Changeset: 96ee870d1d0e for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=96ee870d1d0e
Modified Files:
clients/Tests/MAL-signatures.stable.out
clients/Tests/MAL-signatures.stable.out.int128
gdk/ChangeLog.Dec2016
gdk/gdk.h
gdk/gdk_align.c
gdk/gdk_bat.c
gdk/gdk_batop.c
gdk/gdk_calc.c
gdk/gdk_join.c
gdk/gdk_project.c
gdk/gdk_search.c
monetdb5/modules/kernel/bat5.c
monetdb5/modules/kernel/bat5.mal
Branch: Dec2016
Log Message:
Split the 2 bit tkey field into a 1 bit tkey and a 1 bit tunique field.
The tunique field holds the old tkey&BOUND2BTRUE value.
Also, in bat.setKey, test for uniqueness when setting the property.
diffs (truncated from 812 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
@@ -599,7 +599,7 @@ Ready.
[ "bat", "setColumn", "command
bat.setColumn(b:bat[:any_1],t:str):void ", "BKCsetColumn;", "Give a
logical name to the tail column of a BAT." ]
[ "bat", "setHash", "command bat.setHash(b:bat[:any_1]):bit ",
"BKCsetHash;", "Create a hash structure on the column" ]
[ "bat", "setImprints", "command bat.setImprints(b:bat[:any_1]):bit ",
"BKCsetImprints;", "Create an imprints structure on the column" ]
-[ "bat", "setKey", "command
bat.setKey(b:bat[:any_1],mode:bit):bat[:any_1] ", "BKCsetkey;", "Sets
the 'key' property of the tail column to 'mode'. In 'key' mode, \n the
kernel will silently block insertions that cause a duplicate \n entries
in the head column. KNOWN BUG:when 'key' is set to TRUE, this \n\t\tfunction
does not automatically eliminate duplicates. " ]
+[ "bat", "setKey", "command
bat.setKey(b:bat[:any_1],mode:bit):bat[:any_1] ", "BKCsetkey;", "Sets
the 'key' property of the tail column to 'mode'. In 'key' mode,\n the
kernel will silently block insertions that cause a duplicate\n entry in
the head column." ]
[ "bat", "setName", "command bat.setName(b:bat[:any_1],s:str):void
", "BKCsetName;", "Give a logical name to a BAT. " ]
[ "bat", "setPersistent", "command
bat.setPersistent(b:bat[:any_1]):void ", "BKCsetPersistent;", "Make
the BAT persistent." ]
[ "bat", "setTransient", "command bat.setTransient(b:bat[:any_1]):void
", "BKCsetTransient;", "Make the BAT transient. Returns \n\tboolean
which indicates if the\nBAT administration has indeed changed." ]
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
@@ -705,7 +705,7 @@ Ready.
[ "bat", "setColumn", "command
bat.setColumn(b:bat[:any_1],t:str):void ", "BKCsetColumn;", "Give a
logical name to the tail column of a BAT." ]
[ "bat", "setHash", "command bat.setHash(b:bat[:any_1]):bit ",
"BKCsetHash;", "Create a hash structure on the column" ]
[ "bat", "setImprints", "command bat.setImprints(b:bat[:any_1]):bit ",
"BKCsetImprints;", "Create an imprints structure on the column" ]
-[ "bat", "setKey", "command
bat.setKey(b:bat[:any_1],mode:bit):bat[:any_1] ", "BKCsetkey;", "Sets
the 'key' property of the tail column to 'mode'. In 'key' mode, \n the
kernel will silently block insertions that cause a duplicate \n entries
in the head column. KNOWN BUG:when 'key' is set to TRUE, this \n\t\tfunction
does not automatically eliminate duplicates. " ]
+[ "bat", "setKey", "command
bat.setKey(b:bat[:any_1],mode:bit):bat[:any_1] ", "BKCsetkey;", "Sets
the 'key' property of the tail column to 'mode'. In 'key' mode,\n the
kernel will silently block insertions that cause a duplicate\n entry in
the head column." ]
[ "bat", "setName", "command bat.setName(b:bat[:any_1],s:str):void
", "BKCsetName;", "Give a logical name to a BAT. " ]
[ "bat", "setPersistent", "command
bat.setPersistent(b:bat[:any_1]):void ", "BKCsetPersistent;", "Make
the BAT persistent." ]
[ "bat", "setTransient", "command bat.setTransient(b:bat[:any_1]):void
", "BKCsetTransient;", "Make the BAT transient. Returns \n\tboolean
which indicates if the\nBAT administration has indeed changed." ]
diff --git a/gdk/ChangeLog.Dec2016 b/gdk/ChangeLog.Dec2016
--- a/gdk/ChangeLog.Dec2016
+++ b/gdk/ChangeLog.Dec2016
@@ -4,6 +4,9 @@
* Thu Dec 1 2016 Sjoerd Mullender <sjo...@acm.org>
- The tnokey values must now be 0 if it is not known whether all values
in a column are distinct.
+- The 2-bit tkey field in the bat descriptor has been split into two
+ single bit fields: tkey and tunique. The old tkey&BOUND2BTRUE value
+ is now stored in tunique.
* Wed Oct 26 2016 Sjoerd Mullender <sjo...@acm.org>
- Implemented conversion to str from any type (not just the internal
diff --git a/gdk/gdk.h b/gdk/gdk.h
--- a/gdk/gdk.h
+++ b/gdk/gdk.h
@@ -513,7 +513,6 @@
#define TRUE true
#define FALSE false
#endif
-#define BOUND2BTRUE 2 /* TRUE, and bound to be so */
#define IDLENGTH 64 /* maximum BAT id length */
#define BATMARGIN 1.2 /* extra free margin for new heaps */
@@ -764,7 +763,8 @@ gdk_export int VALisnil(const ValRecord
* // Tail properties
* int ttype; // Tail type number
* str tident; // name for tail column
- * bit tkey; // tail values should be unique?
+ * bit tkey; // tail values are unique
+ * bit tunique; // tail values must be kept unique
* bit tnonil; // tail has no nils
* bit tsorted; // are tail values currently ordered?
* bit tvarsized; // for speed: tail type is varsized?
@@ -824,7 +824,8 @@ typedef struct {
bte shift; /* log2 of bunwidth */
unsigned int
varsized:1, /* varsized (1) or fixedsized (0) */
- key:2, /* duplicates allowed? */
+ key:1, /* no duplicate values present */
+ unique:1, /* no duplicate values allowed */
dense:1, /* OID only: only consecutive values */
nonil:1, /* there are no nils in the column */
nil:1, /* there is a nil in the column */
@@ -890,6 +891,7 @@ typedef struct BATiter {
#define creator_tid S.tid
#define ttype T.type
#define tkey T.key
+#define tunique T.unique
#define tvarsized T.varsized
#define tseqbase T.seq
#define tsorted T.sorted
diff --git a/gdk/gdk_align.c b/gdk/gdk_align.c
--- a/gdk/gdk_align.c
+++ b/gdk/gdk_align.c
@@ -417,8 +417,8 @@ VIEWreset(BAT *b)
b->theap.parentid = 0;
b->batRestricted = BAT_WRITE;
- /* reset BOUND2BTRUE */
b->tkey = BATtkey(v);
+ b->tunique = 0;
/* copy the heaps */
b->theap = tail;
@@ -442,8 +442,8 @@ VIEWreset(BAT *b)
b->batCopiedtodisk = 0;
b->batDirty = 1;
- /* reset BOUND2KEY */
b->tkey = BATtkey(v);
+ b->tunique = 0;
/* make the BAT empty and insert all again */
DELTAinit(b);
diff --git a/gdk/gdk_bat.c b/gdk/gdk_bat.c
--- a/gdk/gdk_bat.c
+++ b/gdk/gdk_bat.c
@@ -88,6 +88,7 @@ BATcreatedesc(oid hseq, int tt, int heap
bn->ttype = tt;
bn->tkey = FALSE;
+ bn->tunique = FALSE;
bn->tnonil = TRUE;
bn->tnil = FALSE;
bn->tsorted = bn->trevsorted = ATOMlinear(tt) != 0;
@@ -909,7 +910,7 @@ setcolprops(BAT *b, const void *x)
/* first value */
b->tsorted = b->trevsorted = ATOMlinear(b->ttype) != 0;
b->tnosorted = b->tnorevsorted = 0;
- b->tkey |= 1;
+ b->tkey = 1;
b->tnokey[0] = b->tnokey[1] = 0;
b->tnodense = 0;
if (b->ttype == TYPE_void) {
@@ -954,7 +955,8 @@ setcolprops(BAT *b, const void *x)
prv = BUNtail(bi, pos - 1);
cmp = atom_CMP(prv, x, b->ttype);
- if (b->tkey == 1 && /* assume outside check if BOUND2BTRUE */
+ if (!b->tunique && /* assume outside check if tunique */
+ b->tkey &&
(cmp == 0 || /* definitely not KEY */
(b->batCount > 1 && /* can't guarantee KEY if unordered */
((b->tsorted && cmp > 0) ||
@@ -1002,7 +1004,7 @@ BUNappend(BAT *b, const void *t, bit for
BATcheck(b, "BUNappend", GDK_FAIL);
assert(!isVIEW(b));
- if ((b->tkey & BOUND2BTRUE) && BUNfnd(b, t) != BUN_NONE) {
+ if (b->tunique && BUNfnd(b, t) != BUN_NONE) {
return GDK_SUCCEED;
}
@@ -1094,7 +1096,7 @@ BUNdelete(BAT *b, oid o)
b->batCount--;
if (b->batCount <= 1) {
/* some trivial properties */
- b->tkey |= 1;
+ b->tkey = 1;
b->tsorted = b->trevsorted = 1;
b->tnosorted = b->tnorevsorted = 0;
if (b->batCount == 0) {
@@ -1185,7 +1187,7 @@ BUNinplace(BAT *b, BUN p, const void *t,
}
} else if (b->tnorevsorted >= p)
b->tnorevsorted = 0;
- if (((b->ttype != TYPE_void) & b->tkey & !(b->tkey & BOUND2BTRUE)) &&
b->batCount > 1) {
+ if (((b->ttype != TYPE_void) & b->tkey & !b->tunique) && b->batCount >
1) {
BATkey(b, FALSE);
}
if (b->tnonil)
@@ -1211,7 +1213,7 @@ BUNreplace(BAT *b, oid id, const void *t
if (id < b->hseqbase || id >= b->hseqbase + BATcount(b))
return GDK_SUCCEED;
- if ((b->tkey & BOUND2BTRUE) && BUNfnd(b, t) != BUN_NONE) {
+ if (b->tunique && BUNfnd(b, t) != BUN_NONE) {
return GDK_SUCCEED;
}
if (b->ttype == TYPE_void) {
@@ -1236,7 +1238,7 @@ void_inplace(BAT *b, oid id, const void
GDKerror("void_inplace: id out of range\n");
return GDK_FAIL;
}
- if ((b->tkey & BOUND2BTRUE) && BUNfnd(b, val) != BUN_NONE)
+ if (b->tunique && BUNfnd(b, val) != BUN_NONE)
return GDK_SUCCEED;
if (b->ttype == TYPE_void)
return GDK_SUCCEED;
@@ -1433,11 +1435,9 @@ BATkey(BAT *b, int flag)
{
BATcheck(b, "BATkey", GDK_FAIL);
assert(b->batCacheid > 0);
+ assert(flag == 0 || flag == 1);
+ assert(!b->tunique || flag);
if (b->ttype == TYPE_void) {
- if (b->tseqbase == oid_nil && flag == BOUND2BTRUE) {
- GDKerror("BATkey: nil-column cannot be kept unique.\n");
- return GDK_FAIL;
- }
if (b->tseqbase != oid_nil && flag == FALSE) {
GDKerror("BATkey: dense column must be unique.\n");
return GDK_FAIL;
@@ -1447,11 +1447,9 @@ BATkey(BAT *b, int flag)
return GDK_FAIL;
}
}
- if (flag)
- flag |= (1 | b->tkey);
- if (b->tkey != flag)
+ if (b->tkey != (flag != 0))
b->batDirtydesc = TRUE;
- b->tkey = flag;
+ b->tkey = flag != 0;
if (!flag)
b->tdense = 0;
else
@@ -2005,10 +2003,10 @@ BATmode(BAT *b, int mode)
* revsorted The column is reversely sorted (descending). If
* also sorted, then all values are equal.
*
- * The "key" property consists of two bits. The lower bit, when set,
- * indicates that all values in the column are distinct. The upper
- * bit, when set, indicates that all values must be distinct
- * (BOUND2BTRUE).
+ * In addition there is a property "unique" that, when set, indicates
+ * that values must be kept unique (and hence that the "key" property
+ * must be set). This property is only used when changing (adding,
+ * replacing) values.
*
* Note that the functions BATtseqbase and BATkey also set more
* properties than you might suspect. When setting properties on a
@@ -2047,8 +2045,7 @@ BATassertProps(BAT *b)
assert(b->ttype >= TYPE_void);
assert(b->ttype < GDKatomcnt);
assert(b->ttype != TYPE_bat);
- /* if BOUND2BTRUE is set, then so must the low order bit */
- assert(!(b->tkey & BOUND2BTRUE) || (b->tkey & 1)); /* tkey != 2 */
+ assert(!b->tunique || b->tkey); /* if unique, then key */
assert(isVIEW(b) ||
b->ttype == TYPE_void ||
BBPfarms[b->theap.farmid].roles & (1 << b->batRole));
diff --git a/gdk/gdk_batop.c b/gdk/gdk_batop.c
--- a/gdk/gdk_batop.c
+++ b/gdk/gdk_batop.c
@@ -367,12 +367,12 @@ BATappend(BAT *b, BAT *n, bit force)
ALIGNapp(b, "BATappend", force, GDK_FAIL);
BATcompatible(b, n, GDK_FAIL, "BATappend");
- if (b->tkey & BOUND2BTRUE) {
- /* if b has the BOUND2BTRUE bit set, only insert
- * values from n that don't already occur in b, and
- * make sure we don't insert any duplicates either; we
- * do this by calculating a subset of n that complies
- * with this */
+ if (b->tunique) {
+ /* if b has the unique bit set, only insert values
+ * from n that don't already occur in b, and make sure
+ * we don't insert any duplicates either; we do this
+ * by calculating a subset of n that complies with
+ * this */
BAT *d, *u;
d = BATdiff(n, b, NULL, NULL, 1, BUN_NONE);
@@ -395,7 +395,7 @@ BATappend(BAT *b, BAT *n, bit force)
}
if (BUNlast(b) + BATcount(n) > BUN_MAX) {
- if (b->tkey & BOUND2BTRUE)
+ if (b->tunique)
BBPunfix(n->batCacheid);
GDKerror("BATappend: combined BATs too large\n");
return GDK_FAIL;
@@ -430,13 +430,13 @@ BATappend(BAT *b, BAT *n, bit force)
if (BATtdense(n) && BATcount(b) + b->tseqbase == f) {
sz += BATcount(b);
BATsetcount(b, sz);
- if (b->tkey & BOUND2BTRUE)
+ if (b->tunique)
BBPunfix(n->batCacheid);
return GDK_SUCCEED;
}
/* we need to materialize the tail */
if (BATmaterialize(b) != GDK_SUCCEED) {
- if (b->tkey & BOUND2BTRUE)
+ if (b->tunique)
BBPunfix(n->batCacheid);
return GDK_FAIL;
}
@@ -463,9 +463,9 @@ BATappend(BAT *b, BAT *n, bit force)
}
b->tdense = n->tdense;
b->tnodense = n->tnodense;
- b->tkey |= (n->tkey & TRUE);
- /* if BOUND2BTRUE, uniqueness is guaranteed above */
- if ((b->tkey & BOUND2BTRUE) == 0) {
+ /* if tunique, uniqueness is guaranteed above */
+ b->tkey = n->tkey | b->tunique;
+ if (!b->tunique) {
b->tnokey[0] = n->tnokey[0];
b->tnokey[1] = n->tnokey[1];
}
@@ -488,9 +488,10 @@ BATappend(BAT *b, BAT *n, bit force)
b->trevsorted = FALSE;
b->tnorevsorted = 0;
}
- if (b->tkey == 1 &&
+ if (!b->tunique && /* uniqueness is guaranteed above */
+ b->tkey &&
(!(BATtordered(b) || BATtrevordered(b)) ||
- n->tkey == 0 || xx == 0)) {
+ !n->tkey || xx == 0)) {
BATkey(b, FALSE);
}
if (b->ttype != TYPE_void && b->tsorted && b->tdense &&
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
https://www.monetdb.org/mailman/listinfo/checkin-list
