Changeset: 67af55b88549 for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=67af55b88549
Modified Files:
monetdb5/modules/mosaic/mosaic.c
monetdb5/modules/mosaic/mosaic.h
sql/backends/monet5/sql.mal
sql/backends/monet5/sql_mosaic.c
sql/scripts/76_mosaic.sql
sql/test/mosaic/Tests/analysis.sql
sql/test/mosaic/Tests/analysis.stable.out
Branch: mosaic
Log Message:
Add 'common_compression' a.k.a. exceptions argument to analysis.
diffs (254 lines):
diff --git a/monetdb5/modules/mosaic/mosaic.c b/monetdb5/modules/mosaic/mosaic.c
--- a/monetdb5/modules/mosaic/mosaic.c
+++ b/monetdb5/modules/mosaic/mosaic.c
@@ -1238,7 +1238,7 @@ MOSjoin(Client cntxt, MalBlkPtr mb, MalS
*/
static int
-makepatterns(uint16_t *patterns, int size, str compressions, BAT* b)
+makepatterns(uint16_t *patterns, int size, str compressions, sht common_mask,
BAT* b)
{
int i,j,k, idx, bit=1, step = MOSAIC_METHODS;
int lim= 8*7*6*5*4*3*2;
@@ -1252,6 +1252,10 @@ makepatterns(uint16_t *patterns, int siz
// Unset corresponding bit if type is not allowed.
UNSET_METHOD(compression_mask, i);
}
+ if ( METHOD_IS_SET(common_mask, i) && !MOSisTypeAllowed(i, b) )
{
+ // Unset corresponding bit if type is not allowed.
+ UNSET_METHOD(common_mask, i);
+ }
}
for( k=0, i=0; i<lim && k <size; i++){
@@ -1264,11 +1268,13 @@ makepatterns(uint16_t *patterns, int siz
idx /= step;
}
+ patterns[k] |= common_mask; // Make sure the common mask is
always present in the to-be-analyzed patterns.
+
// weed out duplicates
for( j=0; j< k; j++)
if(patterns[k] == patterns[j]) break;
if( j < k ) continue;
-
+
#ifdef _MOSAIC_DEBUG_
mnstr_printf(GDKstdout,"#");
for(j=0, bit=1; j < MOSAIC_METHODS; j++){
@@ -1303,7 +1309,7 @@ struct PAT{
};
str
-MOSAnalysis(BAT *b, BAT *btech, BAT *blayout, BAT *boutput, BAT *bratio, BAT
*bcompress, BAT *bdecompress, str compressions)
+MOSAnalysis(BAT *b, BAT *btech, BAT *blayout, BAT *boutput, BAT *bratio, BAT
*bcompress, BAT *bdecompress, str compressions, str common_compressions)
{
unsigned i,j,cases, bid= b->batCacheid;
uint16_t pattern[CANDIDATES];
@@ -1314,8 +1320,12 @@ MOSAnalysis(BAT *b, BAT *btech, BAT *bla
struct PAT pat[CANDIDATES];
+ // create mask of common compressions, e.g. exception methods.
+ sht common_mask = 0;
+ _construct_compression_mask(&common_mask, common_compressions);
+
// create the list of all possible 2^6 compression patterns
- cases = makepatterns(pattern,CANDIDATES, compressions, b);
+ cases = makepatterns(pattern,CANDIDATES, compressions, common_mask, b);
memset(antipattern,0, sizeof(antipattern));
antipatternSize++; // the first pattern aka 0 is always an antipattern.
@@ -1419,7 +1429,7 @@ MOSAnalysis(BAT *b, BAT *btech, BAT *bla
}
for(j=0; j < MOSAIC_METHODS; j++){
- if ( ((MosaicHdr) b->tmosaic->base)->blks[j] == 0) {
+ if ( ((MosaicHdr) b->tmosaic->base)->blks[j] == 0 &&
(!(MOSmethods[j].bit & common_mask))) {
antipattern[antipatternSize++] = pattern[i];
}
}
diff --git a/monetdb5/modules/mosaic/mosaic.h b/monetdb5/modules/mosaic/mosaic.h
--- a/monetdb5/modules/mosaic/mosaic.h
+++ b/monetdb5/modules/mosaic/mosaic.h
@@ -286,7 +286,7 @@ mal_export str MOSthetaselect(Client cnt
mal_export str MOSprojection(Client cntxt, MalBlkPtr mb, MalStkPtr stk,
InstrPtr pci);
mal_export str MOSjoin(Client cntxt, MalBlkPtr mb, MalStkPtr stk, InstrPtr
pci);
mal_export str MOSlayout(BAT *b, BAT *bbsn, BAT *btech, BAT *bcount, BAT
*binput, BAT *boutput, BAT *bproperties);
-mal_export str MOSAnalysis(BAT *b, BAT *btech, BAT *blayout, BAT *output, BAT
*factor, BAT *compress, BAT *decompress, str compressions);
+mal_export str MOSAnalysis(BAT *b, BAT *btech, BAT *blayout, BAT *output, BAT
*factor, BAT *compress, BAT *decompress, str compressions, str
common_compressions);
void MOSupdateHeader(MOStask* task);
void MOSinitHeader(MOStask* task);
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
@@ -546,7 +546,7 @@ pattern mosaicanalysis(sch:str,tbl:str,c
compress:bat[:lng],
decompress:bat[:lng])
address sql_mosaicAnalysis;
-pattern mosaicanalysis(sch:str,tbl:str,col:str)(
+pattern mosaicanalysis(sch:str,tbl:str,col:str,compression:str,
common_compression:str)(
technique:bat[:str],
layout:bat[:json],
outputsize:bat[: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
@@ -160,6 +160,7 @@ sql_mosaicAnalysis(Client cntxt, MalBlkP
int *tech,*output, *factor, *compress, *decompress, *layout;
BAT *bn, *btech, *boutput, *bfactor, *bcompress, *bdecompress, *blayout;
str compressions = NULL;
+ str common_compressions = "raw";
if (msg != MAL_SUCCEED || (msg = checkSQLContext(cntxt)) != NULL)
return msg;
@@ -219,12 +220,14 @@ sql_mosaicAnalysis(Client cntxt, MalBlkP
layout = getArgReference_bat(stk, pci, 1);
*layout = blayout->batCacheid;
- sch = *getArgReference_str(stk, pci, 6);
- tbl = *getArgReference_str(stk, pci, 7);
- col = *getArgReference_str(stk, pci, 8);
- if ( pci->argc == 10){
+ sch = *getArgReference_str(stk, pci, 6);
+ tbl = *getArgReference_str(stk, pci, 7);
+ col = *getArgReference_str(stk, pci, 8);
+ compressions = *getArgReference_str(stk, pci, 9);
+
+ if ( pci->argc == 11) {
// use a predefined collection of compression schemes.
- compressions = *getArgReference_str(stk,pci,9);
+ common_compressions = *getArgReference_str(stk, pci, 10);
}
#ifdef DEBUG_SQL_MOSAIC
@@ -253,7 +256,7 @@ sql_mosaicAnalysis(Client cntxt, MalBlkP
continue;
// perform the analysis
bn =
store_funcs.bind_col(m->session->tr, c, 0);
- msg = MOSAnalysis(bn, btech,
blayout, boutput, bfactor, bcompress, bdecompress, compressions);
+ msg = MOSAnalysis(bn, btech,
blayout, boutput, bfactor, bcompress, bdecompress, compressions,
common_compressions);
BBPunfix(bn->batCacheid);
(void) c;
diff --git a/sql/scripts/76_mosaic.sql b/sql/scripts/76_mosaic.sql
--- a/sql/scripts/76_mosaic.sql
+++ b/sql/scripts/76_mosaic.sql
@@ -11,10 +11,10 @@ create function mosaic.layout(sch string
returns table(technique string, "count" bigint, inputsize bigint, outputsize
bigint,properties json, bsn bigint)
external name sql.mosaiclayout;
-create function mosaic.analysis(sch string, tbl string, col string)
+create function mosaic.analysis(sch string, tbl string, col string,
compression string)
returns table(technique string, layout json, outputsize bigint, factor float,
"compress" bigint, "decompress" bigint)
external name sql.mosaicanalysis;
-create function mosaic.analysis(sch string, tbl string, col string,
compression string)
+create function mosaic.analysis(sch string, tbl string, col string,
compression string, common_compression string)
returns table(technique string, layout json, outputsize bigint, factor float,
"compress" bigint, "decompress" bigint)
external name sql.mosaicanalysis;
diff --git a/sql/test/mosaic/Tests/analysis.sql
b/sql/test/mosaic/Tests/analysis.sql
--- a/sql/test/mosaic/Tests/analysis.sql
+++ b/sql/test/mosaic/Tests/analysis.sql
@@ -8,15 +8,17 @@ insert into tmp4 select * from generate_
-- tmp4 is currently uncompressed
-select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i') order by technique, factor desc;
+select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i', NULL, '') order by technique, factor desc;
-- should be materialized as the graph of a cutoff function.
insert into tmp4 select 10000000 from tmp4;
-select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i') order by technique, factor desc;
+select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i', NULL, '') order by technique, factor desc;
-select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i', 'linear, runlength') order by technique, factor desc;
+select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i', 'linear, runlength', '') order by technique, factor desc;
+
+select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i', 'linear, runlength', 'raw') order by technique, factor desc;
-- Make sure that mosaic.analysis also works correctly on a column with
compression
@@ -24,9 +26,9 @@ set optimizer='mosaic_pipe';
alter table tmp4 alter column i set storage 'runlength, linear';
-select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i') order by technique, factor desc;
+select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i', NULL, '') order by technique, factor desc;
-select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i', 'linear, runlength') order by technique, factor desc;
+select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i', 'linear, runlength', '') order by technique, factor desc;
drop table tmp4;
diff --git a/sql/test/mosaic/Tests/analysis.stable.out
b/sql/test/mosaic/Tests/analysis.stable.out
--- a/sql/test/mosaic/Tests/analysis.stable.out
+++ b/sql/test/mosaic/Tests/analysis.stable.out
@@ -27,8 +27,8 @@ stdout of test 'analysis` in directory '
[ 0 ]
#insert into tmp4 select * from generate_series(0, 10000000);
[ 10000000 ]
-#select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i') order by technique, factor desc;
-% .%4, .%4, ., . # table_name
+#select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i', NULL, '') order by technique, factor desc;
+% .%6, .%6, ., . # table_name
% technique, factor, blocks, elements # name
% clob, double, json, json # type
% 18, 24, 39, 41 # length
@@ -68,8 +68,8 @@ stdout of test 'analysis` in directory '
[ "runlength prefix ", 2.132, "[[{\"prefix\":306}]]",
"[[{\"prefix\":10000000}]]" ]
#insert into tmp4 select 10000000 from tmp4;
[ 10000000 ]
-#select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i') order by technique, factor desc;
-% .%4, .%4, ., . # table_name
+#select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i', NULL, '') order by technique, factor desc;
+% .%6, .%6, ., . # table_name
% technique, factor, blocks, elements # name
% clob, double, json, json # type
% 18, 24, 39, 46 # length
@@ -107,18 +107,26 @@ stdout of test 'analysis` in directory '
[ "runlength frame ", 2.285, "[[{\"frame\":306}]]",
"[[{\"frame\":20000000}]]" ]
[ "runlength linear ", 357142.843, "[[{\"runlength\":1},{\"linear\":1}]]",
"[[{\"runlength\":9999999},{\"linear\":10000001}]]" ]
[ "runlength prefix ", 2.285, "[[{\"prefix\":306}]]",
"[[{\"prefix\":20000000}]]" ]
-#select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i', 'linear, runlength') order by technique, factor desc;
-% .%5, .%5, ., . # table_name
+#select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i', 'linear, runlength', '') order by technique, factor desc;
+% .%6, .%6, ., . # table_name
% technique, factor, blocks, elements # name
% clob, double, json, json # type
% 17, 24, 32, 45 # length
[ "linear ", 350877.187, "[[{\"linear\":2}]]",
"[[{\"linear\":20000000}]]" ]
[ "runlength ", 0.666, "[[{\"runlength\":10000001}]]",
"[[{\"runlength\":20000000}]]" ]
[ "runlength linear ", 357142.843, "[[{\"runlength\":1},{\"linear\":1}]]",
"[[{\"runlength\":9999999},{\"linear\":10000001}]]" ]
+#select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i', 'linear, runlength', 'raw') order by technique, factor desc;
+% .%6, .%6, ., . # table_name
+% technique, factor, blocks, elements # name
+% clob, double, json, json # type
+% 21, 24, 32, 45 # length
+[ "raw linear ", 350877.187, "[[{\"linear\":2}]]",
"[[{\"linear\":20000000}]]" ]
+[ "raw runlength ", 1.999, "[[{\"raw\":1},{\"runlength\":1}]]",
"[[{\"raw\":10000000},{\"runlength\":10000000}]]" ]
+[ "raw runlength linear ", 357142.843,
"[[{\"runlength\":1},{\"linear\":1}]]",
"[[{\"runlength\":9999999},{\"linear\":10000001}]]" ]
#set optimizer='mosaic_pipe';
-#alter table tmp4 alter column i set storage 'dict';
-#select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i') order by technique, factor desc;
-% .%4, .%4, ., . # table_name
+#alter table tmp4 alter column i set storage 'runlength, linear';
+#select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i', NULL, '') order by technique, factor desc;
+% .%6, .%6, ., . # table_name
% technique, factor, blocks, elements # name
% clob, double, json, json # type
% 18, 24, 39, 46 # length
@@ -156,8 +164,8 @@ stdout of test 'analysis` in directory '
[ "runlength frame ", 2.285, "[[{\"frame\":306}]]",
"[[{\"frame\":20000000}]]" ]
[ "runlength linear ", 357142.843, "[[{\"runlength\":1},{\"linear\":1}]]",
"[[{\"runlength\":9999999},{\"linear\":10000001}]]" ]
[ "runlength prefix ", 2.285, "[[{\"prefix\":306}]]",
"[[{\"prefix\":20000000}]]" ]
-#select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i', 'linear, runlength') order by technique, factor desc;
-% .%5, .%5, ., . # table_name
+#select technique, factor, json.filter(layout, 'blks') as blocks,
json.filter(layout, 'elms') as elements from mosaic.analysis('sys', 'tmp4',
'i', 'linear, runlength', '') order by technique, factor desc;
+% .%6, .%6, ., . # table_name
% technique, factor, blocks, elements # name
% clob, double, json, json # type
% 17, 24, 32, 45 # length
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
https://www.monetdb.org/mailman/listinfo/checkin-list
