueryOverride = mutate::AllowQueryOverride::check(_indexEnv.getProperties());
     _always_mark_phrase_expensive = matching::AlwaysMarkPhraseExpensive::check(_indexEnv.getProperties());
+    _create_postinglist_when_non_strict = matching::CreatePostingListWhenNonStrict::check(_indexEnv.getProperties());
 }
 
 void
diff --git a/searchlib/src/vespa/searchlib/fef/ranksetup.h b/searchlib/src/vespa/searchlib/fef/ranksetup.h
index d744b38cc6e..d560614ce39 100644
--- a/searchlib/src/vespa/searchlib/fef/ranksetup.h
+++ b/searchlib/src/vespa/searchlib/fef/ranksetup.h
@@ -70,6 +70,7 @@ private:
     bool                     _compileError;
     bool                     _degradationAscendingOrder;
     bool                     _always_mark_phrase_expensive;
+    bool                     _create_postinglist_when_non_strict;
     vespalib::string         _diversityAttribute;
     uint32_t                 _diversityMinGroups;
     double                   _diversityCutoffFactor;
@@ -222,6 +223,7 @@ public:
         return _degradationAscendingOrder;
     }
     bool always_mark_phrase_expensive() const noexcept { return _always_mark_phrase_expensive; }
+    bool create_postinglist_when_non_strict() const noexcept { return _create_postinglist_when_non_strict; }
     /** get number of hits to collect during graceful degradation in match phase */
     uint32_t getDegradationMaxHits() const {
         return _degradationMaxHits;
diff --git a/searchlib/src/vespa/searchlib/queryeval/blueprint.cpp b/searchlib/src/vespa/searchlib/queryeval/blueprint.cpp
index 639805e116e..6bc125226de 100644
--- a/searchlib/src/vespa/searchlib/queryeval/blueprint.cpp
+++ b/searchlib/src/vespa/searchlib/queryeval/blueprint.cpp
@@ -624,7 +624,7 @@ IntermediateBlueprint::fetchPostings(const ExecuteInfo &execInfo)
     double nextHitRate = execInfo.hitRate();
     for (size_t i = 0; i < _children.size(); ++i) {
         Blueprint & child = *_children[i];
-        child.fetchPostings(ExecuteInfo::create(execInfo.isStrict() && inheritStrict(i), nextHitRate, execInfo.getDoom()));
+        child.fetchPostings(ExecuteInfo::create(execInfo.isStrict() && inheritStrict(i), nextHitRate, execInfo));
         nextHitRate = computeNextHitRate(child, nextHitRate);
     }
 }
diff --git a/searchlib/src/vespa/searchlib/queryeval/executeinfo.cpp b/searchlib/src/vespa/searchlib/queryeval/executeinfo.cpp
index 6a7ca84b72f..27366a9b924 100644
--- a/searchlib/src/vespa/searchlib/queryeval/executeinfo.cpp
+++ b/searchlib/src/vespa/searchlib/queryeval/executeinfo.cpp
@@ -4,7 +4,7 @@
 
 namespace search::queryeval {
 
-const ExecuteInfo ExecuteInfo::TRUE(true, 1.0, nullptr);
-const ExecuteInfo ExecuteInfo::FALSE(false, 1.0, nullptr);
+const ExecuteInfo ExecuteInfo::TRUE(true, 1.0, nullptr, true);
+const ExecuteInfo ExecuteInfo::FALSE(false, 1.0, nullptr, true);
 
 }
diff --git a/searchlib/src/vespa/searchlib/queryeval/executeinfo.h b/searchlib/src/vespa/searchlib/queryeval/executeinfo.h
index 362b0826f67..0095429e609 100644
--- a/searchlib/src/vespa/searchlib/queryeval/executeinfo.h
+++ b/searchlib/src/vespa/searchlib/queryeval/executeinfo.h
@@ -12,18 +12,25 @@ namespace search::queryeval {
  */
 class ExecuteInfo {
 public:
-    ExecuteInfo() noexcept : ExecuteInfo(false, 1.0F, nullptr) { }
+    ExecuteInfo() noexcept : ExecuteInfo(false, 1.0F, nullptr, true) { }
     bool isStrict() const noexcept { return _strict; }
+    bool create_postinglist_when_non_strict() const noexcept { return _create_postinglist_when_non_strict; }
     float hitRate() const noexcept { return _hitRate; }
     bool soft_doom() const noexcept { return _doom && _doom->soft_doom(); }
     const vespalib::Doom * getDoom() const { return _doom; }
     static const ExecuteInfo TRUE;
     static const ExecuteInfo FALSE;
     static ExecuteInfo create(bool strict, const ExecuteInfo & org) noexcept {
-        return {strict, org._hitRate, org.getDoom()};
+        return {strict, org._hitRate, org.getDoom(), org.create_postinglist_when_non_strict()};
+    }
+    static ExecuteInfo create(bool strict, float hitRate, const ExecuteInfo & org) noexcept {
+        return {strict, hitRate, org.getDoom(), org.create_postinglist_when_non_strict()};
     }
     static ExecuteInfo create(bool strict, float hitRate, const vespalib::Doom * doom) noexcept {
-        return {strict, hitRate, doom};
+        return create(strict, hitRate, doom, true);
+    }
+    static ExecuteInfo create(bool strict, float hitRate, const vespalib::Doom * doom, bool postinglist_when_non_strict) noexcept {
+        return {strict, hitRate, doom, postinglist_when_non_strict};
     }
     static ExecuteInfo createForTest(bool strict) noexcept {
         return createForTest(strict, 1.0F);
@@ -32,14 +39,16 @@ public:
         return create(strict, hitRate, nullptr);
     }
 private:
-    ExecuteInfo(bool strict, float hitRate_in, const vespalib::Doom * doom) noexcept
+    ExecuteInfo(bool strict, float hitRate_in, const vespalib::Doom * doom, bool postinglist_when_non_strict) noexcept
         : _doom(doom),
           _hitRate(hitRate_in),
-          _strict(strict)
+          _strict(strict),
+          _create_postinglist_when_non_strict(postinglist_when_non_strict)
     { }
     const vespalib::Doom * _doom;
     float                  _hitRate;
     bool                   _strict;
+    bool                   _create_postinglist_when_non_strict;
 };
 
 }
diff --git a/searchlib/src/vespa/searchlib/queryeval/same_element_blueprint.cpp b/searchlib/src/vespa/searchlib/queryeval/same_element_blueprint.cpp
index 5c795679d48..c93cef47c27 100644
--- a/searchlib/src/vespa/searchlib/queryeval/same_element_blueprint.cpp
+++ b/searchlib/src/vespa/searchlib/queryeval/same_element_blueprint.cpp
@@ -63,7 +63,7 @@ SameElementBlueprint::fetchPostings(const ExecuteInfo &execInfo)
     double hit_rate = execInfo.hitRate() * _terms[0]->hit_ratio();
     for (size_t i = 1; i < _terms.size(); ++i) {
         Blueprint & term = *_terms[i];
-        term.fetchPostings(ExecuteInfo::create(false, hit_rate, execInfo.getDoom()));
+        term.fetchPostings(ExecuteInfo::create(false, hit_rate, execInfo));
         hit_rate = hit_rate * term.hit_ratio();
     }
 }