Changeset: 01d3e88859f6 for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB/rev/01d3e88859f6
Modified Files:
sql/backends/monet5/sql.c
sql/storage/bat/bat_storage.c
Branch: default
Log Message:
Merged with Jan2022
diffs (truncated from 865 to 300 lines):
diff --git a/monetdb5/optimizer/opt_fastpath.c
b/monetdb5/optimizer/opt_fastpath.c
--- a/monetdb5/optimizer/opt_fastpath.c
+++ b/monetdb5/optimizer/opt_fastpath.c
@@ -17,6 +17,8 @@
#include "opt_costModel.h"
#include "opt_dataflow.h"
#include "opt_deadcode.h"
+#include "opt_dict.h"
+#include "opt_for.h"
#include "opt_emptybind.h"
#include "opt_evaluate.h"
#include "opt_garbageCollector.h"
@@ -76,7 +78,10 @@ OPTminimalfastImplementation(Client cntx
optcall(true, OPTinlineImplementation);
optcall(true, OPTremapImplementation);
optcall(bincopy, OPTbincopyfromImplementation);
+ optcall(true, OPTemptybindImplementation);
optcall(true, OPTdeadcodeImplementation);
+ optcall(true, OPTforImplementation);
+ optcall(true, OPTdictImplementation);
optcall(multiplex, OPTmultiplexImplementation);
optcall(generator, OPTgeneratorImplementation);
optcall(malProfileMode, OPTprofilerImplementation);
@@ -117,6 +122,8 @@ OPTdefaultfastImplementation(Client cntx
optcall(true, OPTdeadcodeImplementation);
optcall(true, OPTpushselectImplementation);
optcall(true, OPTaliasesImplementation);
+ optcall(true, OPTforImplementation);
+ optcall(true, OPTdictImplementation);
optcall(true, OPTmitosisImplementation);
optcall(true, OPTmergetableImplementation);
optcall(bincopy, OPTbincopyfromImplementation);
diff --git a/sql/backends/monet5/dict.c b/sql/backends/monet5/dict.c
--- a/sql/backends/monet5/dict.c
+++ b/sql/backends/monet5/dict.c
@@ -233,6 +233,8 @@ DICTcompress_col(Client cntxt, MalBlkPtr
if (!isTable(t))
throw(SQL, "dict.compress", SQLSTATE(42000) "%s '%s' is not
persistent",
TABLE_TYPE_DESCRIPTION(t->type, t->properties),
t->base.name);
+ if (t->system)
+ throw(SQL, "dict.compress", SQLSTATE(42000) "columns from
system tables cannot be compressed");
sql_column *c = find_sql_column(t, cname);
if (!c)
throw(SQL, "dict.compress", SQLSTATE(3F000) "column '%s.%s.%s'
unknown", sname, tname, cname);
@@ -497,7 +499,7 @@ DICTrenumber_intern( BAT *o, BAT *lc, BA
}
/* dense or cheap dense check */
if (!BATtdense(lc) && !(lc->tsorted && lc->tkey && BATcount(lc) ==
offcnt && *(oid*)Tloc(lc, offcnt-1) == offcnt-1)) {
- BAT *nrc = COLnew(0, rc->ttype, offcnt, TRANSIENT);
+ BAT *nrc = COLnew(0, ATOMtype(rc->ttype), offcnt, TRANSIENT);
if (!nrc) {
if (lc != olc)
bat_destroy(lc);
@@ -507,14 +509,26 @@ DICTrenumber_intern( BAT *o, BAT *lc, BA
}
/* create map with holes filled in */
- oid *op = Tloc(nrc, 0);
- oid *ip = Tloc(rc, 0);
+ oid *restrict op = Tloc(nrc, 0);
unsigned char *lp = Tloc(lc, 0);
- for(BUN i = 0, j = 0; i<offcnt; i++) {
- if (lp[j] > i) {
- op[i] = offcnt;
- } else {
- op[i] = ip[j++];
+ if (BATtvoid(rc)) {
+ oid seq = rc->tseqbase, j = 0;
+ for(BUN i = 0; i<offcnt; i++) {
+ if (lp[j] > i) {
+ op[i] = offcnt;
+ } else {
+ op[i] = seq + j;
+ j++;
+ }
+ }
+ } else {
+ oid *ip = Tloc(rc, 0);
+ for(BUN i = 0, j = 0; i<offcnt; i++) {
+ if (lp[j] > i) {
+ op[i] = offcnt;
+ } else {
+ op[i] = ip[j++];
+ }
}
}
BATsetcount(nrc, offcnt);
@@ -685,7 +699,7 @@ DICTthetaselect(Client cntxt, MalBlkPtr
BUN max_cnt = lv->ttype == TYPE_bte?256:(64*1024);
if ((lv->tkey && (op[0] == '=' || op[0] == '!')) || ((op[0] == '<' ||
op[0] == '>') && lv->tsorted && BATcount(lv) < (max_cnt/2))) {
BUN p = BUN_NONE;
- if (ATOMvarsized(lv->ttype))
+ if (ATOMextern(lv->ttype))
v = *(ptr*)v;
if (op[0] == '=' || op[0] == '!') {
p = BUNfnd(lv, v);
@@ -707,7 +721,7 @@ DICTthetaselect(Client cntxt, MalBlkPtr
bn = BATdense(0, 0, 0);
}
} else { /* select + intersect */
- if (ATOMvarsized(lv->ttype))
+ if (ATOMextern(lv->ttype))
v = *(ptr*)v;
bn = BATthetaselect(lv, NULL, v, op);
/* call dict convert */
@@ -767,7 +781,7 @@ DICTselect(Client cntxt, MalBlkPtr mb, M
throw(SQL, "dict.select", SQLSTATE(HY013) MAL_MALLOC_FAIL);
}
- if (ATOMvarsized(lv->ttype)) {
+ if (ATOMextern(lv->ttype)) {
l = *(ptr*)l;
h = *(ptr*)h;
}
diff --git a/sql/backends/monet5/for.c b/sql/backends/monet5/for.c
--- a/sql/backends/monet5/for.c
+++ b/sql/backends/monet5/for.c
@@ -29,6 +29,59 @@ BATnegateprops(BAT *b)
b->tnokey[1] = 0;
}
+BAT *
+FORdecompress_(BAT *o, lng minval, int type, role_t role)
+{
+ BAT *b = COLnew(o->hseqbase, type, BATcount(o), role);
+
+ if (!b)
+ return NULL;
+ BUN cnt = BATcount(o);
+#ifdef HAVE_HGE
+ if (type == TYPE_hge) {
+ if (o->ttype == TYPE_bte) {
+ hge *ov = Tloc(b, 0);
+ bte *iv = Tloc(o, 0);
+ for(BUN i = 0; i<cnt; i++)
+ ov[i] = minval + iv[i];
+ } else {
+ hge *ov = Tloc(b, 0);
+ sht *iv = Tloc(o, 0);
+ for(BUN i = 0; i<cnt; i++)
+ ov[i] = minval + iv[i];
+ }
+ } else
+#endif
+ if (type == TYPE_lng) {
+ if (o->ttype == TYPE_bte) {
+ lng *ov = Tloc(b, 0);
+ bte *iv = Tloc(o, 0);
+ for(BUN i = 0; i<cnt; i++)
+ ov[i] = minval + iv[i];
+ } else {
+ lng *ov = Tloc(b, 0);
+ sht *iv = Tloc(o, 0);
+ for(BUN i = 0; i<cnt; i++)
+ ov[i] = minval + iv[i];
+ }
+ } else if (type == TYPE_int) {
+ if (o->ttype == TYPE_bte) {
+ int *ov = Tloc(b, 0);
+ bte *iv = Tloc(o, 0);
+ for(BUN i = 0; i<cnt; i++)
+ ov[i] = (int) (minval + iv[i]);
+ } else {
+ int *ov = Tloc(b, 0);
+ sht *iv = Tloc(o, 0);
+ for(BUN i = 0; i<cnt; i++)
+ ov[i] = (int) (minval + iv[i]);
+ }
+ }
+ BATsetcount(b, cnt);
+ BATnegateprops(b);
+ return b;
+}
+
str
FORdecompress(Client cntxt, MalBlkPtr mb, MalStkPtr stk, InstrPtr pci)
{
@@ -54,37 +107,46 @@ FORdecompress(Client cntxt, MalBlkPtr mb
throw(SQL, "for.decompress", SQLSTATE(3F000) "for decompress:
invalid type");
}
- BUN cnt = BATcount(o);
- if (tt == TYPE_lng) {
- lng minval = *getArgReference_lng(stk, pci, 2);
+ lng minval = *getArgReference_lng(stk, pci, 2);
- b = COLnew(o->hseqbase, TYPE_lng, cnt, PERSISTENT);
- if (!b) {
- bat_destroy(o);
- throw(SQL, "for.decompress", SQLSTATE(HY013)
MAL_MALLOC_FAIL);
- }
- if (o->ttype == TYPE_bte) {
- lng *ov = Tloc(b, 0);
- bte *iv = Tloc(o, 0);
- for(BUN i = 0; i<cnt; i++)
- ov[i] = minval + iv[i];
- } else {
- lng *ov = Tloc(b, 0);
- sht *iv = Tloc(o, 0);
- for(BUN i = 0; i<cnt; i++)
- ov[i] = minval + iv[i];
- }
- } else {
+ b = FORdecompress_(o, minval, tt, TRANSIENT);
+ if (!b) {
bat_destroy(o);
- throw(SQL, "for.decompress", SQLSTATE(3F000) "offset type %s
not yet implemented", ATOMname(tt));
+ throw(SQL, "for.decompress", SQLSTATE(HY013) MAL_MALLOC_FAIL);
}
bat_destroy(o);
- BATsetcount(b, cnt);
- BATnegateprops(b);
BBPkeepref(*r = b->batCacheid);
return MAL_SUCCEED;
}
+static BAT *
+FORcompress_(BAT *b, lng min_val, lng max_val, role_t role)
+{
+ BAT *o;
+ BUN cnt = BATcount(b);
+
+ if ((max_val-min_val) < GDK_bte_max/2) {
+ o = COLnew(b->hseqbase, TYPE_bte, cnt, role);
+ if (!o)
+ return NULL;
+ bte *ov = Tloc(o, 0);
+ lng *iv = Tloc(b, 0);
+ for(BUN i = 0; i<cnt; i++)
+ ov[i] = (bte)(iv[i] - min_val);
+ } else {
+ o = COLnew(b->hseqbase, TYPE_sht, cnt, role);
+ if (!o)
+ return NULL;
+ sht *ov = Tloc(o, 0);
+ lng *iv = Tloc(b, 0);
+ for(BUN i = 0; i<cnt; i++)
+ ov[i] = (sht)(iv[i] - min_val);
+ }
+ BATsetcount(o, cnt);
+ BATnegateprops(o);
+ return o;
+}
+
static str
FORcompress_intern(char **comp_min_val, BAT **r, BAT *b)
{
@@ -122,23 +184,9 @@ FORcompress_intern(char **comp_min_val,
throw(SQL, "for.compress", SQLSTATE(3F000) "for
compress: for 'for' compression column's cannot have NULL's");
if ((max_val-min_val) > GDK_sht_max)
throw(SQL, "for.compress", SQLSTATE(3F000) "for
compress: too large value spread for 'for' compression");
- if ((max_val-min_val) < GDK_bte_max/2) {
- o = COLnew(b->hseqbase, TYPE_bte, cnt, PERSISTENT);
- if (!o)
- throw(SQL, "for.compress", SQLSTATE(HY013)
MAL_MALLOC_FAIL);
- bte *ov = Tloc(o, 0);
- lng *iv = Tloc(b, 0);
- for(BUN i = 0; i<cnt; i++)
- ov[i] = (bte)(iv[i] - min_val);
- } else {
- o = COLnew(b->hseqbase, TYPE_sht, cnt, PERSISTENT);
- if (!o)
- throw(SQL, "for.compress", SQLSTATE(HY013)
MAL_MALLOC_FAIL);
- sht *ov = Tloc(o, 0);
- lng *iv = Tloc(b, 0);
- for(BUN i = 0; i<cnt; i++)
- ov[i] = (sht)(iv[i] - min_val);
- }
+ o = FORcompress_(b, min_val, max_val, PERSISTENT);
+ if (!o)
+ throw(SQL, "for.compress", SQLSTATE(HY013)
MAL_MALLOC_FAIL);
snprintf(buf, 64, "FOR-" LLFMT, min_val);
} else {
GDKfree(mn);
@@ -149,8 +197,6 @@ FORcompress_intern(char **comp_min_val,
bat_destroy(o);
throw(SQL, "for.compress", SQLSTATE(HY013) MAL_MALLOC_FAIL);
}
- BATsetcount(o, cnt);
- BATnegateprops(o);
*r = o;
return NULL;
}
@@ -188,6 +234,8 @@ FORcompress_col(Client cntxt, MalBlkPtr
if (!isTable(t))
throw(SQL, "for.compress", SQLSTATE(42000) "%s '%s' is not
persistent",
TABLE_TYPE_DESCRIPTION(t->type, t->properties),
t->base.name);
+ if (t->system)
+ throw(SQL, "for.compress", SQLSTATE(42000) "columns from system
tables cannot be compressed");
sql_column *c = find_sql_column(t, cname);
if (!c)
throw(SQL, "for.compress", SQLSTATE(3F000) "column '%s.%s.%s'
unknown", sname, tname, cname);
@@ -236,3 +284,72 @@ FORcompress_col(Client cntxt, MalBlkPtr
}
return msg;
}
+
+int
+FORprepare4append(BAT **noffsets, BAT *b, lng minval, int tt)
+{
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
https://www.monetdb.org/mailman/listinfo/checkin-list
