Changeset: b92e6d5af624 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=b92e6d5af624
Modified Files:
monetdb5/modules/mal/cluster.c
Branch: default
Log Message:
Merge with Oct2014 branch.
diffs (truncated from 2495 to 300 lines):
diff --git a/monetdb5/modules/mal/cluster.c b/monetdb5/modules/mal/cluster.c
--- a/monetdb5/modules/mal/cluster.c
+++ b/monetdb5/modules/mal/cluster.c
@@ -64,161 +64,35 @@
#include <mal_exception.h>
#include "algebra.h"
-static void
-CLUSTER_key_bte( BAT *map, BAT *b)
-{
- bte *bt, *be;
- oid *o;
-
- assert(BUNfirst(map) == 0);
- assert(BUNfirst(b) == 0);
- o = (oid*)Tloc(map, 0);
- bt = (bte*)Tloc(b, 0);
- be = bt + BATcount(b);
- for ( ; bt < be; bt++){
- BUN h = hash_bte(b->T->hash,bt);
- *o++= h;
- }
+#define CLUSTERKEY(TYPE) \
+static void
\
+CLUSTER_key_##TYPE( BAT *map, BAT *b) \
+{
\
+ TYPE *bt, *be;
\
+ oid *o;
\
+
\
+ assert(BUNfirst(map) == 0); \
+ assert(BUNfirst(b) == 0); \
+ o = (oid*)Tloc(map, 0); \
+ bt = (TYPE*)Tloc(b, 0); \
+ be = bt + BATcount(b); \
+ for ( ; bt < be; bt++){ \
+ BUN h = hash_##TYPE(b->T->hash,bt); \
+ *o++= h;
\
+ }
\
}
-static void
-CLUSTER_key_sht( BAT *map, BAT *b)
-{
- sht *bt, *be;
- oid *o;
-
- assert(BUNfirst(map) == 0);
- assert(BUNfirst(b) == 0);
- o = (oid*)Tloc(map, 0);
- bt = (sht*)Tloc(b, 0);
- be = bt + BATcount(b);
- for ( ; bt < be; bt++){
- BUN h = hash_sht(b->T->hash,bt);
- *o++= h;
- }
-}
-
-static void
-CLUSTER_key_oid( BAT *map, BAT *b)
-{
- oid *bt, *be;
- oid *o;
-
- assert(BUNfirst(map) == 0);
- assert(BUNfirst(b) == 0);
- o = (oid*)Tloc(map, 0);
- bt = (oid*)Tloc(b, 0);
- be = bt + BATcount(b);
- for ( ; bt < be; bt++){
- BUN h = hash_oid(b->T->hash,bt);
- *o++= h;
- }
-}
-
-static void
-CLUSTER_key_wrd( BAT *map, BAT *b)
-{
- wrd *bt, *be;
- oid *o;
-
- assert(BUNfirst(map) == 0);
- assert(BUNfirst(b) == 0);
- o = (oid*)Tloc(map, 0);
- bt = (wrd*)Tloc(b, 0);
- be = bt + BATcount(b);
- for ( ; bt < be; bt++){
- BUN h = hash_wrd(b->T->hash,bt);
- *o++= h;
- }
-}
-
-static void
-CLUSTER_key_int( BAT *map, BAT *b)
-{
- int *bt, *be;
- oid *o;
-
- assert(BUNfirst(map) == 0);
- assert(BUNfirst(b) == 0);
- o = (oid*)Tloc(map, 0);
- bt = (int*)Tloc(b, 0);
- be = bt + BATcount(b);
- for ( ; bt < be; bt++){
- BUN h = hash_int(b->T->hash,bt);
- *o++= h;
- }
-}
-
-static void
-CLUSTER_key_lng( BAT *map, BAT *b)
-{
- lng *bt, *be;
- oid *o;
-
- assert(BUNfirst(map) == 0);
- assert(BUNfirst(b) == 0);
- o = (oid*)Tloc(map, 0);
- bt = (lng*)Tloc(b, 0);
- be = bt + BATcount(b);
- for ( ; bt < be; bt++){
- BUN h = hash_lng(b->T->hash,bt);
- *o++= h;
- }
-}
-
+CLUSTERKEY(bte)
+CLUSTERKEY(sht)
+CLUSTERKEY(int)
+CLUSTERKEY(wrd)
+CLUSTERKEY(lng)
#ifdef HAVE_HGE
-static void
-CLUSTER_key_hge( BAT *map, BAT *b)
-{
- hge *bt, *be;
- oid *o;
-
- assert(BUNfirst(map) == 0);
- assert(BUNfirst(b) == 0);
- o = (oid*)Tloc(map, 0);
- bt = (hge*)Tloc(b, 0);
- be = bt + BATcount(b);
- for ( ; bt < be; bt++){
- BUN h = hash_hge(b->T->hash,bt);
- *o++= h;
- }
-}
+CLUSTERKEY(hge)
#endif
-
-static void
-CLUSTER_key_flt( BAT *map, BAT *b)
-{
- flt *bt, *be;
- oid *o;
-
- assert(BUNfirst(map) == 0);
- assert(BUNfirst(b) == 0);
- o = (oid*)Tloc(map, 0);
- bt = (flt*)Tloc(b, 0);
- be = bt + BATcount(b);
- for ( ; bt < be; bt++){
- BUN h = hash_flt(b->T->hash,bt);
- *o++= h;
- }
-}
-
-static void
-CLUSTER_key_dbl( BAT *map, BAT *b)
-{
- dbl *bt, *be;
- oid *o;
-
- assert(BUNfirst(map) == 0);
- assert(BUNfirst(b) == 0);
- o = (oid*)Tloc(map, 0);
- bt = (dbl*)Tloc(b, 0);
- be = bt + BATcount(b);
- for ( ; bt < be; bt++){
- BUN h = hash_dbl(b->T->hash,bt);
- *o++= h;
- }
-}
-
+CLUSTERKEY(oid)
+CLUSTERKEY(flt)
+CLUSTERKEY(dbl)
static void
CLUSTER_key_str( BAT *map, BAT *b)
@@ -265,12 +139,12 @@ CLUSTER_key_str( BAT *map, BAT *b)
break;
}
}
-static str
+static str
CLUSTER_column_any(BAT *nb, BAT *b, BAT *cmap)
{
oid *ct, *ce, o = 0;
BATiter bi= bat_iterator(b);
-
+
ct = (oid *)Tloc(cmap, 0);
ce = ct + BATcount(cmap);
nb->H->heap.dirty = nb->T->heap.dirty= TRUE;
@@ -284,7 +158,7 @@ CLUSTER_column_any(BAT *nb, BAT *b, BAT
}
BATsetcount(nb, BATcount(b));
BATderiveProps(nb, 0);
- if (!(nb->batDirty&2))
+ if (!(nb->batDirty&2))
BATsetaccess(nb, BAT_READ);
return MAL_SUCCEED;
}
@@ -326,7 +200,7 @@ CLUSTER_key( bat *M, const bat *B){
case TYPE_str: CLUSTER_key_str(map,b); break;
default:
throw(MAL, "cluster.key", MAL_MALLOC_FAIL);
-
+
}
BATsave(map); /* dump dirty pages from memory */
BBPunfix(*B);
@@ -347,7 +221,7 @@ typedef struct{
BUN base,limit,nxt;
} Basket;
-str
+str
CLUSTER_map(bat *RB, const bat *B)
{
BUN rng,bsize, bnr=0, h, N= 2; /* number of buckets */
@@ -378,7 +252,7 @@ CLUSTER_map(bat *RB, const bat *B)
map->H->nonil = b->H->nonil;
map->T->nonil = TRUE;
BATmax(b, (ptr) &rng); /* get the maximum hash key , could use mask !*/
- rng++;
+ rng++;
/*
* The key challenge is to determine the number of clusters.
* A large number of clusters benefits subsequent performance,
@@ -397,7 +271,7 @@ CLUSTER_map(bat *RB, const bat *B)
bsize= (rng+N-1) / N;
#ifdef _CLUSTER_DEBUG
N=2; /* for debugging only */
- mnstr_printf(GDKout,"bucket pages %d size %d max %d N %d\n",
+ mnstr_printf(GDKout,"bucket pages %d size %d max %d N %d\n",
(int)MT_npages(), (int)bsize, (int)rng, (int)N);
#endif
basket = (Basket*) GDKzalloc((N+1) * sizeof(Basket));
@@ -459,7 +333,7 @@ CLUSTER_map(bat *RB, const bat *B)
* We change the BAT in place using a temporary copy
* to guide the move.
*/
-str
+str
CLUSTER_apply(bat *bid, BAT *b, BAT *cmap)
{
BAT *nb;
@@ -498,7 +372,7 @@ CLUSTER_apply(bat *bid, BAT *b, BAT *cma
return MAL_SUCCEED;
}
-str
+str
CLUSTER_column( Client cntxt, MalBlkPtr mb, MalStkPtr stk, InstrPtr pci)
{
bat *res =getArgReference_bat(stk, pci, 0);
@@ -522,7 +396,7 @@ CLUSTER_column( Client cntxt, MalBlkPtr
return msg;
}
-str
+str
CLUSTER_table( Client cntxt, MalBlkPtr mb, MalStkPtr stk, InstrPtr pci)
{
BAT *map,*b;
@@ -561,1854 +435,294 @@ CLUSTER_table( Client cntxt, MalBlkPtr m
#include "cluster.h"
#include <mal_exception.h>
-str
-CLS_create_bte( bat *rpsum, bat *rcmap, bat *B, int *Bits, int *offset)
-{
- BAT *psum, *cmap, *b;
- int i, mask = 0, off = *offset;
- int bits = *Bits;
- bte *bt, *be;
- wrd *cnt, *pos, sum, *m;
-
- if (off < 0)
- off = 0;
- if (bits >= (int)sizeof(int)*8 || bits < 0)
- throw(MAL, "cluster.new", TOO_MANY_BITS);
-
- if ((bits) != 0)
- bits--;
- mask = (1<<bits) - 1;
- if ((b = BATdescriptor(*B)) == NULL)
- throw(MAL, "cluster.new", INTERNAL_BAT_ACCESS);
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
https://www.monetdb.org/mailman/listinfo/checkin-list
