Changeset: 29cf41414a05 for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=29cf41414a05
Modified Files:
gdk/gdk_cand.h
gdk/gdk_join.c
gdk/gdk_private.h
gdk/gdk_select.c
sql/test/SQLite_regress/sqllogictest/Tests/select3.test.stable.out
sql/test/SQLite_regress/sqllogictest/Tests/select3.test.stable.out.int128
Branch: candidate-exceptions
Log Message:
Converted BATselect to candidate iterators.
diffs (truncated from 78388 to 300 lines):
diff --git a/gdk/gdk_cand.h b/gdk/gdk_cand.h
--- a/gdk/gdk_cand.h
+++ b/gdk/gdk_cand.h
@@ -98,6 +98,20 @@ canditer_next(struct canditer *ci)
}
return o;
}
+#define canditer_next_dense(ci) ((ci)->next == (ci)->ncand ?
oid_nil : (ci)->seq + (ci)->next++)
+#define canditer_next_mater(ci) ((ci)->next == (ci)->ncand ?
oid_nil : (ci)->oids[(ci)->next++])
+static inline oid
+canditer_next_except(struct canditer *ci)
+{
+ if (ci->next == ci->ncand)
+ return oid_nil;
+ oid o = ci->seq + ci->add + ci->next++;
+ while (ci->add < ci->noids && o == ci->oids[ci->add]) {
+ ci->add++;
+ o++;
+ }
+ return o;
+}
gdk_export BUN canditer_init(struct canditer *ci, BAT *b, BAT *s);
gdk_export oid canditer_peek(struct canditer *ci);
diff --git a/gdk/gdk_join.c b/gdk/gdk_join.c
--- a/gdk/gdk_join.c
+++ b/gdk/gdk_join.c
@@ -431,10 +431,8 @@ mergejoin_void(BAT **r1p, BAT **r2p, BAT
* has a match in r */
assert(ATOMtype(l->ttype) == ATOMtype(r->ttype));
assert(r->tsorted || r->trevsorted);
- assert(sl == NULL || sl->tsorted);
- assert(sr == NULL || sr->tsorted);
assert(BATcount(l) > 0);
- assert(BATtdense(r));
+ assert(rci->tpe == cand_dense);
assert(BATcount(r) > 0);
/* figure out range [lo..hi) of values in r that we need to match */
diff --git a/gdk/gdk_private.h b/gdk/gdk_private.h
--- a/gdk/gdk_private.h
+++ b/gdk/gdk_private.h
@@ -120,10 +120,6 @@ void BBPdump(void) /* never called: for
__attribute__((__visibility__("hidden")));
__hidden Heap *createOIDXheap(BAT *b, bool stable)
__attribute__((__visibility__("hidden")));
-__hidden BAT *doublerange(oid l1, oid h1, oid l2, oid h2)
- __attribute__((__visibility__("hidden")));
-__hidden BAT *doubleslice(BAT *b, BUN l1, BUN h1, BUN l2, BUN h2)
- __attribute__((__visibility__("hidden")));
__hidden void gdk_bbp_reset(void)
__attribute__((__visibility__("hidden")));
__hidden gdk_return GDKextend(const char *fn, size_t size)
diff --git a/gdk/gdk_select.c b/gdk/gdk_select.c
--- a/gdk/gdk_select.c
+++ b/gdk/gdk_select.c
@@ -52,7 +52,14 @@ virtualize(BAT *bn)
bn->tseqbase = 0;
else
bn->tseqbase = * (const oid *) Tloc(bn, 0);
- HEAPfree(&bn->theap, true);
+ if (VIEWtparent(bn)) {
+ BBPunshare(VIEWtparent(bn));
+ BBPunfix(VIEWtparent(bn));
+ bn->theap.parentid = 0;
+ bn->theap.base = NULL;
+ } else {
+ HEAPfree(&bn->theap, true);
+ }
bn->theap.storage = bn->theap.newstorage = STORE_MEM;
bn->theap.size = 0;
bn->ttype = TYPE_void;
@@ -64,69 +71,6 @@ virtualize(BAT *bn)
return bn;
}
-BAT *
-doublerange(oid l1, oid h1, oid l2, oid h2)
-{
- BAT *bn;
- oid *restrict p;
-
- assert(l1 <= h1);
- assert(l2 <= h2);
- assert(h1 <= l2);
- if (l1 == h1 || l2 == h2) {
- return BATdense(0, l1 == h1 ? l2 : l1, h1 - l1 + h2 - l2);
- }
- bn = COLnew(0, TYPE_oid, h1 - l1 + h2 - l2, TRANSIENT);
- if (bn == NULL)
- return NULL;
- BATsetcount(bn, h1 - l1 + h2 - l2);
- p = (oid *) Tloc(bn, 0);
- while (l1 < h1)
- *p++ = l1++;
- while (l2 < h2)
- *p++ = l2++;
- bn->tkey = true;
- bn->tsorted = true;
- bn->trevsorted = BATcount(bn) <= 1;
- bn->tnil = false;
- bn->tnonil = true;
- return bn;
-}
-
-BAT *
-doubleslice(BAT *b, BUN l1, BUN h1, BUN l2, BUN h2)
-{
- BAT *bn;
- oid *restrict p;
- const oid *restrict o;
-
- assert(l1 <= h1);
- assert(l2 <= h2);
- assert(h1 <= l2);
- assert(b->tsorted);
- assert(b->tkey);
- if (b->ttype == TYPE_void)
- return doublerange(l1 + b->tseqbase, h1 + b->tseqbase,
- l2 + b->tseqbase, h2 + b->tseqbase);
- bn = COLnew(0, TYPE_oid, h1 - l1 + h2 - l2, TRANSIENT);
- if (bn == NULL)
- return NULL;
- BATsetcount(bn, h1 - l1 + h2 - l2);
- p = (oid *) Tloc(bn, 0);
- o = (const oid *) Tloc(b, l1);
- while (l1++ < h1)
- *p++ = *o++;
- o = (const oid *) Tloc(b, l2);
- while (l2++ < h2)
- *p++ = *o++;
- bn->tkey = true;
- bn->tsorted = true;
- bn->trevsorted = BATcount(bn) <= 1;
- bn->tnil = false;
- bn->tnonil = true;
- return virtualize(bn);
-}
-
#define HASHloop_bound(bi, h, hb, v, lo, hi) \
for (hb = HASHget(h, HASHprobe((h), v)); \
hb != HASHnil(h); \
@@ -136,19 +80,20 @@ doubleslice(BAT *b, BUN l1, BUN h1, BUN
(*cmp)(v, BUNtail(bi, hb)) == 0))
static BAT *
-hashselect(BAT *b, BAT *s, BAT *bn, const void *tl, BUN maximum, bool phash)
+hashselect(BAT *b, struct canditer *restrict ci, BAT *bn,
+ const void *tl, BUN maximum, bool phash)
{
BATiter bi;
BUN i, cnt;
oid o, *restrict dst;
- BUN l, h;
+ BUN l, h, d = 0;
oid seq;
int (*cmp)(const void *, const void *);
assert(bn->ttype == TYPE_oid);
seq = b->hseqbase;
- l = 0;
- h = BUNlast(b);
+ l = ci->seq - seq;
+ h = canditer_last(ci) + 1 - seq;
if (phash) {
BAT *b2 = BBPdescriptor(VIEWtparent(b));
@@ -158,22 +103,12 @@ hashselect(BAT *b, BAT *s, BAT *bn, cons
"for hash\n",
ALGOBATPAR(b),
ALGOBATPAR(b2));
- l = (BUN) ((b->theap.base - b2->theap.base) >> b->tshift);
- h = l + BATcount(b);
+ d = (BUN) ((b->theap.base - b2->theap.base) >> b->tshift);
+ l += d;
+ h += d;
b = b2;
}
- if (s && BATtdense(s)) {
- /* no need for binary search in s, we just adjust the
- * boundaries */
- if (s->tseqbase + BATcount(s) < seq + (h - l))
- h -= seq + (h - l) - (s->tseqbase + BATcount(s));
- if (s->tseqbase > seq) {
- l += s->tseqbase - seq;
- seq += s->tseqbase - seq;
- }
- s = NULL;
- }
if (BAThash(b) != GDK_SUCCEED) {
BBPreclaim(bn);
return NULL;
@@ -190,11 +125,10 @@ hashselect(BAT *b, BAT *s, BAT *bn, cons
bi = bat_iterator(b);
dst = (oid *) Tloc(bn, 0);
cnt = 0;
- if (s) {
- assert(s->tsorted);
+ if (ci->tpe != cand_dense) {
HASHloop_bound(bi, b->thash, i, tl, l, h) {
- o = (oid) (i - l + seq);
- if (BATcandcontains(s, o)) {
+ o = (oid) (i + seq - d);
+ if (canditer_search(ci, o, false) != BUN_NONE) {
buninsfix(bn, dst, cnt, o,
maximum - BATcapacity(bn),
maximum, NULL);
@@ -203,7 +137,7 @@ hashselect(BAT *b, BAT *s, BAT *bn, cons
}
} else {
HASHloop_bound(bi, b->thash, i, tl, l, h) {
- o = (oid) (i - l + seq);
+ o = (oid) (i + seq - d);
buninsfix(bn, dst, cnt, o,
maximum - BATcapacity(bn),
maximum, NULL);
@@ -230,36 +164,33 @@ hashselect(BAT *b, BAT *s, BAT *bn, cons
/* Imprints select code */
/* inner check */
-#define impscheck(CAND,TEST,ADD) \
- do { \
- e = (BUN) (i+limit-pr_off+off); \
- if (im[icnt] & mask) { \
- if ((im[icnt] & ~innermask) == 0) { \
- while (p < q && o < e) { \
- v = src[o-off]; \
- ADD; \
- cnt++; \
- p++; \
- if (p < q) \
- CAND; \
- } \
- } else { \
- while (p < q && o < e) { \
- v = src[o-off]; \
- ADD; \
- cnt += (TEST) != 0; \
- p++; \
- if (p < q) \
- CAND; \
- } \
- } \
- } else { \
- while (p < q && o < e) { \
- p++; \
- if (p < q) \
- CAND; \
- } \
- } \
+#define impscheck(canditer_next,TEST,ADD) \
+ do { \
+ const oid e = (oid) (i+limit-pr_off+hseq); \
+ if (im[icnt] & mask) { \
+ if ((im[icnt] & ~innermask) == 0) { \
+ while (p < ci->ncand && o < e) { \
+ v = src[o-hseq]; \
+ ADD; \
+ cnt++; \
+ p++; \
+ o = canditer_next(ci); \
+ } \
+ } else { \
+ while (p < ci->ncand && o < e) { \
+ v = src[o-hseq]; \
+ ADD; \
+ cnt += (TEST) != 0; \
+ p++; \
+ o = canditer_next(ci); \
+ } \
+ } \
+ } else { \
+ while (p < ci->ncand && o < e) { \
+ p++; \
+ o = canditer_next(ci); \
+ } \
+ } \
} while (false)
/* main loop for imprints */
@@ -268,38 +199,38 @@ hashselect(BAT *b, BAT *s, BAT *bn, cons
* dcnt is the iterator for dictionary entries
* i is the iterator for the values in imprints
*/
-#define impsloop(CAND,TEST,ADD)
\
+#define impsloop(canditer_next,TEST,ADD) \
do { \
- BUN dcnt, icnt, limit, i, l, e; \
- cchdc_t *restrict d = (cchdc_t *) imprints->dict; \
- uint8_t rpp = ATOMelmshift(IMPS_PAGE >> b->tshift); \
- CAND; \
- for (i = 0, dcnt = 0, icnt = 0; \
- dcnt < imprints->dictcnt && i + off < w + pr_off && p < q;
\
+ BUN dcnt, icnt, limit, i; \
+ const cchdc_t *restrict d = (cchdc_t *) imprints->dict; \
+ const uint8_t rpp = ATOMelmshift(IMPS_PAGE >> b->tshift); \
+ o = canditer_next(ci); \
+ for (i = 0, dcnt = 0, icnt = 0, p = 0; \
+ dcnt < imprints->dictcnt && i <= w - hseq + pr_off && p <
ci->ncand; \
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list
