[bug] NAR answering with precondition_beliefs about "do nothing operator"

[ARCJ137442]

Background Information

I noticed that ONA has a concept of operations called "do nothing operator", which is related to NAL-8 and can be seen at:

OpenNARS-for-Applications/src/Decision.h

Lines 87 to 88 in 96f4d03

	//assumption of failure, also works for "do nothing operator"
	void Decision_Anticipate(int operationID, Term op_term, long currentTime);

In normal code logic, the integer "operator id" corresponding to this "do nothing operator" is 0, and in many places "is do-nothing operator or not" is distinguished by op or !op:

OpenNARS-for-Applications/src/Cycle.c

Lines 385 to 389 in 96f4d03

	int op_id = Memory_getOperationID(&postcondition.term);
	Term op_term = Narsese_getOperationTerm(&postcondition.term);
	conceptProcessID2++;
	for(int j=0; !op_id && j<occurrenceTimeIndex.itemsAmount; j++) //search for op
	{

Some code refers to "traversing all real operators" by explicitly traversing 1 through OPERATIONS_MAX:

OpenNARS-for-Applications/src/Decision.c

Line 423 in 96f4d03

for(int opi=1; opi<=OPERATIONS_MAX && operations[opi-1].term.atoms[0] != 0; opi++)

The Main Issue

The main issue happens in the following code:

OpenNARS-for-Applications/src/NAR.c

Lines 190 to 211 in 96f4d03

	for(int op_k = 0; op_k<OPERATIONS_MAX; op_k++)
	{
	for(int j=0; j<c->precondition_beliefs[op_k].itemsAmount; j++)
	{
	Implication *imp = &c->precondition_beliefs[op_k].array[j];
	if(!Variable_Unify2(unused, &term, &imp->term, true).success)
	{
	continue;
	}
	if(queryCommand && Truth_Expectation(imp->truth) > answerTruthExpThreshold)
	{
	NAR_PrintAnswer(imp->stamp, imp->term, imp->truth, answerOccurrenceTime, imp->creationTime);
	}
	if(Truth_Expectation(imp->truth) >= Truth_Expectation(best_truth))
	{
	best_stamp = imp->stamp;
	best_truth = imp->truth;
	best_term = imp->term;
	answerCreationTime = imp->creationTime;
	}
	}
	}

Notice the int op_k = 0; op_k<OPERATIONS_MAX in the code, which actually means "traverse the previous operator from the do-nothing operator to the last operator"
when the field precondition_beliefs in struct Concept is defined to allow "do nothing operator" in index 0:

OpenNARS-for-Applications/src/Concept.h

Line 57 in 96f4d03

Table precondition_beliefs[OPERATIONS_MAX+1];

So,
Do we need to prohibit traversing "do nothing operator" here?
Like the following:

for(int op_k = 1; op_k<=OPERATIONS_MAX; op_k++)

Or do you want to allow all operations including "do nothing operator"?

for(int op_k = 0; op_k<=OPERATIONS_MAX; op_k++)

(like above)

Or,
Do we need to make a clear distinction between "real operators" and "do nothing operator" and "no operation id found"?

[patham9]

The code does the right thing there, the code piece you reference searches the implication tables for a link which unifies with the users's question. But I agree this index trickery sucks and can potentially be improved at some point. :)

[patham9]

Also: your observation that index 0 is reserved for temporal implications without operation (which is like a do-nothing operator almost) is valid.