diff --git a/src/InputMgr.cc b/src/InputMgr.cc
index 648a933a22..5270c83d5a 100644
--- a/src/InputMgr.cc
+++ b/src/InputMgr.cc
@@ -13,6 +13,18 @@
 
 #include "InputReaderAscii.h"
 
+#include "CompHash.h"
+
+
+class InputHash {
+public:
+	HashKey* valhash;
+	HashKey* idxkey; // does not need ref or whatever - if it is present here, it is also still present in the TableVal.
+};
+
+declare(PDict, InputHash);
+
+
 struct InputMgr::ReaderInfo {
 	EnumVal* id;
 	EnumVal* type;
@@ -24,6 +36,9 @@ struct InputMgr::ReaderInfo {
 	RecordType* rtype;
 	RecordType* itype;
 
+	PDict(InputHash)* currDict;
+	PDict(InputHash)* lastDict;
+
 	};
 
 struct InputReaderDefinition {
@@ -147,10 +162,20 @@ InputReader* InputMgr::CreateReader(EnumVal* id, RecordVal* description)
 	info->itype = idx;
 	Ref(idx);
 	readers.push_back(info);
+	info->currDict = new PDict(InputHash);
+	info->lastDict = new PDict(InputHash);
 
 
-	reader_obj->Init(source, fieldsV.size(), fields);
-	reader_obj->Update();
+	int success = reader_obj->Init(source, fieldsV.size(), idxfields, fields);
+	if ( success == false ) {
+		RemoveReader(id);
+		return 0;
+	}
+	success = reader_obj->Update();
+	if ( success == false ) {
+		RemoveReader(id);
+		return 0;
+	}
 	
 	return reader_obj;
 	
@@ -198,7 +223,6 @@ bool InputMgr::IsCompatibleType(BroType* t)
 	return false;
 	}
 
-
 bool InputMgr::RemoveReader(EnumVal* id) {
 	ReaderInfo *i = 0;
 	for ( vector<ReaderInfo *>::iterator s = readers.begin(); s != readers.end(); ++s )
@@ -215,6 +239,9 @@ bool InputMgr::RemoveReader(EnumVal* id) {
 		return false; // not found
 	}
 
+	i->reader->Finish();
+
+
 	Unref(i->type);
 	Unref(i->tab);
 	Unref(i->itype);
@@ -267,8 +294,7 @@ bool InputMgr::ForceUpdate(EnumVal* id)
 		return false;
 	}
 
-	i->reader->Update();
-	return true;
+	return i->reader->Update();
 }
 
 Val* InputMgr::LogValToIndexVal(int num_fields, const RecordType *type, const LogVal* const *vals) {
@@ -297,6 +323,105 @@ Val* InputMgr::LogValToIndexVal(int num_fields, const RecordType *type, const Lo
 
 }
 
+
+void InputMgr::SendEntry(const InputReader* reader, const LogVal* const *vals) {
+	ReaderInfo *i = FindReader(reader);
+	if ( i == 0 ) {
+		reporter->InternalError("Unknown reader");
+		return;
+	}
+
+	HashKey* idxhash = HashLogVals(i->num_idx_fields, vals);
+	HashKey* valhash = HashLogVals(i->num_val_fields, vals+i->num_idx_fields);
+
+	InputHash *h = i->lastDict->Lookup(idxhash);
+	if ( h != 0 ) {
+		// seen before
+		if ( h->valhash->Hash() == valhash->Hash() ) {
+			// ok, double.
+			i->lastDict->Remove(idxhash);
+			i->currDict->Insert(idxhash, h);
+			return;
+		} else {
+			// updated
+			i->lastDict->Remove(idxhash);
+			delete(h);
+		}
+	}
+
+
+	Val* idxval = LogValToIndexVal(i->num_idx_fields, i->itype, vals);
+	Val* valval;
+	
+	int position = i->num_idx_fields;
+	if ( i->num_val_fields == 1 ) {
+		valval = LogValToVal(vals[i->num_idx_fields]);
+	} else {
+		RecordVal * r = new RecordVal(i->rtype);
+
+		/* if ( i->rtype->NumFields() != (int) i->num_val_fields ) {
+			reporter->InternalError("Type mismatch");
+			return;
+		} */
+
+		for ( int j = 0; j < i->rtype->NumFields(); j++) {
+
+			Val* val = 0;
+			if ( i->rtype->FieldType(j)->Tag() == TYPE_RECORD ) {
+				val = LogValToRecordVal(vals, i->rtype->FieldType(j)->AsRecordType(), &position);
+			} else {
+				val =  LogValToVal(vals[position], i->rtype->FieldType(j)->Tag());
+				position++;
+			}
+			
+			if ( val == 0 ) {
+				reporter->InternalError("conversion error");
+				return;
+			}
+
+			r->Assign(j,val);
+
+		}
+		valval = r;
+	}
+
+	//i->tab->Assign(idxval, valval);
+	HashKey* k = i->tab->ComputeHash(idxval);
+	if ( !k ) {
+		reporter->InternalError("could not hash");
+		return;
+	}
+
+	i->tab->Assign(idxval, k, valval);
+	InputHash* ih = new InputHash();
+	ih->idxkey = k;
+	ih->valhash = valhash;
+
+	i->currDict->Insert(idxhash, ih);
+
+}
+
+void InputMgr::EndCurrentSend(const InputReader* reader) {
+	ReaderInfo *i = FindReader(reader);
+	if ( i == 0 ) {
+		reporter->InternalError("Unknown reader");
+		return;
+	}
+
+	// lastdict contains all deleted entries
+	IterCookie *c = i->lastDict->InitForIteration();
+	InputHash* ih;
+	while ( ( ih = i->lastDict->NextEntry(c )) ) {
+		i->tab->Delete(ih->idxkey);
+	}
+
+	i->lastDict->Clear();
+	delete(i->lastDict);
+
+	i->lastDict = i->currDict;	
+	i->currDict = new PDict(InputHash);
+}
+
 void InputMgr::Put(const InputReader* reader, const LogVal* const *vals) {
 	ReaderInfo *i = FindReader(reader);
 	if ( i == 0 ) {
@@ -418,6 +543,95 @@ Val* InputMgr::LogValToRecordVal(const LogVal* const *vals, RecordType *request_
 
 } 
 
+HashKey* InputMgr::HashLogVals(const int num_elements, const LogVal* const *vals) {
+	int length = 0;
+
+	for ( int i = 0; i < num_elements; i++ ) {
+		const LogVal* val = vals[i];
+		switch (val->type) {
+		case TYPE_BOOL:
+		case TYPE_INT:
+			length += sizeof(val->val.int_val);
+			break;
+
+		case TYPE_COUNT:
+		case TYPE_COUNTER:
+		case TYPE_PORT:
+			length += sizeof(val->val.uint_val);
+		break;
+		
+		case TYPE_DOUBLE:
+		case TYPE_TIME:
+		case TYPE_INTERVAL:
+			length += sizeof(val->val.double_val);
+			break;
+
+		case TYPE_STRING:
+			{
+			length += val->val.string_val->size();
+			break;
+			}
+	
+		case TYPE_ADDR:
+			length += NUM_ADDR_WORDS*sizeof(uint32_t);
+			break;
+
+		default:
+			reporter->InternalError("unsupported type for hashlogvals");
+		}
+		
+	}
+
+	int position = 0;
+	char *data = (char*) malloc(length);
+	for ( int i = 0; i < num_elements; i++ ) {
+		const LogVal* val = vals[i];
+		switch ( val->type ) {
+		case TYPE_BOOL:
+		case TYPE_INT:
+			*(data+position) = val->val.int_val;
+			position += sizeof(val->val.int_val);
+			break;
+
+		case TYPE_COUNT:
+		case TYPE_COUNTER:
+		case TYPE_PORT:
+			*(data+position) = val->val.uint_val;
+			position += sizeof(val->val.uint_val);
+			break;
+	
+		case TYPE_DOUBLE:
+		case TYPE_TIME:
+		case TYPE_INTERVAL:
+			*(data+position) = val->val.double_val;
+			position += sizeof(val->val.double_val);
+			break;
+
+		case TYPE_STRING:
+			{
+			memcpy(data+position, val->val.string_val->c_str(), val->val.string_val->length());
+			position += val->val.string_val->size();
+			break;
+			}
+	
+		case TYPE_ADDR:
+			memcpy(data+position, val->val.addr_val, NUM_ADDR_WORDS*sizeof(uint32_t));
+			position += NUM_ADDR_WORDS*sizeof(uint32_t);
+			break;
+
+		default:
+			reporter->InternalError("unsupported type for hashlogvals2");
+		}
+
+		
+	}
+
+	assert(position == length);
+	return new HashKey(data, length);
+
+
+}
+
 Val* InputMgr::LogValToVal(const LogVal* val, TypeTag request_type) {
 	
 	if ( request_type != TYPE_ANY && request_type != val->type ) {
@@ -495,3 +709,11 @@ InputMgr::ReaderInfo* InputMgr::FindReader(const EnumVal* id)
 	}
 
 
+string InputMgr::Hash(const string &input) {
+	unsigned char digest[16];
+	hash_md5(input.length(), (const unsigned char*) input.c_str(), digest);
+	string out((const char*) digest, 16);
+	return out;
+}
+
+
diff --git a/src/InputMgr.h b/src/InputMgr.h
index d5f732935c..d147fa262a 100644
--- a/src/InputMgr.h
+++ b/src/InputMgr.h
@@ -15,9 +15,10 @@
 
 class InputReader;
 
+
 class InputMgr {
 public:
-    InputMgr();
+	InputMgr();
     
     	InputReader* CreateReader(EnumVal* id, RecordVal* description);
 	bool ForceUpdate(EnumVal* id);
@@ -32,6 +33,9 @@ protected:
 	void Put(const InputReader* reader, const LogVal* const *vals);
 	void Clear(const InputReader* reader);
 	bool Delete(const InputReader* reader, const LogVal* const *vals);
+
+	void SendEntry(const InputReader* reader, const LogVal* const *vals);
+	void EndCurrentSend(const InputReader* reader);
 	
 private:
 	struct ReaderInfo;
@@ -40,6 +44,8 @@ private:
 
 	bool UnrollRecordType(vector<LogField*> *fields, const RecordType *rec, const string& nameprepend);
 
+	HashKey* HashLogVals(const int num_elements, const LogVal* const *vals);
+
 	Val* LogValToVal(const LogVal* val, TypeTag request_type = TYPE_ANY);
 	Val* LogValToIndexVal(int num_fields, const RecordType* type, const LogVal* const *vals);
 	Val* LogValToRecordVal(const LogVal* const *vals, RecordType *request_type, int* position);	
@@ -49,6 +55,9 @@ private:
 	ReaderInfo* FindReader(const EnumVal* id);
 
 	vector<ReaderInfo*> readers;
+
+	string Hash(const string &input);	
+
 };
 
 extern InputMgr* input_mgr;
diff --git a/src/InputReader.cc b/src/InputReader.cc
index 1facc57c7f..494df3fb81 100644
--- a/src/InputReader.cc
+++ b/src/InputReader.cc
@@ -41,20 +41,22 @@ void InputReader::Delete(const LogVal* const *val)
 }
 
 
-bool InputReader::Init(string arg_source, int arg_num_fields,
+bool InputReader::Init(string arg_source, int arg_num_fields, int arg_idx_fields,
 					   const LogField* const * arg_fields) 
 {
 	source = arg_source;
 	num_fields = arg_num_fields;
+	index_fields = arg_idx_fields;
 	fields = arg_fields;
 
 	// disable if DoInit returns error.
-	disabled = !DoInit(arg_source, arg_num_fields, arg_fields);
+	disabled = !DoInit(arg_source, arg_num_fields, arg_idx_fields, arg_fields);
 	return !disabled;
 }
 
 void InputReader::Finish() {
 	DoFinish();
+	disabled = true;
 }
 
 bool InputReader::Update() {
@@ -91,3 +93,10 @@ const char* InputReader::Fmt(const char* format, ...)
 	}
 
 
+void InputReader::SendEntry(const LogVal* const *vals) {
+	input_mgr->SendEntry(this, vals);
+}
+
+void InputReader::EndCurrentSend() {
+	input_mgr->EndCurrentSend(this);
+}
diff --git a/src/InputReader.h b/src/InputReader.h
index 0e93344e1a..b547d29506 100644
--- a/src/InputReader.h
+++ b/src/InputReader.h
@@ -15,7 +15,7 @@ public:
     InputReader();
     virtual ~InputReader();
 	
-	bool Init(string arg_source, int num_fields, const LogField* const* fields);
+	bool Init(string arg_source, int arg_num_fields, int arg_idx_fields, const LogField* const* fields);
     
 	void Finish();
 
@@ -23,7 +23,7 @@ public:
 	
 protected:
     // Methods that have to be overwritten by the individual readers
-	virtual bool DoInit(string arg_source, int num_fields, const LogField* const * fields) = 0;
+	virtual bool DoInit(string arg_source, int arg_num_fields, int arg_idx_fields, const LogField* const * fields) = 0;
 	
 	virtual void DoFinish() = 0;
 
@@ -46,11 +46,16 @@ protected:
 	void Clear();
 	void Delete(const LogVal* const *val);
 
+	void SendEntry(const LogVal* const *vals);
+	void EndCurrentSend();
+	
+
 private:
     friend class InputMgr;
 	
 	string source;
 	int num_fields;
+	int index_fields;
 	const LogField* const * fields;
     
     // When an error occurs, this method is called to set a flag marking the 
diff --git a/src/InputReaderAscii.cc b/src/InputReaderAscii.cc
index e4d581c0d8..e434f7e750 100644
--- a/src/InputReaderAscii.cc
+++ b/src/InputReaderAscii.cc
@@ -21,17 +21,26 @@ InputReaderAscii::InputReaderAscii()
 {
 	//DBG_LOG(DBG_LOGGING, "input reader initialized");
 	file = 0;
+
+	//keyMap = new map<string, string>();
 }
 
 InputReaderAscii::~InputReaderAscii()
 {
+	DoFinish();
 }
 
 void InputReaderAscii::DoFinish()
 {
+	columnMap.empty();
+	if ( file != 0 ) {
+		file->close();
+		delete(file);
+		file = 0;
+	}
 }
 
-bool InputReaderAscii::DoInit(string path, int num_fields, const LogField* const * fields)
+bool InputReaderAscii::DoInit(string path, int num_fields, int idx_fields, const LogField* const * fields)
 {
 	fname = path;
 	
@@ -47,6 +56,9 @@ bool InputReaderAscii::DoInit(string path, int num_fields, const LogField* const
 		Error("could not read first line");
 		return false;
 	}
+
+	this->num_fields = num_fields;
+	this->idx_fields = idx_fields;
 	 
 	// split on tabs...
 	istringstream splitstream(line);
@@ -83,12 +95,10 @@ bool InputReaderAscii::DoInit(string path, int num_fields, const LogField* const
 	if ( wantFields != num_fields ) {
 		// we did not find all fields?
 		// :(
-		Error("wantFields != num_fields");
+		Error("One of the requested fields could not be found in the input data file");
 		return false;
 	}
 	
-
-	this->num_fields = num_fields;
 	
 	// well, that seems to have worked...
 	return true;
@@ -101,6 +111,9 @@ bool InputReaderAscii::DoUpdate() {
 	//
 	
 
+	// new keymap
+	//map<string, string> *newKeyMap = new map<string, string>();
+
 	string line;
 	while ( getline(*file, line ) ) {
 		// split on tabs
@@ -109,10 +122,12 @@ bool InputReaderAscii::DoUpdate() {
 		string s;
 	
 		LogVal** fields = new LogVal*[num_fields];
+		//string string_fields[num_fields];
 
 		unsigned int currTab = 0;
 		unsigned int currField = 0;
 		while ( splitstream ) {
+
 			if ( !getline(splitstream, s, '\t') )
 				break;
 
@@ -146,8 +161,11 @@ bool InputReaderAscii::DoUpdate() {
 			case TYPE_BOOL:
 				if ( s == "T" ) {
 					val->val.int_val = 1;
-				} else {
+				} else if ( s == "F" ) {
 					val->val.int_val = 0;
+				} else {
+					Error(Fmt("Invalid value for boolean: %s", s.c_str()));
+					return false;
 				}
 				break;
 
@@ -173,9 +191,15 @@ bool InputReaderAscii::DoUpdate() {
 				val->val.subnet_val.width = atoi(width.c_str());
 				string addr = s.substr(0, pos);
 				s = addr;
-				// fallthrough
+				// NOTE: dottet_to_addr BREAKS THREAD SAFETY! it uses reporter.
+				// Solve this some other time....
+				val->val.subnet_val.net = dotted_to_addr(s.c_str());
+				break;
+
 				}
 			case TYPE_ADDR: {
+				// NOTE: dottet_to_addr BREAKS THREAD SAFETY! it uses reporter.
+				// Solve this some other time....
 				addr_type t =  dotted_to_addr(s.c_str());
 #ifdef BROv6
 				copy_addr(t, val->val.addr_val);
@@ -193,19 +217,57 @@ bool InputReaderAscii::DoUpdate() {
 			}	
 
 			fields[currMapping.position] = val;
+			//string_fields[currMapping.position] = s;
 
 			currField++;
 		}
 
 		if ( currField != num_fields ) {
-			Error("curr_field != num_fields in DoUpdate");
+			Error("curr_field != num_fields in DoUpdate. Columns in file do not match column definition.");
 			return false;
 		}
 
-		// ok, now we have built our line. send it back to the input manager
-		Put(fields);
+
+		SendEntry(fields);
+
+		/* 
+		string indexstring = "";
+		string valstring = "";
+		for ( unsigned int i = 0; i < idx_fields; i++ ) {
+			indexstring.append(string_fields[i]);
+		}
+
+		for ( unsigned int i = idx_fields; i < num_fields; i++ ) {
+			valstring.append(string_fields[i]);
+		}
+
+		string valhash = Hash(valstring);
+		string indexhash = Hash(indexstring);
+
+		if ( keyMap->find(indexhash) == keyMap->end() ) {
+			// new key
+			Put(fields);
+		} else if ( (*keyMap)[indexhash] != valhash ) {
+			// changed key
+			Put(fields);
+			keyMap->erase(indexhash);
+		} else {
+			// field not changed
+			keyMap->erase(indexhash);
+		}
+
+
+		(*newKeyMap)[indexhash] = valhash;
+		 */
+		
+		for ( unsigned int i = 0; i < num_fields; i++ ) {
+			delete fields[i];
+		}
+		delete [] fields;
 
 	}
 
+
+	EndCurrentSend();
 	return true;
 }
diff --git a/src/InputReaderAscii.h b/src/InputReaderAscii.h
index 551a08b02e..c26a139dcd 100644
--- a/src/InputReaderAscii.h
+++ b/src/InputReaderAscii.h
@@ -30,7 +30,7 @@ public:
     
 protected:
 	
-	virtual bool DoInit(string path, int num_fields,
+	virtual bool DoInit(string path, int arg_num_fields, int arg_idx_fields,
 						const LogField* const * fields);
 	virtual void DoFinish();
 
@@ -42,9 +42,12 @@ private:
 	string fname;
 
 	unsigned int num_fields;
+	unsigned int idx_fields;
 
 	// map columns in the file to columns to send back to the manager
 	vector<FieldMapping> columnMap;
+
+	//map<string, string> *keyMap;
 	
 };
 
diff --git a/src/Val.h b/src/Val.h
index d851be311b..3ae0bc3334 100644
--- a/src/Val.h
+++ b/src/Val.h
@@ -841,6 +841,9 @@ public:
 			timer = 0;
 		}
 
+	HashKey* ComputeHash(const Val* index) const
+		{ return table_hash->ComputeHash(index, 1); }
+
 protected:
 	friend class Val;
 	friend class StateAccess;
@@ -851,8 +854,6 @@ protected:
 	void CheckExpireAttr(attr_tag at);
 	int ExpandCompoundAndInit(val_list* vl, int k, Val* new_val);
 	int CheckAndAssign(Val* index, Val* new_val, Opcode op = OP_ASSIGN);
-	HashKey* ComputeHash(const Val* index) const
-		{ return table_hash->ComputeHash(index, 1); }
 
 	bool AddProperties(Properties arg_state);
 	bool RemoveProperties(Properties arg_state);