Changeset: 55c595a00ec5 for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=55c595a00ec5
Modified Files:
clients/Tests/MAL-signatures.stable.out
clients/Tests/MAL-signatures.stable.out.int128
clients/Tests/exports.stable.out
gdk/gdk_join.c
gdk/gdk_select.c
monetdb5/ChangeLog
monetdb5/modules/kernel/algebra.c
monetdb5/modules/kernel/algebra.h
monetdb5/modules/kernel/algebra.mal
Branch: default
Log Message:
BATjoin and BATleftjoin now optionally don't produce the right output.
diffs (truncated from 1094 to 300 lines):
diff --git a/clients/Tests/MAL-signatures.stable.out
b/clients/Tests/MAL-signatures.stable.out
--- a/clients/Tests/MAL-signatures.stable.out
+++ b/clients/Tests/MAL-signatures.stable.out
@@ -555,7 +555,9 @@ stdout of test 'MAL-signatures` in direc
[ "algebra", "ilikeselect", "command algebra.ilikeselect(b:bat[:str],
cand:bat[:oid], pat:str, esc:str, anti:bit):bat[:oid] ",
"PCRElikeselect1;", "" ]
[ "algebra", "ilikeselect", "command algebra.ilikeselect(b:bat[:str],
cand:bat[:oid], pat:str, anti:bit):bat[:oid] ", "PCRElikeselect4;",
"" ]
[ "algebra", "intersect", "command algebra.intersect(l:bat[:any_1],
r:bat[:any_1], sl:bat[:oid], sr:bat[:oid], nil_matches:bit,
estimate:lng):bat[:oid] ", "ALGintersect;", "Intersection of l
and r with candidate lists (i.e. half of semi-join)" ]
+[ "algebra", "join", "command algebra.join(l:bat[:any_1], r:bat[:any_1],
sl:bat[:oid], sr:bat[:oid], nil_matches:bit, estimate:lng):bat[:oid] ",
"ALGjoin1;", "Join; only produce left output" ]
[ "algebra", "join", "command algebra.join(l:bat[:any_1], r:bat[:any_1],
sl:bat[:oid], sr:bat[:oid], nil_matches:bit, estimate:lng) (X_0:bat[:oid],
X_1:bat[:oid]) ", "ALGjoin;", "Join" ]
+[ "algebra", "leftjoin", "command algebra.leftjoin(l:bat[:any_1],
r:bat[:any_1], sl:bat[:oid], sr:bat[:oid], nil_matches:bit,
estimate:lng):bat[:oid] ", "ALGleftjoin1;", "Left join with candidate
lists; only produce left output" ]
[ "algebra", "leftjoin", "command algebra.leftjoin(l:bat[:any_1],
r:bat[:any_1], sl:bat[:oid], sr:bat[:oid], nil_matches:bit, estimate:lng)
(X_0:bat[:oid], X_1:bat[:oid]) ", "ALGleftjoin;", "Left join with candidate
lists" ]
[ "algebra", "like", "command algebra.like(s:str, pat:str):bit ",
"PCRElike2;", "" ]
[ "algebra", "like", "command algebra.like(s:str, pat:str, esc:str):bit ",
"PCRElike3;", "" ]
diff --git a/clients/Tests/MAL-signatures.stable.out.int128
b/clients/Tests/MAL-signatures.stable.out.int128
--- a/clients/Tests/MAL-signatures.stable.out.int128
+++ b/clients/Tests/MAL-signatures.stable.out.int128
@@ -659,7 +659,9 @@ stdout of test 'MAL-signatures` in direc
[ "algebra", "ilikeselect", "command algebra.ilikeselect(b:bat[:str],
cand:bat[:oid], pat:str, esc:str, anti:bit):bat[:oid] ",
"PCRElikeselect1;", "" ]
[ "algebra", "ilikeselect", "command algebra.ilikeselect(b:bat[:str],
cand:bat[:oid], pat:str, anti:bit):bat[:oid] ", "PCRElikeselect4;",
"" ]
[ "algebra", "intersect", "command algebra.intersect(l:bat[:any_1],
r:bat[:any_1], sl:bat[:oid], sr:bat[:oid], nil_matches:bit,
estimate:lng):bat[:oid] ", "ALGintersect;", "Intersection of l
and r with candidate lists (i.e. half of semi-join)" ]
+[ "algebra", "join", "command algebra.join(l:bat[:any_1], r:bat[:any_1],
sl:bat[:oid], sr:bat[:oid], nil_matches:bit, estimate:lng):bat[:oid] ",
"ALGjoin1;", "Join; only produce left output" ]
[ "algebra", "join", "command algebra.join(l:bat[:any_1], r:bat[:any_1],
sl:bat[:oid], sr:bat[:oid], nil_matches:bit, estimate:lng) (X_0:bat[:oid],
X_1:bat[:oid]) ", "ALGjoin;", "Join" ]
+[ "algebra", "leftjoin", "command algebra.leftjoin(l:bat[:any_1],
r:bat[:any_1], sl:bat[:oid], sr:bat[:oid], nil_matches:bit,
estimate:lng):bat[:oid] ", "ALGleftjoin1;", "Left join with candidate
lists; only produce left output" ]
[ "algebra", "leftjoin", "command algebra.leftjoin(l:bat[:any_1],
r:bat[:any_1], sl:bat[:oid], sr:bat[:oid], nil_matches:bit, estimate:lng)
(X_0:bat[:oid], X_1:bat[:oid]) ", "ALGleftjoin;", "Left join with candidate
lists" ]
[ "algebra", "like", "command algebra.like(s:str, pat:str):bit ",
"PCRElike2;", "" ]
[ "algebra", "like", "command algebra.like(s:str, pat:str, esc:str):bit ",
"PCRElike3;", "" ]
diff --git a/clients/Tests/exports.stable.out b/clients/Tests/exports.stable.out
--- a/clients/Tests/exports.stable.out
+++ b/clients/Tests/exports.stable.out
@@ -728,7 +728,9 @@ str ALGfirstn(Client cntxt, MalBlkPtr mb
str ALGgroupby(bat *res, const bat *gids, const bat *cnts);
str ALGintersect(bat *r1, const bat *lid, const bat *rid, const bat *slid,
const bat *srid, const bit *nil_matches, const lng *estimate);
str ALGjoin(bat *r1, bat *r2, const bat *l, const bat *r, const bat *sl, const
bat *sr, const bit *nil_matches, const lng *estimate);
+str ALGjoin1(bat *r1, const bat *l, const bat *r, const bat *sl, const bat
*sr, const bit *nil_matches, const lng *estimate);
str ALGleftjoin(bat *r1, bat *r2, const bat *l, const bat *r, const bat *sl,
const bat *sr, const bit *nil_matches, const lng *estimate);
+str ALGleftjoin1(bat *r1, const bat *l, const bat *r, const bat *sl, const bat
*sr, const bit *nil_matches, const lng *estimate);
str ALGmaxany(ptr result, const bat *bid);
str ALGmaxany_skipnil(ptr result, const bat *bid, const bit *skipnil);
str ALGminany(ptr result, const bat *bid);
diff --git a/gdk/gdk_join.c b/gdk/gdk_join.c
--- a/gdk/gdk_join.c
+++ b/gdk/gdk_join.c
@@ -111,6 +111,9 @@ joininitresults(BAT **r1p, BAT **r2p, BU
BAT *r1, *r2;
BUN maxsize, size;
+ /* if nil_on_miss is set, we really need a right output */
+ assert(!nil_on_miss || r2p != NULL);
+
lkey |= lcnt <= 1;
rkey |= rcnt <= 1;
@@ -329,13 +332,19 @@ selectjoin(BAT **r1p, BAT **r2p, BAT *l,
"selectjoin", t0);
}
BAT *r1 = COLnew(0, TYPE_oid, lci->ncand * BATcount(bn), TRANSIENT);
- BAT *r2 = COLnew(0, TYPE_oid, lci->ncand * BATcount(bn), TRANSIENT);
- if (r1 == NULL || r2 == NULL) {
+ if (r1 == NULL) {
BBPunfix(bn->batCacheid);
- BBPreclaim(r1);
- BBPreclaim(r2);
return GDK_FAIL;
}
+ BAT *r2 = NULL;
+ if (r2p) {
+ r2 = COLnew(0, TYPE_oid, lci->ncand * BATcount(bn), TRANSIENT);
+ if (r2 == NULL) {
+ BBPunfix(bn->batCacheid);
+ BBPreclaim(r1);
+ return GDK_FAIL;
+ }
+ }
r1->tsorted = true;
r1->trevsorted = lci->ncand == 1;
@@ -343,44 +352,56 @@ selectjoin(BAT **r1p, BAT **r2p, BAT *l,
r1->tkey = BATcount(bn) == 1;
r1->tnil = false;
r1->tnonil = true;
- r2->tsorted = lci->ncand == 1 || BATcount(bn) == 1;
- r2->trevsorted = BATcount(bn) == 1;
- r2->tseqbase = lci->ncand == 1 && BATtdense(bn) ? bn->tseqbase :
oid_nil;
- r2->tkey = lci->ncand == 1;
- r2->tnil = false;
- r2->tnonil = true;
+ if (r2) {
+ r2->tsorted = lci->ncand == 1 || BATcount(bn) == 1;
+ r2->trevsorted = BATcount(bn) == 1;
+ r2->tseqbase = lci->ncand == 1 && BATtdense(bn) ? bn->tseqbase
: oid_nil;
+ r2->tkey = lci->ncand == 1;
+ r2->tnil = false;
+ r2->tnonil = true;
+ }
if (BATtdense(bn)) {
oid *o1p = (oid *) Tloc(r1, 0);
- oid *o2p = (oid *) Tloc(r2, 0);
+ oid *o2p = r2 ? (oid *) Tloc(r2, 0) : NULL;
oid bno = bn->tseqbase;
- BUN q = BATcount(bn);
+ BUN p, q = BATcount(bn);
do {
- for (BUN p = 0; p < q; p++) {
+ for (p = 0; p < q; p++) {
*o1p++ = o;
- *o2p++ = bno + p;
+ }
+ if (o2p) {
+ for (p = 0; p < q; p++) {
+ *o2p++ = bno + p;
+ }
}
o = canditer_next(lci);
} while (!is_oid_nil(o));
} else {
oid *o1p = (oid *) Tloc(r1, 0);
- oid *o2p = (oid *) Tloc(r2, 0);
+ oid *o2p = r2 ? (oid *) Tloc(r2, 0) : NULL;
const oid *bnp = (const oid *) Tloc(bn, 0);
- BUN q = BATcount(bn);
+ BUN p, q = BATcount(bn);
do {
- for (BUN p = 0; p < q; p++) {
+ for (p = 0; p < q; p++) {
*o1p++ = o;
- *o2p++ = bnp[p];
+ }
+ if (o2p) {
+ for (p = 0; p < q; p++) {
+ *o2p++ = bnp[p];
+ }
}
o = canditer_next(lci);
} while (!is_oid_nil(o));
}
BATsetcount(r1, lci->ncand * BATcount(bn));
- BATsetcount(r2, lci->ncand * BATcount(bn));
+ *r1p = r1;
+ if (r2p) {
+ BATsetcount(r2, lci->ncand * BATcount(bn));
+ *r2p = r2;
+ }
BBPunfix(bn->batCacheid);
- *r1p = r1;
- *r2p = r2;
ALGODEBUG fprintf(stderr, "#%s: %s(l=" ALGOBATFMT ","
"r=" ALGOBATFMT ",sl=" ALGOOPTBATFMT ","
"sr=" ALGOOPTBATFMT ",nil_matches=%d)%s %s "
@@ -933,8 +954,10 @@ mergejoin_int(BAT **r1p, BAT **r2p, BAT
/* value occurs multiple times in l, so entry
* in r will be repeated multiple times: hence
* r2 is not key and not dense */
- r2->tkey = false;
- r2->tseqbase = oid_nil;
+ if (r2) {
+ r2->tkey = false;
+ r2->tseqbase = oid_nil;
+ }
/* multiple different values will be inserted
* in r1 (always in order), so not reverse
* ordered anymore */
@@ -949,13 +972,17 @@ mergejoin_int(BAT **r1p, BAT **r2p, BAT
/* multiple different values will be inserted
* in r2 (in order), so not reverse ordered
* anymore */
- r2->trevsorted = false;
- if (nl > 1) {
- /* multiple values in l match multiple
- * values in r, so an ordered sequence
- * will be inserted multiple times in
- * r2, so r2 is not ordered anymore */
- r2->tsorted = false;
+ if (r2) {
+ r2->trevsorted = false;
+ if (nl > 1) {
+ /* multiple values in l match
+ * multiple values in r, so an
+ * ordered sequence will be
+ * inserted multiple times in
+ * r2, so r2 is not ordered
+ * anymore */
+ r2->tsorted = false;
+ }
}
}
if (BATcount(r1) > 0) {
@@ -963,14 +990,17 @@ mergejoin_int(BAT **r1p, BAT **r2p, BAT
* r1, so r1 is not reverse ordered anymore */
r1->trevsorted = false;
/* a new higher value will be added to r2 */
- r2->trevsorted = false;
+ if (r2) {
+ r2->trevsorted = false;
+ }
if (BATtdense(r1) &&
((oid *) r1->theap.base)[r1->batCount - 1] + 1 !=
l->hseqbase + lstart - nl) {
r1->tseqbase = oid_nil;
}
}
- if (BATcount(r2) > 0 &&
+ if (r2 &&
+ BATcount(r2) > 0 &&
BATtdense(r2) &&
((oid *) r2->theap.base)[r2->batCount - 1] + 1 !=
r->hseqbase + rstart - nr) {
r2->tseqbase = oid_nil;
@@ -980,28 +1010,37 @@ mergejoin_int(BAT **r1p, BAT **r2p, BAT
lv = l->hseqbase + lstart - nl;
for (i = 0; i < nl; i++) {
BUN j;
- oid rv;
- rv = r->hseqbase + rstart - nr;
for (j = 0; j < nr; j++) {
APPEND(r1, lv);
- APPEND(r2, rv);
- rv++;
+ }
+ if (r2) {
+ oid rv = r->hseqbase + rstart - nr;
+
+ for (j = 0; j < nr; j++) {
+ APPEND(r2, rv);
+ rv++;
+ }
}
lv++;
}
}
/* also set other bits of heap to correct value to indicate size */
BATsetcount(r1, BATcount(r1));
- BATsetcount(r2, BATcount(r2));
- assert(BATcount(r1) == BATcount(r2));
+ if (r2) {
+ BATsetcount(r2, BATcount(r2));
+ assert(BATcount(r1) == BATcount(r2));
+ }
if (BATcount(r1) > 0) {
if (BATtdense(r1))
r1->tseqbase = ((oid *) r1->theap.base)[0];
- if (BATtdense(r2))
+ if (r2 && BATtdense(r2))
r2->tseqbase = ((oid *) r2->theap.base)[0];
} else {
- r1->tseqbase = r2->tseqbase = 0;
+ r1->tseqbase = 0;
+ if (r2) {
+ r2->tseqbase = 0;
+ }
}
ALGODEBUG fprintf(stderr, "#%s: %s(l=" ALGOBATFMT ","
"r=" ALGOBATFMT ","
@@ -1214,8 +1253,10 @@ mergejoin_lng(BAT **r1p, BAT **r2p, BAT
/* value occurs multiple times in l, so entry
* in r will be repeated multiple times: hence
* r2 is not key and not dense */
- r2->tkey = false;
- r2->tseqbase = oid_nil;
+ if (r2) {
+ r2->tkey = false;
+ r2->tseqbase = oid_nil;
+ }
/* multiple different values will be inserted
* in r1 (always in order), so not reverse
* ordered anymore */
@@ -1230,13 +1271,17 @@ mergejoin_lng(BAT **r1p, BAT **r2p, BAT
/* multiple different values will be inserted
* in r2 (in order), so not reverse ordered
* anymore */
- r2->trevsorted = false;
- if (nl > 1) {
- /* multiple values in l match multiple
- * values in r, so an ordered sequence
- * will be inserted multiple times in
- * r2, so r2 is not ordered anymore */
- r2->tsorted = false;
+ if (r2) {
+ r2->trevsorted = false;
+ if (nl > 1) {
+ /* multiple values in l match
+ * multiple values in r, so an
+ * ordered sequence will be
+ * inserted multiple times in
+ * r2, so r2 is not ordered
+ * anymore */
+ r2->tsorted = false;
+ }
}
}
if (BATcount(r1) > 0) {
@@ -1244,14 +1289,17 @@ mergejoin_lng(BAT **r1p, BAT **r2p, BAT
* r1, so r1 is not reverse ordered anymore */
r1->trevsorted = false;
/* a new higher value will be added to r2 */
- r2->trevsorted = false;
+ if (r2) {
+ r2->trevsorted = false;
+ }
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list
