tsc raw data combination restructure

src/client/inc/tscUtil.h

@@ -40,7 +40,8 @@ extern "C" {
 #define UTIL_TABLE_IS_TMP_TABLE(metaInfo)  \
   (((metaInfo)->pTableMeta != NULL) && ((metaInfo)->pTableMeta->tableType == TSDB_TEMP_TABLE))
 
-#define KvRowNColsThresh 1  // default 1200. TODO: only for test, restore to default value after test finished
+#define KvRowNColsThresh 1  // default 1200
+#define KVRowRatio 0.85     // for NonVarType, we get value from SDataRow directly, while needs readdressing for SKVRow
 
 #pragma pack(push,1)
 // this struct is transfered as binary, padding two bytes to avoid
@@ -96,11 +97,21 @@ typedef struct SVgroupTableInfo {
   SArray     *itemList;   // SArray<STableIdInfo>
 } SVgroupTableInfo;
 
+typedef struct SBlockKeyTuple {
+  TSKEY skey;
+  void* payloadAddr;
+} SBlockKeyTuple;
+
+typedef struct SBlockKeyInfo {
+  int32_t         nBytesAlloc;
+  SBlockKeyTuple* pKeyTuple;
+} SBlockKeyInfo;
+
 int32_t converToStr(char *str, int type, void *buf, int32_t bufSize, int32_t *len);
 
 int32_t tscCreateDataBlock(size_t initialSize, int32_t rowSize, int32_t startOffset, SName* name, STableMeta* pTableMeta, STableDataBlocks** dataBlocks);
 void tscDestroyDataBlock(STableDataBlocks* pDataBlock, bool removeMeta);
-void tscSortRemoveDataBlockDupRows(STableDataBlocks* dataBuf);
+int     tscSortRemoveDataBlockDupRows(STableDataBlocks* dataBuf, SBlockKeyInfo* pBlkKeyInfo);
 
 void tscDestroyBoundColumnInfo(SParsedDataColInfo* pColInfo);
 void doRetrieveSubqueryData(SSchedMsg *pMsg);
@@ -343,22 +354,6 @@ char* strdup_throw(const char* str);
 bool vgroupInfoIdentical(SNewVgroupInfo *pExisted, SVgroupMsg* src);
 SNewVgroupInfo createNewVgroupInfo(SVgroupMsg *pVgroupMsg);
 
-typedef struct {
-  // for SDataRow
-  SSchema*  pSchema;
-  int16_t   sversion;
-  int32_t   flen;
-  // for  SKVRow
-  uint16_t    nCols;
-  uint16_t    size;
-  void*       buf;
-  
-  void*       pDataBlock;
-  SSubmitBlk* pSubmitBlk;
-} SMemRowBuilder;
-
-SMemRow tdGenMemRowFromBuilder(SMemRowBuilder* pBuilder);
-
 #ifdef __cplusplus
 }
 #endif

src/client/inc/tsclient.h

@@ -95,6 +95,43 @@ typedef struct SParsedDataColInfo {
   SBoundColumn   *cols;
 } SParsedDataColInfo;
 
+typedef struct {
+  // for SDataRow
+  SSchema *pSchema;
+  int16_t  sversion;
+  int32_t  flen;
+  // for  SKVRow
+  uint16_t nCols;
+  uint16_t size;
+  void *   buf;
+
+  void *      pDataBlock;
+  SSubmitBlk *pSubmitBlk;
+  uint16_t    allNullLen;
+} SMemRowBuilder;
+
+int FORCE_INLINE initSMemRowBuilder(SMemRowBuilder *pBuilder, SSchema *pSSchema, uint16_t nCols,
+                                    uint16_t allNullColsLen) {
+  ASSERT(nCols > 0);
+  pBuilder->pSchema = pSSchema;
+  pBuilder->allNullLen = allNullColsLen;  //  TODO: get allNullColsLen when creating or altering table meta
+  if (pBuilder->allNullLen == 0) {
+    for (uint16_t i = 0; i < nCols; ++i) {
+      uint8_t type = pSSchema[i].type;
+      int32_t typeLen = TYPE_BYTES[type];
+      ASSERT(typeLen > 0);
+      pBuilder->allNullLen += typeLen;
+      if (TSDB_DATA_TYPE_BINARY == type) {
+        pBuilder->allNullLen += (sizeof(VarDataLenT) + CHAR_BYTES);
+      } else if (TSDB_DATA_TYPE_NCHAR == type) {
+        int len = sizeof(VarDataLenT) + TSDB_NCHAR_SIZE;
+        pBuilder->allNullLen += len;
+      }
+    }
+  }
+  return 0;
+}
+
 typedef struct STableDataBlocks {
   SName       tableName;
   int8_t      tsSource;     // where does the UNIX timestamp come from, server or client
@@ -114,7 +151,8 @@ typedef struct STableDataBlocks {
   // for parameter ('?') binding
   uint32_t       numOfAllocedParams;
   uint32_t       numOfParams;
-  SParamInfo *params;
+  SParamInfo *   params;
+  SMemRowBuilder rowBuilder;
 } STableDataBlocks;
 
 typedef struct {

src/client/src/tscUtil.c

@@ -1639,61 +1639,29 @@ int32_t tscGetDataBlockFromList(SHashObj* pHashList, int64_t id, int32_t size, i
 
   return TSDB_CODE_SUCCESS;
 }
-static FORCE_INLINE uint8_t checkTdRowType(SSchema* pSchema, void* pData, int32_t nCols, int32_t flen,
-                                           uint16_t* nColsNotNull) {
-  ASSERT(pData != NULL);
-  if (nCols < KvRowNColsThresh) {
-    return SMEM_ROW_DATA;
-  }
-  int32_t dataRowLength = flen;
-  int32_t kvRowLength = TD_MEM_ROW_KV_VER_SIZE;
-
-  uint16_t nColsNull = 0;
-  char*   p = (char*)pData;
-  for (int i = 0; i < nCols; ++i) {
-    if (IS_VAR_DATA_TYPE(pSchema[i].type)) {
-      dataRowLength += varDataTLen(p);
-      if (!isNull(p, pSchema[i].type)) {
-        kvRowLength += (sizeof(SColIdx) + varDataTLen(p));
-      } else {
-        ++nColsNull;
-      }
-    } else {
-      if (!isNull(p, pSchema[i].type)) {
-        kvRowLength += (sizeof(SColIdx) + TYPE_BYTES[pSchema[i].type]);
-      } else {
-        ++nColsNull;
-      }
-    }
 
-    // next column
-    p += pSchema[i].bytes;
-  }
 
-  tscDebug("nColsNull %d, nCols: %d, kvRowLen: %d, dataRowLen: %d", (int32_t)nColsNull, nCols, kvRowLength,
-          dataRowLength);
 
-  if (kvRowLength < dataRowLength) {
-    if (nColsNotNull) {
-      *nColsNotNull = nCols - nColsNull;
-    }
-    return SMEM_ROW_KV;
-  }
 
-  return SMEM_ROW_DATA;
-}
-SMemRow tdGenMemRowFromBuilder(SMemRowBuilder* pBuilder) {
+static SMemRow tdGenMemRowFromBuilder(SMemRowBuilder* pBuilder) {
   SSchema* pSchema = pBuilder->pSchema;
   char*    p = (char*)pBuilder->buf;
   int      toffset = 0;
-
-  if(pBuilder->nCols <= 0){
+  uint16_t nCols = pBuilder->nCols;
+
+//  RawRow payload structure:
+//  |<---------- header ------------->|<------- column data array ------->|
+//  |SMemRowType|  dataLen |  nCols   |  colId  | colType | value |...|...|
+//  +-----------+----------+----------+---------------------------------->|
+//  | uint8_t   | uint16_t | uint16_t | int16_t | uint8_t |  ???  |...|...|
+//  +-----------+----------+----------+---------------------------------->|
+  uint8_t  memRowType = payloadType(p);
+  uint16_t nColsNotNull = payloadNCols(p);
+  if (pBuilder->nCols <= 0 || nColsNotNull <= 0) {
     return NULL;
   }
+  ASSERT(nColsNotNull <= nCols);
 
-  uint16_t nColsNotNull = 0;
-  uint8_t  memRowType = checkTdRowType(pSchema, p, pBuilder->nCols, pBuilder->flen, &nColsNotNull);
-  // nColsNotNull = pBuilder->nCols;
   SMemRow* memRow = (SMemRow)pBuilder->pDataBlock;
   memRowSetType(memRow, memRowType);
 
@@ -1702,12 +1670,39 @@ SMemRow tdGenMemRowFromBuilder(SMemRowBuilder* pBuilder) {
     dataRowSetLen(trow, (uint16_t)(TD_DATA_ROW_HEAD_SIZE + pBuilder->flen));
     dataRowSetVersion(trow, pBuilder->sversion);
 
-    p = (char*)pBuilder->buf;
-    for (int32_t j = 0; j < pBuilder->nCols; ++j) {
-      tdAppendColVal(trow, p, pSchema[j].type, pSchema[j].bytes, toffset);
+    p = (char*)payloadBody(pBuilder->buf);
+    uint16_t i = 0, j = 0;
+    while (j < pBuilder->nCols) {
+      if (i >= nColsNotNull) {
+        break;
+      }
+      int16_t colId = *(int16_t*)p;
+      if (colId == pSchema[j].colId) {
+        tdAppendColVal(trow, payloadColValue(p), pSchema[j].type, toffset);
         toffset += TYPE_BYTES[pSchema[j].type];
-      p += pSchema[j].bytes;
+        p = skipToNextEles(p);
+        ++i;
+        ++j;
+      } else if (colId < pSchema[j].colId) {
+        p = skipToNextEles(p);
+        ++i;
+      } else {
+        tdAppendColVal(trow, tdGetNullVal(pSchema[j].type), pSchema[j].type, toffset);
+        toffset += TYPE_BYTES[pSchema[j].type];
+        ++j;
       }
+    }
+
+    while (j < pBuilder->nCols) {
+      tdAppendColVal(trow, tdGetNullVal(pSchema[j].type), pSchema[j].type, toffset);
+      toffset += TYPE_BYTES[pSchema[j].type];
+      ++j;
+    }
+    while (i < nColsNotNull) {
+      p = skipToNextEles(p);
+      ++i;
+    }
+
     pBuilder->buf = p;
   } else if (memRowType == SMEM_ROW_KV) {
     ASSERT(nColsNotNull <= pBuilder->nCols);
@@ -1717,14 +1712,17 @@ SMemRow tdGenMemRowFromBuilder(SMemRowBuilder* pBuilder) {
     kvRowSetNCols(kvRow, nColsNotNull);
     memRowKvSetVersion(memRow, pBuilder->sversion);
 
-    p = (char*)pBuilder->buf;
-    for (int32_t j = 0; j < pBuilder->nCols; ++j) {
-      if(!isNull(p, pSchema[j].type)) {
-        tdAppendKvColVal(kvRow, p, pSchema[j].colId, pSchema[j].type, toffset);
+    p = (char*)payloadBody(pBuilder->buf);
+    int i = 0;
+    while (i < nColsNotNull) {
+      int16_t colId = payloadColId(p);
+      uint8_t colType = payloadColType(p);
+      tdAppendKvColVal(kvRow, payloadColValue(p), colId, colType, toffset);
       toffset += sizeof(SColIdx);
+      p = skipToNextEles(p);
+      ++i;
     }
-      p += pSchema[j].bytes;
-    }
+
     pBuilder->buf = p;
 
   } else {
@@ -1738,11 +1736,12 @@ SMemRow tdGenMemRowFromBuilder(SMemRowBuilder* pBuilder) {
 }
 
 // Erase the empty space reserved for binary data
-static int trimDataBlock(void* pDataBlock, STableDataBlocks* pTableDataBlock, bool includeSchema) {
+static int trimDataBlock(void* pDataBlock, STableDataBlocks* pTableDataBlock, bool includeSchema, SBlockKeyTuple *blkKeyTuple) {
   // TODO: optimize this function, handle the case while binary is not presented
   STableMeta*     pTableMeta = pTableDataBlock->pTableMeta;
   STableComInfo   tinfo = tscGetTableInfo(pTableMeta);
   SSchema*        pSchema = tscGetTableSchema(pTableMeta);
+  SMemRowBuilder* pBuilder = &pTableDataBlock->rowBuilder;
 
   SSubmitBlk* pBlock = pDataBlock;
   memcpy(pDataBlock, pTableDataBlock->pData, sizeof(SSubmitBlk));
@@ -1778,18 +1777,18 @@ static int trimDataBlock(void* pDataBlock, STableDataBlocks* pTableDataBlock, bo
   pBlock->dataLen = 0;
   int32_t numOfRows = htons(pBlock->numOfRows);
 
-  SMemRowBuilder mRowBuilder;
-  mRowBuilder.pSchema = pSchema;
-  mRowBuilder.sversion = pTableMeta->sversion;
-  mRowBuilder.flen = flen;
-  mRowBuilder.nCols = tinfo.numOfColumns;
-  mRowBuilder.pDataBlock = pDataBlock;
-  mRowBuilder.pSubmitBlk = pBlock;
-  mRowBuilder.buf = p;
-  mRowBuilder.size = 0;
+  pBuilder->pSchema = pSchema;
+  pBuilder->sversion = pTableMeta->sversion;
+  pBuilder->flen = flen;
+  pBuilder->nCols = tinfo.numOfColumns;
+  pBuilder->pDataBlock = pDataBlock;
+  pBuilder->pSubmitBlk = pBlock;
+  pBuilder->buf = p;
+  pBuilder->size = 0;
 
   for (int32_t i = 0; i < numOfRows; ++i) {
-    tdGenMemRowFromBuilder(&mRowBuilder);
+    pBuilder->buf = (blkKeyTuple+i)->payloadAddr;
+    tdGenMemRowFromBuilder(pBuilder);
   }
 
   int32_t len = pBlock->dataLen + pBlock->schemaLen;
@@ -1807,9 +1806,8 @@ static int32_t getRowExpandSize(STableMeta* pTableMeta) {
     if (IS_VAR_DATA_TYPE((pSchema + i)->type)) {
       result += TYPE_BYTES[TSDB_DATA_TYPE_BINARY];
     }
-    result += sizeof(SColIdx);
   }
-  result += TD_MEM_ROW_TYPE_SIZE;  // add len of SMemRow flag
+  result += TD_MEM_ROW_KV_TYPE_VER_SIZE;  // add prefix len of KV type SMemRow(we may use SDataRow or SKVRow)
   return result;
 }
 
@@ -1838,13 +1836,16 @@ static void extractTableNameList(SInsertStatementParam *pInsertParam, bool freeB
 
 int32_t tscMergeTableDataBlocks(SInsertStatementParam *pInsertParam, bool freeBlockMap) {
   const int INSERT_HEAD_SIZE = sizeof(SMsgDesc) + sizeof(SSubmitMsg);
-
+  int       code = 0;
   void*     pVnodeDataBlockHashList = taosHashInit(128, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BIGINT), true, false);
   SArray* pVnodeDataBlockList = taosArrayInit(8, POINTER_BYTES);
 
   STableDataBlocks** p = taosHashIterate(pInsertParam->pTableBlockHashList, NULL);
 
   STableDataBlocks* pOneTableBlock = *p;
+
+  SBlockKeyInfo blkKeyInfo = {0};  // share by pOneTableBlock
+  
   while(pOneTableBlock) {
     SSubmitBlk* pBlocks = (SSubmitBlk*) pOneTableBlock->pData;
     if (pBlocks->numOfRows > 0) {
@@ -1858,6 +1859,7 @@ int32_t tscMergeTableDataBlocks(SInsertStatementParam *pInsertParam, bool freeBl
         tscError("0x%"PRIx64" failed to prepare the data block buffer for merging table data, code:%d", pInsertParam->objectId, ret);
         taosHashCleanup(pVnodeDataBlockHashList);
         tscDestroyBlockArrayList(pVnodeDataBlockList);
+        tfree(blkKeyInfo.pKeyTuple);
         return ret;
       }
 
@@ -1878,16 +1880,26 @@ int32_t tscMergeTableDataBlocks(SInsertStatementParam *pInsertParam, bool freeBl
           taosHashCleanup(pVnodeDataBlockHashList);
           tscDestroyBlockArrayList(pVnodeDataBlockList);
           tfree(dataBuf->pData);
+          tfree(blkKeyInfo.pKeyTuple);
 
           return TSDB_CODE_TSC_OUT_OF_MEMORY;
         }
       }
 
-      tscSortRemoveDataBlockDupRows(pOneTableBlock);
-      char* ekey = (char*)pBlocks->data + pOneTableBlock->rowSize*(pBlocks->numOfRows-1);
+      if((code = tscSortRemoveDataBlockDupRows(pOneTableBlock, &blkKeyInfo)) != 0){
+          taosHashCleanup(pVnodeDataBlockHashList);
+          tscDestroyBlockArrayList(pVnodeDataBlockList);
+          tfree(dataBuf->pData);
+          tfree(blkKeyInfo.pKeyTuple);
+        return code;
+      }
+
+      ASSERT(blkKeyInfo.pKeyTuple != NULL && pBlocks->numOfRows > 0);
 
-      tscDebug("0x%"PRIx64" name:%s, tid:%d rows:%d sversion:%d skey:%" PRId64 ", ekey:%" PRId64, pInsertParam->objectId, tNameGetTableName(&pOneTableBlock->tableName),
-          pBlocks->tid, pBlocks->numOfRows, pBlocks->sversion, GET_INT64_VAL(pBlocks->data), GET_INT64_VAL(ekey));
+      SBlockKeyTuple* pLastKeyTuple = blkKeyInfo.pKeyTuple + pBlocks->numOfRows - 1;
+      tscDebug("0x%" PRIx64 " name:%s, tid:%d rows:%d sversion:%d skey:%" PRId64 ", ekey:%" PRId64,
+               pInsertParam->objectId, tNameGetTableName(&pOneTableBlock->tableName), pBlocks->tid, pBlocks->numOfRows,
+               pBlocks->sversion, blkKeyInfo.pKeyTuple->skey, pLastKeyTuple->skey);
 
       int32_t len = pBlocks->numOfRows * (pOneTableBlock->rowSize + expandSize) + sizeof(STColumn) * tscGetNumOfColumns(pOneTableBlock->pTableMeta);
 
@@ -1898,7 +1910,7 @@ int32_t tscMergeTableDataBlocks(SInsertStatementParam *pInsertParam, bool freeBl
       pBlocks->schemaLen = 0;
 
       // erase the empty space reserved for binary data
-      int32_t finalLen = trimDataBlock(dataBuf->pData + dataBuf->size, pOneTableBlock, pInsertParam->schemaAttached);
+      int32_t finalLen = trimDataBlock(dataBuf->pData + dataBuf->size, pOneTableBlock, pInsertParam->schemaAttached, blkKeyInfo.pKeyTuple);
       assert(finalLen <= len);
 
       dataBuf->size += (finalLen + sizeof(SSubmitBlk));
@@ -1926,6 +1938,7 @@ int32_t tscMergeTableDataBlocks(SInsertStatementParam *pInsertParam, bool freeBl
   // free the table data blocks;
   pInsertParam->pDataBlocks = pVnodeDataBlockList;
   taosHashCleanup(pVnodeDataBlockHashList);
+  tfree(blkKeyInfo.pKeyTuple);
 
   return TSDB_CODE_SUCCESS;
 }

src/common/inc/tdataformat.h

@@ -24,6 +24,33 @@
 extern "C" {
 #endif
 
+#pragma pack(push, 1)
+typedef struct {
+  VarDataLenT len;
+  uint8_t     data;
+} SBinaryNullT;
+
+typedef struct {
+  VarDataLenT len;
+  uint32_t    data;
+} SNCharNullT;
+#pragma pack(pop)
+
+extern const uint8_t      BoolNull;
+extern const uint8_t      TinyintNull;
+extern const uint16_t     SmallintNull;
+extern const uint32_t     IntNull;
+extern const uint64_t     BigintNull;
+extern const uint64_t     TimestampNull;
+extern const uint8_t      UTinyintNull;
+extern const uint16_t     USmallintNull;
+extern const uint32_t     UIntNull;
+extern const uint64_t     UBigintNull;
+extern const uint32_t     FloatNull;
+extern const uint64_t     DoubleNull;
+extern const SBinaryNullT BinaryNull;
+extern const SNCharNullT  NcharNull;
+
 #define STR_TO_VARSTR(x, str)                     \
   do {                                            \
     VarDataLenT __len = (VarDataLenT)strlen(str); \
@@ -207,7 +234,7 @@ SDataRow tdDataRowDup(SDataRow row);
 SMemRow tdMemRowDup(SMemRow row);
 
 // offset here not include dataRow header length
-static FORCE_INLINE int tdAppendColVal(SDataRow row, void *value, int8_t type, int32_t bytes, int32_t offset) {
+static FORCE_INLINE int tdAppendColVal(SDataRow row, const void *value, int8_t type, int32_t offset) {
   ASSERT(value != NULL);
   int32_t toffset = offset + TD_DATA_ROW_HEAD_SIZE;
   char *  ptr = (char *)POINTER_SHIFT(row, dataRowLen(row));
@@ -260,6 +287,42 @@ void dataColSetOffset(SDataCol *pCol, int nEle);
 bool isNEleNull(SDataCol *pCol, int nEle);
 void dataColSetNEleNull(SDataCol *pCol, int nEle, int maxPoints);
 
+static FORCE_INLINE const void *tdGetNullVal(int8_t type) {
+  switch (type) {
+    case TSDB_DATA_TYPE_BOOL:
+      return &BoolNull;
+    case TSDB_DATA_TYPE_TINYINT:
+      return &TinyintNull;
+    case TSDB_DATA_TYPE_SMALLINT:
+      return &SmallintNull;
+    case TSDB_DATA_TYPE_INT:
+      return &IntNull;
+    case TSDB_DATA_TYPE_BIGINT:
+      return &BigintNull;
+    case TSDB_DATA_TYPE_FLOAT:
+      return &FloatNull;
+    case TSDB_DATA_TYPE_DOUBLE:
+      return &DoubleNull;
+    case TSDB_DATA_TYPE_BINARY:
+      return &BinaryNull;
+    case TSDB_DATA_TYPE_TIMESTAMP:
+      return &TimestampNull;
+    case TSDB_DATA_TYPE_NCHAR:
+      return &NcharNull;
+    case TSDB_DATA_TYPE_UTINYINT:
+      return &UTinyintNull;
+    case TSDB_DATA_TYPE_USMALLINT:
+      return &USmallintNull;
+    case TSDB_DATA_TYPE_UINT:
+      return &UIntNull;
+    case TSDB_DATA_TYPE_UBIGINT:
+      return &UBigintNull;
+    default:
+      ASSERT(0);
+      return NULL;
+  }
+}
+
 // Get the data pointer from a column-wised data
 static FORCE_INLINE void *tdGetColDataOfRow(SDataCol *pCol, int row) {
   if (IS_VAR_DATA_TYPE(pCol->type)) {
@@ -545,6 +608,47 @@ static FORCE_INLINE void *tdGetMemRowDataOfCol(void *row, int8_t type, int32_t o
   return NULL;
 }
 
+//  RawRow payload structure:
+//  |<---------- header ------------->|<---- body: column data tuple ---->|
+//  |SMemRowType|  dataLen |  nCols   |  colId  | colType | value |...|...|
+//  +-----------+----------+----------+---------------------------------->|
+//  | uint8_t   | uint16_t | uint16_t | int16_t | uint8_t |  ???  |...|...|
+//  +-----------+----------+----------+---------------------------------->|
+
+#define PAYLOAD_NCOLS_LEN sizeof(uint16_t)
+#define PAYLOAD_NCOLS_OFFSET (sizeof(uint8_t) + sizeof(TDRowLenT))
+#define PAYLOAD_HEADER_LEN (PAYLOAD_NCOLS_OFFSET + PAYLOAD_NCOLS_LEN)
+#define PAYLOAD_ID_LEN sizeof(int16_t)
+#define PAYLOAD_ID_TYPE_LEN (sizeof(int16_t) + sizeof(uint8_t))
+
+#define payloadBody(r) POINTER_SHIFT(r, PAYLOAD_HEADER_LEN)
+#define payloadType(r) (*(uint8_t *)(r))
+#define payloadSetType(r, t) (payloadType(r) = (t))
+#define payloadTLen(r) (*(TDRowLenT *)POINTER_SHIFT(r, TD_MEM_ROW_TYPE_SIZE))  // including total header
+#define payloadSetTLen(r, l) (payloadTLen(r) = (l))
+#define payloadNCols(r) (*(TDRowLenT *)POINTER_SHIFT(r, PAYLOAD_NCOLS_OFFSET))
+#define payloadSetNCols(r, n) (payloadNCols(r) = (n))
+
+#define payloadColId(r) (*(int16_t *)(r))
+#define payloadColType(r) (*(uint8_t *)POINTER_SHIFT(r, PAYLOAD_ID_LEN))
+#define payloadColValue(r) POINTER_SHIFT(r, PAYLOAD_ID_TYPE_LEN)
+
+#define payloadColSetId(r, i) (payloadColId(r) = (i))
+#define payloadColSetType(r, t) (payloadColType(r) = (t))
+
+#define payloadKeyAddr(r) POINTER_SHIFT(r, PAYLOAD_HEADER_LEN + PAYLOAD_ID_TYPE_LEN)
+#define payloadTKey(r) (*(TKEY *)(payloadKeyAddr(r)))
+#define payloadKey(r) tdGetKey(payloadTKey(r))
+
+static FORCE_INLINE char *skipToNextEles(char *p) {
+  uint8_t colType = payloadColType(p);
+  if (IS_VAR_DATA_TYPE(colType)) {
+    return POINTER_SHIFT(p, PAYLOAD_ID_TYPE_LEN + varDataTLen(payloadColValue(p)));
+  } else {
+    return POINTER_SHIFT(p, PAYLOAD_ID_TYPE_LEN + TYPE_BYTES[colType]);
+  }
+}
+
 #ifdef __cplusplus
 }
 #endif

src/common/src/tdataformat.c

@@ -18,6 +18,21 @@
 #include "tcoding.h"
 #include "wchar.h"
 
+const uint8_t      BoolNull = TSDB_DATA_BOOL_NULL;
+const uint8_t      TinyintNull = TSDB_DATA_TINYINT_NULL;
+const uint16_t     SmallintNull = TSDB_DATA_SMALLINT_NULL;
+const uint32_t     IntNull = TSDB_DATA_INT_NULL;
+const uint64_t     BigintNull = TSDB_DATA_BIGINT_NULL;
+const uint64_t     TimestampNull = TSDB_DATA_BIGINT_NULL;
+const uint8_t      UTinyintNull = TSDB_DATA_UTINYINT_NULL;
+const uint16_t     USmallintNull = TSDB_DATA_USMALLINT_NULL;
+const uint32_t     UIntNull = TSDB_DATA_UINT_NULL;
+const uint64_t     UBigintNull = TSDB_DATA_UBIGINT_NULL;
+const uint32_t     FloatNull = TSDB_DATA_FLOAT_NULL;
+const uint64_t     DoubleNull = TSDB_DATA_DOUBLE_NULL;
+const SBinaryNullT BinaryNull = {1, TSDB_DATA_BINARY_NULL};
+const SNCharNullT  NcharNull = {4, TSDB_DATA_NCHAR_NULL};
+
 static void tdMergeTwoDataCols(SDataCols *target, SDataCols *src1, int *iter1, int limit1, SDataCols *src2, int *iter2,
                                int limit2, int tRows);
 

src/cq/src/cqMain.c

@@ -502,7 +502,7 @@ static void cqProcessStreamRes(void *param, TAOS_RES *tres, TAOS_ROW row) {
       memcpy((char *)val + sizeof(VarDataLenT), buf, len);
       varDataLen(val) = len;
     }
-    tdAppendColVal(dataRow, val, c->type, c->bytes, c->offset);
+    tdAppendColVal(dataRow, val, c->type, c->offset);
   }
   pBlk->dataLen = htonl(memRowDataTLen(trow));
   pBlk->schemaLen = 0;

src/tsdb/src/tsdbMain.c

@@ -720,7 +720,7 @@ static int tsdbRestoreLastColumns(STsdbRepo *pRepo, STable *pTable, SReadH* pRea
       // OK,let's load row from backward to get not-null column
       for (int32_t rowId = pBlock->numOfRows - 1; rowId >= 0; rowId--) {
         SDataCol *pDataCol = pReadh->pDCols[0]->cols + i;
-        tdAppendColVal(memRowDataBody(row), tdGetColDataOfRow(pDataCol, rowId), pCol->type, pCol->bytes, pCol->offset);
+        tdAppendColVal(memRowDataBody(row), tdGetColDataOfRow(pDataCol, rowId), pCol->type, pCol->offset);
         //SDataCol *pDataCol = readh.pDCols[0]->cols + j;
         void *value = tdGetRowDataOfCol(memRowDataBody(row), (int8_t)pCol->type, TD_DATA_ROW_HEAD_SIZE + pCol->offset);
         if (isNull(value, pCol->type)) {
@@ -742,7 +742,7 @@ static int tsdbRestoreLastColumns(STsdbRepo *pRepo, STable *pTable, SReadH* pRea
         // save row ts(in column 0)
         pDataCol = pReadh->pDCols[0]->cols + 0;
         pCol = schemaColAt(pSchema, 0);
-        tdAppendColVal(memRowDataBody(row), tdGetColDataOfRow(pDataCol, rowId), pCol->type, pCol->bytes, pCol->offset);
+        tdAppendColVal(memRowDataBody(row), tdGetColDataOfRow(pDataCol, rowId), pCol->type, pCol->offset);
         pLastCol->ts = memRowKey(row);
 
         pTable->restoreColumnNum += 1;
@@ -790,7 +790,7 @@ static int tsdbRestoreLastRow(STsdbRepo *pRepo, STable *pTable, SReadH* pReadh,
     STColumn *pCol = schemaColAt(pSchema, icol);
     SDataCol *pDataCol = pReadh->pDCols[0]->cols + icol;
     tdAppendColVal(memRowDataBody(pTable->lastRow), tdGetColDataOfRow(pDataCol, pBlock->numOfRows - 1), pCol->type,
-                   pCol->bytes, pCol->offset);
+                   pCol->offset);
   }
 
   return 0;

src/tsdb/tests/tsdbTests.cpp

@@ -55,10 +55,10 @@ static int insertData(SInsertInfo *pInfo) {
       for (int j = 0; j < schemaNCols(pInfo->pSchema); j++) {
         STColumn *pTCol = schemaColAt(pInfo->pSchema, j);
         if (j == 0) {  // Just for timestamp
-          tdAppendColVal(row, (void *)(&start_time), pTCol->type, pTCol->bytes, pTCol->offset);
+          tdAppendColVal(row, (void *)(&start_time), pTCol->type, pTCol->offset);
         } else {  // For int
           int val = 10;
-          tdAppendColVal(row, (void *)(&val), pTCol->type, pTCol->bytes, pTCol->offset);
+          tdAppendColVal(row, (void *)(&val), pTCol->type, pTCol->offset);
         }
       }
       pBlock->dataLen += dataRowLen(row);