Changeset: 577b502bb38a for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB/rev/577b502bb38a
Modified Files:
gdk/gdk_bat.c
gdk/gdk_select.c
monetdb5/modules/kernel/algebra.c
sql/test/2024/Tests/nextafter.test
sql/test/BugTracker-2019/Tests/duplicates-not-eliminated-long-CASE-stmt.Bug-6697.test
sql/test/miscellaneous/Tests/simple_plans.test
sql/test/rel-optimizers/Tests/merge-ors-base.test
sql/test/rel-optimizers/Tests/merge-ors-multi-col-eq-to-cmp_in.test
sql/test/rel-optimizers/Tests/merge-ors-single-col-eq-to-cmp_in.test
sql/test/rel-optimizers/Tests/merge-unions.test
testing/Mtest.py.in
Branch: default
Log Message:
Merge with Aug2024 branch.
diffs (truncated from 3372 to 300 lines):
diff --git a/gdk/gdk_bat.c b/gdk/gdk_bat.c
--- a/gdk/gdk_bat.c
+++ b/gdk/gdk_bat.c
@@ -1203,6 +1203,9 @@ BUNappendmulti(BAT *b, const void *value
maxvalp = t;
}
}
+ } else {
+ b->tnil = true;
+ b->tnonil = false;
}
p++;
}
@@ -1222,6 +1225,8 @@ BUNappendmulti(BAT *b, const void *value
} else if (ATOMstorage(b->ttype) == TYPE_msk) {
bi.minpos = bi.maxpos = BUN_NONE;
minvalp = maxvalp = NULL;
+ b->tnil = false;
+ b->tnonil = true;
for (BUN i = 0; i < count; i++) {
t = (void *) ((char *) values + (i <<
b->tshift));
mskSetVal(b, p, *(msk *) t);
@@ -1258,12 +1263,16 @@ BUNappendmulti(BAT *b, const void *value
maxvalp = t;
}
}
+ } else {
+ b->tnil = true;
+ b->tnonil = false;
}
p++;
}
nunique = b->thash ? b->thash->nunique : 0;
}
} else {
+ /* inserting nils, unless it's msk */
for (BUN i = 0; i < count; i++) {
gdk_return rc = tfastins_nocheck(b, p, t);
if (rc != GDK_SUCCEED) {
@@ -1276,6 +1285,8 @@ BUNappendmulti(BAT *b, const void *value
p++;
}
nunique = b->thash ? b->thash->nunique : 0;
+ b->tnil = b->ttype != TYPE_msk;
+ b->tnonil = false;
}
MT_lock_set(&b->theaplock);
b->tminpos = bi.minpos;
@@ -1286,8 +1297,6 @@ BUNappendmulti(BAT *b, const void *value
if (b->ttype == TYPE_oid) {
/* spend extra effort on oid (possible candidate list) */
if (values == NULL || is_oid_nil(((oid *) values)[0])) {
- b->tnil = true;
- b->tnonil = false;
b->tsorted = false;
b->trevsorted = false;
b->tkey = false;
@@ -1298,8 +1307,6 @@ BUNappendmulti(BAT *b, const void *value
b->trevsorted = true;
b->tkey = true;
b->tseqbase = count == 1 ? ((oid *) values)[0]
: oid_nil;
- b->tnil = false;
- b->tnonil = true;
} else {
if (!is_oid_nil(b->tseqbase) &&
(count > 1 ||
@@ -1328,8 +1335,6 @@ BUNappendmulti(BAT *b, const void *value
}
for (BUN i = 1; i < count; i++) {
if (is_oid_nil(((oid *) values)[i])) {
- b->tnil = true;
- b->tnonil = false;
b->tsorted = false;
b->trevsorted = false;
b->tkey = false;
@@ -1360,18 +1365,14 @@ BUNappendmulti(BAT *b, const void *value
}
}
} else if (!ATOMlinear(b->ttype)) {
- b->tnil = b->tnonil = false;
b->tsorted = b->trevsorted = b->tkey = false;
} else if (b->batCount == 0) {
if (values == NULL) {
b->tsorted = b->trevsorted = true;
b->tkey = count == 1;
- b->tnil = true;
- b->tnonil = false;
b->tunique_est = 1;
} else {
int c;
- b->tnil = b->tnonil = false;
switch (count) {
case 1:
b->tsorted = b->trevsorted = b->tkey = true;
@@ -1418,11 +1419,7 @@ BUNappendmulti(BAT *b, const void *value
b->tnokey[0] = 0;
b->tnokey[1] = !b->tkey;
b->tunique_est = (double) (1 + b->tkey);
- b->tnil |= values == NULL;
- b->tnonil = false;
} else {
- b->tnil |= values == NULL;
- b->tnonil = false;
b->tsorted = b->trevsorted = b->tkey = false;
}
BATsetcount(b, p);
diff --git a/gdk/gdk_logger.c b/gdk/gdk_logger.c
--- a/gdk/gdk_logger.c
+++ b/gdk/gdk_logger.c
@@ -767,6 +767,18 @@ la_bat_updates(logger *lg, logaction *la
const void *t = BUNtail(vi, p);
if (q < cnt) {
+ if (b->tnosorted == q)
+ b->tnosorted = 0;
+ if (b->tnorevsorted == q)
+ b->tnorevsorted = 0;
+ if (b->tnokey[0] == q ||
+ b->tnokey[1] == q) {
+ b->tnokey[0] = 0;
+ b->tnokey[1] = 0;
+ }
+ b->tkey = false;
+ b->tsorted = false;
+ b->tkey = false;
if (BUNreplace(b, q, t, true)
!= GDK_SUCCEED) {
logbat_destroy(b);
bat_iterator_end(&vi);
diff --git a/gdk/gdk_select.c b/gdk/gdk_select.c
--- a/gdk/gdk_select.c
+++ b/gdk/gdk_select.c
@@ -1375,6 +1375,7 @@ BATrange(BATiter *bi, const void *tl, co
BAT *pb = NULL;
int c;
int (*atomcmp) (const void *, const void *) = ATOMcompare(bi->type);
+ BATiter bi2 = *bi;
if (tl && (*atomcmp)(tl, ATOMnilptr(bi->type)) == 0)
tl = NULL;
@@ -1388,26 +1389,36 @@ BATrange(BATiter *bi, const void *tl, co
/* keep locked while we look at the property values */
MT_lock_set(&bi->b->theaplock);
- if (bi->minpos != BUN_NONE)
+ if (bi->sorted && (bi->nonil || atomcmp(BUNtail(*bi, 0),
ATOMnilptr(bi->type)) != 0))
+ minval = BUNtail(*bi, 0);
+ else if (bi->revsorted && (bi->nonil || atomcmp(BUNtail(*bi, bi->count
- 1), ATOMnilptr(bi->type)) != 0))
+ minval = BUNtail(*bi, bi->count - 1);
+ else if (bi->minpos != BUN_NONE)
minval = BUNtail(*bi, bi->minpos);
else if ((minprop = BATgetprop_nolock(bi->b, GDK_MIN_BOUND)) != NULL)
minval = VALptr(minprop);
- if (bi->maxpos != BUN_NONE) {
- maxval = BUNtail(*bi, bi->maxpos);
+ if (bi->sorted && (bi->nonil || atomcmp(BUNtail(bi2, bi->count - 1),
ATOMnilptr(bi->type)) != 0)) {
+ maxval = BUNtail(bi2, bi->count - 1);
+ maxincl = true;
+ } else if (bi->revsorted && (bi->nonil || atomcmp(BUNtail(bi2, 0),
ATOMnilptr(bi->type)) != 0)) {
+ maxval = BUNtail(bi2, 0);
+ maxincl = true;
+ } else if (bi->maxpos != BUN_NONE) {
+ maxval = BUNtail(bi2, bi->maxpos);
maxincl = true;
} else if ((maxprop = BATgetprop_nolock(bi->b, GDK_MAX_BOUND)) != NULL)
{
maxval = VALptr(maxprop);
maxincl = false;
}
bool keep = false; /* keep lock on parent bat? */
- if (minprop == NULL || maxprop == NULL) {
+ if (minval == NULL || maxval == NULL) {
if (pb != NULL) {
MT_lock_set(&pb->theaplock);
- if (minprop == NULL && (minprop = BATgetprop_nolock(pb,
GDK_MIN_BOUND)) != NULL) {
+ if (minval == NULL && (minprop = BATgetprop_nolock(pb,
GDK_MIN_BOUND)) != NULL) {
keep = true;
minval = VALptr(minprop);
}
- if (maxprop == NULL && (maxprop = BATgetprop_nolock(pb,
GDK_MAX_BOUND)) != NULL) {
+ if (maxval == NULL && (maxprop = BATgetprop_nolock(pb,
GDK_MAX_BOUND)) != NULL) {
keep = true;
maxval = VALptr(maxprop);
maxincl = true;
@@ -1418,20 +1429,20 @@ BATrange(BATiter *bi, const void *tl, co
}
}
- if (minprop == NULL && maxprop == NULL) {
+ if (minval == NULL && maxval == NULL) {
range = range_inside; /* strictly: unknown */
- } else if (maxprop &&
+ } else if (maxval &&
tl &&
((c = atomcmp(tl, maxval)) > 0 ||
((!maxincl || !li) && c == 0))) {
range = range_after;
- } else if (minprop &&
+ } else if (minval &&
th &&
((c = atomcmp(th, minval)) < 0 ||
(!hi && c == 0))) {
range = range_before;
} else if (tl == NULL) {
- if (minprop == NULL) {
+ if (minval == NULL) {
c = atomcmp(th, maxval);
if (c < 0 || ((maxincl || !hi) && c == 0))
range = range_atstart;
@@ -1441,7 +1452,7 @@ BATrange(BATiter *bi, const void *tl, co
c = atomcmp(th, minval);
if (c < 0 || (!hi && c == 0))
range = range_before;
- else if (maxprop == NULL)
+ else if (maxval == NULL)
range = range_atstart;
else {
c = atomcmp(th, maxval);
@@ -1452,7 +1463,7 @@ BATrange(BATiter *bi, const void *tl, co
}
}
} else if (th == NULL) {
- if (maxprop == NULL) {
+ if (maxval == NULL) {
c = atomcmp(tl, minval);
if (c >= 0)
range = range_atend;
@@ -1462,7 +1473,7 @@ BATrange(BATiter *bi, const void *tl, co
c = atomcmp(tl, maxval);
if (c > 0 || ((!maxincl || !li) && c == 0))
range = range_after;
- else if (minprop == NULL)
+ else if (minval == NULL)
range = range_atend;
else {
c = atomcmp(tl, minval);
@@ -1472,13 +1483,13 @@ BATrange(BATiter *bi, const void *tl, co
range = range_contains;
}
}
- } else if (minprop == NULL) {
+ } else if (minval == NULL) {
c = atomcmp(th, maxval);
if (c < 0 || ((maxincl || !hi) && c == 0))
range = range_inside;
else
range = range_atend;
- } else if (maxprop == NULL) {
+ } else if (maxval == NULL) {
c = atomcmp(tl, minval);
if (c >= 0)
range = range_inside;
@@ -1907,11 +1918,12 @@ BATselect(BAT *b, BAT *s, const void *tl
else
pb = NULL;
pbi = bat_iterator(pb);
- /* use hash only for equi-join, and then only if b or its
- * parent already has a hash, or if b or its parent is
- * persistent and the total size wouldn't be too large; check
- * for existence of hash last since that may involve I/O */
- if (equi || antiequi) {
+ /* use hash only for equi-join if the bat is not sorted, but
+ * only if b or its parent already has a hash, or if b or its
+ * parent is persistent and the total size wouldn't be too
+ * large; check for existence of hash last since that may
+ * involve I/O */
+ if ((equi || antiequi) && !bi.sorted && !bi.revsorted) {
double cost = joincost(b, 1, &ci, &havehash, &phash, NULL);
if (cost > 0 && cost < ci.ncand) {
wanthash = true;
@@ -2011,7 +2023,7 @@ BATselect(BAT *b, BAT *s, const void *tl
}
}
- if (!havehash && (bi.sorted || bi.revsorted || oidxh != NULL)) {
+ if (bi.sorted || bi.revsorted || (!havehash && oidxh != NULL)) {
BUN low = 0;
BUN high = bi.count;
diff --git a/monetdb5/mal/Tests/dataflow01.maltest
b/monetdb5/mal/Tests/dataflow01.maltest
--- a/monetdb5/mal/Tests/dataflow01.maltest
+++ b/monetdb5/mal/Tests/dataflow01.maltest
@@ -4,7 +4,7 @@ b:= bat.new(:lng);
k:=0;
INT_MAX := 2147483647;
dbgmsk_restore := mdb.getDebug();
-dbgmsk_unset := 8+8388608;
+dbgmsk_unset := 8+2;
dbgmsk_keep := calc.xor(INT_MAX,dbgmsk_unset);
dbgmsk_set := calc.and(dbgmsk_restore,dbgmsk_keep);
mdb.setDebug(dbgmsk_set);
diff --git a/monetdb5/mal/Tests/tst901a.maltest
b/monetdb5/mal/Tests/tst901a.maltest
--- a/monetdb5/mal/Tests/tst901a.maltest
+++ b/monetdb5/mal/Tests/tst901a.maltest
@@ -3,7 +3,7 @@ function foo();
b:= bat.new(:lng);
INT_MAX := 2147483647;
dbgmsk_restore := mdb.getDebug();
-dbgmsk_unset := 8+8388608;
+dbgmsk_unset := 8+2;
dbgmsk_keep := calc.xor(INT_MAX,dbgmsk_unset);
dbgmsk_set := calc.and(dbgmsk_restore,dbgmsk_keep);
_______________________________________________
checkin-list mailing list -- checkin-list@monetdb.org
To unsubscribe send an email to checkin-list-le...@monetdb.org
