Changeset: d638889e4eea for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=d638889e4eea
Modified Files:
monetdb5/optimizer/opt_generator.c
sql/backends/monet5/generator/generator.c
Branch: default
Log Message:
New optimizer for generate series
It now also recognizes simple casting over series,
not materializing them, but introducing a new series
with the casted bounds.
Also fixed growing rangejoin activities.
diffs (truncated from 338 to 300 lines):
diff --git a/monetdb5/optimizer/opt_generator.c
b/monetdb5/optimizer/opt_generator.c
--- a/monetdb5/optimizer/opt_generator.c
+++ b/monetdb5/optimizer/opt_generator.c
@@ -25,118 +25,142 @@
* Series generating module for integer, decimal, real, double and timestamps.
*/
+#define errorCheck(P,MOD,I) \
+setModuleId(P, generatorRef);\
+typeChecker(cntxt->fdout, cntxt->nspace, mb, P, TRUE);\
+if(P->typechk == TYPE_UNKNOWN){\
+ setModuleId(P,MOD);\
+ typeChecker(cntxt->fdout, cntxt->nspace, mb, P, TRUE);\
+ setModuleId(series[I], generatorRef);\
+ setFunctionId(series[I], seriesRef);\
+ typeChecker(cntxt->fdout, cntxt->nspace, mb, series[I], TRUE);\
+}\
+pushInstruction(mb,P);
-static int
-assignedOnce(MalBlkPtr mb, int varid)
-{
- InstrPtr p;
- int i,j, c=0;
-
- for(i = 1; i< mb->stop; i++){
- p = getInstrPtr(mb,i);
- for( j = 0; j < p->retc; j++)
- if( getArg(p,j) == varid){
- c++;
- break;
- }
- }
- return c == 1;
-}
-static int
-useCount(MalBlkPtr mb, int varid)
-{
- InstrPtr p;
- int i,j, d,c=0;
-
- for(i = 1; i< mb->stop; i++){
- p = getInstrPtr(mb,i);
- d= 0;
- for( j = p->retc; j < p->argc; j++)
- if( getArg(p,j) == varid)
- d++;
- c += d > 0;
- }
- return c;
-}
+#define casting(TPE)\
+ k= getArg(p,1);\
+ p->argc = p->retc;\
+ q= newStmt(mb,calcRef,TPE##Ref);\
+ setArgType(mb,q,0,TYPE_##TPE);\
+ pushArgument(mb,q,getArg(series[k],1));\
+ p = pushArgument(mb,p, getArg(q,0));\
+ q= newStmt(mb,calcRef,TPE##Ref);\
+ setArgType(mb,q,0,TYPE_##TPE);\
+ pushArgument(mb,q,getArg(series[k],2));\
+ p = pushArgument(mb,p, getArg(q,0));\
+ if( p->argc == 4){\
+ q= newStmt(mb,calcRef,TPE##Ref);\
+ setArgType(mb,q,0,TYPE_##TPE);\
+ pushArgument(mb,q,getArg(series[k],3));\
+ p = pushArgument(mb,p, getArg(q,0));\
+ }\
+ setModuleId(p,generatorRef);\
+ setFunctionId(p,parametersRef);\
+ setVarUDFtype(mb,getArg(p,0));\
+ series[getArg(p,0)] = p;\
+ pushInstruction(mb,p);
int
OPTgeneratorImplementation(Client cntxt, MalBlkPtr mb, MalStkPtr stk, InstrPtr
pci)
{
- InstrPtr p,q;
- int i,j,k, actions=0, used, cases, blocked;
+ InstrPtr p,q, *old, *series;
+ int i, k, limit, actions=0;
+ str m;
+ str bteRef = getName("bte",3);
+ str shtRef = getName("sht",3);
+ str intRef = getName("int",3);
+ str lngRef = getName("lng",3);
+ str fltRef = getName("flt",3);
+ str dblRef = getName("dbl",3);
(void) cntxt;
(void) stk;
(void) pci;
- for( i=1; i < mb->stop; i++){
- p = getInstrPtr(mb,i);
+ series = (InstrPtr*) GDKzalloc(sizeof(InstrPtr) * mb->vtop);
+ old= mb->stmt;
+ limit = mb->stop;
+ if ( newMalBlkStmt(mb,2 * mb->ssize) < 0) {
+ GDKfree(series);
+ return 0;
+ }
+
+ for( i=0; i < limit; i++){
+ p = old[i];
+ if ( p->token == ENDsymbol){
+ pushInstruction(mb,p);
+ break;
+ }
if ( getModuleId(p) == generatorRef && getFunctionId(p) ==
seriesRef){
- /* found a target for propagation */
- used = 0;
- if ( assignedOnce(mb, getArg(p,0)) ){
- cases = useCount(mb, getArg(p,0));
- blocked = 0;
- for( j = i+1; j< mb->stop && blocked == 0; j++){
- q = getInstrPtr(mb,j);
- if ( getModuleId(q) == algebraRef &&
getFunctionId(q) == subselectRef && getArg(q,1) == getArg(p,0)){
- setModuleId(q, generatorRef);
- typeChecker(cntxt->fdout,
cntxt->nspace, mb, q, TRUE);
- used++;
- } else
- if ( getModuleId(q) == algebraRef &&
getFunctionId(q) == thetasubselectRef && getArg(q,1) == getArg(p,0)){
- setModuleId(q, generatorRef);
- typeChecker(cntxt->fdout,
cntxt->nspace, mb, q, TRUE);
- used++;
- } else
- if ( getModuleId(q) == algebraRef &&
getFunctionId(q) == leftfetchjoinRef && getArg(q,2) == getArg(p,0)){
- // projection over a series
- setModuleId(q, generatorRef);
- typeChecker(cntxt->fdout,
cntxt->nspace, mb, q, TRUE);
- used++;
- } else
- if ( getModuleId(q) == algebraRef &&
getFunctionId(q) == joinRef && (getArg(q,2) == getArg(p,0) || getArg(q,3) ==
getArg(p,0))){
- // projection over a series
- setModuleId(q, generatorRef);
- typeChecker(cntxt->fdout,
cntxt->nspace, mb, q, TRUE);
- if(q->typechk == TYPE_UNKNOWN){
- setModuleId(q,
algebraRef);
-
typeChecker(cntxt->fdout, cntxt->nspace, mb, q, TRUE);
- } else
- used++;
- } else
- if ( getModuleId(q) == sqlRef &&
getFunctionId(q) == putName("exportValue",11) &&
isaBatType(getArgType(mb,p,0)) ){
- // interface expects scalar
type only, not expressable in MAL signature
- blocked++;
- mb->errors++;
- showException(cntxt->fdout,
MAL, "generate_series", "internal error, generate_series is a table producing
function");
- }else
- if ( getModuleId(q) == languageRef &&
getFunctionId(q) == passRef && getArg(q,1) == getArg(p,0))
- // nothing happens in this
instruction
- used++;
- else {
- // check for use without
conversion
- for(k = q->retc; k < q->argc;
k++)
- if( getArg(q,k) == getArg(p,0)){
- blocked++;
- }
- // materialize a copy and
re-use where appropriate
- }
+ series[getArg(p,0)] = p;
+ setModuleId(p, generatorRef);
+ setFunctionId(p, parametersRef);
+ typeChecker(cntxt->fdout, cntxt->nspace, mb, p, TRUE);
+ pushInstruction(mb,p);
+ } else
+ if ( getModuleId(p) == algebraRef && getFunctionId(p) ==
subselectRef && series[getArg(p,1)]){
+ errorCheck(p,algebraRef,getArg(p,1));
+ } else
+ if ( getModuleId(p) == algebraRef && getFunctionId(p) ==
thetasubselectRef && series[getArg(p,1)]){
+ errorCheck(p,algebraRef,getArg(p,1));
+ } else
+ if ( getModuleId(p) == algebraRef && getFunctionId(p) ==
leftfetchjoinRef && series[getArg(p,2)]){
+ errorCheck(p,algebraRef,getArg(p,2));
+ } else
+ if ( getModuleId(p) == algebraRef && getFunctionId(p) ==
joinRef && series[getArg(p,2)] ){
+ errorCheck(p,algebraRef,getArg(p,2));
+ } else
+ if ( getModuleId(p) == algebraRef && getFunctionId(p) ==
joinRef && series[getArg(p,3)]){
+ errorCheck(p,algebraRef,getArg(p,3));
+ } else
+ if ( getModuleId(p) == sqlRef && getFunctionId(p) ==
putName("exportValue",11) && isaBatType(getArgType(mb,p,0)) ){
+ // interface expects scalar type only, not expressable
in MAL signature
+ mb->errors++;
+ showException(cntxt->fdout, MAL, "generate_series",
"internal error, generate_series is a table producing function");
+ }else
+ if ( getModuleId(p) == batcalcRef && getFunctionId(p) == bteRef
&& series[getArg(p,1)] && p->argc == 2 ){
+ casting(bte);
+ } else
+ if ( getModuleId(p) == batcalcRef && getFunctionId(p) == shtRef
&& series[getArg(p,1)] && p->argc == 2 ){
+ casting(sht);
+ } else
+ if ( getModuleId(p) == batcalcRef && getFunctionId(p) == intRef
&& series[getArg(p,1)] && p->argc == 2 ){
+ casting(int);
+ } else
+ if ( getModuleId(p) == batcalcRef && getFunctionId(p) == lngRef
&& series[getArg(p,1)] && p->argc == 2 ){
+ casting(lng);
+ } else
+ if ( getModuleId(p) == batcalcRef && getFunctionId(p) == fltRef
&& series[getArg(p,1)] && p->argc == 2 ){
+ casting(flt);
+ } else
+ if ( getModuleId(p) == batcalcRef && getFunctionId(p) == dblRef
&& series[getArg(p,1)] && p->argc == 2 ){
+ casting(dbl);
+ } else
+ if ( getModuleId(p) == languageRef && getFunctionId(p) ==
passRef )
+ pushInstruction(mb,p);
+ else {
+ // check for use without conversion
+ for(k = p->retc; k < p->argc; k++)
+ if( series[getArg(p,k)]){
+ m = getModuleId(p);
+ setModuleId(p, generatorRef);
+ typeChecker(cntxt->fdout, cntxt->nspace, mb, p,
TRUE);
+ if(p->typechk == TYPE_UNKNOWN){
+ setModuleId(p,m);
+ typeChecker(cntxt->fdout,
cntxt->nspace, mb, p, TRUE);
+ setModuleId(series[getArg(p,k)],
generatorRef);
+ setFunctionId(series[getArg(p,k)],
seriesRef);
+ typeChecker(cntxt->fdout,
cntxt->nspace, mb, series[getArg(p,k)], TRUE);
}
- // fix the original, only when all use cases
are replaced by the overloaded function
- if(used == cases && blocked == 0){
- setModuleId(p, generatorRef);
- setFunctionId(p, parametersRef);
- typeChecker(cntxt->fdout,
cntxt->nspace, mb, p, TRUE);
- }
- if( used)
- actions++;
-#ifdef VLT_DEBUG
- mnstr_printf(cntxt->fdout,"#generator target %d
cases %d used %d error %d\n",getArg(p,0), cases, used, p->typechk);
-#endif
}
+ pushInstruction(mb,p);
}
}
+ for (i++; i < limit; i++)
+ pushInstruction(mb, old[i]);
+ GDKfree(old);
+ GDKfree(series);
+
#ifdef VLT_DEBUG
printFunction(cntxt->fdout,mb,0,LIST_MAL_ALL);
#endif
diff --git a/sql/backends/monet5/generator/generator.c
b/sql/backends/monet5/generator/generator.c
--- a/sql/backends/monet5/generator/generator.c
+++ b/sql/backends/monet5/generator/generator.c
@@ -968,13 +968,14 @@ str VLTgenerator_join(Client cntxt, MalB
}
#define VLTrangeExpand() \
-{ bln= BATextend(bln,BATgrows(bln));\
+{ limit+= cnt * (limit/(done?done:1)+1);\
+ bln= BATextend(bln, limit);\
if( bln == NULL){\
BBPreleaseref(blow->batCacheid);\
BBPreleaseref(bhgh->batCacheid);\
throw(MAL,"generator.rangejoin",MAL_MALLOC_FAIL);\
}\
- brn= BATextend(brn,BATgrows(brn));\
+ brn= BATextend(brn, limit);\
if( brn == NULL) {\
BBPreleaseref(blow->batCacheid);\
BBPreleaseref(bhgh->batCacheid);\
@@ -982,7 +983,6 @@ str VLTgenerator_join(Client cntxt, MalB
}\
ol = (oid*) Tloc(bln,BUNfirst(bln)) + c;\
or = (oid*) Tloc(brn,BUNfirst(brn)) + c;\
- limit= BATcapacity(bln);\
}
/* The operands of a join operation can either be defined on a generator */
@@ -998,7 +998,7 @@ str VLTgenerator_join(Client cntxt, MalB
throw(MAL,"generator.rangejoin","Illegal range");\
vlow = (TPE*) Tloc(blow,BUNfirst(blow));\
vhgh = (TPE*) Tloc(bhgh,BUNfirst(bhgh));\
- for( ; cnt >0; cnt--, o++,vlow++,vhgh++){\
+ for( ; cnt >0; cnt--, done++, o++,vlow++,vhgh++){\
f1 = f + floor(ABS(*vlow-f)/ABS(s)) * s;\
if ( f1 < *vlow ) f1+= s;\
w = (BUN) floor(ABS(f1-f)/ABS(s));\
@@ -1015,7 +1015,7 @@ str VLTgenerator_rangejoin(Client cntxt,
{
BAT *blow = NULL, *bhgh = NULL, *bln = NULL, *brn= NULL;
bit li,ri;
- BUN limit, cnt,c =0;
+ BUN limit, cnt, done=0, c =0;
oid o= 0, *ol, *or;
int tpe, incr=0;
InstrPtr p = NULL;
@@ -1059,35 +1059,37 @@ str VLTgenerator_rangejoin(Client cntxt,
/* The actual join code for generators be injected here */
switch(tpe){
case TYPE_bte: VLTrangejoin(bte,abs); break;
- case TYPE_sht: //VLTrangejoin(sht,abs); break;
- { sht f,f1,l,s; sht *vlow,*vhgh; BUN w;
- f = *getArgReference_sht(stk,p, 1);
- l = *getArgReference_sht(stk,p, 2);
+ case TYPE_sht: VLTrangejoin(sht,abs); break;
+ case TYPE_int: VLTrangejoin(int,abs); break;
+ case TYPE_lng: //VLTrangejoin(lng,llabs); break;
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
https://www.monetdb.org/mailman/listinfo/checkin-list
