Identify Exception Line Number

[shawnj]

Is there a way to identify what line number caused an exception other than wrapping each individual line with a try/catch?

If I do something like the following I only get the general error message. Best I can tell, the other members in the Exception object do not give the line number (as it does in Visual Studio):

OnBarUpdate()
{
try
{
//stuff
}
catch(Exception e)
{
Print(e.Message)
}
}


PS. A side note: A Programming Forum should not strip out text formatting such as leading spaces and tabs. It makes reading code examples very difficult.

thanks,
shawnj

[NinjaTrader_Dierk]

>> Is there a way to identify what line number caused an exception other than wrapping each individual line with a try/catch?
None I'm aware of.

[mrlogik]

shawn

Formatting is not stripped, you just have to surround your code with the proper tag.

OnBarUpdate()
{
    try
    {
        //stuff
    }
    catch(Exception e)
    {
        Print(e.Message)
    }
}

If you're getting an exception in your code, you really only need to check for
1. Divide by 0
2. Access to a null object.

Also, you want to be sure you're not referencing a bar number that doesn't exist. Try placing this line first in your OnBarUpdate() method.

if(CurrentBar < whateveryourlongestperiodis)
    return;

to avoid any bar referencing problems.

[NinjaTrader_Dierk]

Second thought: I believe Exception.ToString() or Exception.StackTrace might hold the info you are looking for.

[shawnj]

Originally Posted by NinjaTrader_Dierk

Second thought: I believe Exception.ToString() or Exception.StackTrace might hold the info you are looking for.

Thanks Dierk. I think e.Message is basically the same thing as Exception.ToString(). StackTrace holds promise but it does not give line number. I'm thinking in Visual Studio you have to run in debug mode to get the line number. I haven't tried (yet) to see if running NinjaTrader with debug mode enabled will give the line number. I don't think I can afford the cpu cost running realtime with debug mode enabled though.

[shawnj]

Thanks mrlogik for the format code tip:

TestFormat()
{
   //Space Works!
//No tab, best I can tell. :(
}

I'm getting a very intermittent "index out of bounds" exception. I'm doing a bunch of index reads and writes to a container class.

I think I Am going to have to wrap each line with a try/catch.

[mrlogik]

Any time you reference an object / array / whatever... anything with a bound you should check to make sure you don't go out of bounds.

is it on a DataSeries? make sure your period (how far back you go) is not greater than the # of bars loaded on the chart.

[shawnj]

Thanks mrlogik. I'm using a List<MyClass>.

Of course I could wrap each access of the List with something like:

if ( (index >= 0) && (index < MyList.Count) )
{
MyList[index].DoSomething;
}

But as I've written my logic, I don't think index should ever be less than 0 or greater than Count-1. So I would like to understand better what is going in my code to very intermittently cause it.

Well, I guess I could just do something like the following:

if ( (index >= 0) && (index < MyList.Count) )
{
MyList[index].DoSomething;
}
else
{
Print("Index error at this location. Dump of members: ");
}

But then it might be better to just write:

try
{
MyList[index].DoSomething;
}
catch(Exception e)
{
Print("Exception at this location. Dump of members: ");
Print(e.Message);
}

[mrlogik]

For some reason I feel the

MyList.Count
should be
MyList.Count - 1

I've played with array lists in the past and have found this to always be the case; I just cant remember why :-)

I would keep it like this for now see what the exception is (probably index out of bounds), then change it to what I *think* the root of the problem is (Count - 1)

Please keep me updated; I'm curious

Thanks

[shawnj]

Originally Posted by mrlogik

For some reason I feel the
MyList.Count
should be
MyList.Count - 1

Well this is relatively fresh in my mind. The last index in a collection is Count-1. So there are basically two ways to test for this:

1) if( index <= Count-1 )

or

2) if( index < Count ) //Slight more efficient I'm guessing.

[mrlogik]

I agree, but for some code segment I use, I have to use (index < Count - 1) otherwise I get an exception. Are you doing anything like [index + 1] referencing in you loop?

Depending on what you're doing, you could use a foreach statement, this way you can avoid using an index entirely.