Changeset: 5d781336bfd2 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=5d781336bfd2
Modified Files:
gdk/gdk_batop.c
Branch: Dec2016
Log Message:
Merge with Jun2016 branch.
diffs (179 lines):
diff --git a/gdk/gdk_batop.c b/gdk/gdk_batop.c
--- a/gdk/gdk_batop.c
+++ b/gdk/gdk_batop.c
@@ -58,10 +58,8 @@ static gdk_return
insert_string_bat(BAT *b, BAT *n, int force)
{
BATiter ni; /* iterator */
- int tt; /* tail type */
size_t toff = ~(size_t) 0; /* tail offset */
BUN p, q; /* loop variables */
- oid o = 0; /* in case we're appending */
const void *tp; /* tail value pointer */
unsigned char tbv; /* tail value-as-bte */
unsigned short tsv; /* tail value-as-sht */
@@ -71,18 +69,20 @@ insert_string_bat(BAT *b, BAT *n, int fo
var_t v; /* value */
size_t off; /* offset within n's string heap */
+ assert(b->ttype == TYPE_str);
+ /* only transient bats can use some other bat's string heap */
+ assert(b->batRole == TRANSIENT ||
+ b->tvheap->parentid == abs(b->batCacheid));
if (n->batCount == 0)
return GDK_SUCCEED;
ni = bat_iterator(n);
tp = NULL;
- tt = b->ttype;
- if (tt == TYPE_str &&
- (!GDK_ELIMDOUBLES(b->tvheap) || b->batCount == 0) &&
+ if ((!GDK_ELIMDOUBLES(b->tvheap) || b->batCount == 0) &&
!GDK_ELIMDOUBLES(n->tvheap) &&
b->tvheap->hashash == n->tvheap->hashash &&
/* if needs to be kept unique, take slow path */
(b->tkey & BOUND2BTRUE) == 0) {
- if (b->batRole == TRANSIENT) {
+ if (b->batRole == TRANSIENT || b->tvheap == n->tvheap) {
/* If b is in the transient farm (i.e. b will
* never become persistent), we try some
* clever tricks to avoid copying:
@@ -90,14 +90,16 @@ insert_string_bat(BAT *b, BAT *n, int fo
* string heap with n;
* - otherwise, if b's string heap and n's
* string heap are the same (i.e. shared),
- * we leave it that way;
+ * we leave it that way (this includes the
+ * case that b is persistent and n shares
+ * its string heap with b);
* - otherwise, if b shares its string heap
* with some other bat, we materialize it
* and we will have to copy strings.
*/
bat bid = b->batCacheid;
- if (b->batCount == 0) {
+ if (b->batCount == 0 && b->tvheap != n->tvheap) {
if (b->tvheap->parentid != bid) {
BBPunshare(b->tvheap->parentid);
} else {
@@ -151,8 +153,8 @@ insert_string_bat(BAT *b, BAT *n, int fo
/* make sure we get alignment right */
toff = (toff + GDK_VARALIGN - 1) &
~(GDK_VARALIGN - 1);
assert(((toff >> GDK_VARSHIFT) << GDK_VARSHIFT)
== toff);
- /* if in "force" mode, the heap may be shared
when
- * memory mapped */
+ /* if in "force" mode, the heap may be
+ * shared when memory mapped */
if (HEAPextend(b->tvheap, toff +
n->tvheap->size, force) != GDK_SUCCEED) {
toff = ~(size_t) 0;
goto bunins_failed;
@@ -182,25 +184,25 @@ insert_string_bat(BAT *b, BAT *n, int fo
}
switch (b->twidth) {
case 1:
- tt = TYPE_bte;
+ b->ttype = TYPE_bte;
tp = &tbv;
break;
case 2:
- tt = TYPE_sht;
+ b->ttype = TYPE_sht;
tp = &tsv;
break;
#if SIZEOF_VAR_T == 8
case 4:
- tt = TYPE_int;
+ b->ttype = TYPE_int;
tp = &tiv;
break;
case 8:
- tt = TYPE_lng;
+ b->ttype = TYPE_lng;
tp = &v;
break;
#else
case 4:
- tt = TYPE_int;
+ b->ttype = TYPE_int;
tp = &v;
break;
#endif
@@ -208,13 +210,11 @@ insert_string_bat(BAT *b, BAT *n, int fo
assert(0);
}
b->tvarsized = 0;
- b->ttype = tt;
}
}
if (toff == 0 && n->twidth == b->twidth) {
/* we don't need to do any translation of offset
- * values, nor do we need to do any calculations for
- * the head column, so we can use fast memcpy */
+ * values, so we can use fast memcpy */
memcpy(Tloc(b, BUNlast(b)), Tloc(n, 0),
BATcount(n) * n->twidth);
BATsetcount(b, BATcount(b) + BATcount(n));
@@ -275,7 +275,26 @@ insert_string_bat(BAT *b, BAT *n, int fo
break;
}
bunfastapp(b, tp);
- o++;
+ }
+ } else if (b->tkey & BOUND2BTRUE) {
+ BUN i = BUNlast(b);
+ /* if no duplicate values allowed, insert one-by-one */
+ BATloop(n, p, q) {
+ tp = BUNtvar(ni, p);
+ if (BUNfnd(b, tp) == BUN_NONE) {
+ bunfastapp(b, tp);
+ if (b->thash) {
+ HASHins(b, i, tp);
+ }
+ i++;
+ }
+ }
+ } else if (b->tvheap->free < n->tvheap->free / 2) {
+ /* if b's string heap is much smaller than n's string
+ * heap, don't bother checking whether n's string
+ * values occur in b's string heap */
+ BATloop(n, p, q) {
+ bunfastapp(b, BUNtvar(ni, p));
}
} else {
/* Insert values from n individually into b; however,
@@ -332,19 +351,14 @@ insert_string_bat(BAT *b, BAT *n, int fo
} else {
bunfastapp(b, tp);
}
- o++;
}
}
- if (toff != ~(size_t) 0) {
- b->tvarsized = 1;
- b->ttype = TYPE_str;
- }
+ b->tvarsized = 1;
+ b->ttype = TYPE_str;
return GDK_SUCCEED;
bunins_failed:
- if (toff != ~(size_t) 0) {
- b->tvarsized = 1;
- b->ttype = TYPE_str;
- }
+ b->tvarsized = 1;
+ b->ttype = TYPE_str;
return GDK_FAIL;
}
@@ -464,10 +478,7 @@ BATappend(BAT *b, BAT *n, bit force)
b->tnodense = r;
}
}
- if (b->ttype == TYPE_str &&
- (b->batCount == 0 || !GDK_ELIMDOUBLES(b->tvheap)) &&
- !GDK_ELIMDOUBLES(n->tvheap) &&
- b->tvheap->hashash == n->tvheap->hashash) {
+ if (b->ttype == TYPE_str) {
if (insert_string_bat(b, n, force) != GDK_SUCCEED)
return GDK_FAIL;
} else {
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list
