database/mariadb-columnstore-engine
1
0
Fork 0
mirror of https://github.com/mariadb-corporation/mariadb-columnstore-engine.git synced 2025-07-30 19:23:07 +03:00
Code Activity

MCOL-4580 extent elimination for dictionary-based text/varchar types

Browse Source 
The idea is relatively simple - encode prefixes of collated strings as
integers and use them to compute extents' ranges. Then we can eliminate
extents with strings.

The actual patch does have all the code there but miss one important
step: we do not keep collation index, we keep charset index. Because of
this, some of the tests in the bugfix suite fail and thus main
functionality is turned off.

The reason of this patch to be put into PR at all is that it contains
changes that made CHAR/VARCHAR columns unsigned. This change is needed in
vectorization work.
This commit is contained in:
Serguey Zefirov
2022-02-04 11:55:09 +00:00
parent a66a8dfabf
commit 53b9a2a0f9
54 changed files with 698 additions and 227 deletions
Download Patch File Download Diff File Expand all files Collapse all files

32
dbcon/joblist/batchprimitiveprocessor-jl.cpp
View File

@ -50,6 +50,8 @@ using namespace messageqcpp;
using namespace rowgroup;
using namespace joiner;
//#define XXX_BATCHPRIMPROC_TOKENS_RANGES_XXX
namespace joblist
{
BatchPrimitiveProcessorJL::BatchPrimitiveProcessorJL(const ResourceManager* rm)
@ -152,6 +154,21 @@ void BatchPrimitiveProcessorJL::addFilterStep(const pDictionaryStep& step)
cc->setBatchPrimitiveProcessor(this);
cc->setQueryUuid(step.queryUuid());
cc->setStepUuid(uuid);
#if defined(XXX_BATCHPRIMPROC_TOKENS_RANGES_XXX)
if (filterSteps.size() > 0)
{
size_t stepsIndex = filterSteps.size() - 1;
SCommand prevCC = filterSteps[stepsIndex];
ColumnCommandJL* pcc = dynamic_cast<ColumnCommandJL*>(prevCC.get());
DictStepJL* ccc = dynamic_cast<DictStepJL*>(cc.get());
if (pcc && ccc)
{
filterSteps[stepsIndex].reset(
new ColumnCommandJL(*pcc, *ccc)); // column command will use same filters.
}
}
#endif
filterSteps.push_back(cc);
filterCount++;
needStrValues = true;
@ -443,6 +460,7 @@ void BatchPrimitiveProcessorJL::getElementTypes(ByteStream& in, vector<ElementTy
if (*validCPData)
{
in >> *lbid;
in >> tmp64;
*min = (int64_t)tmp64;
in >> tmp64;
@ -712,8 +730,9 @@ bool BatchPrimitiveProcessorJL::countThisMsg(messageqcpp::ByteStream& in) const
}
if (data[offset] != 0)
offset += (data[offset + CP_FLAG_AND_LBID] * 2) + CP_FLAG_AND_LBID +
1; // skip the CP data with wide min/max values (16/32 bytes each)
offset += (data[offset + CP_FLAG_AND_LBID + 1] * 2) + CP_FLAG_AND_LBID + 1 +
1; // skip the CP data with wide min/max values (16/32 bytes each). we also skip
// cpFromDictScan flag.
else
offset += CP_FLAG_AND_LBID; // skip only the "valid CP data" & LBID bytes
}
@ -750,9 +769,10 @@ void BatchPrimitiveProcessorJL::deserializeAggregateResult(ByteStream* in, vecto
}
void BatchPrimitiveProcessorJL::getRowGroupData(ByteStream& in, vector<RGData>* out, bool* validCPData,
uint64_t* lbid, int128_t* min, int128_t* max,
uint32_t* cachedIO, uint32_t* physIO, uint32_t* touchedBlocks,
bool* countThis, uint32_t threadID, bool* hasWideColumn,
uint64_t* lbid, bool* fromDictScan, int128_t* min,
int128_t* max, uint32_t* cachedIO, uint32_t* physIO,
uint32_t* touchedBlocks, bool* countThis, uint32_t threadID,
bool* hasWideColumn,
const execplan::CalpontSystemCatalog::ColType& colType) const
{
uint64_t tmp64;
@ -789,6 +809,8 @@ void BatchPrimitiveProcessorJL::getRowGroupData(ByteStream& in, vector<RGData>*
{
in >> *lbid;
in >> tmp8;
*fromDictScan = tmp8 != 0;
in >> tmp8;
*hasWideColumn = (tmp8 > utils::MAXLEGACYWIDTH);
if (UNLIKELY(*hasWideColumn))
{