Add tests to check if ANALYZE is choosing common non-periodic samples for the stat4...

FossilOrigin-Name: 175842997af134138784bff6f8e93573deb5b36b

diff --git a/manifest b/manifest
index 3855207e56bceb932249e08771e086fb6e5ed1ef..9680d50eec801eb61f7045b438799b6b666001e1 100644 (file)
--- a/manifest
+++ b/manifest
@@ -1,5 +1,5 @@
-C Fix\sa\sproblem\swith\susing\sstat4\sdata\sto\sestimate\sthe\snumber\sof\srows\sscanned\sby\sa\srange\sconstraint\son\sthe\ssecond\sor\ssubsequent\scolumn\sof\sany\sindex\swhere\san\saffinity\stransformation\smust\sbe\sapplied\sto\sthe\sconstraint\sargument.
-D 2013-09-02T07:16:40.805
+C Add\stests\sto\scheck\sif\sANALYZE\sis\schoosing\scommon\snon-periodic\ssamples\sfor\sthe\sstat4\stable.
+D 2013-09-02T11:52:11.512
 F Makefile.arm-wince-mingw32ce-gcc d6df77f1f48d690bd73162294bbba7f59507c72f
 F Makefile.in 5e41da95d92656a5004b03d3576e8b226858a28e
 F Makefile.linux-gcc 91d710bdc4998cb015f39edf3cb314ec4f4d7e23
@@ -308,7 +308,7 @@ F test/analyze5.test 765c4e284aa69ca172772aa940946f55629bc8c4
 F test/analyze6.test 19151da2c4e918905d2081b74ac5c4d47fc850ab
 F test/analyze7.test bb1409afc9e8629e414387ef048b8e0e3e0bdc4f
 F test/analyze8.test 093d15c1c888eed5034304a98c992f7360130b88
-F test/analyze9.test c4717447731bf9bb5751e0785d6337883094db3c
+F test/analyze9.test bce50d7448d4c51e1a2fd8efdcf895c2e5042e76
 F test/analyzeA.test 1a5c40079894847976d983ca39c707aaa44b6944
 F test/async.test 1d0e056ba1bb9729283a0f22718d3a25e82c277b
 F test/async2.test c0a9bd20816d7d6a2ceca7b8c03d3d69c28ffb8b
@@ -1109,7 +1109,7 @@ F tool/warnings-clang.sh f6aa929dc20ef1f856af04a730772f59283631d4
 F tool/warnings.sh fbc018d67fd7395f440c28f33ef0f94420226381
 F tool/wherecosttest.c f407dc4c79786982a475261866a161cd007947ae
 F tool/win/sqlite.vsix 97894c2790eda7b5bce3cc79cb2a8ec2fde9b3ac
-P c94933f13208f3f7d6d1b117ce6d2821100655a4
-R 15f429912bd32e56367e01c1c9d0d2ba
+P c21f58d84859e479a6cc619671a0df48b2f9692e
+R 9fc61f7137863d5244ff2fd6cc214c12
 U dan
-Z fd6aacb9a694e6de7810fd7923f14ee2
+Z 6c1b0e52f8f47aae31db001aa8668b6b

diff --git a/manifest.uuid b/manifest.uuid
index 26c3d8ab8e631b0921369200f1832683b8d3981c..569ff735b6a26756f7f02d1b991d13e68e55e5fd 100644 (file)
--- a/manifest.uuid
+++ b/manifest.uuid
@@ -1 +1 @@
-c21f58d84859e479a6cc619671a0df48b2f9692e
\ No newline at end of file
+175842997af134138784bff6f8e93573deb5b36b
\ No newline at end of file

diff --git a/test/analyze9.test b/test/analyze9.test
index bbd4a4429b9554465d7f95c50ee24577323d65b8..415d69a56cbf107ef292dfb751b48243064bad31 100644 (file)
--- a/test/analyze9.test
+++ b/test/analyze9.test
@@ -611,6 +611,99 @@ do_eqp_test 13.2.2 {
   SELECT * FROM t1 WHERE a='ott' AND b<'10' AND c=1
 } {/SEARCH TABLE t1 USING INDEX i1/}
 
-finish_test
+#-------------------------------------------------------------------------
+# By default, 16 non-periodic samples are collected for the stat4 table.
+# The following tests attempt to verify that the most common keys are
+# being collected.
+#
+proc check_stat4 {tn} {
+  db eval ANALYZE
+  db eval {SELECT a, b, c, d FROM t1} {
+    incr k($a)
+    incr k([list $a $b])
+    incr k([list $a $b $c])
+    if { [info exists k([list $a $b $c $d])]==0 } { incr nRow }
+    incr k([list $a $b $c $d])
+  }
+
+  set L [list]
+  foreach key [array names k] {
+    lappend L [list $k($key) $key]
+  }
+
+  set nSample $nRow
+  if {$nSample>16} {set nSample 16}
+
+  set nThreshold [lindex [lsort -decr -integer -index 0 $L] [expr $nSample-1] 0]
+  foreach key [array names k] {
+    if {$k($key)>$nThreshold} {
+      set expect($key) 1
+    }
+    if {$k($key)==$nThreshold} {
+      set possible($key) 1
+    }
+  }
+
 
+  set nPossible [expr $nSample - [llength [array names expect]]]
+
+  #puts "EXPECT: [array names expect]"
+  #puts "POSSIBLE($nPossible/[array size possible]): [array names possible]"
+  #puts "HAVE: [db eval {SELECT test_decode(sample) FROM sqlite_stat4 WHERE idx='i1'}]"
+
+  db eval {SELECT test_decode(sample) AS s FROM sqlite_stat4 WHERE idx='i1'} {
+    set seen 0
+    for {set i 0} {$i<4} {incr i} {
+      unset -nocomplain expect([lrange $s 0 $i])
+      if {[info exists possible([lrange $s 0 $i])]} {
+        set seen 1
+        unset -nocomplain possible([lrange $s 0 $i])
+      }
+    }
+    if {$seen} {incr nPossible -1}
+  }
+  if {$nPossible<0} {set nPossible 0}
+
+  set res [list [llength [array names expect]] $nPossible]
+  uplevel [list do_test $tn [list set {} $res] {0 0}]
+}
+
+drop_all_tables
+do_test 14.1.1 {
+  execsql {
+    CREATE TABLE t1(a,b,c,d);
+    CREATE INDEX i1 ON t1(a,b,c,d);
+  }
+  for {set i 0} {$i < 160} {incr i} {
+    execsql { INSERT INTO t1 VALUES($i,$i,$i,$i) }
+    if {($i % 10)==0} { execsql { INSERT INTO t1 VALUES($i,$i,$i,$i) } }
+  }
+} {}
+check_stat4 14.1.2
+
+do_test 14.2.1 {
+  execsql { DELETE FROM t1 }
+  for {set i 0} {$i < 1600} {incr i} {
+    execsql { INSERT INTO t1 VALUES($i/10,$i/17,$i/27,$i/37) }
+  }
+} {}
+check_stat4 14.2.2
+
+do_test 14.3.1 {
+  for {set i 0} {$i < 10} {incr i} {
+    execsql { INSERT INTO t1 VALUES($i*50,$i*50,$i*50,$i*50) }
+    execsql { INSERT INTO t1 VALUES($i*50,$i*50,$i*50,$i*50) }
+    execsql { INSERT INTO t1 VALUES($i*50,$i*50,$i*50,$i*50) }
+    execsql { INSERT INTO t1 VALUES($i*50,$i*50,$i*50,$i*50) }
+    execsql { INSERT INTO t1 VALUES($i*50,$i*50,$i*50,$i*50) }
+    execsql { INSERT INTO t1 VALUES($i*50,$i*50,$i*50,$i*50) }
+    execsql { INSERT INTO t1 VALUES($i*50,$i*50,$i*50,$i*50) }
+    execsql { INSERT INTO t1 VALUES($i*50,$i*50,$i*50,$i*50) }
+    execsql { INSERT INTO t1 VALUES($i*50,$i*50,$i*50,$i*50) }
+    execsql { INSERT INTO t1 VALUES($i*50,$i*50,$i*50,$i*50) }
+  }
+} {}
+check_stat4 14.3.2
+
+finish_test