change signature array in a std::set ! lookup performance is now log(N). About 10x performance can be expected on cpimport containing varchars.

Signed-off-by: Patrice Linel <plinel@mendel-master2.cm.cluster>

writeengine/bulk/we_colbufmgr.cpp

@@ -45,7 +45,7 @@ namespace
 {
 // Minimum time to wait for a condition, so as to periodically wake up and
 // check the global job status, to see if the job needs to terminate.
-const int COND_WAIT_SECONDS = 3;
+const int COND_WAIT_SECONDS = 1;
 }
 
 namespace WriteEngine

writeengine/dictionary/we_dctnry.cpp

@@ -29,6 +29,7 @@
 #include <cstdlib>
 #include <cstring>
 #include <vector>
+#include <set>
 #include <sstream>
 #include <inttypes.h>
 #include <iostream>
@@ -108,7 +109,6 @@ Dctnry::Dctnry() :
            &m_endHeader,  HDR_UNIT_SIZE);
     m_curFbo  = INVALID_NUM;
     m_curLbid = INVALID_LBID;
-    memset(m_sigArray, 0, MAX_STRING_CACHE_SIZE * sizeof(Signature));
     m_arraySize = 0;
 
     clear();//files
@@ -130,14 +130,16 @@ Dctnry::~Dctnry()
  ******************************************************************************/
 void Dctnry::freeStringCache( )
 {
-    for (int i = 0; i < m_arraySize; i++)
+    std::set<Signature,sig_compare>::iterator it;
+    for (it=m_sigArray.begin(); it!=m_sigArray.end(); it++)
     {
-        delete [] m_sigArray[i].signature;
-        m_sigArray[i].signature = 0;
+	Signature sig  = *it;
+        delete [] sig.signature;
+        sig.signature = 0;
     }
 
-    memset(m_sigArray, 0, MAX_STRING_CACHE_SIZE * sizeof(Signature));
     m_arraySize = 0;
+    m_sigArray.clear();
 }
 
 /*******************************************************************************
@@ -161,7 +163,6 @@ int  Dctnry::init()
     m_curOp = 0;
     memset( m_curBlock.data, 0, sizeof(m_curBlock.data));
     m_curBlock.lbid = INVALID_LBID;
-    memset(m_sigArray, 0, MAX_STRING_CACHE_SIZE * sizeof(Signature));
     m_arraySize = 0;
 
     return NO_ERROR;
@@ -623,16 +624,14 @@ int Dctnry::openDctnry(const OID& dctnryOID,
  ******************************************************************************/
 bool Dctnry::getTokenFromArray(Signature& sig)
 {
-    for (int i = 0; i < (int)m_arraySize ; i++ )
-    {
-        if (sig.size == m_sigArray[i].size)
-        {
-            if (!memcmp(sig.signature, m_sigArray[i].signature, sig.size))
-            {
-                sig.token = m_sigArray[i].token;
+	std::set<Signature,sig_compare>::iterator it;
+	it = m_sigArray.find(sig);
+	if ( it  == m_sigArray.end()){
+		return false;
+	}else{
+		Signature sigfound  = *it;
+		sig.token = sigfound.token;
 		return true;
-            }//endif sig compare
-        }//endif size compare
 	}
 
 	return false;
@@ -1329,7 +1328,7 @@ void  Dctnry::preLoadStringCache( const DataBlock& fileBlock )
         memcpy(aSig.signature, &fileBlock.data[offBeg], len);
         aSig.token.op    = op;
         aSig.token.fbo   = m_curLbid;
-        m_sigArray[op - 1] = aSig;
+        m_sigArray.insert(aSig);
 
         offEnd           = offBeg;
         hdrOffsetBeg    += HDR_UNIT_SIZE;
@@ -1368,7 +1367,7 @@ void  Dctnry::addToStringCache( const Signature& newSig )
     memcpy(asig.signature, newSig.signature, newSig.size );
     asig.size      = newSig.size;
     asig.token     = newSig.token;
-    m_sigArray[m_arraySize] = asig;
+    m_sigArray.insert(asig);
     m_arraySize++;
 }
 
@@ -1461,7 +1460,7 @@ int  Dctnry::updateDctnry(unsigned char* sigValue, int& sigSize,
         sig.signature = new unsigned char[sigSize];
         memcpy (sig.signature, sigValue, sigSize);
         sig.token = token;
-        m_sigArray[m_arraySize] = sig;
+        m_sigArray.insert(sig);
         m_arraySize++;
     }
 

writeengine/dictionary/we_dctnry.h

@@ -56,6 +56,17 @@ typedef struct Signature
     Token token;
 } Signature;
 
+struct sig_compare {
+	bool operator() (const Signature& a, const Signature& b) const {
+		if (a.size == b.size){
+			return memcmp(a.signature,b.signature,a.size)<0;}
+		else if (a.size<b.size){
+			return true;
+		}else{ return false;}
+	}
+}; 
+
+
 /**
  * @brief Class to interface with dictionary store files.
  */
@@ -285,7 +296,7 @@ protected:
     virtual void  closeDctnryFile(bool doFlush, std::map<FID, FID>& oids);
     virtual int   numOfBlocksInFile();
 
-    Signature    m_sigArray[MAX_STRING_CACHE_SIZE]; // string cache
+    std::set<Signature,sig_compare> m_sigArray;
     int          m_arraySize;                       // num strings in m_sigArray
 
     // m_dctnryHeader  used for hdr when readSubBlockEntry is used to read a blk

writeengine/dictionary/we_dctnrystore.cpp

@@ -133,7 +133,7 @@ const int  DctnryStore::updateDctnryStore(unsigned char* sigValue,
         sig.signature = new unsigned char[sigSize];
         memcpy (sig.signature, sigValue, sigSize);
         sig.token = token;
-        m_dctnry.m_sigArray[m_dctnry.m_arraySize] = sig;
+        m_dctnry.m_sigArray.insert(sig) = sig;
         m_dctnry.m_arraySize++;
     }