Changeset: 3a598360f206 for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=3a598360f206
Branch: unlock
Log Message:
Merge unlock-cands branch into unlock.
diffs (truncated from 1542 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
@@ -9512,7 +9512,6 @@ stdout of test 'MAL-signatures` in direc
[ "status", "vm_maxsize", "command status.vm_maxsize(X_1:lng):void ",
"SYSsetvm_maxsize;", "" ]
[ "str", "ascii", "command str.ascii(X_1:str):int ",
"STRascii;", "" ]
[ "str", "endsWith", "command str.endsWith(X_1:str, X_2:str):bit ",
"STRSuffix;", "" ]
-[ "str", "epilogue", "command str.epilogue():void ", "STRepilogue;",
"" ]
[ "str", "insert", "command str.insert(X_1:str, X_2:int, X_3:int,
X_4:str):str ", "STRinsert;", "" ]
[ "str", "length", "command str.length(X_1:str):int ",
"STRLength;", "" ]
[ "str", "like", "command str.like(X_1:str, X_2:str):bit ",
"STRlikewrap2;", "" ]
@@ -9525,7 +9524,7 @@ stdout of test 'MAL-signatures` in direc
[ "str", "ltrim", "command str.ltrim(X_1:str):str ",
"STRLtrim;", "" ]
[ "str", "nbytes", "command str.nbytes(X_1:str):int ",
"STRBytes;", "" ]
[ "str", "prefix", "command str.prefix(X_1:str, X_2:int):str ",
"STRprefix;", "" ]
-[ "str", "prelude", "command str.prelude():void ", "STRprelude;",
"" ]
+[ "str", "prelude", "command str.prelude():void ", "strPrelude;",
"" ]
[ "str", "r_search", "command str.r_search(X_1:str, X_2:str):int ",
"STRReverseStrSearch;", "" ]
[ "str", "repeat", "command str.repeat(X_1:str, X_2:int):str ",
"STRrepeat;", "" ]
[ "str", "replace", "command str.replace(X_1:str, X_2:str, X_3:str,
X_4:str):str ", "PCREreplace_wrap;", "" ]
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
@@ -12880,7 +12880,6 @@ stdout of test 'MAL-signatures` in direc
[ "status", "vm_maxsize", "command status.vm_maxsize(X_1:lng):void ",
"SYSsetvm_maxsize;", "" ]
[ "str", "ascii", "command str.ascii(X_1:str):int ",
"STRascii;", "" ]
[ "str", "endsWith", "command str.endsWith(X_1:str, X_2:str):bit ",
"STRSuffix;", "" ]
-[ "str", "epilogue", "command str.epilogue():void ", "STRepilogue;",
"" ]
[ "str", "insert", "command str.insert(X_1:str, X_2:int, X_3:int,
X_4:str):str ", "STRinsert;", "" ]
[ "str", "length", "command str.length(X_1:str):int ",
"STRLength;", "" ]
[ "str", "like", "command str.like(X_1:str, X_2:str):bit ",
"STRlikewrap2;", "" ]
@@ -12893,7 +12892,7 @@ stdout of test 'MAL-signatures` in direc
[ "str", "ltrim", "command str.ltrim(X_1:str):str ",
"STRLtrim;", "" ]
[ "str", "nbytes", "command str.nbytes(X_1:str):int ",
"STRBytes;", "" ]
[ "str", "prefix", "command str.prefix(X_1:str, X_2:int):str ",
"STRprefix;", "" ]
-[ "str", "prelude", "command str.prelude():void ", "STRprelude;",
"" ]
+[ "str", "prelude", "command str.prelude():void ", "strPrelude;",
"" ]
[ "str", "r_search", "command str.r_search(X_1:str, X_2:str):int ",
"STRReverseStrSearch;", "" ]
[ "str", "repeat", "command str.repeat(X_1:str, X_2:int):str ",
"STRrepeat;", "" ]
[ "str", "replace", "command str.replace(X_1:str, X_2:str, X_3:str,
X_4:str):str ", "PCREreplace_wrap;", "" ]
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
@@ -161,6 +161,7 @@ gdk_return BATjoin(BAT **r1p, BAT **r2p,
gdk_return BATkey(BAT *b, bool onoff);
bool BATkeyed(BAT *b);
gdk_return BATleftjoin(BAT **r1p, BAT **r2p, BAT *l, BAT *r, BAT *sl, BAT *sr,
bool nil_matches, BUN estimate) __attribute__((__warn_unused_result__));
+gdk_return BATmaskedcands(BAT *dense_cands, BAT *masked, bool selected);
void *BATmax(BAT *b, void *aggr);
void *BATmax_skipnil(BAT *b, void *aggr, bit skipnil);
BAT *BATmergecand(BAT *a, BAT *b);
@@ -168,7 +169,7 @@ void *BATmin(BAT *b, void *aggr);
void *BATmin_skipnil(BAT *b, void *aggr, bit skipnil);
gdk_return BATmode(BAT *b, bool transient);
void BATmsync(BAT *b);
-gdk_return BATnegcands(BAT *cands, BAT *odels);
+gdk_return BATnegcands(BAT *dense_cands, BAT *odels);
BAT *BATnil_grp(BAT *l, BAT *g, BAT *e, BAT *s);
bool BATordered(BAT *b);
bool BATordered_rev(BAT *b);
diff --git a/gdk/gdk.h b/gdk/gdk.h
--- a/gdk/gdk.h
+++ b/gdk/gdk.h
@@ -991,6 +991,30 @@ typedef var_t stridx_t;
#include "gdk_atoms.h"
+
+/* candidates by design are ordered oid lists, besides native oid bats
+ * there are
+ * void bats for dense oid lists,
+ * negative oid lists
+ * masked oid lists
+ */
+
+#define CAND_NEGOID 0
+#define CAND_MSK 1
+
+typedef struct {
+ uint64_t
+ type:1,
+ mask:1;
+} ccand_t;
+
+#define CCAND(b) ((ccand_t*)b->tvheap->base)
+#define complex_cand(b) (b->ttype == TYPE_void && b->tvheap != NULL)
+#define negoid_cand(b) (b->ttype == TYPE_void && b->tvheap != NULL &&
CCAND(b)->type == CAND_NEGOID)
+#define mask_cand(b) (b->ttype == TYPE_void && b->tvheap != NULL &&
CCAND(b)->type == CAND_MSK)
+#define ccand_first(b) (b->tvheap->base+sizeof(ccand_t))
+#define ccand_free(b) (b->tvheap->free-sizeof(ccand_t))
+
/* return the oid value at BUN position p from the (v)oid bat b
* works with any TYPE_void or TYPE_oid bat */
static inline oid
@@ -1011,16 +1035,17 @@ BUNtoid(BAT *b, BUN p)
if (b->ttype == TYPE_oid || b->tvheap == NULL) {
return o;
}
+ assert(!mask_cand(b));
/* exceptions only allowed on transient BATs */
assert(b->batRole == TRANSIENT);
/* make sure exception area is a reasonable size */
- assert(b->tvheap->free % SIZEOF_OID == 0);
- BUN nexc = (BUN) (b->tvheap->free / SIZEOF_OID);
+ assert(ccand_free(b) % SIZEOF_OID == 0);
+ BUN nexc = (BUN) (ccand_free(b) / SIZEOF_OID);
if (nexc == 0) {
/* no exceptions (why the vheap?) */
return o;
}
- const oid *exc = (oid *) b->tvheap->base;
+ const oid *exc = (oid *) ccand_first(b);
if (o < exc[0])
return o;
if (o + nexc > exc[nexc - 1])
diff --git a/gdk/gdk_aggr.c b/gdk/gdk_aggr.c
--- a/gdk/gdk_aggr.c
+++ b/gdk/gdk_aggr.c
@@ -982,7 +982,7 @@ mskCountOnes(BAT *b, struct canditer *ci
BUN cnt = 0, ncand = ci->ncand;
if (ci->tpe == cand_dense && BATcount(b)) {
- const uint32_t *restrict src = Tloc(b, (ci->seq - b->hseqbase)
/ 32);
+ const uint32_t *restrict src = mask_cand(b)?ccand_first(b)+
(ci->seq - b->hseqbase) / 32:Tloc(b, (ci->seq - b->hseqbase) / 32);
int bits = (ci->seq - b->hseqbase) % 32;
if (bits + ncand <= 32) {
if (ncand == 32)
@@ -1028,7 +1028,7 @@ BATsum(void *res, int tp, BAT *b, BAT *s
GDKerror("%s\n", err);
return GDK_FAIL;
}
- if (ATOMstorage(b->ttype) == TYPE_msk) {
+ if (ATOMstorage(b->ttype) == TYPE_msk || mask_cand(b)) {
ncand = mskCountOnes(b, &ci);
switch (tp) {
case TYPE_bte:
diff --git a/gdk/gdk_bat.c b/gdk/gdk_bat.c
--- a/gdk/gdk_bat.c
+++ b/gdk/gdk_bat.c
@@ -134,6 +134,8 @@ BATcreatedesc(oid hseq, int tt, bool hea
sizeof(bn->tvheap->filename),
nme, ".theap", NULL);
}
+ } else {
+ assert(bn->theap == NULL);
}
char name[MT_NAME_LEN];
snprintf(name, sizeof(name), "heaplock%d", bn->batCacheid); /* fits */
@@ -1596,7 +1598,7 @@ BUNfnd(BAT *b, const void *v)
BATcheck(b, BUN_NONE);
if (!v)
return r;
- if (b->ttype == TYPE_void && b->tvheap != NULL) {
+ if (complex_cand(b)) {
struct canditer ci;
canditer_init(&ci, NULL, b);
return canditer_search(&ci, * (const oid *) v, false);
@@ -2350,8 +2352,8 @@ BATassertProps(BAT *b)
/* candidate list with exceptions */
assert(b->batRole == TRANSIENT);
assert(b->tvheap->free <= b->tvheap->size);
- assert(b->tvheap->free % SIZEOF_OID == 0);
- if (b->tvheap->free > 0) {
+ assert((negoid_cand(b) && ccand_free(b) %
SIZEOF_OID == 0) || mask_cand(b));
+ if (negoid_cand(b) && ccand_free(b) > 0) {
const oid *oids = (const oid *)
b->tvheap->base;
q = b->tvheap->free / SIZEOF_OID;
assert(oids != NULL);
diff --git a/gdk/gdk_batop.c b/gdk/gdk_batop.c
--- a/gdk/gdk_batop.c
+++ b/gdk/gdk_batop.c
@@ -1458,7 +1458,7 @@ BATslice(BAT *b, BUN l, BUN h)
goto doreturn;
}
- if (b->ttype == TYPE_void && b->tvheap != NULL) {
+ if (complex_cand(b)) {
/* slicing a candidate list with exceptions */
struct canditer ci;
canditer_init(&ci, NULL, b);
diff --git a/gdk/gdk_cand.c b/gdk/gdk_cand.c
--- a/gdk/gdk_cand.c
+++ b/gdk/gdk_cand.c
@@ -18,6 +18,8 @@ BATiscand(BAT *b)
return true;
if (ATOMtype(b->ttype) != TYPE_oid)
return false;
+ if (complex_cand(b))
+ return true;
if (b->ttype == TYPE_void && is_oid_nil(b->tseqbase))
return false;
return BATtordered(b) && BATtkey(b);
@@ -466,14 +468,14 @@ canditer_init(struct canditer *ci, BAT *
.s = s,
};
- if (s->ttype == TYPE_void) {
+ if (s->ttype == TYPE_void && !mask_cand(s)) {
assert(!is_oid_nil(ci->seq));
if (s->tvheap) {
- assert(s->tvheap->free % SIZEOF_OID == 0);
- ci->nvals = s->tvheap->free / SIZEOF_OID;
+ assert(ccand_free(s) % SIZEOF_OID == 0);
+ ci->nvals = ccand_free(s) / SIZEOF_OID;
if (ci->nvals > 0) {
ci->tpe = cand_except;
- ci->oids = (const oid *) s->tvheap->base;
+ ci->oids = (const oid *) ccand_first(s);
} else {
/* why the vheap? */
ci->tpe = cand_dense;
@@ -481,7 +483,13 @@ canditer_init(struct canditer *ci, BAT *
} else {
ci->tpe = cand_dense;
}
+ } else if (s->ttype == TYPE_void && mask_cand(s)) {
+ ci->tpe = cand_mask;
+ ci->mask = (const uint32_t *) ccand_first(s);
+ ci->seq = s->hseqbase;
+ ci->nvals = (cnt + 31U) / 32U;
} else if (s->ttype == TYPE_msk) {
+ assert(0);
ci->tpe = cand_mask;
ci->mask = (const uint32_t *) s->theap->base;
ci->seq = s->hseqbase;
@@ -680,7 +688,7 @@ canditer_init(struct canditer *ci, BAT *
assert(i >= 0); /* there should be a set bit */
ci->firstbit += i;
cnt -= i;
- ci->mskoff = s->hseqbase + (ci->mask - (const uint32_t *)
s->theap->base) * 32U;
+ ci->mskoff = s->hseqbase + (ci->mask - (const uint32_t *)
(mask_cand(s)?ccand_first(s):s->theap->base)) * 32U;
ci->seq = ci->mskoff + ci->firstbit;
ci->hseq = ci->seq;
ci->nextbit = ci->firstbit;
@@ -1254,6 +1262,7 @@ BATnegcands(BAT *dense_cands, BAT *odels
const char *nme;
Heap *dels;
BUN lo, hi;
+ ccand_t *c;
assert(BATtdense(dense_cands));
assert(dense_cands->ttype == TYPE_void);
@@ -1276,17 +1285,21 @@ BATnegcands(BAT *dense_cands, BAT *odels
strconcat_len(dels->filename, sizeof(dels->filename),
nme, ".theap", NULL);
- if (HEAPalloc(dels, hi - lo, sizeof(oid), 0) != GDK_SUCCEED) {
+ if (HEAPalloc(dels, hi - lo + (sizeof(ccand_t)/sizeof(oid)),
sizeof(oid), 0) != GDK_SUCCEED) {
GDKfree(dels);
return GDK_FAIL;
}
+ c = (ccand_t*)dels->base;
+ c->type = CAND_NEGOID;
dels->parentid = dense_cands->batCacheid;
- dels->free = sizeof(oid) * (hi - lo);
+ dels->free = sizeof(ccand_t) + sizeof(oid) * (hi - lo);
if (odels->ttype == TYPE_void) {
+ oid *r = (oid*)(dels->base + sizeof(ccand_t));
for (BUN x = lo; x < hi; x++)
- ((oid *) dels->base)[x - lo] = x + odels->tseqbase;
+ r[x - lo] = x + odels->tseqbase;
} else {
- memcpy(dels->base, Tloc(odels, lo), dels->free);
+ oid *r = (oid*)(dels->base + sizeof(ccand_t));
+ memcpy(r, Tloc(odels, lo), dels->free);
}
dense_cands->batDirtydesc = true;
dense_cands->tvheap = dels;
@@ -1298,6 +1311,56 @@ BATnegcands(BAT *dense_cands, BAT *odels
return GDK_SUCCEED;
}
+gdk_return
+BATmaskedcands(BAT *dense_cands, BAT *masked, bool selected)
+{
+ const char *nme;
+ Heap *msks;
+ ccand_t *c;
+
+ assert(BATtdense(dense_cands));
+ assert(dense_cands->ttype == TYPE_void);
+ assert(dense_cands->batRole == TRANSIENT);
+ assert(masked->ttype == TYPE_msk);
+
+ if (BATcount(masked) == 0)
+ return GDK_SUCCEED;
+
+ nme = BBP_physical(dense_cands->batCacheid);
+ if ((msks = (Heap*)GDKzalloc(sizeof(Heap))) == NULL ||
+ (msks->farmid = BBPselectfarm(dense_cands->batRole,
dense_cands->ttype, varheap)) < 0){
+ GDKfree(msks);
+ return GDK_FAIL;
+ }
+ strconcat_len(msks->filename, sizeof(msks->filename),
+ nme, ".theap", NULL);
+
+ if (HEAPalloc(msks, (BATcount(masked)+31)/32 +
(sizeof(ccand_t)/sizeof(int)), sizeof(int), 0) != GDK_SUCCEED) {
+ GDKfree(msks);
+ return GDK_FAIL;
+ }
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
https://www.monetdb.org/mailman/listinfo/checkin-list
