Changeset: 26239957e3d7 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=26239957e3d7
Modified Files:
monetdb5/modules/mal/mosaic.c
monetdb5/modules/mal/mosaic.h
sql/backends/monet5/sql.mal
sql/backends/monet5/sql_mosaic.c
sql/scripts/76_compression.sql
Branch: mosaic
Log Message:
First step into mosaic layout analysis
The layout can be obtained as an ordinary SQL table
for further analysis
diffs (170 lines):
diff --git a/monetdb5/modules/mal/mosaic.c b/monetdb5/modules/mal/mosaic.c
--- a/monetdb5/modules/mal/mosaic.c
+++ b/monetdb5/modules/mal/mosaic.c
@@ -74,13 +74,30 @@ MOSdumpTask(Client cntxt,MOStask task)
}
str
-MOSlayout(Client cntxt, BAT *b, BAT *btech, BAT *bcount, BAT *binput, BAT
*boutput, BAT *bproperties)
+MOSlayout(Client cntxt, BAT *b, BAT *btech, BAT *bcount, BAT *binput, BAT
*boutput, BAT *bproperties, str compressionscheme)
{
MOStask task=0;
+ int i,ret,bid;
+ BAT *bn= NULL;
task= (MOStask) GDKzalloc(sizeof(*task));
if( task == NULL)
throw(SQL,"mosaic",MAL_MALLOC_FAIL);
+
+ if(compressionscheme){
+ //create a tempory compressed column
+ for( i = 0; i< MOSAIC_METHODS; i++)
+ task->filter[i]= strstr(compressionscheme,
MOSfiltername[i]) != 0;
+ bid = b->batCacheid;
+ MOScompressInternal(cntxt, &ret, &bid, task,0,FALSE);
+ if( ret == 0)
+ throw(MAL,"mosaic.layout","Compression failed");
+ bn = BATdescriptor(ret);
+ if( bn == NULL)
+ throw(MAL,"mosaic.layout", RUNTIME_OBJECT_MISSING);
+ b = bn;
+ }
+
MOSinit(task,b);
MOSinitializeScan(cntxt,task,0,task->hdr->top);
while(task->start< task->stop){
@@ -110,6 +127,8 @@ MOSlayout(Client cntxt, BAT *b, BAT *bte
MOSadvance_frame(cntxt,task);
}
}
+ if( bn)
+ BBPunfix(bn->batCacheid);
return MAL_SUCCEED;
}
@@ -1330,7 +1349,7 @@ makepatterns(int *patterns, int size, st
int candidate[MOSAIC_METHODS]= {0};
for( i = 0; i < MOSAIC_METHODS; i++)
- candidate[i] = strstr(compressions,MOSfiltername[i]) != 0;
+ candidate[i] = compressions == NULL ||
strstr(compressions,MOSfiltername[i]) != 0;
for( k=0, i=0; i<lim && k <size; i++){
patterns[k]=0;
@@ -1669,7 +1688,7 @@ MOSoptimize(Client cntxt, MalBlkPtr mb,
bid = *getArgReference_int(stk,pci,1);
MOSblocklimit = 100000;
- for( i = 0; i < 1024; i++)
+ for( i = 0; i < CANDIDATES; i++)
xf[i]= -1;
for( i = 1; i< cases; i++){
diff --git a/monetdb5/modules/mal/mosaic.h b/monetdb5/modules/mal/mosaic.h
--- a/monetdb5/modules/mal/mosaic.h
+++ b/monetdb5/modules/mal/mosaic.h
@@ -217,7 +217,7 @@ mosaic_export str MOSthetasubselect(Clie
mosaic_export str MOSleftfetchjoin(Client cntxt, MalBlkPtr mb, MalStkPtr stk,
InstrPtr pci);
mosaic_export str MOSjoin(Client cntxt, MalBlkPtr mb, MalStkPtr stk, InstrPtr
pci);
mosaic_export str MOSdump(Client cntxt, MalBlkPtr mb, MalStkPtr stk, InstrPtr
pci);
-mosaic_export str MOSlayout(Client cntxt, BAT *b, BAT *btech, BAT *bcount, BAT
*binput, BAT *boutput, BAT *bproperties);
+mosaic_export str MOSlayout(Client cntxt, BAT *b, BAT *btech, BAT *bcount, BAT
*binput, BAT *boutput, BAT *bproperties, str compressionscheme);
mosaic_export str MOSsliceInternal(Client cntxt, bat *slices, BUN size, BAT
*b);
mosaic_export int MOSanalyseInternal(Client cntxt, int threshold, MOStask
task, bat bid);
mosaic_export void MOSanalyseReport(Client cntxt, BAT *b, BAT *btech, BAT
*output, BAT *factor, str compressions);
diff --git a/sql/backends/monet5/sql.mal b/sql/backends/monet5/sql.mal
--- a/sql/backends/monet5/sql.mal
+++ b/sql/backends/monet5/sql.mal
@@ -477,6 +477,13 @@ pattern storage()(
address sql_storage
comment "return a table with storage information ";
+pattern mosaicLayout(sch:str,tbl:str,col:str,compressionscheme:str)(
+ technique:bat[:oid,:str],
+ count:bat[:oid,:lng],
+ inputsize:bat[:oid,:lng],
+ outputsize:bat[:oid,:lng],
+ properties:bat[:oid,:str])
+address sql_mosaicLayout;
pattern mosaicLayout(sch:str,tbl:str,col:str)(
technique:bat[:oid,:str],
count:bat[:oid,:lng],
diff --git a/sql/backends/monet5/sql_mosaic.c b/sql/backends/monet5/sql_mosaic.c
--- a/sql/backends/monet5/sql_mosaic.c
+++ b/sql/backends/monet5/sql_mosaic.c
@@ -31,6 +31,7 @@ sql_mosaicLayout(Client cntxt, MalBlkPtr
str sch = 0, tbl = 0, col = 0;
BAT *bn, *btech, *bcount, *binput, *boutput, *bproperties;
int *tech,*count,*input,*output, *properties;
+ str compressionscheme= NULL;
if (msg != MAL_SUCCEED || (msg = checkSQLContext(cntxt)) != NULL)
return msg;
@@ -80,12 +81,16 @@ sql_mosaicLayout(Client cntxt, MalBlkPtr
throw(SQL,"mosaicLayout", MAL_MALLOC_FAIL);
}
BATseqbase(bproperties,0);
- properties = getArgReference_bat(stk, pci, 3);
- *properties = boutput->batCacheid;
+ properties = getArgReference_bat(stk, pci, 4);
+ *properties = bproperties->batCacheid;
- sch = *getArgReference_str(stk, pci, 4);
- tbl = *getArgReference_str(stk, pci, 5);
- col = *getArgReference_str(stk, pci, 6);
+ sch = *getArgReference_str(stk, pci, 5);
+ tbl = *getArgReference_str(stk, pci, 6);
+ col = *getArgReference_str(stk, pci, 7);
+ if ( pci->argc == 9){
+ // use a predefined collection of compression schemes.
+ compressionscheme = *getArgReference_str(stk,pci,8);
+ }
#ifdef DEBUG_SQL_MOSAIC
mnstr_printf(cntxt->fdout, "#mosaic layout %s.%s.%s \n", sch, tbl, col);
@@ -112,12 +117,17 @@ sql_mosaicLayout(Client cntxt, MalBlkPtr
continue;
// perform the analysis
bn =
store_funcs.bind_col(m->session->tr, c, 0);
- MOSlayout(cntxt, bn, btech,
bcount, binput, boutput, bproperties);
+ MOSlayout(cntxt, bn, btech,
bcount, binput, boutput, bproperties, compressionscheme);
BBPunfix(bn->batCacheid);
(void) c;
}
}
}
+ BBPkeepref(*tech);
+ BBPkeepref(*count);
+ BBPkeepref(*input);
+ BBPkeepref(*output);
+ BBPkeepref(*properties);
return MAL_SUCCEED;
}
@@ -131,7 +141,6 @@ sql_mosaicAnalysis(Client cntxt, MalBlkP
str sch = 0, tbl = 0, col = 0;
int *tech,*output, *factor;
BAT *bn, *btech, *boutput, *bfactor;
- int k;
str compressions = NULL;
if (msg != MAL_SUCCEED || (msg = checkSQLContext(cntxt)) != NULL)
diff --git a/sql/scripts/76_compression.sql b/sql/scripts/76_compression.sql
--- a/sql/scripts/76_compression.sql
+++ b/sql/scripts/76_compression.sql
@@ -31,6 +31,15 @@ returns table (
properties string -- additional information
) external name sql."mosaicLayout";
+create function mosaic.layout(sch string, tbl string, col string,
compressionscheme string)
+returns table (
+ "technique" string, -- any of the built-in compressors + header
+ "count" bigint, -- entries covered
+ inputsize bigint, -- original storage footprint
+ outputsize bigint, -- after compression
+ properties string -- additional information
+ ) external name sql."mosaicLayout";
+
-- provide an analysis of the possible mosaic storage layouts
create function mosaic.analysis(sch string, tbl string, col string)
returns table(
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list
